Department of Health and Human Services
Departmental Appeals Board
RESEARCH INTEGRITY ADJUDICATIONS PANEL
SUBJECT: Mikulas Popovic, M.D.,
DATE: November 3, 1993 Ph.D.
Docket No. A-93-100
Decision No. 1446
DECISION
Mikulas Popovic, M.D., Ph.D. requested a hearing on the finding of
the
Public Health Service's Office of Research Integrity (ORI) that he
had
engaged in scientific misconduct when reporting scientific methods
and
data in a 1984 paper. ORI deemed the alleged misconduct
"relatively
minor" and proposed responsive measures which may be
similarly
described: rather than debarment from federal research
funding, ORI
proposed to require only a three-year period of "close
supervision" of
Dr. Popovic's laboratory work.
In spite of the narrow focus of this proceeding, this case has
compelled
us to parse a record reflecting years of investigations, thousands
of
pages of documents and lawyers' briefs, a hearing which lasted 12
days,
and the time, attention, and disagreement of dozens of
scientists,
investigators, and lawyers -- all focused essentially on the
meaning
which we should give a handful of words and notations contained in
one
heavily-edited paper written by a scientist with limited English
skills
during a volatile period of scientific discovery a decade ago.
The
paper in question, it is undisputed, made a major and
lasting
contribution to establishing that a retrovirus was the etiological
agent
of AIDS -- even assuming the most ambitious reading of error into
the
parts in question here.
One might anticipate that from all this evidence, after all the sound
and
fury, there would be at least a residue of palpable wrongdoing.
That is not
the case. On reviewing this lengthy record and all the
evidence and
argument related to each of ORI's allegations, and
notwithstanding the
vigorous efforts of ORI counsel, we find that ORI
was simply unable to prove
by a preponderance of the evidence that Dr.
Popovic is guilty of scientific
misconduct -- even under standards first
promulgated years after the paper
was published.
As we describe below at length -- in fact, in detail which represents
the
first time after all these years that there has been a
comprehensive,
independent review of all the evidence, tested in the
fire of full
confrontation by Dr. Popovic -- ORI's findings,
individually and
collectively, contained fundamental flaws: for
example, ambiguous
language is read one way among other credible
possibilities, and such choices
in turn become untested assumptions
underlying other evidence and expert
testimony.
How could it happen that such a massive effort produced no
substantial
evidence of its premise? Part of the answer may be that
early on,
investigators' attention was on controversies related
to
misappropriation of the French LAV virus, the conduct of
other
scientists, and a patent dispute -- matters which, it is important
to
note, are not in dispute here but which were regarded as
extremely
serious. This dispute is largely vestigial.
Undoubtedly, many factors
may have played a part: the long and
disputatious course of this case,
difficulties involved in reconstructing
events that occurred years
earlier, repetitive reviews, layering and
bootstrapping of evidence,
complexity of the issues, and human nature.
Whatever the cause, we can
assure both ORI and Dr. Popovic of this: we
have carefully considered
each and every argument and item of evidence in
this record --
individually and collectively -- and we have concluded that
the record
does not support a finding of scientific misconduct.
Immediately following is a summary of our decision. After that, we
give
an overview of the legal and factual framework before proceeding to
our
detailed analysis.
SUMMARY
The ORI findings of scientific misconduct involve a paper published
in
1984 in Science magazine, "Detection, Isolation, and
Continuous
Production of Cytopathic Retroviruses (HTLV-III) from Patients
with AIDS
and Pre-AIDS." Dr. Popovic was the first listed author and
principal
researcher for the experimental work reported. This paper and
three
companion papers published in the same issue of Science are regarded
as
a "tour de force" of science. The paper in question is regarded as
a
seminal work, possibly the most important paper in virology in the
20th
century. Together the four papers established that a retrovirus
was the
cause of acquired immune deficiency syndrome (AIDS). The
Popovic
Science paper reported on a cell system which could continuously
produce
this retrovirus. Dr. Popovic's success in establishing this
cell system
permitted more detailed study of the virus as well as development
of the
blood test needed to ensure the safety of donated blood supplies.
The Popovic Science paper was subject to intense scrutiny as part of
an
investigation lasting three years. This investigation focused
primarily
on allegations made in a 1989 newspaper article that the
laboratory
where Dr. Popovic was employed, the Laboratory of Tumor Cell
Biology at
the National Institutes of Health (NIH), had misappropriated an
isolate
of the AIDS virus from the Pasteur Institute in France. In a
final
report issued December 29, 1992, ORI found that one sentence and
seven
data points in the Popovic Science paper had been falsified.
1/
However, ORI stated:
The confirmed scientific misconduct on the part of Dr.
Popovic
is relatively minor, does not invalidate the findings of
his
breakthrough research, and should not preclude his employment
as
a scientist.
ORI Final Report at 61. Dr. Popovic appealed the ORI findings.
Under the guidelines for these cases, we held an evidentiary hearing.
ORI
had the burden of proving by a preponderance of the evidence both
that Dr.
Popovic committed scientific misconduct, and that ORI's
proposed
administrative action, close supervision, was appropriate.
Based on the record before us, we conclude that ORI did not prove by
a
preponderance of the evidence that Dr. Popovic committed
scientific
misconduct. Specifically, ORI did not prove that the Science
paper
contains untrue statements or data, much less that it
contains
intentional falsifications.
A. GENERAL FLAWS IN ORI'S CASE
We found the following general flaws in the evidence
ORI
presented:
o Each of the findings of scientific misconduct was based on reading
in
a particular way language which, in context, was merely
ambiguous.
ORI's reading of that language was developed when the focus of
the
investigation was on entirely different issues. Overall, the
testimony
at the hearing did not support ORI's reading. Testimony by
the expert
scientific advisors on whom ORI relied in reaching its
ultimate
conclusions merely reflected ORI's reading rather than a
critical
examination of the actual language in context.
o The opinions given at the hearing by ORI's scientific advisors
were
based on misunderstandings about what was involved in the research
at
issue and what Dr. Popovic and others had said. While these
advisors
were experts with impressive credentials, they were not asked by ORI
to
conduct a first-hand investigation. Their information was
largely
derived from the ORI investigators' understandings, which in
some
instances ignored or misrepresented evidence in the record.
Moreover,
none of these advisors had direct experience isolating a
novel
retrovirus. Since their opinions were based on a number of
erroneous
assumptions, they were largely irrelevant. Testimony by the
one
retrovirologist who testified for ORI who could be considered
a
disinterested expert supported Dr. Popovic's case more than ORI's.
o These advisors drew unreasonable inferences from the
non-scientific
evidence in the case. For example, one scientific
advisor inferred that
since the English in other papers by Dr. Popovic was
"pretty good," he
was not credible in claiming that he could not understand
the nuances of
the language at issue. This inference is
unreasonable. Unrebutted
evidence shows that some papers by Dr. Popovic
had been translated by
others from his native language, that some papers were
heavily edited,
and that Dr. Popovic's English skills were in fact limited at
the time
he drafted the Science papers (which also was subjected to
substantial
editing).
o ORI and its experts also did not have a clear idea of the
proper
legal and scientific standards to apply to Dr. Popovic's
conduct. They
faulted him simply for doing things differently from how
they would have
done things. In evaluating Dr. Popovic's conduct
against their own,
they also applied their experience in other disciplines in
a way that
failed to take into account the nature of the experiments reported
in
the Science paper at issue and the status of AIDS research at the
time.
o ORI gave an importance to the matters at issue here which is
not
justified when the paper is examined as a whole. None of the
matters
here has any significance to the validity of the major conclusions
of
the paper. ORI's advisors were clearly frustrated by the
difficulties
in addressing the issues raised by the 1989 newspaper
article. This may
have led them to consider out of context the method
and data at issue
here.
Next we summarize our determinations in relation to each of
ORI's
particular findings.
B. ORI'S PARTICULAR FINDINGS
ORI'S FIRST FINDING CONCERNS THE FOLLOWING SENTENCE, IN A
PARAGRAPH
DESCRIBING INFECTION OF A PARENTAL T-CELL LINE: "THE
CONCENTRATED
FLUIDS WERE FIRST SHOWN TO CONTAIN PARTICLE-ASSOCIATED RT
[REVERSE
TRANSCRIPTASE]." ORI found that this sentence was false
because ORI
found that cell cultures from individual patients with AIDS or
pre-AIDS
were not shown to be RT positive before the cell cultures were used
to
infect the T-cell line. This finding was based on ORI's reading
the
term "concentrated fluids" to mean individual patient samples and
the
"first" to be intended to establish a priority in the timing of the
RT
tests compared to the infection.
1. We conclude that ORI's first finding is not
supported by a
preponderance of the evidence, for the following reasons:
o The evidence as a whole shows that the "first shown" sentence
does
not necessarily have the meaning ORI ascribed to it. ORI's
reading
ignored Dr. Popovic's and others' consistent explanations of what
was
meant by the term "concentrated fluids" in that sentence: the
fluids
that were first pooled and then concentrated for each of the
three
infections of the same cell line recorded in Dr. Popovic's
notebook. At
the hearing, ORI presented no evidence to support its
assumption that
the fluids were concentrated before pooling -- indeed, some
evidence ORI
presented undercuts that assumption because it indicates that it
would
have made more sense to pool the fluids before concentrating
them.
ORI's reading of the purpose of the word "first" in that sentence
is
also not the only reasonable reading, when considered in context.
The
point of the paragraph was to establish that the virus had
been
transmitted from patient fluids to the parental T-cell line.
o While ORI established generally that there would have been
some
reason to test for RT activity before pooling patient cultures, ORI
did
not establish that Dr. Popovic would have viewed not doing so
as
illogical. Dr. Popovic's explanation of why he pooled fluids
from the
patient cultures is reasonable, not directly rebutted, and supported
by
other evidence which shows it is a technique that he would likely
have
used in a practical sense to isolate a suspected retrovirus.
o ORI did not establish that Dr. Popovic drafted the "first
shown"
sentence, that his attention was drawn to it during the editing
process,
or that, even if he noticed it, he would have recognized that it
might
have been misinterpreted by others. The evidence shows that many
others
were involved in editing the paper, that the sentence may have
resulted
from a purely editorial change, and that Dr. Popovic did not have
the
skills in English to recognize subtle distinctions in meaning.
o ORI did not show that the paragraph in which the sentence appears
is
the methodological section of the paper, or that Dr. Popovic would
have
had a motive to falsify his methods. The paragraph in question is
not
the only instance in the paper of virus isolation, so it simply does
not
have the importance in the context of the overall paper that
ORI
ascribed to it. The main complaint of ORI's witnesses was that
the
pooling technique used in infecting the parental cell line was not
good
science. Contrary to what ORI argued, however, the paragraph
in
question does not hide the fact that more than one patient sample
was
used to infect the parental T-cell line. We find it highly
unlikely
that Dr. Popovic would have revealed that he used a less than
ideal
technique and then chosen an ambiguous at best method of
misrepresenting
the logic of that technique.
SECOND, ORI FOUND THAT SIX "ND" ENTRIES IN TABLES 1 AND 2 IN THE
PAPER
FALSELY REPORTED THE EXPERIMENTAL RESULTS, GIVEN THE DEFINITION "ND,
NOT
DONE" (WHICH APPEARS IN FINE PRINT IN THE MIDDLE OF THE LEGEND TO
TABLE
1). ORI interpreted "not done" as meaning "not performed" (in the
sense
of not even attempted). ORI concluded from the laboratory records
that
the experiments reported as "ND" had been performed.
.2. We
conclude that ORI's second finding is not
supported by a preponderance
of the evidence, for the following reasons:
o The evidence shows that both ND and "not done" are subject
to
differing interpretations, some of which would accurately represent
the
results of the reported experiments. ORI did not substantiate its
view
that "not done" can only mean "not performed."
o Persuasive testimony by expert retrovirologists corroborates
Dr.
Popovic's testimony that his use of "ND" involved a reasonable
judgment
of how to report the experimental results. This expert
testimony also
showed that reporting the entries exactly as they appeared in
the
laboratory records would not have made any difference in how
a
retrovirologist would evaluate the data in the tables.
o The expert testimony on which ORI relied was based on reading
the
technician's notes in a way which is not consistent with her
testimony
on what she meant, which fails to consider the nature of the
experiments
reported, and which fails to consider other experimental
results. These
other results, of parallel tests, contradict ORI's
expert advisors'
reading of the laboratory notes to mean that there was no
virus
expression.
ORI'S THIRD FINDING WAS THAT A 10% ENTRY IN TABLE 1 WAS FALSE BECAUSE
IT
WAS INCONSISTENT WITH THE TECHNICIAN'S RECORDED READING OF ASSAY
RESULTS
ON A SLIDE AND BECAUSE THE 10% DID NOT APPEAR IN ANY LABORATORY
RECORDS
THAT WERE AVAILABLE IN 1990 WHEN THE INVESTIGATION BEGAN.
3. We conclude that ORI's third finding is not
supported by a
preponderance of the evidence, for the following reasons:
o ORI's finding on the 10% entry is also based on a misreading --
in
this instance a misreading of both the technician's notes "very
few
cells positive for rabbit antibody" and a written explanation
the
technician gave during the investigation. The technician's
testimony
shows that ORI's expert, in determining that likely 100% of the
cells
left on the slide were positive, had misunderstood the
technician's
explanation of what she did. Moreover, the evidence shows
that a 100%
value is unlikely, since the reading for the same culture eight
days
later was less than that.
o Persuasive evidence shows that Dr. Popovic also read the slide
and
that the 10% is likely the true value. ORI presented no evidence
that
would indicate the 10% is an impossible, or even an unlikely value.
o The absence of a written note for the 10% is not significant,
in
light of all of the circumstances here. These circumstances include
the
fact that a number of years had passed since the experiments
were
performed. ORI failed to establish any motive for Dr. Popovic to
have
fabricated this one insignificant data point.
4. We conclude that ORI did not prove by a
preponderance of the
evidence that there was any pattern of conduct to
falsify methods and
data.
o Finally, contrary to what ORI and its scientific advisors found,
we
do not find here a pattern of conduct of falsifying methods or data
to
make the experiments reported in the Science paper look more
rigorous
than they in fact were. ORI established at most two
ambiguities (rather
than untruths) in the paper: the "first shown" in
the disputed sentence
and the "not done" in the legend to Table 1.
While we recognize that a
researcher deliberately intending to mislead a
reader might cleverly
choose to do so through introduction of subtle
ambiguities, we conclude
that ORI did not establish such a scenario is likely
here:
o ORI did not establish that Dr.
Popovic's English skills
were sufficient so that he would have
understood the ways in
which the words "first shown" and "not done"
might have been
misleading to a reader who interpreted them as ORI
did.
o ORI did not establish that Dr.
Popovic drafted the
"first shown" language or retained it in the paper
after
recognizing an ambiguity. Since he honestly disclosed in
the
paper that he used fluids from more than one patient to
infect
the parental T-cell line, we think it unlikely that he then in
a
calculated way tried to enhance the logic of doing so.
o ORI's advisors were under the
impression that the effect
of each of the ND entries was to slant the
data toward the
hypothesis of the paper and to mislead the reader about
the
rigorousness of the experiments. The evidence does not
support
such a conclusion. None of the data points at issue (one
of
which was recorded as positive in the technician's notebook)
had
any significance to the conclusions of the paper.
Persuasive
testimony from retrovirologists -- representing the
researchers
who potentially would be the most critical readers of the
paper
-- was that they would not have evaluated the experiment
any
differently if Dr. Popovic had reported these results exactly
as
they appeared in the notebook. On the other hand, the
testimony
from the non-retrovirologists supports the reasonableness of
Dr.
Popovic's judgment that doing this would have been misleading
to
others.
LEGAL FRAMEWORK
The Department of Health and Human Services has the
discretionary
authority to protect the integrity of research it funds by
taking
administrative actions against those who have engaged in
scientific
misconduct. In 1992, the Departmental Appeals Board, in the
Office of
the Secretary, was given responsibility for hearing appeals
from
findings of scientific misconduct made by ORI. See 57 Fed. Reg.
53,125
(1992). Under the applicable guidelines, a Research
Integrity
Adjudications Panel is appointed to decide each appeal. 2/
The guidelines provide for a de novo review. What this means is
that
our decision is not a review of what ORI did during its investigation
or
whether what ORI found was reasonable based on the evidence
ORI
considered. Rather, we held a 12-day evidentiary hearing during
which
both sides had an opportunity to present testimony from witnesses
and
documentary exhibits. This hearing was Dr. Popovic's first
opportunity
to confront and cross-examine witnesses against him and to test
the
expert opinions on which ORI relied. 3/ Our decision is based
solely on
the evidence admitted into the record before us.
The purpose of this proceeding was not to revisit all of the
issues
addressed in the ORI Report. Under the guidelines, the Panel is
to
determine whether ORI proved the findings of scientific misconduct
which
ORI made, and which were appealed, and to determine the
appropriateness
of the proposed administrative action. This narrowed
the case
considerably from what was involved earlier in the
investigation.
The guidelines require ORI to prove scientific misconduct by
a
preponderance of the evidence. This means evidence that is
more
convincing than the opposing evidence and shows as a whole
that
misconduct was more probable than not.
In this decision, we evaluate the evidence presented against
the
regulatory definition of scientific misconduct published in 1989,
and
find it lacking. That definition states:
Misconduct or Misconduct in Science means
fabrication,
falsification, plagiarism or other practices that
seriously
deviate from those that are commonly accepted within
the
scientific community for proposing, conducting or
reporting
research. It does not include honest error or
honest
differences in interpretations or judgments of data.
42 C.F.R. . 50.102.
There was considerable controversy in this case over whether
this
definition is properly applied to intramural research conducted
in
1983-84, such as the research at issue here. However, since ORI
failed
to present evidence adequate to show that the 1989 regulatory
definition
was met, and since there was no evidence presented to suggest any
more
stringent earlier standard, we do not need to resolve this
controversy
or to decide exactly what would have been understood in 1983-84
to be
scientific misconduct in intramural research. 4/
ORI argued here that to meet the 1989 definition of scientific
misconduct,
ORI did not need to prove that Dr. Popovic had an intent to
deceive.
ORI relied in part on what it said was the legal definition
of
"falsification" (but in fact was the legal definition of
"false").
Black's Law Dictionary (5th Ed. 1979 at 540, cited by ORI)
indicates
that the term "false" may sometimes be used to encompass a thing
that is
untrue by mistake or accident, with no intent to deceive.
ORI's
argument begs the question of whether the term "falsification" as
used
in the regulatory definition of misconduct encompasses mere mistaken
or
accidental untruths. The regulatory definition itself
indicates
otherwise since it specifically excludes "honest error or
honest
differences in interpretations or judgments of data."
Moreover, ORI's argument that scientific misconduct
encompasses
unintentional conduct is contrary to a statement in the 1990
Guidelines
for the Conduct of Research at NIH. That statement refers to
scientific
misconduct as "fabrication, falsification, plagiarism, or
other
practices motivated by intent to deceive." Exhibit (Ex.) P-20 at
15.
This interpretation is consistent with reading the exclusion of
"honest
error or honest differences in opinion" as meaning that the
1989
definition encompasses as falsification only conduct intended
to
deceive. 5/
In any event, while we discuss below why we conclude that ORI did
not
prove intentional deception here, we also discuss why we conclude
that
ORI did not prove by a preponderance of the evidence that the
disputed
statements or data were untrue at all. Thus, even if we agreed
with ORI
(which we do not) that unintentional errors in insignificant details
in
a paper would constitute scientific misconduct under the
1989
definition, we would not reach a different result here.
FACTUAL FRAMEWORK
We discuss here the factual framework needed to help the reader
understand
the context of this dispute. 6/
A. THE STATUS OF AIDS RESEARCH IN THE EARLY 1980'S
In the early part of 1981, information on AIDS was limited. The
initial
cases recognized a severe immunodeficiency in some homosexual
men.
Later it was learned that AIDS was occurring in other populations
such
as hemophiliacs and intravenous drug users. Prior to the
publication of
the Popovic Science paper, AIDS was recognized as a fatal
disease which
was epidemic. Scientists and clinicians had considerable
questions and
concerns about the cause of this disease. By mid-1983,
the scientific
community recognized that AIDS was a transmissible agent and
that the
"pathogen was in the blood supply." Tr. at 1965, 1968-69
(Blattner).
People were developing AIDS as a result of blood transfusions,
causing
anxiety in the medical community and in the public at
large.
Consequently, there was strong motivation to find the cause of AIDS
in
order to attempt to save lives and protect the blood supply.
Working with tissue and blood samples from patients with AIDS at this
time
was considered dangerous. Many laboratories in this country
and
elsewhere would not permit clinical specimens from AIDS patients in
the
laboratory. Nonetheless, scientists around the world embarked on
a
concentrated, intense search for the causative agent of this disease.
Numerous theories were being circulated, such as that a fungus, a
common
virus, or homosexual activities were the cause of AIDS. NIH
announced a
month or two before the Science paper at issue here was published
that a
fungus was the cause of AIDS. In late 1982, Dr. Robert Gallo,
chief of
the Laboratory of Tumor Cell Biology (LTCB), NIH, and Dr. Max
Essex,
from Harvard, began to theorize about the cause of AIDS and whether
it
might be caused by a human retrovirus. 7/ Little by little the LTCB
got
into AIDS research. By mid-1983 the research efforts in the LTCB
were
intense and by mid-1984, AIDS dominated that laboratory's research.
Dr. Sodroski testified that prior to the publication of the
Science
papers, there was skepticism in the scientific community that
a
retrovirus was the cause of AIDS. Dr. Sodroski indicated that the
four
Science papers resulting from the LTCB research reported
multiple
isolations, showed the virus established in a stable cell line
for
diagnosis, and provided serological data showing that there was a
high
percentage of people in the at-risk group or who had AIDS that
had
antibodies to this virus. Dr. Sodroski stated that as a result of
the
Science papers, he, as a retrovirologist, became convinced that
a
retrovirus was the cause of AIDS. Dr. Blattner, an
epidemiologist
actively involved since the early 1980s in AIDS research,
stated that
the Science papers proved unequivocally that the cause of AIDS
was a
human retrovirus. Tr. at 1986.
B. DR. POPOVIC AND HIS
WORK
Dr. Popovic was trained as a physician in Czechoslovakia and in
1971
received a Ph.D. in cytopathology from the Cancer Research
Institute,
Slovak Academy, in Bratislava, Czechoslovakia. His
postdoctoral
training in cell biology was at Wallenberg Laboratory,
University of
Uppsala (Sweden).
Dr. Popovic is a cell biologist. 8/ He began his scientific career
in
Czechoslovakia under the tutelage of Professor Jan Svoboda,
considered
one of the "Fathers of Retrovirology." Tr. at 1836-37
(Gartner).
Together they worked with the avian sarcoma virus.
Dr. Popovic initially came to the LTCB in 1980 as an American
Cancer
Society Fellow and worked initially on HTLV-I, a retrovirus
responsible
for human T-cell lymphoma leukemia. There were few people
at that time
who were involved in the study of human retrovirology.
During
1982-1984, Dr. Popovic was a Visiting Associate in the LTCB, and
then
from 1984-1989, he was a Visiting Scientist at the LTCB. Dr.
Popovic,
who became a United States citizen in 1984, has published
approximately
163 papers during his career as a research scientist and has
been listed
as first author on 31 of those papers.
Dr. Popovic, because of his belief that the cause of AIDS might be a
human
retrovirus, used the techniques he learned in Czechoslovakia to
isolate and
grow a new retrovirus. The work in the LTCB with the first
human
retrovirus isolates, HTLV-I and HTLV-II, in the 1970s and 1980s
provided
invaluable background information for the later work with AIDS.
This work
provided information on "ways to detect [human retroviruses]
efficiently,
ways to grow the right kinds of cells, and . . . to know
that T-lymphocytes
were the right kinds of cells to grow." Tr. at
736-37 (Sodroski).
Since there was no conclusive evidence that a retrovirus was the
causative
agent of AIDS, Dr. Popovic's work on AIDS involved a search
for an unknown
virus. Searching for an unknown virus is not like
working with a known
virus. It involves searching for clues much like
detective work or
attempting to put pieces of a large puzzle together.
It involves using a
variety of laboratory methods to first determine
whether a virus is present
and using other methods to isolate that virus
and propagate it.
Dr. Sodroski indicated that in most body fluids from infected
individuals,
the amount of starting material that is infectious is very
small. Thus,
a scientist wants to be able to first propagate the virus
by making the virus
make more of itself in tissue culture, because it is
easier to detect virus
once there are a lot of infected cells making a
lot of viral protein.
However, the major difficulty with the AIDS virus
is that it is cytopathic,
meaning it kills the target cells used to
establish a culture, and it will
grow in some target cells and not in
others. To overcome these
obstacles and succeed in growing such a virus
requires not only science, but
"art" and probably even "luck", or a
"green thumb." Tr. at 782
(Sodroski); Tr. at 1977 (Blattner).
C. WHAT THE SCIENCE PAPER WAS ABOUT
We have attached as Appendix A of this decision the Science paper
at
issue. As explained in the abstract to the paper, the paper reported
on
development of a "cell system" for the "reproducible detection of
human
T-lymphotropic retroviruses (HTLV family) from patients with
the
acquired immunodeficiency syndrome (AIDS) or with signs or symptoms
that
frequently precede AIDS (pre-AIDS)." The cells were described
as
"specific clones from a permissive human neoplastic T-cell line,"
some
of which "permanently grow and continuously produce large amounts
of
virus after infection with cytopathic (HTLV-III) variants of
these
viruses." The abstract explained that the significance of the
reported
research is that the cell system "opens the way to the routine
detection
of HTLV-III and related cytopathic variants of HTLV in patients"
and
"provides large amounts of virus for detailed molecular
and
immunological analyses." Ex. H-5 at 497.
The paper focuses on experimental results for those clones of
the
permissive cell line which were found to permanently grow
and
continuously produce large amounts of virus. One paragraph of
the
three-page paper discusses the preliminary infection of the
parental
T-cell line (from which the clones were developed). Dr.
Popovic's
original handwritten draft of the paper did not discuss this
infection
at all. This infection was discussed only because of a later
decision
to develop an AIDS test using the variant, later designated
HTLV-IIIB,
growing on one of the clones (designated H9) after it was infected
by
exposure to concentrated virus from the infected parental T-cell
line.
The paper also describes, in somewhat greater detail, the infection
of
the cell clones with isolates from individual patients.
The authors of the paper do not purport to have isolated one virus (or
one
variant of a virus) as the cause of AIDS. Instead, the abstract
refers
to cytopathic variants of viruses. The paper itself states:
"These new
HTLV isolates are collectively designated HTLV-III, although
it is not yet
proved that they are identical." Ex. H-5 at 498. The
paper
reports a number of different isolates. 9/
Experimental results reported in the tables in the paper are results
of
infection of the clones with the virus variant produced by the
parental
T-cell line, at 6 and 14 days after infection, and with
individual
patient isolates.
Next we turn to our detailed analysis of each of the three major
findings
of scientific misconduct.
DETAILED ANALYSIS
A. THE DISPUTED SENTENCE
The key sentence at issue appears in a paragraph describing how
the
parental cell line was infected, prior to the cloning and infecting
of
the clones. We reproduce here most of the first four sentences of
that
paragraph to preserve the immediate context in which the allegedly
false
sentence appears:
The cell line HT was tested for HTLV before being infected
in
vitro and was negative by all criteria including lack
of
proviral sequences (32). Continuous production of HTLV-III
was
obtained after repeated exposure of parental HT cells . . .
to
concentrated culture fluids harvested from short-term
cultures
of T cells (grown with TCGF) obtained from patients with AIDS
or
pre-AIDS. The concentrated fluids were first shown to
contain
particle-associated RT. When cell proliferation
declined,
usually 10 to 20 days after exposure to the culture fluids,
the
fresh (uninfected) HT cells were added to the cultures.
Culture
fluids from the infected parental cell line were positive
for
particulate RT activity, . . . .
Ex. H-5 at 498 (emphasis added; reference omitted). 10/
Dr. Popovic's testimony, and his notebooks show, that he: 1) exposed
the
parental T-cell line on November 15, 1983 to concentrated culture
fluid
harvested from cultures of three patients' primary cells; 2) exposed
the
parental cell line on November 22, 1983 to concentrated culture
fluid
from the same three patients' cultures; and 3) exposed the parental
cell
line on January 2, 1984 to concentrated culture fluid from seven
more
patients' cultures. Ex. H-19 at 33, 34, 44; Ex. H-157 at 29-35;
Tr. at
2397-98. 11/
ORI's finding that the underlined sentence is false was based on
ORI's
reading of that sentence as meaning that Dr. Popovic tested
individual
patient samples from each of the ten patients for RT activity,
before
"pooling" those samples to infect the cell line, and showed that all
of
them were positive for RT activity. See, e.g., ORI post-hearing br.
at
46-49. It is undisputed that each of the ten individual patient
samples
was not found positive for RT activity before being used to infect
the
cell line. See Tr. at 2302-04. Dr. Popovic testified,
however, that
after he first pooled and then concentrated the fluids he had
harvested
from individual patient cultures, he sent part of the
pooled
concentrated culture fluid for RT analysis at the same time as he
used
another part of the concentrated fluid to infect the cell line; he
said
that he had obtained the results (which were positive) from Dr.
M.G.
Sarngadharan, after exposing the cell line. Tr. at 2498.
ORI alleged that there were two falsifications in the sentence: "one
that
it was not first shown; and two, that [RT results for individual
patient
samples] were not all positive." Tr. at 235. ORI
presented
evidence which it said showed that many of the ten individual
samples
tested negative for RT, that, at most, only one of the ten samples
was
tested and shown positive for RT before it was pooled, and that only
six
of the ten samples contained any AIDS virus. 12/
We conclude that ORI has not proved by a preponderance of the evidence
its
finding that Dr. Popovic falsified the disputed sentence, for the
following
reasons:
ORI's finding depends on a reading of the
disputed
sentence which is not the only reasonable
reading.
ORI's reading that "concentrated fluids"
means
individual patient samples ignores unrebutted
testimony
and other evidence that the fluids from the
patient
samples were first pooled and then
concentrated.
Moreover, in context, the "first" can reasonably be
read
as intended to convey that the RT activity
was
associated with the concentrated patient fluids
(and
therefore a virus had been transmitted from the
fluids
to the cell line).
ORI's reliance on its experts' opinions on the
meaning
of the sentence is misplaced since, for the most
part,
the experts did not independently determine the
meaning
of the sentence. Their testimony as a whole
supports a
conclusion that, in context, the statement is
merely
ambiguous.
We find Dr. Popovic's testimony about what he in
fact
did to be credible since that testimony is
consistent
with the laboratory notebooks, is corroborated
by
others' testimony, and is unrebutted.
ORI did not show either that Dr. Popovic added
the
sentence in question or that in the editing process
he
was made aware of the addition of the sentence
and
should have known it may have been misinterpreted.
ORI overrated the significance of the sentence
in
arguing that Dr. Popovic had a motive to falsify
it.
ORI did not establish that the paragraph was the
key
methodological section of the paper, nor that
Dr.
Popovic would have viewed what he in fact did
as
illogical and lied about it to make his
experiment
appear more rigorous.
1. ORI misread the sentence.
In our view, the greatest weakness in ORI's case is that it is
dependent
on reading the disputed sentence a particular way, which is not the
only
reasonable reading of the sentence. Indeed, ORI's view that
the
sentence falsely states that each individual patient sample was
tested
for RT activity before pooling and shown positive ignores the context
in
which the sentence appears.
The context refers to "repeated exposure" (more than one exposure) of
the
parental T-cells to "concentrated culture fluids" (more than one
fluid)
harvested from short-term cultures obtained from "patients with
AIDS or
pre-AIDS" (more than one patient). ORI reads the plural
"fluids" as
referring to fluids from individual patient samples because
it assumes that
the fluids were concentrated before they were "pooled."
However, ORI provided
no evidence to support that assumption. The
wording of the sentence is
not inconsistent with RT testing of different
fluids obtained after patient
cultures were first pooled and then
concentrated and used for different
exposures (which is what Dr. Popovic
has consistently stated that he
did). Indeed, if the writer intended to
suggest testing of each
individual patient culture either before pooling
or before exposing the cell
line, one would expect different wording
(for example, "Each patient culture
tested positive for RT activity
before the exposure.").
Similarly, ORI's view that the disputed sentence is false because
the
tests for RT activity were performed after the cell line was infected
is
contingent on reading of the word "first" as establishing a priority
in
the timing of RT testing compared to the timing of infecting the
cell
line. 13/ In context, however, the "first" can reasonably be read
as
having a different purpose. Two sentences later the reader learns
that
culture fluids from the infected cell line were tested for RT
activity
after infection. Thus, the significance of the "first" in the
disputed
sentence can reasonably be seen as establishing a priority in the
RT
activity as occurring in the concentrated culture fluids harvested
from
the patient cultures before those fluids were used to infect the
cell
line. It is not unambiguously clear that it means that the RT
tests
were performed before the cell line was exposed to the
concentrated
culture fluids.
ORI did not establish that the timing of the tests for RT activity in
the
concentrated culture fluids relative to the timing of the infection
would
have any significance critical to the paper's conclusions. 14/ On
the
other hand, tests which would indicate that RT activity was present
in the
patient fluids and not in the cell line are significant to the
paper's
conclusions. The paper needed to establish that the retrovirus
was
transmitted to the cell line from the patient culture fluids and was
not
already present in the cell line. Indeed, the paragraph in
question
begins by referring to tests which indicated that the cell line
was
negative for HTLV before infection. 15/
That one possible intent of the word "first" was merely to indicate
that
RT activity was present in the fluids used to infect the cell line
is
supported by a comparison of drafts seven and eight of the
Science
paper. Draft seven contains the following sentence:
Continuous production of HTLV-III was obtained after
repeated
exposure of parental HT cells . . . to concentrated
culture
fluids positive for particulate reverse transcriptase
(RT),
which was harvested from short term cultured T-cells
originated
from patients with lymphadenopathy and AIDS.
Ex. H-12 at 6.
This sentence does not contain the words "first shown." The
editorial
change which appears on draft eight and which created the
disputed
sentence, considered in context, clarifies and emphasizes evidence
that
the retrovirus was present in the concentrated culture fluids used
to
infect the cell line. The most likely reason for emphasizing
this
evidence is that it supports the overall conclusion that a
retrovirus
had been transmitted to the cell line from the patient fluids.
In sum, ORI's reading of the disputed sentence (on which ORI's
allegation
depends) fails to consider the sentence in light of the
context, purpose, and
wording of the paragraph as a whole. The disputed
sentence, while
ambiguous, can reasonably be read as consistent with
what the evidence shows,
in fact, occurred. Dr. Popovic conceded during
the ORI investigation,
after ORI had pointed out that it was reading the
"first" to mean before the
pooling and infection, that the sentence (as
interpreted by ORI) could be
seen as imprecise because RT positivity of
the fluids was not "shown" until
after the pooling and infection. But
the fact that the sentence could
reasonably be read another way is
important, for two reasons.
First, if Dr. Popovic were intentionally trying to mislead the reader
into
believing that each of the individual patient samples had tested
positive for
RT activity before being pooled, one would have expected
him to do it in a
more direct and unambiguous way. Second, ORI took the
position that,
even if Dr. Popovic did not add the "first shown"
sentence, he should be
found guilty of scientific misconduct because he
had the responsibility to
change the sentence to ensure its accuracy.
The fact that the sentence is
ambiguous -- and that the one word "first"
might add imprecision if read a
particular way -- supports our
conclusion that failure to change the sentence
did not amount to a
falsification. This is one reason why (particularly
in light of Dr.
Popovic's English-speaking abilities at the time) we find
credible Dr.
Popovic's testimony that, if he saw this change when editing the
galley
proofs of the paper, it did not impress him as false.
2. The expert testimony does not establish that
the sentence is
false.
ORI argued that we should find the disputed sentence to be false
because
testimony from its experts supports its view that the disputed
sentence
should be interpreted to mean that individual patient samples
were
tested and shown RT positive before the pooling. But, in fact,
the
experts' testimony was not dispositively supportive.
ORI did not argue that the expert testimony showed that the plain
wording
of the sentence makes it subject to only one reasonable
interpretation.
Nor does the cited testimony support a conclusion that
ORI's is the only
reasonable interpretation.
We found the testimony on this sentence by the scientific advisors who
had
been involved in ORI's investigation (Drs. Richards, Berns, and
Schaffer) to
be either irrelevant or unpersuasive -- in spite of these
experts' impressive
credentials, for the following reasons:
Based on their testimony as a whole, as well as
their
own descriptions of their role in the investigation,
we
find that these advisors did not form their
opinions
independently, after consideration of all of
the
relevant evidence. See, e.g., Ex. H-73 at 1; Tr.
at
468-70, 481, 512 (Richards); Tr. at 1042, 1062
(Berns);
Tr. at 1578-80 (Schaffer). They had reached
conclusions
based on assumptions derived from how ORI
investigators
presented the issues to them and based on
partial
evidence. They appeared unable to reevaluate
those
conclusions when those assumptions were questioned
--
perhaps due to their involvement in the investigation
of
the origin of HTLV-IIIB, frustration over
the
difficulties of resolving this question, and
association
with the ORI findings. See, e.g., Tr. at
1639-44;
1660-65 (Schaffer).
None of these experts appeared to have
carefully
examined the language and context of the
disputed
sentence, or to have independently tested it against
the
record. During their direct testimony on the
sentence,
Drs. Richards and Schaffer each conclusorily
testified
that the sentence was false because it "says" or
"said"
that the individual patient samples had been shown to
be
RT positive before they were used to infect the
cell
line. Tr. at 498-500 (Richards); Tr. at
1491-92
(Schaffer). They appeared unaware of
possible
ambiguities. For example, Dr. Schaffer's
explanation of
her interpretation assumed that each
concentrated
culture fluid was from an individual patient.
Tr. at
1502-03; 1505-06. Dr. Berns appeared to be under
the
impression that Dr. Popovic had conceded that
the
sentence falsely stated that all of the
individual
patient samples had been tested before being
pooled.
Tr. at 1045, 1064. In other words, the opinions
these
experts provided were based on assumptions about
what
the sentence meant, rather than on the wording,
context,
or underlying facts.
When Dr. Schaffer's attention was focused on the
exact
language in the paper, she acknowledged that it
was
ambiguous. Tr. at 1660-62. Dr. Richards admitted
that,
in reading the sentence to say that the RTs were
all
done and were all positive, he did not know when
the
samples became concentrated fluids. Tr. at 500.
In some instances, these experts drew
unreasonable
inferences from non-scientific evidence. For
example,
based on her examination of other papers on which
Dr.
Popovic was an author, Dr. Schaffer discounted his
claim
that he had been hampered in drafting and editing
the
Science paper because English was not his
native
language. She decided that, since the English in
the
other papers was "pretty good," he was not
credible.
Tr. at 1529-30. However, she lacked factual
information
critical to evaluating his skills; she admitted at
the
hearing that she did not know whether these other
papers
reflected his personal English skills or instead
either
were written in his native language and translated
into
English or were heavily edited by others. Tr. at
1599.
16/
Dr. Martin, another of ORI's expert witnesses, simply cannot be
considered
disinterested. He acknowledged that he had had a dispute
with Dr. Gallo
regarding the research involved here, and his testimony
evidenced sympathy to
the French researchers' claims that Dr. Gallo had
failed to give them
appropriate credit. Tr. at 1337; 1359-62; 1311-12;
1316; 1328-29.
As discussed below, his testimony showed a frustration
with what he saw as a
lack of detail in the paragraph at issue that
derived from what he wanted to
know for his particular purposes and what
Dr. Gallo had said to him about the
Science paper, rather than
reflecting an impartial analysis based on reading
the paper as a whole.
Tr. at 1333-63. For example, Dr. Martin's
interpretation seemed to
assume that patient donor cells had been
cocultivated with the T-cells,
rather than that the T-cells had been exposed
to cell-free supernatants
(fluids) from the patient cultures, as the paper
itself describes. See
Tr. at 1324.
We found Dr. Sodroski, the only retrovirologist other than Dr. Martin
who
testified for ORI, to be not only objective, but highly credible
and
persuasive. When viewed as a whole, however, his testimony
supported
Dr. Popovic's case more than ORI's case. On direct
examination, Dr.
Sodroski testified with respect to the term "concentrated
fluids" in the
disputed sentence that he believed that the term referred to
the
"concentrated culture fluids harvested from short-term cultures
of
T-cells grown with TCGF, or T-cell growth factor, which were
obtained
from patients with AIDS or pre-AIDS." Tr. at 74l. (That
is, the
concentrated culture fluids referred to in the preceding
sentence.) He
then testified that the basis for his conclusion was that
"the word
`first' suggests that the demonstration of particle-associated RT
was
associated with the actual short-term cultures of the
T-cells
themselves." Tr. at 741. This testimony supports our
analysis above
that, in context, the importance of the sentence is that it
establishes
an association between RT activity and the patient cultures
necessary to
show that the retrovirus was transmitted to the cell line from
patient
cultures.
ORI counsel did ask Dr. Sodroski: "Does that sentence then mean
that
before the pooling the patient samples were RT positive?" Tr. at
744.
He answered: "Yes." Tr. at 744. His responses to
subsequent
questions, however, indicated that this opinion was based on
several
assumptions and factors. He assumed the RT testing would have
been done
on individual patient samples before pooling because pooling
samples
with no virus would dilute the amount of virus, and because he
assumed
that RT results for the cultures could have been obtained within 2.5
to
3 hours, so that culture viability would not suffer much from
waiting
for the results. Tr. at 744-45; 749; 759. We discuss
below the
question of the "logic" of pooling prior to RT testing. With
respect to
waiting for RT results, Dr. Sodroski indicated that, if he had to
send
for results to another laboratory and wait three days to obtain
results,
this would affect his view of whether culture viability would
be
affected. Tr. at 759. As our analysis discusses, the evidence
shows
that Dr. Popovic was sending materials to another laboratory for
RT
testing, so that culture viability would suffer from waiting
for
results.
Dr. Sodroski also indicated that he read the term "concentrated
culture
fluids" to refer to individual patient samples because the word
"fluids"
is plural. Tr. at 772; 775-76. When informed of the
possibility that
supernatants from individual patient cultures were first
pooled and then
concentrated, and that there were three such concentrated
culture fluids
used for the "repeated exposure" of the cell line, he
acknowledged that
the term "concentrated fluids" was ambiguous. Tr. at
789. 17/
Finally, testimony for Dr. Popovic by Dr. Malkovsky, who had the
most
extensive experience of any of the witnesses in isolating and
growing
the AIDS virus, also indicated that the sentence was
ambiguous,
depending on whether the fluids were first pooled and then
concentrated
or were tested for RT before the pooling. Tr. at
2235-36. He viewed
the editorial change which created the "first shown"
sentence as not
changing the meaning, but as stressing more the fact that the
fluids
were positive for RT. Tr. at 2235. Contrary to what ORI
argued, we do
not find that the mere fact that, like Dr. Popovic, Dr.
Malkovsky was
born in Czechoslovakia indicates that he was biased in Dr.
Popovic's
favor.
In sum, ORI did not establish through a preponderance of the
credible
testimony that ORI's reading is the only reasonable reading of
the
disputed sentence.
3. Dr. Popovic's testimony on what he in fact did is credible.
ORI also suggested that, even if the sentence was intended to mean
that
pooled concentrated fluids were shown to contain particle-associated
RT,
we should find the sentence false. ORI argued that there were
two
reasons why we should not find credible Dr. Popovic's assertions that
he
had sent such fluids to Dr. Sarngadharan's laboratory in Maryland for
RT
testing. First, ORI argued, Dr. Prem Sarin, at the LTCB, was capable
of
doing RT tests and had performed such tests for Dr. Popovic on
patient
samples. Second, ORI asserted, there is no documentary evidence
that
Dr. Sarngadharan had tested such concentrated fluids and found
them
positive.
We find credible Dr. Popovic's testimony that he sent materials to
Dr.
Sarngadharan for RT testing, rather than relying on Dr. Sarin to
perform
the tests, for the following reasons:
We found Dr. Popovic credible as a witness based on
a
number of factors, including his demeanor during
the
hearing, the straightforward way in which he
testified,
others' testimony about his basic integrity (which
was
wholly unrebutted), and our review of his
past
statements.
o ORI's major argument concerning Dr.
Popovic's credibility was
that that he allegedly lied in a written submission
to ORI in stating
that seven or eight of the pooled cultures were from
patients confirmed
to be HIV positive. (HIV is the current designation for
AIDS virus
variants.) ORI relied on an analysis by Roche Diagnostics of
aliquots
of the ten pool samples, which found HIV virus in only six of the
ten
samples. What ORI ignored was that one of the four other samples
was
identified as being from a patient who "seroconverted between June
1984
and June 1985." Ex. H-79 at 468. Thus, since at least seven
patients
could reasonably be said to have been confirmed to be HIV positive,
Dr.
Popovic's statement was not inaccurate.
Dr. Popovic's contemporaneous notebooks and those of
his
technician indicate that, during this time period,
he
was regularly sending materials to Dr. Sarngadharan
for
RT testing. See, e.g., Ex. H-19 at 29, 33, 44.
18/
They also show specifically that he sent some
materials
to Dr. Sarngadharan for RT testing on the dates of
the
first and third pool infections. Ex. H-19 at 33,
44.
Dr. Sarngadharan's testimony before ORI (admitted
here
by agreement of the parties) corroborates Dr.
Popovic's
testimony. Ex. H-50 at 15-19; see also Ex. H-52
at 17,
38.
Dr. Popovic's testimony that he had other evidence
of
the presence of a retrovirus in the samples (such
as
multinucleated cells) which led him to question
Dr.
Sarin's results is consistent with unrebutted
testimony
on the kinds of clues on which a researcher in
general,
and Dr. Popovic in particular, would rely in trying
to
detect and isolate a retrovirus. Moreover,
Dr.
Sarngadharan was instrumental in developing the RT
test
and an expert on it. Ex. H-50 at 7-9.
ORI would have us find against Dr. Popovic on the basis that there is
a
lack of any primary data showing RT results for the three
concentrated
culture fluids. There are several reasons why we do not
make such a
finding here. First, there is some data in the notebooks
which may be
relevant; the problem is simply in interpreting those
data. Codes were
used on materials sent to Dr. Sarngadharan so that his
testing would be
blind. Usually, these codes were just numbers,
starting with one,
corresponding to samples sent on a particular date.
Dr. Sarngadharan
would generally call Dr. Popovic with the results. Ex.
H-50 at 15-18.
Dr. Popovic's notes, showing RT results with numbers and
various dates,
are difficult to interpret because the code is not always
clear. For
example, RT results from Dr. Sarngadharan corresponding to
the date of
the first pool infection, November 15, 1983, include a positive
result
for code number 3, which is identified only by a notation which
looks
like a "17" (in contrast to several other code numbers which
have
identifiers corresponding to individual patient samples mentioned in
the
notebook). Ex. H-19 at 57. 19/ Dr. Popovic has said that this
may be
the RT result for the first concentrated culture fluid (which
was
harvested from cultures from the same three patients as the
second
concentration). While his notes do not specifically identify
"17" as
the pool, neither do they rule this out as a possibility. It
is,
moreover, credible that each of the three pooled and
concentrated
culture fluids would have tested positive for RT activity, since
it is
undisputed that aliquots from one or more of the patient samples used
to
create each of the three concentrated culture fluids contained the
AIDS
virus.
Second, the particular circumstances here such as the passage of time
and
the manner in which Dr. Popovic and his technician kept notes make
inferences
which we might otherwise draw from a lack of primary data
unreasonable
here.
Third, it is not clear that Dr. Popovic, rather than Dr.
Sarngadharan,
would have had the responsibility, if any, to retain records
recording
the results of experiments Dr. Sarngadharan performed. ORI
did not
rebut testimony that the contract between Dr. Sarngadharan's
laboratory
and the NIH required him to retain records for only three
years. More
than five years had passed before ORI investigators
requested such data,
and Dr. Sarngadharan testified as to circumstances in
his laboratory
which made retention of old records difficult. Ex. H-50
at 34.
Dr. Popovic's testimony that he first pooled and then concentrated
the
supernatants (fluids) from the patient cultures is also credible,
for
the following reasons:
The protocol he wrote for these experiments in
September
1983 states: "If not sufficient volume of
culture
fluids; pool together several samples. Use both
(conc.)
culture fluids and cells for infection." Ex. H-19
at
16-17; see also Ex. H-45 at 15-21 (12/1/90
preliminary
response).
Dr. Sodroski, in describing what a
retrovirologist
typically would do, indicated that a researcher
would
usually concentrate the fluids before testing them
for
RT activity since it gives a better result. Tr.
at
677-78.
Dr. Sodroski also indicated that concentration would
be
done in two steps. Tr. at 678. In light of this,
it
simply would make more sense first to pool the
culture
fluids and then to concentrate them since this
would
require fewer steps of concentration than if
each
individual culture fluid were first concentrated.
Dr. Sarngadharan explained that the fluids would
have
been first pooled and then concentrated, with part
used
for infection and part sent for RT testing, and that
it
would have been contrary to the purpose of pooling
(to
save scarce patient material) to have first sent
parts
of the individual patient cultures for RT testing.
Ex.
H-52 at 34-35.
Finally, we find credible Dr. Popovic's testimony that, if there was
some
imprecision in the sentence, he was not aware of it at the time.
20/ As
we discuss next, ORI did not prove by a preponderance of the
evidence either
that Dr. Popovic drafted the sentence or had his
attention directed to it
during the editing process. Even more
important, ORI did not prove by a
preponderance of the evidence that Dr.
Popovic had a motive to falsify the
sentence.
4. ORI did not prove either that Dr. Popovic
drafted the sentence
or that he became aware of any imprecision in the
sentence during the
editing process.
During the course of the investigation, Dr. Popovic provided copies
of
eight of the drafts of the Science paper at issue, including
his
handwritten draft and later typed or xeroxed versions which
show
corrections in other persons' handwriting. These drafts were
referred
to by number at the hearing as though they were consecutive drafts,
but
there possibly was at least one intervening draft, which is missing.
21/
We nonetheless use the draft numbers used by the parties.
The unrebutted testimony shows that, after preparing a rough
handwritten
draft of the paper and having it typed, Dr. Popovic traveled to
a
conference in Utah on or about March 18, 1984 to present a paper.
While
at the conference, he received a phone call from Dr. Gallo asking him
to
return. Meanwhile, Dr. Gallo began editing the draft and it
went
through several versions before Dr. Popovic returned to the
Laboratory
of Tumor Cell Biology on March 27, 1984. Dr. Gallo told Dr.
Popovic
upon his return that if he was not able to complete editing of the
paper
within a very short period of time, it might not be published with
the
other three papers. Over the next few days, several others
(including
Dr. Gallo, Dr. Sarngadharan, and Anna Mazzuca, who acted as an
editorial
assistant at the LTCB) assisted Dr. Popovic in editing the
paper. The
paper was submitted to Science on the morning of March 30,
1984.
Neither the paragraph nor the specific sentence at issue here appear
in
Dr. Popovic's handwritten draft. Draft seven contains the
sentence
quoted above, which appears on draft eight as two sentences
(including
the "first shown" sentence). Since there is no marking on
draft seven
indicating such a change, and no one has produced any intervening
draft,
it is impossible to determine who made the change.
ORI would have us infer that Dr. Popovic made the change because
the
unedited sentence in draft seven was added in response to a direction
by
Dr. Gallo in draft five to "give method" and Dr. Popovic was the
person
responsible for the method. See Ex. H-10 at 6. It is
equally
reasonable, however, to infer that the change which created the
"first
shown" sentence was made by someone else as an editorial change.
The
clause in draft seven beginning "which was harvested from
short-term
cultured T-cells" is a misplaced modifier which could mistakenly
be read
as modifying "RT" (the antecedent noun), rather than as
modifying
"concentrated culture fluids." Moreover, this clause had a
singular
verb form "was," but clearly was intended to relate back to the
plural
noun "fluids." In our view, this makes it just as likely that
the
change in draft eight adding the "first shown" sentence was made
by
someone skilled in English, as that it was made by Dr. Popovic.
ORI presented no evidence that the "first shown" sentence was
specifically
brought to Dr. Popovic's attention, either before the paper
was submitted to
Science or when he read the galley. Editing of the
galleys was done
under unusual circumstances, due to the nature of the
four papers, and Dr.
Popovic's participation was limited to reviewing
the galleys at the house of
a Science editor on a weekend day. In a
period of about an hour, he not
only reviewed the galleys of his paper,
but also briefly reviewed the galleys
of two of the other papers on
which he was an author. Tr. at 2301; see
also Tr. at 1717-19 (Kulstad).
While Dr. Popovic himself admitted that he would have liked more time
to
review the galleys, ORI did not establish that Dr. Popovic had
any
meaningful control over the editorial process, other than
potentially
withdrawing his name from the paper. One of ORI's own
witnesses
acknowledged that this would be "inconceivable," given the nature
of the
papers and Dr. Popovic's contribution. Tr. at 465
(Richards). We
agree. The editorial process was clearly adequate;
none of the
allegations here calls into question the major conclusions of the
paper.
We doubt that any paper could withstand entirely the scrutiny this
paper
has received. ORI sought to establish that the pressures to
publish
were created or fabricated by Dr. Gallo and did not justify any
failure
to ensure precision in every aspect of the paper. We found
persuasive,
however, Dr. Popovic's testimony that, as a physician, the
primary
pressure he felt was that publishing the papers could save lives.
Under the circumstances, including the subtle nature of any imprecision
in
the "first shown" sentence and the fact that English is not Dr.
Popovic's
native language, we find credible Dr. Popovic's assertion that
he did not
recognize any inaccuracy in the sentence. 22/
Thus, even if we viewed the sentence as inaccurate, rather than
merely
ambiguous, we would not find that it was falsified by Dr. Popovic.
5. ORI did not show that Dr. Popovic had a
motive to falsify the
sentence.
ORI took the position that Dr. Popovic had a motive to falsify
the
sentence because it would make the methodology of the
reported
experiment appear more rigorous if the RT tests had been performed
on
each of the individual patient samples before they were pooled.
ORI
relied on testimony that it would be more logical to test for
RT
activity in patient samples before pooling them because adding
samples
with no virus in them to the pool would "dilute" the pool and reduce
the
chances of infecting the cell line with the virus.
We reject this analysis for two reasons. First, ORI's position
would
require a finding that Dr. Popovic viewed the "first shown" sentence
as
a part of the methodology reported in the paper and viewed RT testing
as
an integral part of his methodology. Second, ORI's position
would
require a finding that Dr. Popovic thought that what he did
was
illogical, so that he should misrepresent it to make it appear
more
logical and therefore more rigorous. The record does not support
such
findings.
In our view, characterizing the paragraph in question as the
key
methodology section in the paper takes that paragraph out of
context.
As discussed above, the paper's focus was on the development of a
cell
system to permanently grow and continuously produce large amounts
of
virus in spite of the cytopathic effects of the virus. The
most
important contribution of this paper was identification of
the
appropriate target cells, and identification of the most
permissive
clones of that cell line. Dr. Sodroski testified that --
the key to success I think here was that the fact that
the
cells, the right target cells for the virus could be
propagated
in the culture, allowing the virus to replicate . . . to a
level
that was detectable . . . .
Tr. at 674; see also Tr. at 1982-83, 2002 (Dr. Blattner). He
stated
that the importance of establishing continuous production of the
virus
was that it could be used to further characterize the virus, as well
as
to provide a reagent for the purpose of establishing diagnostics.
Tr.
at 780-81.
The testimony established that there were certain standard methods
for
infecting a cell line with a virus: cell free transmission (or
exposure)
and cocultivation. Cocultivation is mentioned several
paragraphs before
the one in question and a reference is given. 23/
Both cell free
transmission and cocultivation are referred to and described
in somewhat
greater detail in the legend to Table 2 in the paper. Thus,
the key
contribution to "methodology" in the paper was the identification of
the
right target cells to obtain continuous production of the virus, not
the
particulars of how to infect a cell line with a virus.
Moreover, contrary to what ORI suggested, there is no apparent reason
why
Dr. Popovic would have thought it important to the paper's
conclusions to go
into the details of how he pooled viruses together
when infecting the
parental T-cell line. 24/ Indeed, the
contemporaneous evidence (the
various drafts of the papers) indicates
that he did not originally include
this paragraph. He added it after a
decision had been made that the
isolate later called HTLV-IIIB would be
used for the AIDS test. This
decision ultimately gave the infection of
the parental cell line an
importance it would not otherwise have had.
With respect to how Dr. Popovic
would have likely viewed the paragraph
at the time in light of the
conclusions in the paper, however, the
evidence shows that ORI overrated the
importance of the paragraph.
An examination of the paper itself reveals that the infection described
in
this paragraph is not the only one described in the paper.
When
commenting on this paragraph, Dr. Sodroski noted the examples in
the
paper of infection of sub-clones of the HT cell line with
primary
patient isolates that were not pooled. Thus, he said, "there
are other
examples in the paper of a successful propagation of an HIV isolate
on a
permanent T-cell line . . . that, apart from the paragraph in
question,
would support the major claims of the paper." Tr. at
779-80. Indeed,
he considered the paper a "tour de force" of science
because "not only
do you see multiple examples of virus isolations, you also
see that the
virus is established in a stable cell line, which allows for
diagnosis."
Tr. at 734. He stated that this evidence, together with the
serological
data in the paper and the accompanying papers, was convincing
evidence
that a retrovirus was the etiologic agent of AIDS, rather than
an
opportunistic infection, which could have been the interpretation
when
you only have sporadic examples of virus isolation, as had
been
published by French researchers. Tr. at 734-36.
As a result of the allegations regarding misappropriation of the
French
isolate, the paragraph took on inordinate importance to
the
investigators. The evidence does not show that this paragraph
would
have had importance to other retrovirologists generally or to
Dr.
Popovic in particular, at the time it was written. Dr. Sodroski
noted
that, in light of the overall evidence supporting the paper's
major
conclusions, the paragraph in question would be less important
to
"perhaps 99 percent of investigators or more." Tr. at 781. ORI
relied
on testimony that the methodology might be important to
investigators
attempting to isolate viruses from other patients. Dr.
Sodroski
recognized this, but noted that "the paper does give examples of
those
types of isolations, and passaging virus onto permanent T-cell lines,
in
which pooling does not appear to have been done." Tr. at 781.
Thus, he
said, "I don't know that the specific details that we're focusing on
in
this paragraph really had much practical import, in terms of the
actual
advance of science, or the advance of diagnostics in the AIDS
field."
Tr. at 781.
Thus, we conclude that, in describing this paragraph as a
methodology
section of the paper, ORI overrates its significance and
exaggerates Dr.
Popovic's possible motivation for falsifying it.
We further conclude that, even if the paragraph is properly viewed
as
intended to set out a methodology so that the experiment described
could
be reproduced, ORI did not show that the "first shown" sentence
should
be considered as intended to represent part of that methodology and
to
misrepresent the "rigorousness" of the methodology.
For its position that RT testing was an integral part of the
methodology
presented in the paper, ORI relied on the fact that in a 1990
textbook
chapter describing how to propagate the AIDS virus in a
neoplastic
T-cell line, co-authored by Dr. Popovic, the following
statement
appears:
Culture fluids to be concentrated for use as inocula
should
contain at least 100,000 cpm RT activity/ml
(before
concentration).
Ex. H-89 at 18. 25/
When questioned about the textbook and whether it was saying one must
have
positive RT activity before using a patient sample to infect a cell
line, Dr.
Sodroski said that he would interpret it that "to make the
assay optimal" one
would want the specified level of RT activity. Tr.
at 754. On
cross-examination, he said that the impression the book
chapter gave was that
"a certain level of virus in the samples used to
infect the cells would
increase the efficiency of success in the
experiment." Tr. at
765. He explained that an RT test "doesn't really
tell you what the
infectivity of a particular virus preparation is . . .
So I find setting
those kinds of thresholds to be relatively arbitrary
anyway, because one can
have a very high reverse transcriptase and have
a very low infectivity, . . .
virus that's sitting around for a long
period of time can lose its
infectivity without necessarily losing
ability to detect reverse
transcriptase from the culture." Tr. 765-66;
see also 782-83
(Sodroski); 2241-42 (Malkovsky). 26/
Dr. Sodroski also acknowledged that it would be fair to say that
the
understanding of methods for culturing and propagating HIV has
improved
between 1983 and 1990, and that the value in the textbook was
perhaps
based on practical experience between 1983 and 1990. Tr. at
767. The
issue here is how in 1983-1984 Dr. Popovic viewed the
importance of RT
testing before using patient samples to infect a cell
line. ORI's
position that Dr. Popovic considered this paragraph an
important part of
the methodology presented in the paper is not supported by
the record.
Indeed, since Dr. Popovic did not indicate in the paper that RT
was
first shown for the individual patient isolates described there,
the
contemporaneous evidence suggests that he did not ascribe
particular
importance to testing for RT before infection. 27/
To show that Dr. Popovic had a motive to falsify the statement,
however,
ORI relied on testimony that it would be more logical to test for
RT
activity prior to pooling patient samples because adding samples with
no
virus in them would "dilute" the virus in the pool. For example,
Dr.
Sodroski testified that testing for RT activity before pooling --
does provide a more rational appearance to the pooling.
It
would not make sense to pool samples that didn't contain
virus,
or for which there was no evidence of virus. Adding those
to
the pool would only dilute out the virus preparation. So
it
would make more sense to have first shown that there
was
evidence of virus in any sample added to a pool.
Tr. at 744-45. However, when asked whether it would change
scientists'
opinion of the rigorousness of the experiment, Dr. Sodroski
indicated
that pooling was simply not very rigorous to begin with because it
could
never be reproduced in all of its details again because each isolate
is
unique at that time. Tr. 745-46. Yet, the relevant paragraph
in the
paper does not hide the fact that more than one patient sample was
used
to infect the parental cell line; the paragraph states that
the
"concentrated culture fluids" used for "repeated exposure"
were
"harvested from . . . cultures . . . obtained from patients with AIDS
or
pre-AIDS." Ex. H-5 at 499. We do not find it credible that Dr.
Popovic
would honestly reveal in the paper the lack of rigorousness inherent
in
using more than one patient culture, yet risk his career by
deliberately
falsifying the timing of obtaining RT results.
Dr. Popovic acknowledged that it would have been preferable to
have
established RT activity in each of the individual cultures
before
pooling them, but provided a reasonable explanation for why he did
not,
which is consistent with testimony by others. First, he said
that
pooling would increase the multiplicity of infection. ORI argued
that
we should reject this reasoning, citing testimony by Dr. Sodroski.
Dr.
Sodroski said that it was not credible that Dr. Popovic would pool
to
increase the multiciplicity of infection because: "To me, knowing
that
there were a certain number of samples that . . . didn't contain
at
least high amounts of HIV, that those samples were added
to
virus-positive samples, certainly one would not expect the
multiplicity
of infection to increase." Tr. at 748. This
testimony is based on the
assumption that Dr. Popovic knew he was adding
samples that did not
contain high amounts of HIV. 28/ When he conducted
these experiments,
however, he did not even know for sure what he was looking
for. He
testified that he had information from clinical physicians
and
morphological evidence (such as giant multinucleated cells) that led
him
to believe the patient samples contained a retrovirus. See, e.g.,
Tr.
at 2303, 2415-19; see also Tr. at 2163 (Read-Connole). Moreover,
he
testified persuasively and without rebuttal that it is well
established
in retrovirology that concentration can increase multiciplicity
of
infection by twenty to fifty fold. He gave the example that if,
under
standard conditions, the increase would be thirty fold, pooling
and
concentrating samples could potentially increase the multiplicity
of
infection fifteen fold, even if only half of the samples contained
the
virus. Tr. at 2495-98.
Dr. Popovic also explained that, based on past studies, he knew
that
retroviruses were heterogeneous and that some strains might be more
apt
to infect a T-cell line than others. Thus, by increasing the
variety of
strains to which he exposed the cell line, he could increase his
chances
of success. Ex. H-153. Dr. Sodroski testified that, even
if some
pooled samples did not contain high amounts of HIV, "you
could
potentially add very low levels of minor virus variance to the
pool"
which would increase "virus diversity." Tr. at 748-49. Dr.
Sodroski
also testified that it is now known that some strains of HIV grow
better
in T-cell lines than others. Tr. at 715-17. Thus, Dr.
Sodroski's
testimony supported Dr. Popovic's on the diversity question;
ORI
presented no testimony that rebutted Dr. Popovic's on this point.
Finally, unrebutted testimony established that Dr. Popovic would have
used
pooling when working under his mentor, Dr. Svoboda (Tr. at
1949-51), and that
Dr. Popovic is the type of intuitive scientist who,
while engaged in hunting
for a novel retrovirus, would likely try
experiments which others (with
hindsight) might view as illogical (see,
e.g., Tr. at 1985-86). In our
opinion, these factors make it unlikely
that Dr. Popovic would have viewed
what he did as illogical when he
wrote the paper and would have felt a need
to misrepresent his method to
make it appear more "rigorous" to others.
In sum, ORI did not prove by a preponderance of the evidence that
the
disputed sentence was intentionally falsified or even that it
was
untrue.
B. ND ENTRIES IN TABLES 1 AND 2
Table 1 of the Science paper reports on the response of cloned
T-cell
populations to infection with HTLV-III. Results for eight
different
clones at 6 and 14 days after infection are shown for the
following:
total cell number; percent of multinucleated cells; percent of
positive
cells as shown by immunofluorescence assay (IFA) 29/ against both
rabbit
antiserum to HTLV-III (diluted 1:2000) and patient serum (from
patient
E.T.); and reverse transcriptase (RT) activity. Table 2 of the
paper
reports on isolation of HTLV-III from patients with AIDS and
pre-AIDS.
Information is given for five patient isolates regarding the
patient
diagnosis and origin of the sample. Virus expression is
reported
through RT activity; percent positive cells in IFAs against
rabbit
antiserum and serum from patient E.T.; and electron microscopy.
See
Appendix A, the Science paper, for these tables.
At issue here is the use of "ND" in Table 1 for four data points
reporting
6-day IFA results for four different clones against patient
serum E.T., and
the use of "ND" in Table 2 for two data points for
patient isolate
S.N.: the IFA against E.T. serum and the electron
microscopy
result. The question of scientific misconduct here centers
around the
definition "ND, not done." Ex. H-5 at 498. This
definition
appears in fine print in the middle of the lengthy legend to Table
1,
but does not appear in the legend to Table 2.
ORI argued that the ND entries are false based on its conclusion that
the
phrase "not done" means that an experiment was "not performed" and
that the
laboratory notebooks demonstrate that these experiments were,
in fact,
performed. See, e.g., ORI post-hearing br. at 73, 83. ORI
argued
that, instead of recording in the tables the actual results of
the
experiments, Dr. Popovic deliberately concealed the actual results
by listing
the data entries as "not done." ORI post-hearing br. at 77,
88.
The parties' presentations on this issue represented a battle of
experts
concerning the meaning and use of "ND" and "ND, not done" in
scientific
papers -- a battle in which much of ORI's own experts' testimony
did not
support its position. While there may be an immediate,
commonsense
reaction that "not done" means that you did not do it (what ever
"it"
is), even a commonsense reaction is ambiguous in this context --
what
did you not do, any experiment at all or any quantification of
its
results. After extensive expert testimony, we found that "ND" and
"ND,
not done" might mean something different from what a
commonsense
reaction might tell one and in fact mean various things when used
in
scientific papers (but in general convey that there is no
meaningful
data to report).
Dr. Popovic has consistently stated that, at the time he drafted
the
tables, and put in the "NDs," he intended to convey the meaning that
the
data was inconclusive, unquantifiable, or not determinable. See,
e.g.,
Tr. at 2310-19; 2447-49; 2451-54; and 2471-77; Ex. H-157 at
19-22. He
said he could not recall putting the definition "ND, not
done" in the
legend, but argued that, in any event, it was not inaccurate to
state
that the experiments were "not done," given the various
possible
meanings of that phrase. Tr. at 2309; 2352-53. Dr.
Popovic produced
drafts of the paper which show that the definition was not
in his
original handwritten draft, but appeared first in draft seven.
Compare
Exs. H-6 and H-12. Unrebutted testimony shows that a number of
other
individuals were involved in editing the drafts by that point.
ORI presented evidence which it said showed that Dr. Popovic most
likely
drafted the legend, including the definition, and that, in any event,
he
was responsible for it. We do not need to decide here whether
Dr.
Popovic in fact drafted the definition in the legend, in light of
our
conclusions on the threshold issue of whether the ND entries were
false.
On the basis of the record before us, we conclude that ORI did not show
by
a preponderance of the evidence that the ND entries in Table 1 and 2
were
inaccurate or misleading, much less that they were
intentional
falsification. We conclude instead that any dispute over
the validity
of the ND entries is based on honest differences in
interpretation or
judgment of data.
We reach our conclusions about the ND entries for the following reasons:
o ORI's finding depends on a reading of
"not done" as "not
performed," but the weight of the evidence does not
support ORI's
contention that it is commonly accepted within the scientific
community
that the phrase "not done" means "not performed."
o Dr. Popovic's testimony was credible
that he did not use ND to
mean "not performed," but instead meant to convey
that the data was
inconclusive, unquantifiable, or not determinable. He
has consistently
asserted that he used the ND in that sense, and what he did
in this
paper was not inconsistent with what he did in other papers, contrary
to
what ORI argued.
o Dr. Popovic satisfactorily explained
that the use of ND here
represented a reasonable judgment because it would
have been misleading
to the reader to report exactly what appeared in the
laboratory
notebooks. Dr. Popovic's testimony was supported by
persuasive expert
testimony.
o The expert testimony on which ORI
relied to interpret the
results in the laboratory notebooks indicates that
these experts'
opinions were not based on accurate facts; the testimony
indicated that
the ORI experts had failed to take into account the entire
parameters of
the experiments performed; failed to consider the subjective
nature of
the IFAs or the imprecise nature of an EM and to read the results
in
light of the results of parallel experiments showing virus
expression;
and, in many instances, misconstrued what certain entries in
the
laboratory notebooks meant.
o ORI did not establish that Dr. Popovic
had any motive to
misrepresent or falsify the data, and the expert testimony
indicates he
would not have had a motive. In the opinion of
experienced
retrovirologists, the use of the NDs rather than the precise
notebook
entries did not add anything to the paper nor affect the
overall
conclusions.
1. ORI's construction of "not done" as meaning
"not performed" is
not supported by the record.
ORI's basis for arguing that Dr. Popovic falsified the ND entries
on
Tables 1 and 2 of the Science paper is its construction of
the
definition, "ND, not done", in the legend to Table 1. ORI reads
"not
done" as unequivocally meaning "not performed" (by which ORI means
not
even attempted). ORI determined that the ND entries were
falsifications
because they were represented in the tables as "ND, not done,"
yet the
experiments were performed. ORI concurred with previous
investigatory
findings rejecting Dr. Popovic's explanation that by "ND" he
meant "not
determinable" or "not done properly," although ORI also stated
that it
did not believe that "the `ND' entries served to enhance
substantially
either the methodological rigor or the robustness of the
findings of the
paper." ORI Final Report at 9-10, 55.
Therefore, ORI's position is dependent on reading "ND, not done"
a
particular way. However, the testimony and evidence shows that
ORI's
reading is not the only reasonable reading of this term.
First, Dr. Popovic pointed to dictionary definitions of the term
"done"
which indicate that "not performed" is not the only meaning and that
one
possible meaning is "not completed." ORI did not offer any
dictionary
definition which would indicate that "not performed" is the
only
meaning, or even that it is a preferred meaning, of "not done."
30/ Nor
did ORI specifically find that, if "not done" were used to mean
"not
completed", the ND entries would be false. Thus, at the very
least, ORI
had to establish that the commonly accepted usage of "not done" in
the
scientific community at the time of the conduct was different from
the
ordinary meaning, and, in fact, was the meaning ORI advanced here.
In
addition, to show that Dr. Popovic intentionally falsified the
results,
ORI would have to show that Dr. Popovic was aware of that special
usage.
For its position that "ND, not done" meant "not performed," ORI relied
on
testimony from five witnesses (Dr. Richards, Dr. Hadley, Dr. Martin,
Dr.
Watkins, and Dr. Gartner). 31/ For the following reasons, we find
that
ORI's reliance on this testimony is misplaced:
o Dr. Richards was one of a group of
scientists who advised NIH
during the investigation. The advisors did
not themselves undertake to
investigate and ascertain whether the statements
in issue were false.
Tr. at 513-15. Dr. Richards testified that the
advisors had concluded
that the ND data points appeared to be falsified
because ND is defined
in the footnote to the paper as meaning not done, which
is a term of art
meaning the experiments were not performed. Tr. at
438, 514. We give
little weight to his testimony for two reasons.
First, his opinions
were developed based on selected information given to him
by ORI. See
also Tr. at 421, 467-70; 492; 510; 512-13. Second,
Dr. Richards readily
acknowledged that he did not have a background in cell
biology,
virology, or retrovirology. Tr. at 479-81; 503-04;
506-07. We find,
therefore, that he could not reliably comment on
whether "not done" had
a particular meaning in these disciplines. 32/
We also note that Dr.
Richards stated that there are no generally accepted
standard symbols to
use when you might have ambiguous results or a problem
with a particular
experiment. Tr. at 520.
o Dr. Hadley (a psychologist) led the
initial investigation. She
was not accepted as an expert witness for
purposes of this proceeding.
Dr. Hadley's testimony does not establish a
community standard.
She was asked:
Q: On the basis of the
investigation
conducted at ORI, what is your understanding
of
the term not done to mean?
She testified as follows:
A: It means the experiment was not
done. It was
not attempted. It was not performed.
Tr. at 936. 33/ This testimony merely states an
ultimate
conclusion and does not provide us with any clear basis for
Dr.
Hadley's "understanding." In the absence of any
such
information, this testimony has little utility in
establishing
that there was in fact a community standard that "not
done" only
meant "not performed" at the time used in the Science
paper.
o Dr. Martin (who was qualified as an
expert retrovirologist)
testified, when first questioned on the NDs, that
every time he sees an
ND "it can either mean not done or not detected.
It could be either
one." Tr. at 1344. He further stated that it
meant "that particular
experimental part of the table, for some reason, it's
usually a
technical reason, contamination with mold, for example, the bane
of
virologists, or some problem akin to that, just wiped out
that
particular experimental result, or the experiment just wasn't --
was
overlooked. We all make mistakes, and it just wasn't --" Tr.
at 1344.
It was only after he was specifically asked whether he meant it was
"not
performed" that he answered, "It was not performed. Correct." Tr.
at
1344. Since Dr. Martin did not state that "not done" meant
"not
performed" until prompted to do so by counsel (and since, as
discussed
above, Dr. Martin may not have been disinterested), we give
more
credence to his first reaction about the meaning of the NDs in
the
tables. That testimony supports Dr. Popovic's position that
ND,
regardless of how it is defined, can be used not only where
an
experiment is not even attempted, but also where there are
technical
difficulties which prevent it from being properly completed.
o Dr. Watkins, an ORI investigator,
testified that "[t]he
tradition in science, the use of not done means that an
experiment was
not done, not that it was done, but not interpretable, or not
finished,
or something of that nature." Tr. at 1414. Dr. Watkins
did not testify
about his basis for reaching this conclusion. Since he
was not
qualified in these proceedings as an expert (in any relevant
discipline)
and did not explain the basis for his conclusion, we give little,
if
any, weight to his opinion.
o ORI further relied on testimony of Dr.
Gartner, an expert
retrovirologist, as establishing that it was commonly
accepted within
the scientific community that "not done" means "not
performed." The
transcript page ORI cited (1845) does not contain any
remarks by Dr.
Gartner on this particular subject. The only reference
to "not done" is
in a question to Dr. Gartner asking her to assume that for a
particular
patient sample no IFA was done. In response, Dr. Gartner
asks, "What do
you mean by not done?" and counsel answers, "[t]he test was
not
performed." Tr. at 1846. Dr. Gartner then indicates that to
her
"that's a significant difference." Tr. at 1846. Obviously if
Dr.
Gartner thought that "not done" commonly meant "not performed,"
she
would not have asked ORI counsel what he meant by "not done" and
would
not have stated that there is a significant difference between the
two
terms. Other testimony by Dr. Gartner more directly supported
Dr.
Popovic. She testified that she had used ND to mean "not done,
not
determined meaning I couldn't come to a definitive conclusion." Tr.
at
1883. While she stated that she would feel obliged to explain to
the
reader where there was an "ambiguity" or some technical problem,
she
also indicated that the journals sometimes treat that type
of
explanation as "irrelevant." Tr. at 1883-84. Finally, she
testified
that the scientific community had different ways of explaining that
they
do not have a data point, and that whichever definition of "ND" was
used
"[m]ost of us realize that what that means is that there was . . .
no
meaningful data generated, at least that's what it means to me. . .
."
Tr. at 1887-88.
Thus, ORI's reliance on these witnesses' testimony is misplaced;
either
that testimony, considered as a whole, supports Dr. Popovic more
than
ORI, or that testimony is conclusory opinions by individuals who did
not
have relevant expertise and who did not clearly articulate any basis
for
their conclusions.
Testimony from other expert witnesses presented at the hearing
supports
Dr. Popovic's position and establishes that both ND and "not done"
have
many meanings other than the narrow meaning relied on by ORI.
o Dr. Huth, ORI's expert witness in the
publication of scientific
papers and the standards in the scientific
community with respect to
editing and publication of scientific papers,
stated that "there can be
a lot of ambiguity about [ND]. This new style
manual that I'm working
on with the Council for Biology Editors . . . has an
array of possible
meanings, not determined, not done." Tr. at
1200. While he testified
that he felt that an author using ND has the
obligation to inform the
reader of the usage of ND, nevertheless his
testimony indicates that he
did not ascribe the limited meaning to ND that
ORI did -- that an
experiment was not even attempted. Dr. Huth's
testimony indicates that
ND and "not done" were ambiguous. Tr. 1200-01;
1228-29; 1239-40.
o Dr. Schachman, an expert in accepted
practices for conducting
and reporting biological research, testified that he
was familiar with
the use of the abbreviation "ND" and that it usually means
"not
determined," which he interprets "as not determinable, so that
an
experiment is done, and one doesn't know how to handle the data"
because
one could not quantify the sample, "the number didn't make sense, or
you
couldn't do the experiment sufficiently precisely to warrant a
number."
Tr. at 1274-75. He also testified that he usually took for
granted the
definition of ND. Tr. at 1275.
o Ruth Kulstad, an editor of scientific
journals for almost 30
years, and the editor of the Science paper here,
testified that the
abbreviation ND had no standard meaning and that "it means
all sorts of
different things to different people." Tr. at 1723.
She testified that
she quite often found that abbreviation used without a
definition. She
further stated that she did not think anyone
(particularly scientists)
really cared how the term was defined. Tr. at
1723, 1742-43.
o Dr. Malkovsky, an expert
retrovirologist, testified that his
understanding of the meaning of ND both
generally and within the
specific context of Table 1, was that either the
experiment could not be
performed for various technical reasons or, if it was
performed, it
provided inconclusive data which was impossible to
interpret. Tr. at
2228; see also Tr. at 2249-51. He stated "[i]t
simply means I don't
know, so it conveys sort of zero information." Tr.
at 2228. He further
explained that, while a scientist may have an
obligation to report what
is meant by ND, "ND, not done" has many different
interpretations. Tr.
at 2253. Dr. Malkovsky also testified that
"if you don't get [a]
technically satisfactory result, some people would say
not done, some
people would say not determined, some people would basically
not comment
on it at all." Tr. at 2251.
o Dr. Berns, one of ORI's expert
witnesses and scientific
advisors, testified that there was no distinction,
in his understanding
between "not done" and "not determined." Tr. at
1049, 1096. Moreover,
his testimony showed that he was senior author on
a paper which reported
experiments as "ND, not done" in a table while at the
same time
describing those experiments in the discussion section of the paper
as
though they had been performed. Tr. at 1101-03; see Ex.
D-18. Thus,
on its face, this paper would support a conclusion that
"not done" is
used not only to mean "not performed." 34/
Finally, while ORI submitted several exhibits that suggested
the
scientific community may have tried to establish standard
abbreviations
or standard meanings for the abbreviations, this evidence does
not show
that the scientific community had, during the time period in
question,
established a standard meaning for the use of the abbreviation ND
or the
term "not done" to mean "not performed." See Exs. H-102, H-109,
H-110.
In sum, ORI did not establish by a preponderance of the evidence that
"not
done" meant only "not performed" (not even attempted). Instead,
the
evidence shows that both ND and "not done" had a variety of
meanings.
2. Dr. Popovic's testimony about what he meant
in the paper is
credible.
There are a number of reasons why we find credible Dr. Popovic's
testimony
that he did not use the NDs as meaning "not performed" and was
not aware that
the definition used in the paper might have suggested
that meaning to some
scientists. As discussed above, we found Dr.
Popovic in general to be
credible, based on our observation of him at
the hearing and unrebutted
testimony about his integrity. Moreover, he
provided a persuasive
explanation of why he put the NDs, rather than the
exact entries in the
laboratory notebooks, and ORI did not establish any
motive for him to falsify
the results. In this section, we discuss two
additional reasons why we
found him credible on this issue: first, he
has been consistent in
explaining what he meant by the NDs (even in a
statement made before this
investigation began), and second, ORI did not
establish that he had knowingly
used ND to mean "not performed" in other
papers.
Our review of the record indicates that Dr. Popovic has
consistently
maintained that he meant to convey a different meaning than that
the
experiments were not performed. Ex. H-48 at 43-48; Ex. H-157 at
19-22;
Ex. 49 at 5-7; Tr. at 2308; 2471-75. In the first written
response to
ORI in our record that specifically deals with the subject of the
use of
"ND, not done" in the paper, Dr. Popovic asserted that he had
explained
from the beginning that, at the time the paper was written, "ND"
"was
equal to `not done,' `undeterminable,' or `not detected,'" in his
mind
and that he did not "understand that `ND' could have at least
three
different meanings when [he] prepared the original manuscript
tables
[since he] did not fully appreciate the distinction among
`not
detected,' `not determined,' and `not done'." Ex. H-48 at 43.
Dr. Popovic's explanation is corroborated by other evidence in
the
record. We find particularly persuasive that in a memorandum to
Dr.
Gallo, dated March 25, 1986, Dr. Popovic stated that the EM data
for
isolate S.N. in the Science paper "were not done properly." Ex.
H-61.
This evidence predates the investigation into this matter and
was
prepared outside of the confines of this dispute. This
memorandum
indicates that Dr. Popovic did not intend to suggest that
the
experiments were not performed at all.
Also, we reviewed the use of ND by Dr. Popovic in papers (other that
the
Science paper in dispute here) on which he was first author that
were
submitted into the record. 35/ Ex. P-42, P-43, P-50, P-51.
Dr. Popovic
did not use the term "ND, not done" in any paper in the record
for which
he was listed as first author, other than the paper at issue.
He used
the term "ND-not determined" in Exhibit P-50, and "ND-none detected"
in
Exhibit P-51; however, this paper was issued after the paper
in
question. Ex. P-51; Tr. at 2346-47 (Popovic). In the other two
papers
on which Dr. Popovic was first author, he used the term "NT-not
tested"
to reflect when experiments were not performed. 36/ If Dr.
Popovic had
truly intended to mislead the reader into thinking that the
experiments
in question here had not been even attempted, we think it more
likely
that he would have used "NT-not tested," rather than the more
ambiguous
"ND, not done."
ORI determined that Dr. Popovic had used the terms NT and
ND
interchangeably to mean "not performed." ORI post-hearing brief at
80.
This determination was based on an analysis of two of Dr.
Popovic's
papers that used ND as "not done." Ex. H-84 at 94. One
of these papers
was published in the journal Neoplasma in 1970, and one was
published in
the International Journal of Cancer in 1969. We do not
find this
analysis persuasive, for several reasons. First, ORI deduced
that the
experiments in the papers were not in fact performed based solely on
its
analysis of what the paper said. Ex. H-84 at 93-94. ORI did
not
apparently have, and did not present to us, any independent evidence
of
whether the experiments were in fact performed. Second,
unrebutted
testimony by Dr. Popovic establishes that his articles in
Neoplasma were
written in his native language and then translated into
English by
language experts, as were many of his early articles that appeared
in
other journals. Tr. at 2287-88. Also, he does not appear as
either the
first or the senior author on the article in the International
Journal
of Cancer. Ex. P-48 at 46. Absent evidence that the
translators or
other authors discussed with him how they were defining ND in
these
papers, we cannot reasonably infer that Dr. Popovic intended to use
ND
to mean "not tested" in these papers. Finally, in deciding what
Dr.
Popovic's practice would have been in 1984, we find articles
published
closer to 1984 to be more relevant than these papers published more
than
a dozen years before.
In sum, we found credible Dr. Popovic's testimony to the effect that
he
did not mean to convey to the reader the erroneous impression that
none
of the experiments reported as ND were even attempted.
3. The ND entries represented a reasonable
judgment by Dr. Popovic
as to how to report the actual experimental
results.
Among other data, Table 1 contains the percentage of
immunofluorescence
positive cells for eight clones against rabbit and patient
antisera at 6
and 14 days after infection. Dr. Popovic reported as ND
the IFA results
for clones H9, H17, H31, and H35 for patient serum E.T. at 6
days after
infection. Ms. Read-Connole was the technician who prepared
and first
read the IFA slides. Her laboratory notes record her readings
of IFA
slides against eight antisera for 17 different cell cultures; most
of
them were T cell-clones (either infected or uninfected) or
patient
samples. She recorded results for: three different
dilutions of rabbit
antisera; normal rabbit serum; sera from AIDS patients;
serum from Dr.
Popovic (MIKA); and one control positive serum against HTLV-I
virus.
See Appendix for Table 1 and for Ms. Read-Connole's laboratory
notebook
page 30. While generally positive results were reported as
a
percentage, no percentages were recorded for clones H9, H17, H31,
and
H35 at 6 days. For clone H9, Ms. Read Connole recorded "+"; for
clones
H17 and H31, she recorded "-". The record is unclear whether
clone H35
was not tested against the patient antisera or simply not
evaluated
(although it is clear that there was a slide prepared for clone
H35).
Ex. H-63; Tr. at 2168-69 (Read-Connole); Tr. at 1546 (Schaffer); and
Tr.
at 2315-16 (Popovic).
To explain why his assessment of the 6-day results for these four
clones
was unquantifiable or not determinable, Dr. Popovic testified
that in
light of the other data reported in Table 1 for these clones, i.e.,
the
relatively low number of total cells, the presence of
multinucleated
cells, the reaction against the rabbit antisera, and the
reverse
transcriptase activity, he knew that the clones were positive for
virus
production. Tr. at 2310-16.
For clone H9 he explained that due to the cytopathic effect of the
virus,
the cells were damaged and the slide was of "low quality," but he
stated that
he knew from other data that the virus infected the clone.
Tr. at
2310-12. Dr. Popovic gave a similar assessment of the quality of
the
preparation for clone H17 and noted that it was recorded as "-"
because it
was not possible to determine viral activity from the IFA
assay although all
other "parameters clearly indicate that there was a
virus production."
Tr. at 2313. Dr. Popovic stated that, with respect
to the patient sera
IFA, the inability to detect viral activity in clone
H17 was artifactual data
which was misleading. With regard to clone H31
Dr. Popovic testified
that --
The phenomenon was precisely the same as for H9,
H17,
the quality of the slides were low and it
wasn't
possible to quantify. And for that reason, I put
ND.
However, all other data . . . clearly showed that
there
is a virus production.
Tr. at 2314-15. 37/
For clone H35, Dr. Popovic testified that it had not been "evaluated."
Tr.
at 2315. He stated --
So, obviously, I put ND because the problem was again the
same,
there were not sufficient cells to make a good, such quality
of
preparation which one could interpret with a confidence
and
could evaluate, could put there a number.
Tr. at 2316.
Dr. Popovic further testified that not detecting the AIDS virus in some
of
these assays was not truly a negative result, i.e., signifying "no
presence
of the AIDS virus." Tr. at 2446. When discussing the IFA
results
reported in Ms. Read-Connole's notes [Ex. H-22 at 30], he stated
--
It says only that that particular slide which was evaluated,
the
person couldn't with confidence tell that it was a
fluorescent
cells, that is what it says. It does not says that it
is
negative for the virus.
. . . .
No. What it [the minus value] says that with the viral
antigens
which were in the cells, there were no reaction with
that
partic[ular] sera. Not that it was virus or not virus
was
there. The data which we obtained were inconclusive, that
is
what it says.
. . . .
You should take into consideration other reactivity which
we
paralleled, other data which we paralleled with this
experiment.
If you take in consideration together, obviously, this data
are
inconclusive and one cannot claim that it is negative. It
is,
immunofluorescence is a subjective matter. . . .
Tr. at 2446-48.
Table 2 reports RT, IFA, and Electron Microscopy (EM) results
for
individual patient isolates. 38/ See Appendix A for Table 2.
At issue
are the ND entries for patient S.N. for IFA and EM. 39/
Dr. Popovic testified that there were conflicting IFA results for
patient
S.N. since it was negative in one experiment and positive in
another.
Tr. 2317-18; Ex. H-22 at 19, lines 2 and 4. Moreover, since
other data
for this patient isolate were positive for virus activity,
Dr. Popovic
regarded it as "misleading" to report as negative the
failure of one assay to
detect virus activity. Tr. at 2317-18.
Similarly, for the electron
microscopy results, Dr. Popovic testified
that, in light of the other
positive data, as a "very imprecise
technique . . . one can readily miss the
presence of virus in the cells.
. . . Obviously, I have to question the
validity of the electron
microscopy observation." Tr. at 2319.
The expert retrovirologists assessed the results reported as ND
together
with other data presented in Tables 1 and 2. Dr. Gartner
testified that
in the context of Table 1, the ND entries would not make any
difference
in evaluating the data. She testified --
They would mean that there's no data for that, those points.
I
guess it wouldn't mean very much to me because when I look at
a
paper, the primary thing I'm looking for is whether or not
the
evidence presented supports the conclusions the author
is
drawing. So I would focus my attention on some of the
other
things. For example, I would look down at the
reverse
transcriptase activity and at day 14 I see a value of
two
million cpm of RT that is such definitive evidence to have
RT
activity at that level is so unmistakable, I would say
this
person has something.
. . . .
It [the RT activity] is under the H9 clone, but the others
are
also very high. The level of activity there is unmistakable
and
also I'd probably go back and look at, if I look at the NDs,
I'd
say well why did this person get, why is there an ND here and
I
would notice . . . there were too few cells to evaluate would
be
the first thing that would come to my mind, especially
since
they've described here a cytopathic virus.
So I would look at the total cell number . . . the cell
numbers
in those particular clones [H9, H17, H31, and H35] at day
six
were all very low.
. . . .
. . . So I guess I'd conclude that the reason why a
person
couldn't evaluate this, if that was indeed what ND meant
as
opposed to not doing the test, that it was because there
were
too few cells . . .
. . . .
No, it wouldn't make any difference to me [if instead of
ND
those had all been just negative]. Again I would take the
whole
. . . . Looking at the day 14 . . . that's such
definitive
evidence that to me it's overwhelming that this is real.
Tr. at 1826-28; see Tr. at 1904-05 (Blattner); Tr. at 1955 (Svoboda).
Dr. Gartner testified that since the notebook entries indicated
a
quantification in almost all cases for all three dilutions of
rabbit
antisera but no quantification for patient sera for these four
clones,
this suggested to her that the person doing the assay was not able
to
come up with a definitive number for these clones. Tr. at
1831-32. She
testified that whether it was appropriate to report
results for these
clones as ND is "an individual thing" which "depends on
what one's
individual terminology is, the way one keeps track of
one's
observations." Tr. at 1832. She also testified that the
more
meaningful results were against the rabbit antisera because of
the
variability in patient antisera with regard to level of antibody
to
viral components. 40/ Tr. at 1833-36. Dr. Gartner discussed in
detail
her assessment of the IFA results for clones H9, H17, and H31
as
recorded by Ms. Read-Connole. Tr. at 1830-36; 1866-70. Her
testimony
overall corroborated Dr. Popovic's explanation. In addition,
Dr.
Gartner testified that she had concluded that the "investigator
[could
draw] no meaningful conclusion . . . from reading" those
slides. Tr.
at 1905. Because of the low cell counts for day 6,
she thought that
"there was difficulty in interpreting the results."
Tr. at 1904-05.
She affirmed that the data reported in the Table were
adequate for her
to evaluate the results of the experiments. Tr. at
1905.
Dr. Malkovsky testified that he construed the NDs as meaning either
that
the experiments could not be performed for technical reasons or
if
performed, the results were inconclusive and could not be
interpreted.
In his opinion, it did not matter how ND might be defined.
Tr. at 2228.
Dr. Malkovsky testified that he found Table 1 consistent with
the
results reported in Ms. Read-Connole's notebook since the IFA
results
for the clones in question had not been quantified. Tr. at
2232-33,
2250. Moreover, he stated that in light of the other data, he
would
expect the 6-day data for clones H9, H17, H31, and H35 to be
difficult
to interpret. Tr. at 2254.
For Table 2, Dr. Gartner testified that the NDs for IFA and EM results
for
patient S.N. in Table 2 were not "meaningful." Tr. at 1813, 1820.
She
stated --
. . . it's not significant. When I look at this table the
point
I see which distinguishes it from Table 1 is that there
was
isolation of this virus from more that one individual and
that
it was possible to detect that virus from the culture
material
that was used from more than one individual. So that
value to
me, that's insignificant to me.
Tr. at 1829.
In general, Dr. Gartner compared EM "with respect to novel virus hunt,
as
a needle in a haystack." Tr. at 1820. She agreed with the
analogy
that it was like slicing a loaf of raisin bread and if you happened
not
to get a raisin, it did not mean the loaf was not raisin bread. Tr.
at
1820-21. Dr. Sodroski indicated that alone EM was not a
very
"efficient" technique since there may be a relatively low number
of
infected cells to search for. Tr. at 667, 672-73; see also Tr.
at
1954-55 (Svoboda).
In assessing all the ND entries for both Tables 1 and 2, what is
striking
from Dr. Gartner's testimony is that in the circumstance here,
where one is
searching for evidence of a novel retrovirus, each assay or
test only
provides a clue which may or may not be useful. Tr. at
1785-1821
(Gartner). She emphasized that no one test or assay was
"good" by
itself. Rather, she testified:
. . . . Everything is done in the context of the whole,
the
total. You can't evaluate any single piece of data in a
vacuum
when you're working in this kind of system because there are
too
many other potential explanations for that finding.
But when you have several clues that all point to the
same
direction, then it's believable. But no, I would not
evaluate
any of these methods independently.
Tr. at 1821.
The failure to detect the presence of the virus from any one test was
not
particularly significant. The retrovirologists emphasized that
these
were subjective assays or tests, dependent in large measure on
the
investigator's interpretations. Based on unrebutted testimony
from
these retrovirologists as to the nature of IFAs or EMs when
an
investigator is looking for evidence of a novel retrovirus, we find
that
whether or not the results had been literally transposed from
Ms.
Read-Connole's notes (or in the case of the EM reported as
negative),
the tables would have conveyed the same critical information
concerning
the conclusion as to whether there was persuasive proof of
viral
expression. Thus, we conclude that since other data reported
showed
that a culture was positive for virus, it was a reasonable judgment
for
Dr. Popovic to report as ND experimental results for tests where
virus
activity had not been detected but which in the context of
other
available data Dr. Popovic judged to be not quantifiable or
artifactual.
4. The expert testimony which ORI presented did
not prove that the
ND entries were inaccurate, misleading, or false.
We discuss in this section why we found that the expert testimony
which
ORI presented did not outweigh the testimony by Dr. Popovic and
his
witnesses and therefore did not establish that the ND entries
were
inaccurate, misleading, or false. We found the testimony of
ORI's
witnesses deficient in the following respects:
o ORI presented no evidence from a
retrovirologist directly
rebutting the testimony from Dr. Popovic and his
witnesses about why the
ND entries were a reasonable judgment as to how to
report the
experimental results.
o The testimony of ORI's witnesses was
based on the assumption
that "ND, not done" in the tables meant that the
experiments had not
been performed (in the sense of not even
attempted). Dr. Schaffer's
testimony was premised on reading the ND
entries as equivalent to not
performed. Tr. at 1542, 1556, 1558-61,
1563-64, and 1636. Similarly,
Dr. Berns viewed the "ND, not done" here
as meaning only not performed.
Tr. at 1048-49, 1054-55, and 1072. 41/
This is also apparent from the
interpretation given by Dr. Berns to Dr.
Popovic's written statement in
a 1986 memorandum to Dr. Gallo that the EM for
patient S.N. was not done
properly. Dr. Berns concluded that this
statement was inconsistent with
reporting this test as "not done" in the
paper, which to him meant that
the test was not performed at all. Ex.
H-61; Tr. at 1054-57. Dr.
Martin's testimony also suggests that he
reviewed the notebook entries
against what was reported in the table,
specifically Table 1, only in
terms that "not done" meant only not
performed. Tr. at 1348. 42/
o ORI's experts failed to take into
account the subjective and/or
unreliable nature of the IFAs and EMs. As
we discussed above, there was
relevant testimony from expert retrovirologists
that IFA assays are
highly subjective in nature and that EM results were
circumstantial
evidence at best. Tr. at 1791, 1812-16, 1818-20, 1824-25
(Gartner);
2251 (Malkovsky); 667, 672-73 (Sodroski); 1952-55 (Svoboda).
Failure to
consider the nature of these results led ORI's experts to
misinterpret
Ms. Read-Connole's notes to mean that virus was not present.
43/ In
contrast, the retrovirologists indicated that if there were
other tests
performed at the same time that conclusively showed the presence
of
virus, one would conclude that virus was in fact present.
o ORI's experts failed to consider the
other evidence Dr. Popovic
had that there was virus expression. As we
found above, the expert
retrovirologists would have derived from the tables
with the ND entries
information adequate to assess the experimental results
reported there;
they considered the ND entries to be consistent with what
other data in
the table indicated about the slides at the 6-day point in the
case of
the IFAs against patient sera E.T., or about virus expression in
the
patient S.N. isolate as compared to virus expression in the
other
individual patient isolates. Dr. Schaffer testified, with respect
to
the IFA experiments in Table 1, that the results for those clones made
a
difference to her because "if the data were done, I would like to
know
if it's not determinable . . . ." Tr. at 1674-75. This
statement,
however, reflects interest in information that the
expert
retrovirologists testified they could derive from Table 1 as a
whole.
o ORI's experts misunderstood the entries
in Ms. Read-Connole's
notebook. 44/ Tr. at 1048, 1057, and 1072-74
(Berns); Tr. at 1347-48
(Martin); Tr. at 1540-42, 1563, 1611, and 1615-16
(Schaffer). While Dr.
Schaffer was skeptical of Dr. Popovic's reading
of the laboratory
results for IFAs as conflicting for patient S.N., Dr.
Schaffer's
testimony indicates that she did not really understand the
entries. Tr.
at 1601-02; 1616-18. Dr. Berns indicated that he had
no information
from Ms. Read-Connole about what she meant by her notations of
"+" and
"-" and no entry for the clones H9, H17, H31 and H35,
respectively. Tr.
at 1067, 1111; see also Tr. at 1675 (Schaffer).
He stated that negative
to him meant "she didn't see anything lighting
up." Tr. at 1111. This
is inconsistent with what Ms.
Read-Connole's testimony shows about what
one would see on IFA slides with a
cytopathic virus. Tr. at 2156-57;
see also Tr. at 1813-15
(Gartner).
Moreover, ORI did not show that Dr. Popovic was obligated to report
the
results exactly as they appear on the notebook pages. ORI's
experts
gave conflicting testimony about how a scientist in Dr.
Popovic's
position should report the results of these experiments in a
scientific
paper. The experts had differing opinions about:
1) whether or not
these tests should be repeated (Tr. at 1351-54
(Martin); and 1561,
1678-79 (Schaffer); 1051 (Berns)); 2) whether the "+" or
"-" entries
could be intermingled with the percentage figures (Tr. at
1109-10
(Berns); 1542-46 (Schaffer)); 3) whether the "+" and "-" entries
should
have been quantified (Tr. 1543, 1673-77 (Schaffer)); and 4)
whether
there needed to be a footnote explanation of why the slides could not
be
read and therefore results could not be determined (Tr. at
506
(Richards)). We conclude therefore that these witnesses were
testifying
to their own personal standards, rather than to standards for
reporting
data commonly accepted in the scientific community at the time
the
Science paper was published.
5. ORI did not establish that Dr. Popovic had
any motive to falsify
when he used ND in Tables 1 and 2.
ORI contended that Dr. Popovic intentionally concealed the actual
results
of the experiments in Tables 1 and 2 by listing certain data
points as
ND. ORI post-hearing br. at 77-78. ORI, however, concluded
that
the "misreporting . . . did not affect the conclusions of
the
experiment." ORI Final Report at 55-56. The heart of ORI's
argument
here is that "ND, not done" can only mean that the experiments were
not
performed and, therefore, to report experiments as not being
performed
when they were actually performed is an intentional act of
deception.
Moreover, ORI reasoned that we can infer intent to deceive because
the
overall effect of using the NDs is to make the paper look
better
scientifically. ORI post-hearing br. at 78.
We conclude, for the following reasons, that ORI did not prove by
a
preponderance of the evidence that Dr. Popovic had any motive to
falsify
when he used ND in these tables:
o As we discussed above, there is no basis for us to conclude that
"not
done" means only "not performed." We found that Dr. Popovic used
ND in
these tables for the data points at issue to convey that the data
was
inconclusive, unquantifiable, or not determinable. Moreover,
Dr.
Popovic's use of the term ND to mean not determinable here
certainly
conveyed to the reader that the results of these experiments were
not
clearly positive and that they were at best inconclusive.
Furthermore,
Dr. Popovic satisfactorily explained that the use of ND here
represented
a reasonable judgment on his part because to report the results
exactly
as recorded in Ms. Read-Connole's notebooks would have been
misleading
to the reader.
o We find it significant that there was ample testimony from
expert
retrovirologists stating that the NDs in the tables conveyed to
them
zero information. Dr. Malkovsky testified that the 6-day data was
not
important for interpretation of Table 1; and the NDs here conveyed
zero
information. Tr. at 2228. While he found this data useful to
show the
dynamics of the infection, he did not find this data "biologically .
. .
important" and stated that he would probably omit the 6-day
data,
showing only the "optimum" data at 14 days. Tr. at 2229.
Dr. Malkovsky
also testified that IFAs were the least reliable biological
method and
that the 6-day data was "not really adding anything to [Table
1]." Tr.
at 2251. Moreover, he indicated that a reader would
infer from the face
of Table 1 that there might be a difficulty in
determining the results
of an IFA at day 6 because the paper indicates for
the clones in
question that the yield of cells at this time was incredibly
low, which
would indicate massive infection and a lot of damaged cells.
Tr. at
2254; see also Tr. at 1826-28 and 1904-05 (Gartner). Dr.
Gartner
testified that the ND entries for the IFA and EM results are
not
meaningful or significant to her in evaluating the data reported
in
Table 2. Tr. at 1828-29.
o ORI did not present expert testimony to rebut Dr. Malkovsky's and
Dr.
Gartner's evaluation of the ND entries in the overall context of
the
data presented in these tables. Since these entries have no effect
on
the experts' assessment of either the 6-day data or the data as a
whole,
we find that ORI did not show that these entries made the paper
look
better scientifically.
o We do not think it likely that the definition "ND, not
done,"
appearing as it does in extremely fine print in the middle of a
long
legend, could be significantly misleading to the reader. Finally,
the
fact that this definition did not appear until draft seven, late in
the
drafting process, undercuts the notion that this was a
calculated
attempt to mislead.
In sum, ORI did not prove by a preponderance of the evidence that the
ND
entries were intentionally falsified, or even untrue.
C. THE 10% ENTRY FOR THE 6-DAY IFA RESULT FOR CLONE H35
Table 1 of the Science paper reports on the response of cloned
T-cell
populations to infection with HTLV-III. Results for eight
different
clones at 6 and 14 days after infection are shown for the
following:
total cell number; percent of multinucleated cells; percent of
positive
cells as shown by immunofluorescence assay (IFA) against both
rabbit
antiserum to HTLV III (diluted 1:2000) and patient serum (from
patient
E.T.); and reverse transcriptase activity. The entry in Table 1
for the
IFA results against rabbit antiserum for clone H35 at 6 days is
10%.
Ms. Read-Connole's laboratory notes dated February 29, 1984 contain
the
following entry for 6 days after infection for clone H35:
very few cells positive for rabbit antibody
See Appendix B. This notation is written across columns for IFA
assays
with dilutions of the rabbit antiserum of 1:500, 1:1000, and
1:2000.
Dr. Popovic testified that he reread the slides prepared for the
IFA
assay, after Ms. Read-Connole had read them, and quantified the
results
for the 1:2000 dilution for clone H35 at 10%. Tr. at 2320-23;
see also
Exs. H-48 at 49; H-157 at 20-21.
ORI determined that the 10% reported on Table 1 for clone H35 was
"best
characterized as falsification in reporting research." ORI found
that
the contemporaneous laboratory notes indicated that "substantially
more
than 10% were positive." ORI Report at 55-56. ORI argued
that --
Because there is no numerical or other laboratory data
to
support the "10%" entry in Table 1 in the face of
directly
contradictory data and Dr. Popovic's explanations are
not
credible, the "10%" entry is false and misrepresents the
results
of the laboratory data in Ms. Read-Connole's notebook.
ORI post-hearing br. at 89.
We conclude that ORI did not establish by a preponderance of the
evidence
that the 10% entry was intentionally falsified or even untrue,
for the
following reasons:
o Dr. Popovic's testimony that he reread
the slide was credible
and is corroborated by other evidence in the
record.
o Contrary to what ORI suggested, the
record shows that, given the
imprecise and subjective nature of the IFA
assay, it was appropriate for
Dr. Popovic to reread and quantify this data
point, notwithstanding Ms.
Read-Connole's notes.
o Dr. Popovic's testimony that he was
able to place a value of 10%
on the results is credible since ORI presented
no affirmative evidence
that its true value was different and since his
testimony is
corroborated by other evidence in the record.
o ORI did not show that Dr. Popovic had a
motive to falsify this
one insignificant data point.
o ORI's reliance on its expert testimony
is misplaced since that
testimony relied on readings of Ms. Read-Connole's
laboratory notes
which are inconsistent with her own explanation of the notes
and which
are not the only reasonable reading of those notes.
o Under the particular circumstances
here, we find no basis to
draw a negative inference from the absence of
laboratory notes
substantiating the 10%.
o ORI's other arguments are not persuasive.
1. Dr. Popovic's testimony that he reread
the slide was credible.
As discussed above, we found Dr. Popovic generally to be credible, and
his
testimony on rereading the slide is corroborated. Ms.
Read-Connole
testified that the IFA slides reported in Table 1 were from the
"very
first antibody to HIV and this reacted very, very nicely and it
was
quite spectacular and I thought [Dr. Popovic] should see it." Tr.
at
2159. She also testified that it was her general practice to
write
notes to Dr. Popovic in her notebook concerning such results and
to
leave the slides in a refrigerator. Tr. at 2158-59; see Ex. H-22
at
29-31. For these IFAs, her notes singled out certain slides as
a
"[g]ood positive" and directed Dr. Popovic's attention to these
slides.
Ex. H-22 at 31; Tr. at 2158 (Read-Connole).
ORI would have us discount Ms. Read-Connole's testimony on the basis
that,
having collaborated with Dr. Popovic, it was in her interest to
support his
position. Since her testimony is consistent with her
contemporaneous
notes, however, which indicate that she clearly expected
Dr. Popovic to look
at the slides, we do not see any reason to discount
her testimony here solely
on the basis of any interest she might have in
seeing Dr. Popovic
exonerated. Moreover, the testimony of other
scientists who appeared
for ORI indicated that a senior scientist like
Dr. Popovic would reread the
slides after a technician had read them.
See, e.g., Tr. at 318-19
(Goldberger); Tr. at 476 (Richards); Tr. at
1075 (Berns). Here, where
the slides in question were IFAs from the
first tests against the putative
AIDS virus and overall produced data
which the retrovirologists who testified
before us considered
overwhelming evidence of the etiological agent of AIDS,
we find it
almost inconceivable that Dr. Popovic would not have read the
slides, as
he testified that he did.
Thus, we find credible Dr. Popovic's testimony that he reread
the
slides.
2. ORI did not show that it was inappropriate
for Dr. Popovic to
have reread the slides.
ORI said that it was inappropriate for Dr. Popovic to reread slides
and
assign a value where Ms. Read-Connole had not been able to do
so.
However, several of ORI's own witnesses testified that they
would
consider it appropriate generally for a senior scientist to reread
a
slide read by a technician, such as Ms. Read-Connole. See, e.g., Tr.
at
318-19 (Goldberger); Tr. at 474-76 (Richards); Tr. at 1075
(Berns);
accord, Tr. at 1816 (Gartner). 45/
Dr. Richards indicated that he would consider reevaluating the
slide
appropriate, but that he personally would feel uncomfortable doing
so
unless he had discussed the change with the technician or the
technician
agreed. Tr. at 476, 508. While Ms. Read-Connole
testified that Dr.
Popovic had not discussed the 10% value with her at the
time, she
testified at the hearing that she believed that the 10% value for
clone
H35 at 6 days was correct. Tr. at 2170-72. More important,
she also
stated that she had seen Table 1 in Dr. Popovic's handwritten draft
and
that in preparing the table Dr. Popovic would have been working from
her
notes as well as his own notes. Tr. at 2211-12. ORI presented
no
evidence that she objected to the 10% at the time.
Dr. Schaffer also raised questions about Dr. Popovic rereading the
slide,
based on her opinion that Ms. Read-Connole was an expert at
reading IFAs and
that, if she could not assign a value, neither could
Dr. Popovic. Tr.
at 1601-13. In effect, Dr. Schaffer inferred from the
fact that Ms.
Read-Connole did not assign a value that the slide was
unreadable. 46/
This is not a reasonable inference for several reasons.
First, while Ms.
Read-Connole was clearly quite skilled at evaluating
IFAs, her testimony
indicates that she was very conservative in
assigning values when reading IFA
slides. Ms. Read-Connole testified
that --
I think I'm probably one of the best people around to read
them
because I'm very cautious about my readings. I don't like
to
say oh, gee, that's positive without having some other basis
for
believing that the value might be correct or if something
is
indeed not clearcut, I don't want to put it down that it is.
. . . .
A certain percentage when I'm not sure as you can see by
my
pluses in the notebooks.
. . . .
No [she could not make a reading for clone H35], because as
I
explained earlier there was a lot of debris in that first
rabbit
antisera that we had and due to the cytopathic effect of
the
virus infecting the cells in the early stages or a short
time
period after the cells were infected, there was enough
debris
that I could not definitely determine the number of cells
that
were indeed positive to get a positive reading. But I
was
noting that there were positive cells.
Tr. at 2187-89.
Second, the unrebutted testimony indicates that reading an IFA slide
when
detecting a retrovirus involves judgment. In discussing Dr.
Popovic's
quantification in light of Ms. Read-Connole's notebook entry,
Dr. Gartner
testified that --
. . . . But no I wouldn't have any problem with that
because
again, especially with some of the work I've done
with
cytopathic retroviruses like HIV, you don't have the luxury
of
the optimal situation. You have to go with what you can see
and
so, no, I wouldn't have any problem with it.
It's a very subjective assay . . . .
Tr. at 1816. Dr. Gartner's description of what would make a
slide
difficult to read but still possible to evaluate when working with
a
cytopathic retrovirus is consistent with what both Dr. Popovic and
Ms.
Read-Connole said about this clone H35 slide. Ms.
Read-Connole
described the problem with reading such slides as distinguishing
among
the various greens to pick out the true fluorescing cells. Tr.
at
2157-58. There was no scientific evidence presented by ORI to
rebut
this. Since Dr. Schaffer's testimony failed to address this, it
is
possible that she did not consider the judgment involved in reading
IFA
assays of a retrovirus with cytopathic effects, because her work
was
primarily with a different type of virus (herpes simplex). Tr. at
1464
(Schaffer); Tr. at 1801 (Gartner).
Third, we found persuasive Dr. Popovic's testimony that he
felt
comfortable assigning a value, where Ms. Read-Connole had not,
because
there were three wells in the slide with different dilutions of
rabbit
antiserum which gave him confidence in the 10% value he
assigned. Dr.
Popovic testified that --
I would like also just to remind that it says positive
for
rabbit antibody, also those rabbit antibody was tested on
three
wells. So one could evaluate 1 to 500, 1 to 1,000
dilutions
wells, also 1 to 2,000 dilution wells, while in the case
of
human, using human sera in that wells, it was only one well
[per
human sera].
So what gave me confidence that . . . the 10 percent is
correct,
because I went through evaluation of and quantified each
of
these wells and determine that 10 percent is for the data
point
1 to 2,000.
. . . .
usually there were no big discrepancy, but in this case I
sit
down and I evaluated myself since it was possible to come
to
some quantitative value in this particular case with
rabbit
antibodies, that's what I did.
Tr. at 2321-23; see also Tr. at 2440, 2442.
ORI asserted that Dr. Popovic's testimony on rereading the slide is
not
credible since the other datapoints in the table correspond to
Ms.
Read-Connole's notebook entries. This argument is not persuasive
since
only the data points Ms. Read-Connole in fact quantified in her
notes
are repeated in the tables exactly as recorded in the notebook.
Since
she did not actually quantify an IFA reading for clone H35 at 6
days,
the use in the table of a result quantified by Dr. Popovic would not
be
inconsistent with what Dr. Popovic otherwise did in constructing
the
table. Moreover, his testimony provides a rational basis for why
he
would assign a value for rabbit antibody, where there were
three
different wells, but would treat other non-quantified results as
not
determinable. 47/
It is important to note here that this is not a circumstance
where
recorded experimental data was altered and the altered data reported
as
the true value. There was no quantification initially recorded by
Ms.
Read- Connole and the quantification is consistent with what
Ms.
Read-Connole and Dr. Popovic say about the slide.
Thus, we conclude that ORI did not prove that it was inappropriate for
Dr.
Popovic to reread the slide and assign a value, notwithstanding the
fact that
his technician had not done so.
3. Dr. Popovic's testimony that the 10% value
was based on a
reasonable judgment at the time is credible.
ORI did not show by a preponderance of the evidence that the 10% is
false
or inaccurate. ORI relied solely on Ms. Read-Connole's notebook
and its
witnesses' interpretation of her prior statements for its
assertion that the
10% was false. ORI presented no other evidence
either that an accurate
reading of the slide would have given a
different value or that the 10% was a
scientifically impossible or
unlikely result. Dr. Popovic, on the other
hand, presented evidence
that the 10% is not an impossible or unlikely
result, given results of
other experiments performed at or around the same
time.
While we would agree with ORI that an IFA quantification could
not
properly be derived from other available data, ORI's suggestion that
Dr.
Popovic's testimony indicates he fabricated the 10% based on other
data
misconstrues Dr. Popovic's point when discussing other data. As
Dr.
Popovic asserted, other data that is available is consistent with
an
actual quantification at 10% and therefore tends to support
his
assertion that he read the slide and appropriately assigned the
10%
value. The 10% value is consistent with the other data for clone
H35.
Dr. Popovic testified that clone H35 was clearly positive at 6 days
in
light of the presence of giant multinucleated cells and the
relatively
high RT value. Tr. at 2323. Also, he testified that,
since the 14-day
data was "very clear cut and conclusive," there could not
have been very
low or virus negative cell population at the 6-day
point. Tr. at
2323-24. Moreover, Ms. Read-Connole testified that
the 10% value was in
the range of the values obtained with other clones in
later experiments.
Tr. at 2173-78; Ex. H-24 at April 30, 1984 et seq.
While we do not
think that values for other clones are particularly
persuasive, no
evidence was presented by ORI which even suggests that 10% was
an
inaccurate, impossible, or unlikely value for clone H35.
4. ORI did not establish that Dr. Popovic had a
motive to falsify
the 10%.
Another reason why we find credible Dr. Popovic's testimony that the
10%
was based on his rereading of the slide is that ORI did not
establish
any motive for Dr. Popovic to fabricate this data point.
While ORI
discounted Dr. Popovic's claim that the 6-day data had
little
significance, this is just what the evidence here shows. There
is
persuasive evidence that the 6-day data could have been omitted from
the
Science paper altogether. Tr. at 2228-30, 2251-55
(Malkovsky). The
data presented for 14 days was "overwhelming" evidence
of viral activity
in the clones. Tr. at 1828 (Gartner); Tr. at 2228-29
(Malkovsky).
Dr. Popovic said that the 6-day data was reported to show the
progression
of infection of the clones. Tr. at 2472. ORI argued that
this
gave the 10% importance and that Dr. Popovic had a motive to
falsify the 10%
because it made the paper seem more rigorous than if Dr.
Popovic had reported
the result as indeterminable (as Dr. Schaffer and
Dr. Berns said it ought to
have been reported based on their
interpretation of Ms. Read-Connole's
notes). Dr. Malkovsky testified,
however, that this data was useful to
show the dynamics of the
infection, but unnecessary overall, and that it made
no difference in
interpreting the data reported on Table 1 whether there was
a value
reported for clone H35 at 6 days or whether it was reported
as
indeterminable. Tr. at 2229-30. Dr. Malkovsky also pointed out
that
the data for any one clone was not really important for the meaning
of
the table since there were eight clones reported. Tr. at 2230.
Indeed,
unrebutted testimony establishes not only that the 6-day data was
not
necessary to the conclusions of the paper, but that 6-day data
other
than the data for the rabbit antiserum indicated the presence of
virus.
Tr. at 2253-54 (Malkovsky); Tr. at 2171-72 (Read-Connole).
Thus, we conclude that ORI did not prove that Dr. Popovic had any
motive
to falsify this data point.
5. ORI relied on incorrect readings of Ms.
Read-Connole's
laboratory notes.
In asserting that the 10% entry was false and misrepresented
the
laboratory data, ORI erroneously concluded that Ms.
Read-Connole's
notation was "directly contradictory data." ORI
post-hearing br. at 89.
This conclusion was based on testimony, primarily by
Dr. Schaffer, which
was in turn based on her interpretation of Ms.
Read-Connole's notebook
in light of statements made by Ms. Read-Connole in a
letter sent to an
ORI investigator in 1992.
As mentioned above, Ms. Read-Connole's notebook entry for clone H35 at
6
days was --
very few cells positive for rabbit antibody
In her letter, Ms. Read-Connole explained that her notation was
"two
separate statements." One that the slide had "very few cells" was
"a
comment on the number of cells on the slide." She explained that
this
was consistent with the other information available for that slide
--
the "percentage of multi-nucleated giant cells" and the "cell count"
on
day six. Ms. Read-Connole also evaluated the slide explaining that
"the
cells that were on the slide were positive with the rabbit
poly-clonal
antisera" but that she "could not make an adequate estimation of
the
number of positive cells." Ex. H-63, Letter from Read-Connole to
Healy
dated May 13, 1992.
Dr. Schaffer testified that --
. . . So what this means is that if you wash off a whole
pile
of cells you got to guess that a lot of them that you washed
off
were positive and some negatives or whatever, but every one
of
the cells, or all of the cells that were left, were positive,
so
what this indicates is if you were looking at this slide
you
would say there were not many cells but all the cells that
were
there were positive.
100 percent of the cells? [question]
That was the implication that certainly the great majority
of
the cells that were on the slide were positive.
Tr. at 1552-53.
Dr. Schaffer testified that the 10% quantification "has got to be
a
figment of his [Dr. Popovic's] imagination" and that her guess was
that
the actual value was "closer to 100." Tr. at 1553. Dr.
Schaffer read
Ms. Read-Connole's notes, as clarified by Ms. Read-Connole's
1992 letter
(Exhibit H-63), as meaning that "whatever cells remained were
all
positive." Tr. at 1553.
Dr. Schaffer relied for her opinion on her own interpretation of
Ms.
Read-Connole's statements, without actually speaking with
Ms.
Read-Connole. We see no reasonable basis in the record for
interpreting
Ms. Read- Connole's statement as Dr. Schaffer does.
Indeed, such a
reading would be contrary to other recorded data about clone
H35. Ms.
Read-Connole's letter merely indicates that she could not
definitely
quantify a positive reading, but that there were positive
cells. At the
hearing, Ms. Read-Connole testified that while she could
not "definitely
determine the number of cells that were indeed positive," she
was
"noting that there were positive cells." Tr. at 2189. Her
testimony
shows that she did not intend to suggest in her 1992 letter that
all of
the remaining cells were positive. Tr. at 2168-69; 2189.
ORI would have us disregard Ms. Read-Connole's testimony at the hearing
on
the basis that she is biased in Dr. Popovic's favor since she
collaborated
with him. We see no reason why we should rely on Dr.
Schaffer's
interpretation instead. Dr. Schaffer's interpretation of
statements
made by Ms. Read-Connole in 1992 is inconsistent with the
full text of the
1992 letter. Moreover, we find Ms. Read-Connole's
testimony reliable, for
several reasons. First, it is consistent with
what her letter actually
said. Second, her testimony is consistent with
what the expert
retrovirogists said about the condition of the 6-day
slides. Third, and
most important, it is not inconsistent with her
contemporaneous notes.
The notes themselves are the best evidence of what Ms.
Read-Connole's
observation was in 1984 and those notes are ambiguous as shown
by ORI's
expert testimony. While Dr. Schaffer read them as meaning all
of the
cells were positive, Dr. Berns testified that --
[t]he way I read it when I saw it originally was that there
were
very few cells positive, which essentially means a
negative
result to me. . . . .
That's where I would have thought not determinable would
have
been an appropriate notation.
Tr. at 1050-51. Dr. Richards testified that the notebook
is
"ambivalent. It can be read either way." Tr. at 505.
Thus, we conclude that the 10% value is not contradictory to
Ms.
Read-Connole's notes.
6. No negative inference from lack of primary
data is appropriate
under the particular circumstances here.
The 10% value is not reflected in the notebooks produced during
the
investigation. Dr. Popovic indicated that he would have noted
his
reading of the slide and that such notes would have been discarded
once
the Science paper had been published. Tr. at 2457-59; Ex. H-157 at
21.
ORI asserted that we should draw a negative inference here due to the
lack
of primary data to support the 10% value. ORI post-hearing br. at
90;
ORI post-hearing reply at 31. No such negative inference is
properly
drawn, however. This case is clearly distinguishable from the
decision
in Proposed Debarment of Dr. C. David Bridges, DAB No. 1232
(1991), where the
Hearing Officer concluded that it was reasonable to
draw a negative inference
from the lack of adequate primary data for an
entire set of reported
experiments since there was other persuasive
evidence of plagiarism and since
the research was under a grant
requiring data retention. Here, there is
a great deal of primary data
available for review. To draw a negative
inference and find misconduct
because primary data is not available for one
insignificant data point
is unwarranted.
Other factors for why we consider such an inference unreasonable under
the
particular circumstances here are: (1) Dr. Popovic credibly
testified that
his prior practice as head of laboratories where he
worked had been to rely
on technicians for notetaking; (2) Dr. Popovic
and Ms. Read-Connole both
testified that they exchanged and maintained
notes on looseleaf paper so it
is unlikely that all notes would survive;
(3) Dr. Popovic was working with an
infectious virus under controlled
circumstances requiring that IFA slides be
read in the dark so that
notetaking was difficult; and (4) Dr. Popovic's
laboratory and office
were moved several times during this time period.
ORI did not assert
that there were definitive standards applicable here which
required
maintenance of all experimental data.
We also find it significant here that the 10% entry appears in the
first
handwritten draft of the Science paper, since this draft was
prepared
shortly after these slides were first read by Ms.
Read-Connole. She
read the slides on February 29 and March 1 of
1984. The handwritten
draft was prepared several weeks later. Ms.
Read-Connole testified that
she saw this table, and ORI presented no evidence
that would indicate
that she disagreed with the 10% at the time.
Thus, we decline to draw a negative inference from the lack of
primary
data here.
7. ORI's other arguments are not persuasive.
In support of its findings that the 10% entry was falsified, ORI relied
on
the "inconsistency of Dr. Popovic's numerous and
changing
explanations." ORI post-hearing reply at 37. ORI
discussed Dr.
Popovic's explanations made in written submissions earlier in
the
investigatory process that 10% was the equivalent of "very few
cells"
and that he had perhaps averaged his and Ms. Read-Connole's reading
to
develop the 10%. Ex. H-84 at 96; Ex. H-157 at 21.
At the hearing, Dr. Popovic explained that he had attempted to explain
the
basis for the 10% before he had fully recalled the meaning of
Ms.
Read-Connole's notes. Tr. at 2456-58, 2524-26. We do not find
Dr.
Popovic's explanations to be numerous or changing, or to call
his
credibility into question under the circumstances here. It
is
reasonable that for one insignificant data point of the paper it
would
be difficult to remember the precise circumstances after seven
years.
We note here that Ms. Read-Connole did not clarify the meaning of
her
notation until May of 1992. Ex. H-63. We find credible Dr.
Popovic's
statement that at first he did not remember what happened with
clone
H35, but that later he recalled that there had been difficulty
in
quantifying low IFA values with the first rabbit antisera and that
he
had quantified the 1:2000 dilution at 10%. The original
investigation
here focused on allegations considered to be far more serious,
so it is
understandable if Dr. Popovic did not focus his powers of recall on
this
issue until it became clear that it might become a basis for an
action
against him.
In sum, we conclude that ORI did not prove by a preponderance of
the
evidence that Dr. Popovic intentionally falsified the 10% value
for
clone H35, or that it was even untrue.
D. ORI'S OTHER ARGUMENTS
1. ORI's other arguments were not properly raised here.
During the course of the hearing and in post-hearing briefs, ORI
attempted
to raise other issues concerning the accuracy of the Science
paper. We
found introduction of these issues to be inappropriate and
untimely. In
effect, ORI was attempting to reopen allegations of
misconduct on which it
had not made findings of misconduct in the final
report which served as the
basis for this proceeding. While ORI
attempted to characterize these
matters as merely going to Dr. Popovic's
credibility, ORI had not even
identified these matters as part of the
pattern of conduct it said (in the
Offer of Proof submitted several
weeks prior to the hearing) it would show to
prove Dr. Popovic's intent.
Thus, Dr. Popovic did not have fair notice that
these issues would be
addressed, and we ruled that they were outside the
scope of this
hearing. We nonetheless discuss these issues briefly here
since ORI
continued to press them and since they were raised in a public
forum.
48/
2. ORI did not show that the paper lacked necessary detail.
ORI contended that the methodology in the paper was not detailed enough
to
be reproducible and that this lack of detail was intended to obscure
the
"pooling." First, we fail to see how the paper can be said to
obscure
the fact that Dr. Popovic pooled samples when the paragraph in
question
reveals that the infection of the parental cell line was done
with fluids
harvested from "patients with AIDS or pre-AIDS." Ex. H-5 at
498.
This clearly reveals that it was not an isolate from a single
patient which
grew on the parental cell line.
The criticism that the paragraph lacked essential detail on methodology
is
based in part on ORI's erroneous view that pooling should be
considered an
essential part of the methodology. As discussed above,
however, the
paper also gives instances of infection with single patient
isolates and
these infections are discussed in somewhat greater detail
in the legend to
Table 2. The question of reproducibility was raised
primarily in
testimony by Dr. Martin. For reasons discussed above, we
found him not
to be disinterested generally. His specific testimony on
this point was
not persuasive since he admitted that he had not tried to
reproduce the
experiment, and that he was "overlaying this by what I
know in 1993, in terms
of how these systems worked." Tr. at 1334.
Moreover, he limited his
discussion of the methodology of the paper to
the one paragraph on the
infection of the parental cell line (apparently
because, in response to an
inquiry from him, Dr. Gallo had drawn
attention solely to that
paragraph). Tr. at 1322-23, 1361. Dr. Martin
seemed unaware that
further details and references on infecting a cell
line through cell-free
transmission and cocultivation were given
elsewhere in the paper. Also,
he acknowledged that, while he was a
stickler for details, lack of detail is
often a problem in scientific
papers. Tr. at 1345-46, 1360, 1363.
Many of the witnesses who
testified here -- including ORI witnesses -- said
that Science magazine
in particular has space limitations that mean that
Science papers are
more apt to lack detail and be ambiguous than some other
journals. See,
e.g., Tr. at 792 (Sodroski); Tr. at 1642 (Schaffer).
ORI presented no evidence that any scientist who had tried to
reproduce
the experiment had failed, and the weight of the evidence in the
record
is that the methodology in the paper was sufficient. See, e.g.,
Tr. at
803 (Sodroski); Tr. at 2236 (Malkovsky).
Dr. Sodroski's testimony on reproducibility supported Dr.
Popovic's
position. Dr. Sodroski testified that one could never exactly
reproduce
this type of experiment since it is a "unique event . . . because
the
viruses that are present in that patient that particular time may not
be
the same viruses present in that patient in . . . two or three days"
but
that does not mean that attempting to establish a permanent cell line
as
was done here is "intrinsically unreproducible." Tr. at
777-79. He
viewed the methodology of the paper as adequate and said
that, if he had
a problem, he would have telephoned Dr. Popovic. Tr. at
797; see also
Tr. at 1642 (Schaffer). Dr. Martin admitted that he had
never contacted
Dr. Popovic to inquire about the methods. Tr. at
1360-61.
In sum, ORI did not prove by a preponderance of the evidence that
the
paper lacked sufficient methodological detail or obscured the
pooling
technique.
3. ORI did not prove that the pool did not in fact exist.
ORI also tried to suggest belatedly that the pool did not in fact
exist
(even though the ORI report discussed the pool as though it did
exist).
The existence of the pool is established in Dr. Popovic's
notebook,
however, which not only records the three infections with
multiple
culture fluids but also refers to pooling in the protocol for the
series
of experiments and to "pool" or "pooled" as a culture in the
laboratory
after the infection. Ex. H-19 at 16-17, 33, 34, 40, 44,
58. Although
ORI implied through questioning at the hearing that maybe
the notebooks
were not authentic, ORI produced absolutely no evidence that
would call
the authenticity into question. Yet, ORI has had possession
of the
notebooks for years and the opportunity to subject them to
forensic
tests to date them. Moreover, the notebooks have various
indicia of
authenticity, including the interspersing of notes from Dr.
Popovic and
his technician Ms. Read-Connole and information obtained from
others.
We also note that Dr. Popovic's descriptions of how he infected
the
parental cell line have been consistent, even when these
descriptions
were given before the investigation. See, e.g., Ex. H-42
(1985
memorandum to Dr. Gallo). Ms. Read-Connole testified that Dr.
Popovic
had informed her of the pool's existence in early 1984. Tr. at
2201
(Read-Connole). 49/
ORI even went so far as to argue at the hearing, however, that the
Roche
analysis established "conclusively" that there was no pool -- a
position
which would reopen the allegation of misappropriation of the
French
isolate. This was not only extremely prejudicial to Dr. Popovic
but a
misstatement of the results of the Roche analysis. The Roche
analysis
did not conclude that there was misappropriation of the French
isolate;
the Roche analysis concluded that the most likely explanation for
the
striking similarity between the strain called HTLV-IIIB (as grown
on
clone H9) and one of the French isolates (different from the one
they
said they were providing to the LTCB) was contamination, both in
the
French laboratory and at the LTCB, by a variant that was
particularly
virulent. ORI would apparently have us infer that Dr.
Popovic
intentionally contaminated the HT cell line with the French isolate
and
therefore would have us disregard his testimony on the pool.
Even if this were properly an issue before us, we would not draw such
an
inference based on what ORI presented to us, for the following
reasons:
The experts on whom ORI relied for its report did
not
draw such an inference.
Dr. Popovic did not have a fair opportunity to
present
evidence on whether such an inference is reasonable.
The published article describing the Roche
analysis
contains qualifications and assumptions that have
not
been adequately addressed in this proceeding. See
Ex.
H-79.
The relevant expert testimony ORI did present calls
into
question the reasonableness of such an inference.
Dr.
Schaffer testified that the problem with pooling
was
that what could come out of a pool might be
a
recombinant different from any virus that went into
the
pool. Tr. at 1626. Dr. Sodroski's testimony
implies
that some samples may not be amenable to PCR
analysis.
Tr. at 800. He also testified that one can
initiate an
infection with only one infectious particle if the
right
target cells are used, and that the viruses present in
a
sample taken at one time might not be the same as
the
viruses in a sample from the same patient taken two
or
three days later. Tr. at 777-79, 782-83. Together
with
the acknowledged fact that one of the patients
whose
sample did not yield any HIV virus in the Roche
analysis
seroconverted between June 1984 and June 1985,
this
testimony renders suspect any conclusions based on
PCR
analysis of aliquots of patient samples which were
not
the exact ones used in the pool even if they are
from
the same patients.
Evidence regarding circumstances in the LTCB
and
contaminations elsewhere makes accidental
contamination
at least as likely a possibility as
deliberate
contamination. See, e.g., Ex. H-79; Tr. at
1089
(Berns).
ORI did not establish any motive for Dr. Popovic
to
deliberately substitute the French isolate for
the
results of his pooling experiment, and we see
no
apparent reason why he would have done so, given
the
number of other isolates which he successfully grew
on
T-cell lines.
.4. ORI did not prove that the paper inaccurately
referred to
patients with AIDS or pre-AIDS.
Finally, ORI suggested that the paper inaccurately referred to
the
concentrated culture fluids as being from "patients with AIDS
or
pre-AIDS"; ORI argued that some of the individual patient samples
were
from patients who were identified in the LTCB's log of patient
samples
solely as "hemophiliacs" or "homosexuals" and who therefore should
not
be considered as "pre-AIDS." This argument is based primarily on
Dr.
Martin's testimony that a patient has "pre-AIDS" only if the patient
has
certain clinical symptoms and does not include "at risk" groups.
See
Tr. at 1327. In our view, this testimony attempts to impose a
more
current definition of pre-AIDS than the definition in the paper
itself.
The abstract of the paper defines pre-AIDS as "signs or symptoms
that
frequently precede AIDS . . . ." Ex. H-5 at 497. Dr.
Sodroski and
others testified that, at the time the paper was written, the
term
"pre-AIDS" could reasonably have included persons because of
their
status, rather than specific clinical symptoms. Tr. at
620-22
(Sodroski); 846-47 (Hadley). Moreover, although the sample log
on which
ORI relied contains only cursory descriptions of the patients,
the
notebooks as a whole indicate that samples from a variety of
patients
were being provided to the LTCB either because of their diagnosis
as
having AIDS, their clinical symptoms, or their relationship with
a
patient who had AIDS. Thus, we do not find use of the term
"pre-AIDS"
to be inaccurate.
In sum, even if we considered these issues to be properly raised
here,
what evidence ORI did present and rely on for its arguments would
not
make a difference in our decision.
CONCLUSION
For the reasons explained above, we conclude that ORI did not prove by
a
preponderance of the evidence that Dr. Popovic engaged in
scientific
misconduct by intentionally falsifying certain methods or data
reported
in the .Popovic Science paper, or even prove that the methods and
data
at issue were untrue. Consequently, we conclude that ORI's
findings are
not supported and the proposed administrative actions are not
justified.
Judith
A.
Ballard
Norval
D.
(John)
Settle
Cecilia
Sparks
Ford
Presiding
Panel
Member.
APPENDIX A
Exhibit H-5
Popovic, M., Sarngadharan, M.G., Read, E. and Gallo, R.C.,
"Detection,
Isolation, and Continuous Production of Cytopathic
Retroviruses
(HTLV-III) from Patients with AIDS and Pre-AIDS," Science
Vol. 224:
497-500 (4 May 1984).*
* Reprinted with permission of Science.
APPENDIX B
Pages 29, 30, and 31 from laboratory notebook #3 of
Elizabeth
Read-Connole reproduced from Exhibit H-22..
APPENDIX C
Background information concerning the witnesses appearingfor ORI and
Dr.
Popovic..
Appendix C
ORI's Witness
Ken Berns
Professor and chairman of microbiology,
and
professor of pediatrics, at Cornell
Medical
College. Previously professor and chairman
of
immunology and medical microbiology,
and
professor of pediatrics, at the University
of
Florida College of Medicine in
Gainesville.
M.D. and Ph.D. degrees from Johns
Hopkins
University. Peer reviewer on the virology
study
section of the Division of Research Grants
for
four years; on American Cancer Society
virology,
microbiology panel for four years; on
National
Science Foundation genetic biology panel
for
four years.
Accepted as an expert in the field of virology and
with
expertise in proposing, conducting and reporting research.
(Tr.
at 1041.) CV at Ex. H-94.
Robert Goldberger
Adjunct professor at N.Y.U. Rusk Institute
for
Rehabilitation Medicine and
conducting
self-employed research (for last two
years);
past Provost at Columbia University for eight
or
nine years. Director of intramural research
at
NIH 1978-1981; twenty years at the NIH
as
scientist or science administrator,
including
acting as chief of the laboratory
of
biochemistry at the National Cancer
Institute
for 6 years, supervising about 100
scientists;
and at the Arthritis Institute. Educated
at
Harvard University (undergraduate major
in
biochemistry), New York University
(medical
school); Mount Sinai Hospital
(internship;
postdoctoral fellowship in biochemistry).
Recognized as an expert in the standards of conduct that
apply
to NIH scientists during the time period he was employed
there.
(Tr. at 292.) CV at Ex. H-33.
Suzanne Hadley
Ph.D. in Experimental Social and
Clinical
Psychology, U.Mass.; M.A. in
Experimental
Psychology, Hollins College; B.S.,
Middle
Tennessee State University. Performed
research
as graduate student, one year
post-doctoral
fellowship at NIMH, subsequently employed
at
NIMH as Research Psychologist starting in
1978,
then as science administrator from 1983
through
1989. Starting in 1987, Misconduct
Policy
officer.
Not accepted as an expert witness. (Tr. at 539.) CV at
Ex.
H-1.
Edward Huth
Part-time editor for electronic journal of
the
American Association for the Advancement
of
Science, entitled On-Line Journal of
Current
Clinical Trials. Undergraduate degree
from
Wesleyan; M.D., internship and residency
in
internal medicine at University of
Pennsylvania.
Eight years of research in kidney function
and
metabolic diseases; five years at
Women's
Medical College; part-time editing for Annals
of
Internal Medicine of the American College
of
Physicians, becoming full-time in 1965;
chief
editor 1971 until retirement in
1990.
Associated with National Library of Medicine
and
NIH. Chaired committee that produced
style
manual for Cancer Biology editors;
founding
member of International Committee of
Medical
Journal Editors (intended to develop
consensus
for scientific paper format),
including
involvement in drafting statement of
authorship
criteria; author of background paper
on
scientific authorship for Institute of Medicine.
Accepted as an expert in the publication of scientific
papers
and upon the knowledge of the standards in the
scientific
community with respect to the editing and publication
of
scientific papers including the years 1983 and 1984. (Tr.
at
1182-83.) CV at Ex. H-99.
Malcolm Martin
Chief of laboratory of molecular microbiology
at
NIAID. M.D., 1962, from Yale. Following
two
years of clinical training at University
of
Rochester, joined NIH staff.
Investigated
cancer-causing viruses at NIH
1968-1978.
Changed focus to retrovirology starting
1975.
Involved with AIDS research starting
1983;
consulted with World Health Organization
during
past year as virus expert.
Evaluates
microbiological employees for genetics board
of
Howard Hughes Medical Institute; member of
board
for Lucille Markey Foundation, which
sponsors
biomedical research. Presented NIAID
research
to Ninth International Congress of AIDS.
Accepted as an expert in virology and retrovirology, and in
the
standards for publication of papers that were followed at
the
NIH during his employment there. (Tr. at 1317, 1319.)
CV at
Ex. H-95.
William Raub
Currently science advisor to the
Administrator
of the Environmental Protection
Agency.
November 1991 to November 1992,
special
assistant for health affairs in the Office
of
Science and Technology Policy as part of
the
Executive Office of the President. 1986
to
1991, Deputy Director, NIH; 1989 to 1991
Acting
Director NIH. 1984 to 1986, NIH Deputy
Director
for Intramural Research; 1979 to 1984,
Associate
Director for Extramural Research and
Training.
From 1979 through 1986, performed basically
same
function (supervising grant policy
including
oversight of peer review system).
Accepted as an expert in the field of accepted standards
of
conduct in the scientific community, with testimony to
be
weighted according to circumstances of job experience
(i.e.,
focus on health science administration rather than
primary
research direction). (Tr. at 338; see 332 for Dr.
Popovic's
objection.) CV at Ex. H-146.
Frederick Richards
Emeritus Professor and Senior Research
Scientist
at Yale University. 1948, B.S. in
Chemistry,
MIT; 1952, Ph.D. in Biochemistry,
Harvard.
After post-doctoral study 1952-1955, joined
Yale
faculty. 1956 through 1971, Director,
Jane
Coffin Childs Memorial Fund for
Medical
Research. Member and past
president:
Biophysical Society and American Society
for
Biochemistry and Molecular Biology;
member
American Philosophical Society, American
Academy
of Arts and Sciences, National Academy
of
Sciences. Published about 150
scientific
papers; has reviewed papers as editor of
various
scientific journals.
Accepted as an expert in the general scientific methods
commonly
accepted in the scientific community. (Tr. at
410.) CV at Ex.
H-64.
Priscilla Schaffer
Chief of division of molecular genetics
and
chief of the laboratory of tumor virus
genetics
at the Dana Farber Cancer Institute of
Harvard
Medical School; herpes simplex virus
geneticist
and molecular biologist. Full professor
at
Harvard Medical School since 1981.
Graduate
studies in virology at Cornell
1964-1969;
post-doctorate fellowship in molecular
virology
at Baylor 1969-1971. Published more than
100
papers in scientific journals, two on
relation
between herpes simplex virus and AIDS.
Peer
reviewer of about 5 articles per month;
member
of editorial board of Virology and Journal
of
Virology.
Accepted as an expert in the field of virology as well as
the
general standards in the scientific community in 1983 and
1984.
(Tr. at 1475.) CV at Ex. H-88.
Joseph Sodroski
1981 to present, researcher at Dana
Farber
Cancer Institute; 1990 to present,
associate
professor of pathology, Harvard Medical
School.
Undergraduate degree from Allentown College
of
St. Francis de Sales; M.D. in 1980
from
Jefferson Medical College,
Philadelphia;
internship in internal medicine at New
England
Deaconess Hospital, followed by
post-doctoral
fellowship at Dana Farber Cancer
Institute.
Published more than 90 publications mostly
on
retrovirology or cancer-causing
viruses;
contributions to 17 books and
monographs;
reviews approximately 10 papers per month.
Accepted as an expert in the general field of
retrovirology.
(Tr. at 610.) CV at Ex. H-98.
Clyde Watkins
Acting director of Division of
Research
Investigations at Office of Research
Integrity.
Previously deputy director of the Office
of
Scientific Integrity, for 14 months;
before
that, a senior scientist at OSI. Also
faculty
member at Hershey Medical Center for
several
years and from 1982 to 1987 on the faculty
of
the Medical College of Georgia. Bachelor's
in
biology and Ph.D. in physiology
from
Pennsylvania State University College
of
Medicine.
Not offered or accepted as an expert. CV not referenced
in
transcript.
Thomas White
Vice President of Research and Development
for
Roche Molecular Systems, previously
Senior
Director of Research and Development.
Performed
laboratory analysis under contract with OSI.
Accepted as an expert in molecular evolution
(not
retrovirology). (Tr. at 549.) CV at Ex. H-76.
Patricia Woolf
Currently lectures at Princeton
Molecular
Biology Department on responsible conduct
of
research in molecular biology; serves
on
corporate boards; and performs
independent
scholarly research on subject of fraud
in
science or misconduct in research. Ph.D.
from
Johns Hopkins; B.S. in chemistry from
Purdue
University. Past member of Council of
Biology
Editors; member of International Committee
of
Medical Journal Editors. Past board member
of
Hustings Center for studies of ethics
in
research and past member of National Academy
of
Sciences panel on responsible conduct
of
research; member of award panel for Committee
of
Scientific Freedom and Responsibility of
the
American Association for the Advancement
of
Science.
Qualified with expertise in the field of standards and
behavior
in scientific communication. (Tr. at 1124.) CV at
Ex. H-116.
Dr. Popovic's Witness
William Blattner
B.A. and M.D. from Washington University in
St.
Louis. Diplomate of the American Board
of
Internal Medicine with
subspecialty
certification in clinical oncology.
Internship
and residency at Strong Memorial
Hospital,
University of Rochester; senior
assistant
resident at Cornell New York Hospital
and
Memorial-Sloan Kettering Cancer Institute;
two
year fellowship in epidemiology, National
Cancer
Institute; two years clinical oncology
training.
Section head of National Cancer
Institute
epidemiology program from 1981; chief of
viral
epidemiology since 1987. Co-Editor in Chief
of
Journal of Acquired Immunodeficiency Syndromes.
Accepted as an expert in the field of viral epidemiology.
(Tr.
at 1963.) CV at Ex. P-57.
Robert Gallo
Chief of Laboratory of Tumor Cell Biology
and
head of Section of Cellular Control
Mechanisms,
National Cancer Institute, National
Institutes
of Health. M.D. from Thomas
Jefferson
University Medical School; clinical clerkship
in
metabolism at Yale University; internship
and
residency in medicine at University of
Chicago;
NIH from 1965, as clinical associate,
senior
investigator, section chief and branch chief.
Not offered or accepted as an expert. CV not referenced
in
transcript.
Suzanne Gartner
Retrovirologist at Henry M. Jackson
Foundation,
a private foundation administered through
the
Uniform Services University. Ph.D. in
cancer
biology from Stanford University School
of
Medicine in 1984. 1971-1974,
research
technician in human leprosy and murine
leukemia
retroviruses at Stanford Research
Institute.
1975-1984, research technician and
graduate
student at Stanford Medical School,
researching
novel human retroviruses in Hodgkins
and
non-Hodgkins lymphomas. 1984-1989, HIV
research
in Robert Gallo's laboratory. Published
38
scientific papers.
Accepted as an expert in retrovirology and in the tests used
for
identification of HIV infections. (Tr. at 1784.) CV at
Ex.
P-58.
Ruth Kulstad
Managing editor of Clinical Chemistry,
journal
of the American Association of
Clinical
Chemistry. Editor of scientific journals
since
early 1960s; founding editor of New
Biologist;
Senior Editor of Science starting in
1981.
Honors degree in zoology from King's
College,
University of Durham in England.
Accepted as an expert in the publication of scientific
papers.
(Tr. at 1706.) CV at Ex. P-59.
Miraslov Malkovsky
Associate Professor in Department of
Medical
Microbiology and Immunology at the
medical
school of the University of Wisconsin in
Madison
since 1989. M.D., Ph.D., Member of the
Royal
College of Pathologists. Previously,
senior
scientist and head of Virology for
British
research company, one year; eight years
at
Medical Research Council in England;
faculty
member at Charles University, Prague.
Accepted as an expert in immunology and microbiology
and
retrovirology. (Tr. at 2226.) CV at Ex. P-60 and
P-60A.
Mikulas Popovic
M.D. in 1965 (Czechoslovakia); Ph.D.
from
National Cancer Institute, Slovak Academy
of
Sciences in Bratislava, 1971;
International
Cancer Fellowship from Swedish Cancer
Society,
1973; post-doctoral training in Cell
Biology,
University of Uppsala. Later Senior
Scientist
and Chief of Laboratory of Viral Oncogenesis
at
Cancer Research Institute, Bratislava;
1978,
International Cancer Fellowship under
World
Health Organization. Guest worker in
Dr.
Gallo's laboratory, 1980-1989. 1989-1990,
New
Primate Research Institute at New Mexico
State
University.
Not offered or accepted as an expert. CV at Ex. P-1.
Elizabeth Read-Connole
B.A., Mary Baldwin College, 1974; M.S.
in
microbiology, University of Maryland,
1992.
1978-1980, temporary position in
NIAID;
1908-1982, Division of Cancer
Treatment,
National Cancer Institute; 1982 to
present,
biologist in Laboratory of Tumor Cell
Biology,
National Cancer Institute. Assisted
Popovic
starting September or October of
1983;
officially assigned in February 1984.
Not offered or accepted as an expert. CV not referenced
in
transcript.
Howard Schachman
Professor Emeritus at University of
California,
Berkeley, Department of Molecular and
Cell
Biology; Chairman of Public Affairs
Committee
for American Society for Biochemistry
and
Molecular Biology. Bachelor's, master's,
Ph.D.
and two honorary degrees in biochemistry
and
chemistry. Published several hundred
papers,
wrote a book, served on editorial boards
of
Biochemistry, Journal of Biological
Chemistry,
Analytical Biochemistry, Annual Reviews
of
Biochemistry, Journal of Virology;
founding
editor, Biochemical and Biophysical
Research
Communications. Member of National Academy
of
Sciences (NAS) since 1968; past president
of
American Society for Biochemistry and
Molecular
Biology, and of Federation of American
Societies
for Experimental Biology (FASEB).
Participated
in drafting scientific fraud procedures
of
Association of American Universities; served
on
scientific responsibility panel of NAS.
Accepted as an expert in the field of accepted practices
for
conducting and reporting biological research. (Tr. at
1265.)
CV at Ex. P-62.
Jan Svoboda
Currently employed at Institute of
Electrogenics
in Prague; on immunologic panel of World
Health
Organization; member of Czech Academy
of
Sciences. Undergraduate degree for studies
in
natural sciences from Charles
University,
Prague, 1957; Ph.D. from Institute
of
Experimental Biogenetics for work in DNA and
RSV
transformation of mammalian cells;
1966,
doctorate in science for work in
oncogenic
effects of tumor viruses in
heterogenous
species. Has supervised graduate
students.
Founding editor of Somatic Cell
Genetics,
General Virology.
Accepted as an expert in the field of retrovirology. (Tr.
at
1945.) CV at Ex. P-63 and P-63A.
1. In this decision, we use the acronym ORI to refer not only to
the
current Office of Research Integrity, but also to other
investigative
offices which worked on this matter and which the Office of
Research
Integrity replaced, including the Office of Scientific
Integrity.
2. Although the guidelines permit appointment of a scientist to
the
panel, neither party requested that a scientist be appointed to
the
panel in this case.
3. See Appendix C for a brief description of each
witnesses'
background, the area of expertise (if any) in which they were
qualified
at the hearing, and the party for whom they testified.
4. Early in these proceedings, Dr. Popovic filed a motion to
dismiss
raising substantial questions concerning the fairness of applying
the
1989 definition, which established requirements for extramural
research,
to intramural research conducted in 1983-84. We ruled that
ORI had to
prove both that any reasonable researcher in Dr. Popovic's
position
would have considered Dr. Popovic's conduct to be scientific
misconduct
at the time and that the 1989 definition was met. ORI
presented
extensive testimony at the hearing in an effort to prove what would
have
constituted scientific misconduct in 1983-84. Most of this
evidence was
conclusory and merely went to undisputed general propositions,
such as
that intentional falsification of scientific data and methods has
never
been acceptable.
5. Dr. Richards, one of ORI's scientific advisors, testified
that
"honest error" is not a basis for sanction -- "honest meaning that
at
the time the manuscript was submitted, it was thought to be
correct."
Panel Hearing Transcript (Tr.) at 467; see also Tr. at 1127
(Woolf); Tr.
at 1064 (Berns); Tr. at 1247 (Huth).
6. The factual framework here is derived from stipulations
entered
into by the parties and from testimony of Drs. Blattner (Tr. at
1964-64,
1967-69, 1977, 1986), Gartner (Tr. at 1797-98, 1806-07,
1817-18,
1836-39, 1890), Popovic (Tr. at 2258, 2260-63), Sodroski (Tr. at
615-16,
646-50, 684-85, 687, 700-01, 733-37, 782), Svoboda (Tr. at
1943-44,
1947-51), and Gallo (Tr. at 2018-21, 2087, 2090-91). Much of
this
testimony was addressed in proposed findings of fact supported
by
citations to the record submitted by Dr. Popovic to which ORI did
not
directly respond with any citations to the record contrary to
our
factual statements here. Therefore, we do not consider these
factual
matters in dispute.
7. A retrovirus is a virus whose genetic information is encoded
in
ribonucleic acid (RNA). A retrovirus reproduces by creating
a
deoxyribonucleic acid (DNA) copy of its RNA. HIV, the name by which
the
AIDS virus is now known, is a retrovirus. See Glossary of
scientific
terms agreed to by the parties at 3 and 5.
8. The study of retrovirology is a subspeciality of virology;
the
process by which a retrovirus replicates is different from the
usual
process of virus replication. Moreover, within the field of
virology
and the subspeciality of retrovirology, there is a
further
specialization among those working in these areas: some
scientists are
cell biologists who study the virus in the context of the
total living
cell and attempt to detect, isolate, and propagate individual
viruses or
retroviruses; others are molecular biologists who concentrate on
the
subcomponents of viruses such as DNA, RNA, and nucleic acid.
9. ORI seemed to ignore this point, as though the paper (together
with
the papers published at the same time) were claiming that AIDS
was
caused by HTLV-IIIB (a variant designation which is not even
mentioned
in the paper). The HTLV-IIIB variant, as grown on clone H9,
became the
basis for the AIDS blood test, and is the variant which has a
striking
molecular similarity with a contaminant of the LAV variant
French
researchers had reported on and allegedly sent to the LTCB.
Unrebutted
testimony here establishes that Dr. Popovic had advocated using
an
individual patient isolate, and that the decision to use the
HTLV-IIIB
variant was made by others on the basis that its growth was
more
advanced.
10. Reverse transcriptase (RT) is an enzyme protein found
in
retroviruses. Glossary at 5. Retroviruses use reverse
transcription to
convert their genetic material (RNA) into DNA in the process
of
producing more virus. Tr. at 642-43 (Sodroski); see also Tr. at
1799
(Gartner). The LTCB had developed an assay to detect RT. A
positive RT
is a good indication that a retrovirus is present, but a negative
RT
does not mean that a retrovirus is not present; virus expression
might
simply be below the level of detection. Tr. at 1800-07
(Gartner).
11. ORI's report accepted this scenario as true. At the
hearing,
however, ORI raised allegations for the first time concerning
whether
the pool existed and whether the notebooks were authentic. We
agree
with Dr. Popovic that ORI's timing and manner of raising
these
allegations were unfair since it did not allow adequate opportunity
for
Dr. Popovic to prepare a response. We have nonetheless discussed
these
allegations below since the record as developed -- including
testimony
by ORI's own witnesses -- calls these allegations into question
and
since they might otherwise unfairly harm Dr. Popovic's
reputation,
having been raised in a public forum.
12. The evidence on the RT tests was from the notebooks of Dr.
Prem
Sarin at the LTCB. The evidence on whether the samples contained
AIDS
virus was from polymerase chain reaction (PCR) analysis on aliquots
of
patient samples provided by ORI to Roche Diagnostics as part of the
ORI
investigation.
13. While the word "first" may suggest a priority in time, this
is not
its only meaning. One part of the definition of the word "first"
is
"before any or some other person or thing (as in time, space, rank,
or
importance): as the first thing to be mentioned . . . . "
Webster's
Third New International Dictionary at 856.
14. As discussed below, ORI relied on testimony that it would
be
"logical" to test for RT activity before pooling samples and using
them
to infect the pool (because positive RT would indicate presence of
the
virus). Dr. Popovic presented persuasive evidence, however, that
given
his training and practices, he would not have viewed what he in fact
did
as illogical, even if it was not optimal. In any event, even if
it
would have been more logical to test for RT first, ORI presented
no
evidence that reporting that the tests were performed on part of each
of
the three concentrated culture fluids after another part of each
was
used to infect the cell line would have undercut the conclusions of
the
article.
15. One of ORI's experts, Dr. Martin, testified that "the
first
sentence basically says that by the criteria and tests that
were
available . . . the recipient cell was clean, that is to say it
didn't
contain any other virus." Tr. at 1324. He explained that
this was
important "because if the recipient cell contains . . . a virus,
then
what would be . . . isolated at the end of the procedure, would be
this
so-called contaminant virus and not the real cause of the . .
.
disease." Tr. at 1325.
16. Dr. Popovic testified that some of these papers were written
in
his native language and translated and that some were written in
English
and heavily edited by others. Tr. at 2287-89. Others who
knew him in
the relevant time period testified about his difficulties with
the
language, and our observation of him over the two and a half weeks
of
hearing confirmed that he has continuing difficulties with syntax
and
the nuances of the English language, even though his basic English
is
good.
17. Dr. Sodroski testified that it is "not uncommon that there
are
ambiguities in methodologies recorded in journals like Science that
are
very pressed for space," so he did not think the ambiguity in the
term
"concentrated culture fluids" was particularly troublesome. Tr.
at
792-93.
18. The notebooks use the name "Sarang" rather than
"Sarngadharan",
but both parties agreed that these names referred to the same
person.
While the original protocol for the experiments, dated September
10,
1983, called for RT testing before infection, Dr. Popovic testified
that
he started sending tests to Dr. Sarngadharan in October or
November
1983, and the notebooks support this claim, as does Dr.
Sarngadharan's
testimony during the investigation. Tr. at 2495; Ex.
H-19; Ex. H-50 at
15-19.
19. Dr. Schaffer was apparently under the erroneous assumption
that
this was an undated entry. Tr. at 1501.
20. In addition to the other factors we discuss below, Dr.
Popovic's
credibility on this point is also enhanced by testimony by
Dr.
Sarngadharan, who was involved in editing this paper. He testified
that
"first shown" meant that the concentrated culture fluids were
shown
positive before the infection. He interpreted this, however,
as
consistent with what actually occurred since, even if the result
is
found later than the infection, "what you find out is on a result of
a
sample before going on to the cell . . . ." Thus, he said,
"That
positive first only means that the sample was positive prior to going
on
to these cells." Ex. H-52 at 33-38.
21. Alternatively, some changes could have been made directly
by
whoever typed the drafts.
22. ORI argued that we should find Dr. Popovic responsible for
any
inaccuracy in the disputed sentence because it was repeated in a
patent
application filed around the time the paper was published and
because
Dr. Popovic, by signing the application, had certified that it was
not
false. Unrebutted testimony, however, shows that Dr. Popovic did
not
actually review the application before signing it but relied on
others,
including lawyers for NIH, to accurately reflect the paper in
the
application. Tr. at 2325-27; 2487-90. Since ORI did not prove
that Dr.
Popovic was aware of the inaccuracy in the paper, or that he
reviewed
the patent application, the fact that he certified the application
does
not help ORI's case here. Moreover, ORI's implication that he had
a
motive to falsify the application is unwarranted, since at that
time
researchers could not share in the patent proceeds.
23. Even Dr. Martin, who faulted the paper for lack of
detail,
testified that, when the paper came out, use of the process
of
cocultivation in isolating viruses had "long been appreciated." Tr.
at
1324.
24. Dr. Svoboda testified that pooling (the aspect of the
experiments
on which ORI focused) "is not a real technique. It's a
procedure used
in cases when you don't have enough virus." Tr. at
1949-50.
25. In its post-hearing brief, ORI misquotes this as stating that
one
"must show a finite amount of reverse transcriptase activity
(100,000
cpm) before concentration." ORI post-hearing br. at 71.
This appears
to be a particularly egregious example of ORI substituting
its
paraphrase for an exact quotation. ORI's substitution of the
word
"must" for the word "should" changes the meaning.
26. ORI also relied on Dr. Martin's testimony describing RT
testing as
a "go, no go test". Tr. at 1330, 1337-38. The first
time he used this
description, however, Dr. Martin simply said "I imagine"
that's "sort of
a go, no-go kind of test." Tr. at 1330. Dr.
Malkovsky testified that
his laboratory sometimes performed RT testing before
pooling and
sometimes did not. Tr. at 2240. We give more weight
to what is in fact
done in laboratories, than to what someone "imagines"
would be done.
27. See, e.g., Ex. H-5 at 499 (legend to Table 2, referring
to
experiments exposing cells of H9 clones to "concentrated culture
fluids
positive for particulate RT activity").
28. Dr. Malkovsky testified that his laboratory had pooled
samples,
sometimes without testing for RT first. He said that if
the
concentration of virus in some of the fluids were high, diluting
them
with non-virus containing fluids would not make any difference in
the
final effect. He also said that "it is very difficult by any
other
methods [with] the exception of co-culture, to decide that the virus
is
not present" and that measuring RT activity was sometimes
impossible,
even when a retrovirus is present. Tr. at 2239-42; see also
Tr. at
2244, 2248.
29. Immunofluorescence Assays (IFAs) can detect the presence of
a
specific viral protein in cell cultures or primary patient cells
by
using fluorescent dye molecules that link directly or indirectly
to
antibodies to that viral protein. Tr. at 1807-18 (Gartner).
Ms.
Read-Connole indicated that she performed IFA assays using
eight-well
slides, placing a concentration of cells in each well and then
adding
the antibody and fluorescence after certain preparatory steps.
IFA
slides are read under a fluorescent microscope to determine
what
proportion, if any, of the cells counted react. IFA slides have
a
positive reaction when the cells light up, or fluoresce, bright
yellow
green. Tr. at 2155-58 (Read Connole); Tr. at 1830 (Gartner).
30. We find some merit to Dr. Popovic's claim that,
as a non-native
English speaker, he would not be as familiar as a native with
subtle
nuances of the language and might be more apt to rely on
dictionary
definitions.
31. Several of ORI's record citations in support of
its proposition
are to unrelated testimony which does not even mention the
issue
presented. See ORI post-hearing br. at 74, citing Tr. at 610,
937, and
1845.
32. Some of his conclusions were apparently based on
discussions
with his colleagues, but those colleagues were not even
identified, and
we have no way of evaluating their qualifications and
credibility.
Moreover, Dr. Richards seemed to hold Dr. Popovic to his own
standard of
explaining in footnotes why there is no data entry for a
table. See,
e.g., Tr. at 520. Yet, it was undisputed that Science
is a journal with
strict space limitations, and the legend here is already
very detailed
and lengthy.
33. ORI cited to Dr. Hadley's testimony on page 937
of the
transcript. That page, however, does not actually refer to
statements
about the meaning of the term "not done."
34. When it was brought to his attention that
experiments were
reported as "not done" in the table but the body of the
paper described
the experiments as having been performed, Dr. Berns called
this a
mistake which he would characterize as an "honest error." Tr. at
1102.
Dr. Berns' testimony did not establish, however, that usage of the
term
in the paper was a mistake, since he admitted he simply could not
recall
whether or not the experiment was in fact performed. Tr. at
1102. ORI
tried to distinguish his paper from the one at issue here by
saying that
the reader would not be misled since the contradiction was
apparent on
the face of the paper. ORI post-hearing reply at 65.
We note, however,
that the legend to Table 1 in the Science paper describes
the
preparation of the IFA slides in such a way as to suggest that they
were
prepared for each of the clones at both 6 and 14 days. Thus, we do
not
see how the NDs in Table 1 are any more misleading than the NDs in
Dr.
Berns' paper. In fact, Dr. Richards testified, when asked about
the
legend to Table 1, that "in this kind of a table, where you have a
whole
series of cells, it is almost inconceivable that the protocol would
be
set up to omit certain of the samples, that doesn't make any
sense,
because that isn't the way those things are done." Tr. at
515-16.
35. We reviewed only those papers on which Dr.
Popovic was first
author because ORI did not establish that, for any of the
papers where
Dr. Popovic was not the first author, he would have actually
performed
the experiments where ND was used or that he was responsible
for
drafting the legend.
36. Dr. Popovic had initially pointed out that there
were several
papers in which he had used the term "ND, not done" as meaning
either
not determinable or that the experiment was not finished because
the
results were inconclusive, yet ORI here did not specifically rebut
those
statements here. Ex. H-48 at 46-47.
37. Unreported 6-day data for other patient serum
showed a positive
reaction for clones H9 and H31, and was recorded as a minus
for H17.
Dr. Popovic testified that in his opinion the two patient sera
were
"equivalent," which further indicated that the 6-day IFAs for
these
clones were "uninterpretable." Tr. at 2315.
38. Electron Microscopy is the use of a high powered
microscope to
look for evidence of virus expression in cultured material or
tissue.
Thin slices of the biological material are examined using an
electron
microscope to look for the presence of virus particles, particularly
the
budding of the virus particle, and other morphological evidence.
Tr.
1818-21 (Gartner).
39. Data were inadvertently reversed for two isolates
so that the ND
which Dr. Popovic intended to report for patient S.N. is in
the wrong
horizontal column in Table 2. There is no dispute that the ND
was
actually intended for the EM results for patient S.N. and that result
is
what ORI found was falsified.
40. ORI asserted that Dr. Popovic "ignores the fact
that the rabbit
serum used in the IFA experiments may contain impurities,
thereby
producing an anomalous result." ORI post-hearing br. at
77. ORI
provided no record citations for this proposition.
Moreover, in
response to ORI's questions about the rabbit antisera, Dr.
Gartner's
testimony was that presumably the experiment would have been done
in a
way so that the rabbit antiserum would give a better result than
the
patient serum. Tr. at 1833-36; 1867-79; 1897-1901. ORI
presented no
evidence here that the inoculation of the rabbit was not done
using
standard procedures, nor any evidence that Dr. Popovic knew there
was
any question about impurities in the rabbit serum (which the
record
indicates was prepared at another laboratory) which would
cause
anomalous results. There is no reason to conclude that any
difficulty
quantifying low positive results with this rabbit antisera meant
that
other results were anomalous.
41. Dr. Berns considered Respondent's explanation to
the ORI
scientific advisors that ND in his mind meant not determinable
as
opposed to not performed, but said the problem they had with that
"was
that it was clearly labeled with a result. . . ." Tr. at
1048. His
reasoning was that, if one considered indeterminable the ND
entries
corresponding to results reported as "-" in the notebook, one would
be
invalidating the results for the controls, which were also reported
as
"-". Tr. at 1048. We did not find this reasoning persuasive,
for
several reasons. First, ORI presented no evidence that reading
the
slides for the controls would have presented the same difficulties
as
reading the slides for the clones, where cell death was a
problem.
Second, there were no conflicting data for the controls, as there
were
for the other results Ms. Read-Connole recorded as "-". Third,
this
testimony did not take into account evidence that Dr. Popovic reread
the
slides, and that Ms. Read-Connole was cautious in quantifying
IFA
results.
42. In questioning its witnesses, ORI presumed that
the ND entries
meant only that no test had been performed (i.e., attempted)
and
therefore, that any indication from the notebook entries that a test
was
performed meant the use of ND in these tables was a falsification.
In
our opinion, this slant misfocused these experts' testimony.
43. To the extent that ORI's experts misread "-" in
Ms.
Read-Connole's notebook as meaning only that no virus was present,
this
supports Dr. Popovic's view that to report the "-" which appeared in
the
notebook would have been misleading to individuals not experienced
in
evaluating these tests for a cytopathic retrovirus.
44. In contrast, Dr. Gartner, an expert
retrovirologist, understood
from Ms. Read-Connole's notes what the controls
were; Dr. Gartner's
testimony was consistent with Dr. Popovic's concerning
the meaning of
the results recorded for the clones in question.
45. ORI relied on Dr. Gartner's testimony that "a
scientist
`wouldn't make a change in someone else's data'." ORI's
post-hearing
br. at 90, citing Tr. at 1881. ORI takes this statement
out of context.
Dr. Gartner testified that, if she independently changed a
technician's
reading of a slide, she would not make that change in the
technician's
notebook but would record it "in my own notebook." Tr. at
1881.
46. We note, however, that (in contrast to her later
testimony that
the slide was unreadable), Dr. Schaffer initially expressed
the opinion
that the notes indicated that the slide was probably not
"easily
interpretable." Tr. at 1546.
47. Ms. Read-Connole was unwilling to quantify IFA
tests where the
results were not clearcut. The "+" she used "designates
a weak positive
reaction (i.e., [less than] 10%)." Ex. H-32C at
3. Dr. Popovic stated
that the "+" for clone H9 against the E.T.
antisera had been based on
only one unequivocally positive cell. Ex.
H-157 at 20. Therefore, it
is not inconsistent with the recorded notes
for the IFA results that Dr.
Popovic could quantify the IFA results for clone
H35 against the rabbit
antisera. One would conclude from Ms.
Read-Connole's precision and
skill at reading IFA that if it were merely a
weak positive, she would
have noted that. Rather, she stated that the
slide was positive but did
not quantify it. In context, Dr. Popovic's
quantification is consistent
with the other IFA results recorded by Ms.
Read-Connole.
48. We have considered all of the evidence and
arguments presented
by the parties. If we do not specifically mention a
particular argument
or piece of evidence that means we consider it either
covered generally
in our detailed analysis or irrelevant to the issues we
need to decide
here.
49. ORI presented some evidence that pooling would
not be a good
technique because you might add inhibitors or viruses other
than the one
you are trying to isolate. Tr. at 480 (Richards); Tr. at
1506-07
(Schaffer). The mere fact that there might be some problems
with the
technique, however, does not mean that it was a technique Dr.
Popovic
would not have used, under the circumstances here. Dr.
Sodroski
testified that, while pooling was not optimal, "it appears that it
may
have been used here in a practical sense, to try to get some
viruses
that could grow on a cell line." Tr. at 746. This is
consistent with
the September 1983 protocol in Dr. Popovic's laboratory
notebook, which
says "If not sufficient volume of culture fluids; pool
together several
samples." Ex. H-19 at