Milward v. Acuity Specialty Products Group, Inc.

United States Court of Appeals
For the First Circuit

No. 09-2270

BRIAN K. MILWARD AND LINDA J. MILWARD,

Plaintiffs, Appellants,

v.

ACUITY SPECIALTY PRODUCTS GROUP, INC., ET AL.,

Defendants, Appellees.

APPEAL FROM THE UNITED STATES DISTRICT COURT
FOR THE DISTRICT OF MASSACHUSETTS

[Hon. George A. O'Toole, Jr., U.S. District Judge]

Before

Lynch, Chief Judge,
Lipez and Howard, Circuit Judges.

Steve Baughman Jensen, with whom Allen Stewart, P.C., James
Gotz, and Kreindler & Kreindler were on brief, for appellants.
Raphael Metzger, Gregory Coolidge, and Metzger Law Group on
brief for the Council for Education and Research on Toxins, et al.,
amici curiae.
Joseph J. Leghorn, with whom Nixon Peabody LLP was on brief,
for appellees.

March 22, 2011
 LYNCH, Chief Judge. Brian and Linda Milward brought

negligence claims against defendant chemical companies alleging

that the rare type of leukemia that Brian Milward suffers, Acute

Promyelocytic Leukemia (APL), was caused by his routine workplace

exposure to benzene-containing products that had been manufactured

or supplied by defendants. Milward worked as a refrigeration

technician and asserted that he was exposed to benzene from 1973

until the time he filed this complaint and jury demand in October

2007. He had been diagnosed with APL in October 2004.

At defendants' request, the district court bifurcated the

suit into two phases. The first phase concerned whether the expert

opinion offered by plaintiffs on "general causation" was admissible

under Federal Rule of Evidence 702. "'General causation' exists

when a substance is capable of causing a disease." Restatement

(Third) of Torts: Liability for Physical and Emotional Harm § 28

cmt. c(3) (2010) ("Restatement"). If plaintiffs' expert evidence

had been ruled admissible, the second phase would have considered

all other issues, including negligence, exposure, and the "specific

causation" of Milward's leukemia. "'Specific causation' exists

when exposure to an agent caused a particular plaintiff's disease."

Id. § 28 cmt. c(4).

This case never reached the second phase. The district

court ruled that the testimony of plaintiffs' expert on general

causation, Dr. Martyn Smith, was inadmissible under Federal Rule of

-2-
Evidence 702. The court so ruled after reviewing written

statements and materials and conducting a four-day evidentiary

hearing in which it heard testimony from plaintiffs' experts Dr.

Smith, a toxicologist, and Dr. Carl Cranor, an expert on scientific

methodology; and from defendants' experts Dr. David Garabrant, an

epidemiologist, Dr. David Pyatt, a toxicologist, and Dr. John

Bennett, a pathologist. The district court, in a detailed opinion,

ruled that "Dr. Smith's proffered testimony that exposure to

benzene can cause APL lacks sufficient demonstrated scientific

reliability to warrant its admission under Rule 702." Milward v.

Acuity Specialty Prods. Grp., Inc., 664 F. Supp. 2d 137, 140 (D.

Mass. 2009). The court entered final judgment for defendants and

plaintiffs timely appealed.

The appellate standard of review for Rule 702 rulings is

abuse of discretion. Gen. Elec. Co. v. Joiner, 522 U.S. 136, 146

(1997). "This standard is not monolithic: within it, embedded

findings of fact are reviewed for clear error, questions of law are

reviewed de novo, and judgment calls are subjected to classic

abuse-of-discretion review." Ungar v. Palestine Liberation Org.,

599 F.3d 79, 83 (1st Cir. 2010); see also Baker v. Dalkon Shield

Claimants Trust, 156 F.3d 248, 251-52 (1st Cir. 1998) (noting these

three dimensions of the abuse of discretion standard in reviewing

exclusion of expert testimony).

-3-
 We reverse the district court's exclusion of Dr. Smith's

general causation testimony. Cf. Ruiz-Troche v. Pepsi Cola of P.R.

Bottling Co., 161 F.3d 77 (1st Cir. 1998) (reversing exclusion of

expert testimony); Dalkon Shield, 156 F.3d 248 (same). Dr. Smith's

testimony is admissible. We stress that it is up to the jury to

decide whether to accept his opinion that exposure to benzene can

cause APL--a proposition that plaintiffs must prove by a

preponderance of the evidence.

I.

The Supreme Court in Daubert v. Merrell Dow

Pharmaceuticals, Inc., 509 U.S. 579 (1993), vested in trial judges

a gatekeeper function, requiring that they assess proffered expert

scientific testimony for reliability before admitting it.1 The

Court held that Rule 702 displaced the "general acceptance" test of

Frye v. United States, 293 F. 1013 (D.C. Cir. 1923), under which

"the admissibility of an expert opinion or technique turned on its

'general acceptance' vel non within the scientific community."

Ruiz-Troche, 161 F.3d at 80. Under Rule 702:

If scientific, technical, or other specialized
knowledge will assist the trier of fact to
understand the evidence or to determine a fact
in issue, a witness qualified as an expert by
knowledge, skill, experience, training, or
education, may testify thereto in the form of
an opinion or otherwise, if (1) the testimony

1
Kumho Tire Co. v. Carmichael, 526 U.S. 137 (1999),
clarified that courts have this function with respect to all expert
testimony, not just scientific.

-4-
 is based upon sufficient facts or data, (2)
the testimony is the product of reliable
principles and methods, and (3) the witness
has applied the principles and methods
reliably to the facts of the case.

Fed. R. Evid. 702.

The Daubert Court identified four factors that might

assist a trial court in determining the admissibility of an

expert's testimony: "(1) whether the theory or technique can be and

has been tested; (2) whether the technique has been subject to peer

review and publication; (3) the technique's known or potential rate

of error; and (4) the level of the theory or technique's acceptance

within the relevant discipline." United States v. Mooney, 315 F.3d

54, 62 (1st Cir. 2002) (citing Daubert, 509 U.S. at 593-94). These

factors "do not constitute a 'definitive checklist or test.'"

Kumho Tire Co. v. Carmichael, 526 U.S. 137, 150 (1999) (emphasis

omitted) (quoting Daubert, 509 U.S. at 593). Given that "there are

many different kinds of experts, and many different kinds of

expertise," these factors "may or may not be pertinent in assessing

reliability, depending on the nature of the issue, the expert's

particular expertise, and the subject of his testimony." Id.

Exactly what is involved in "reliability" was not and

could not have been filled out by Daubert. Rather, the answers

must come from developing case law in adjudicating individual

controversies. "[T]he question of admissibility 'must be tied to

the facts of a particular case.'" Beaudette v. Louisville Ladder,

-5-
Inc., 462 F.3d 22, 25-26 (1st Cir. 2006) (quoting Kumho Tire, 526

U.S. at 150).

Although Daubert stated that trial courts should focus

"on principles and methodology, not on the conclusions that they

generate," Daubert, 509 U.S. at 595, the Court subsequently

clarified that this focus "need not completely pretermit judicial

consideration of an expert's conclusions," Ruiz-Troche, 161 F.3d at

81 (citing Joiner, 522 U.S. at 146). In Joiner, the Court

explained that "conclusions and methodology are not entirely

distinct from one another" and "nothing in either Daubert or the

Federal Rules of Evidence requires a district court to admit

opinion evidence that is connected to existing data only by the

ipse dixit of the expert." Joiner, 522 U.S. at 146. Expert

testimony may be excluded if there is "too great an analytical gap

between the data and the opinion proffered." Id. "[T]rial judges

may evaluate the data offered to support an expert's bottom-line

opinions to determine if that data provides adequate support to

mark the expert's testimony as reliable." Ruiz-Troche, 161 F.3d at

81.

This does not mean that trial courts are empowered "to

determine which of several competing scientific theories has the

best provenance." Id. at 85. "Daubert does not require that a

party who proffers expert testimony carry the burden of proving to

the judge that the expert's assessment of the situation is

-6-
correct." Id. The proponent of the evidence must show only that

"the expert's conclusion has been arrived at in a scientifically

sound and methodologically reliable fashion." Id.; see also United

States v. Vargas, 471 F.3d 255, 265 (1st Cir. 2006). The object of

Daubert is "to make certain that an expert, whether basing

testimony on professional studies or personal experience, employs

in the courtroom the same level of intellectual rigor that

characterizes the practice of an expert in the relevant field."

Kumho Tire, 526 U.S. at 152.

So long as an expert's scientific testimony rests upon

"'good grounds,' based on what is known," Daubert, 509 U.S. at 590,

it should be tested by the adversarial process, rather than

excluded for fear that jurors will not be able to handle the

scientific complexities, id. at 596. "Vigorous cross-examination,

presentation of contrary evidence, and careful instruction on the

burden of proof are the traditional and appropriate means of

attacking shaky but admissible evidence." Id.; see also Currier v.

United Techs. Corp., 393 F.3d 246, 252 (1st Cir. 2004).

II.

It is uncontested that Dr. Smith's opinion about the

causal link between benzene and APL satisfies certain requirements

of Rule 702. His opinion would "assist the trier of fact to

understand the evidence or to determine a fact in issue." Fed. R.

Evid. 702. And Dr. Smith is "a witness qualified as an expert by

-7-
knowledge, skill, experience, training, or education." Id. He is

acknowledged as a leading expert on the study of the toxic effects

of chemicals and drugs on the human body, with particular emphasis

on the mechanisms by which benzene and its metabolites cause damage

to both cells and the human organism as a whole. The research in

Dr. Smith's laboratory, which is funded by the National Institutes

of Health, focuses on the causes of leukemia and lymphoma, and he

has authored or co-authored over 215 articles in peer-reviewed

journals in the field of toxicology.

The question before us is whether the district court

abused its discretion in concluding that the other requirements of

Rule 702, concerning the reliability of Dr. Smith's opinion, were

not met. We will first discuss some basic facts about leukemia,

the weight of the evidence methodology, and Dr. Smith's use of that

methodology, and we will then turn to an evaluation of the district

court's ruling.

Leukemia is a cancer of the blood cells. There are

different types of leukemia, which are generally classified in two

ways. The first classification is between leukemia's acute and

chronic forms: acute leukemia is characterized by a rapid increase

in the number of immature blood cells, while chronic leukemia is

characterized by the excessive buildup of relatively mature but

abnormal white blood cells. The second classification is between

the types of stem cells affected: leukemia can be either "myeloid"

-8-
or "lymphoid." Combining these two classifications provides a

total of four main categories of leukemia: acute myeloid leukemia

(AML); chronic myeloid leukemia (CML); acute lymphoid leukemia

(ALL); and chronic lymphoid leukemia (CLL).2 Within each of these

categories, there are typically several subcategories.

The general category of AML can be subdivided in more

than one way. Under the common French-American-British

classification system used by the parties, subtypes are classified

morphologically according to the degree of differentiation along

different cell lines and the extent of cell maturation. This

classification system identifies subtypes by convention as M0

through M7.3

Brian Milward's leukemia, APL, is subtype M3 and is an

extremely rare disease. APL accounts for only five to ten percent

of all cases of AML, which is itself rare, with an annual incidence

of 3.5 cases per 100,000 people. APL is characterized by a

deficiency of mature blood cells in the myeloid cell line and an

excess of immature cells called promyelocytes.

2
There are also some types of leukemia that are considered
to be outside of this four-part classification scheme.
3
The World Health Organization has adopted a different
classification system that utilizes not only morphological
characteristics, but also genetic, immunophenotypic, biologic, and
clinical characteristics to define specific disease entities that
have clinical and biological relevance. See generally James W.
Vardiman et al., The World Health Organization (WHO) Classification
of the Myeloid Neoplasms, 100 Blood 2292 (2002).

-9-
 APL is in part caused by the chromosomal translocation of

a gene known as the retinoic acid receptor-alpha gene (RARá) on

chromosome 17.4 Although APL and the other subtypes of AML have

been the subject of extensive research, there is not yet a

scientific consensus as to the causes of the genetic translocation

that induces APL.

Dr. Smith's opinion is that what is known about both AML

and APL supports the inference that exposure to benzene can cause

APL. He reached this opinion using a "weight of the evidence"

methodology in which he considered five lines of evidence drawn

from the peer-reviewed scientific literature on leukemia and

benzene. We first discuss the reliability of this methodology in

general, and then turn to Dr. Smith's application of it.

A. The Reliability of the Weight of the Evidence Methodology

Dr. Smith's opinion was based on a "weight of the

evidence" methodology in which he followed the guidelines

articulated by world-renowned epidemiologist Sir Arthur Bradford

Hill in his seminal methodological article on inferences of

causality. See Arthur Bradford Hill, The Environment and Disease:

Association or Causation?, 58 Proc. Royal Soc'y Med. 295 (1965).

4
In approximately 95% of cases of APL, RARá is involved in
a reciprocal translocation with the promyelocytic leukemia gene
(PML) on chromosome 15--a translocation denoted as
t(15;17)(q22;q12)--which creates a fusion gene known as PML-RARá.
In the remaining cases of APL, RARá translocates and fuses with one
of four other genes.

-10-
 Hill's article explains that one should not conclude that

an observed association between a disease and a feature of the

environment (e.g., a chemical) is causal without first considering

a variety of "viewpoints" on the issue. These viewpoints include:

the strength or frequency of the association; the consistency of

the association in varied circumstances; the specificity of the

association; the temporal relationship between the disease and the

posited cause; the dose response curve between them; the biological

plausibility of the causal explanation given existing scientific

knowledge; the coherence of the explanation with generally known

facts about the disease; the experimental data that relates to it;

and the existence of analogous causal relationships. See id. at

295-99.5

Although Hill identified nine viewpoints, it is generally

agreed that this list is not exhaustive and that no one type of

evidence must be present before causality may be inferred. For

example, when a group from the National Cancer Institute was asked

to rank the different types of evidence, it concluded that "[t]here

5
See also Sheldon Krimsky, The Weight of Scientific
Evidence in Policy and Law, 95 Am. J. Pub. Health S129, S129 (2005)
(explaining that the term "weight of the evidence" is used "to
characterize a process or method in which all scientific evidence
that is relevant to the status of a causal hypothesis is taken into
account").

-11-
should be no such hierarchy."6 Michele Carbon et al., Modern

Criteria to Establish Human Cancer Etiology, 64 Cancer Res. 5518,

5522 (2004); see also Sheldon Krimsky, The Weight of Scientific

Evidence in Policy and Law, 95 Am. J. Pub. Health S129, S130

(2005).

This "weight of the evidence" approach to making causal

determinations involves a mode of logical reasoning often described

as "inference to the best explanation," in which the conclusion is

not guaranteed by the premises.7 See Bitler v. A.O. Smith Corp.,

391 F.3d 1114, 1124 n.5 (10th Cir. 2004). As explained by

plaintiffs' expert on methodology Dr. Cranor, Distinguished

6
This point was also emphasized by Hill, who cautioned in
his article:
None of my nine viewpoints can bring
indisputable evidence for or against the
cause-and-effect hypothesis and none can be
required as a sine qua non. What they can do,
with greater or less strength, is to help us
to make up our minds on the fundamental
question--is there any other way of explaining
the set of facts before us, is there any other
answer equally, or more, likely than cause and
effect?
Austin Bradford Hill, The Environment and Disease: Association or
Causation?, 58 Proc. Royal Soc'y Med. 295, 299 (1965).
7
"Unlike a logical inference made by deduction where one
proposition can be logically inferred from other known
propositions, and unlike induction where a generalized conclusion
can be inferred from a range of known particulars, inference to the
best explanation--or 'abductive inferences'--are drawn about a
particular proposition or event by a process of eliminating all
other possible conclusions to arrive at the most likely one, the
one that best explains the available data." Bitler v. A.O. Smith
Corp., 391 F.3d 1114, 1124 n.5 (10th Cir. 2004).

-12-
Professor of Philosophy at the University of California, Riverside,

inference to the best explanation can be thought of as involving

six general steps, some of which may be implicit. The scientist

must (1) identify an association between an exposure and a disease,

(2) consider a range of plausible explanations for the association,

(3) rank the rival explanations according to their plausibility,

(4) seek additional evidence to separate the more plausible from

the less plausible explanations, (5) consider all of the relevant

available evidence, and (6) integrate the evidence using

professional judgment to come to a conclusion about the best

explanation.

In this mode of reasoning, the use of scientific judgment

is necessary. "No algorithm exists for applying the Hill

guidelines to determine whether an association truly reflects a

causal relationship or is spurious." Restatement § 28 cmt. c(3).

Because "[n]o scientific methodology exists for this process . . .

reasonable scientists may come to different judgments about whether

such an inference is appropriate." Id. § 28 reporters' note

cmt. c(4).

The fact that the role of judgment in the weight of the

evidence approach is more readily apparent than it is in other

methodologies does not mean that the approach is any less

scientific. No matter what methodology is used, "an evaluation of

data and scientific evidence to determine whether an inference of

-13-
causation is appropriate requires judgment and interpretation."

Id. § 28 cmt. c(1).8 The use of judgment in the weight of the

evidence methodology is similar to that in differential diagnosis,

see Cruz v. Bridgestone/Firestone N. Am. Tire, LLC, 388 F. App'x

803, 806-07 (10th Cir. 2010) (explaining that differential analysis

in general is best characterized as a process of reasoning to the

best explanation), which we have repeatedly found to be a reliable

method of medical diagnosis, see Granfield v. CSX Transp., Inc.,

597 F.3d 474, 486 (1st Cir. 2010); Dalkon Shield, 156 F.3d at 253.

Defendants argue that "regardless of its level of

acceptance in the scientific community, a pure 'weight of the

evidence' approach like that utilized by Dr. Smith . . . is hardly

the type of reliable scientific evidence contemplated by Daubert."9

8
The fact that epidemiology relies on statistical methods
does not avoid the use of judgment, as "[e]ven sampling error,
which is analyzed using quantitative statistical methods, only
provides a range of outcomes (associations) that might have been
produced by sampling error even if there is no association between
the agent and disease. Thus, interpreting the results of
epidemiologic studies requires informed judgment and is subject to
uncertainty." Restatement (Third) of Torts: Liability for Physical
and Emotional Harm § 28 reporters' note cmt. c(3) (2010).
9
Defendants draw our attention to a Fifth Circuit case,
excluding testimony based on a weight of the evidence methodology,
in which the court explained:
We are also unpersuaded that the 'weight of
the evidence' methodology these experts use is
scientifically acceptable for demonstrating a
medical link between Allen's EtO exposure and
brain cancer. Regulatory and advisory bodies
such as IARC, OSHA and EPA utilize a 'weight
of the evidence' method to assess the
carcinogenicity of various substances in human

-14-
No serious argument can be made that the weight of the evidence

approach is inherently unreliable. Rather, admissibility must turn

on the particular facts of the case. See, e.g., Cruz, 388 F. App'x

at 807 (explaining that expert testimony based on "inference to the

best explanation" may be admissible, but that there was no error in

the district court's finding that the expert's specific theory did

not have sufficient scientific support). Here, the question is

whether Dr. Smith, in reaching his opinion, applied the methodology

with "the same level of intellectual rigor" that he uses in his

scientific practice. Kumho Tire, 526 U.S. at 152.

B. Dr. Smith's Application of the Methodology

In concluding that the weight of the evidence supported

the conclusion that benzene can cause APL, Dr. Smith relied on his

knowledge and experience in the field of toxicology and molecular

epidemiology and considered five bodies of evidence drawn from the

peer-reviewed scientific literature on benzene and leukemia.

beings and suggest or make prophylactic rules
governing human exposure. This methodology
results from the preventive perspective that
the agencies adopt in order to reduce public
exposure to harmful substances. The agencies'
threshold of proof is reasonably lower than
that appropriate in tort law . . . .
Allen v. Pa. Eng'g Corp., 102 F.3d 194, 198 (5th Cir. 1996).
However, the Fifth Circuit did not, as defendants contend, hold
"that the 'weight-of-the evidence' approach is per se unreliable."
Rather, the court rejected its use in that case--a case in which it
found that the experts' conclusion was "at best weakly supported,
if not contradicted, by the evidence on which they rely," and in
which the experts "all declined to say that they would subject
their findings to the test of peer review for publication." Id.

-15-
 First, Dr. Smith considered the near-consensus among

governmental agencies, experts, and active researchers in the field

that benzene can cause AML as a class. The existence of this

causal connection has been established since the late 1970s. See

Bernard D. Goldstein & Gisela Witz, Benzene, in Environmental

Toxicants: Human Exposures and Their Health Effects 459, 478

(Morton Lippmann ed., 3rd ed. 2009). Dr. Smith noted that

epidemiological studies have found a statistically significant

increased incidence of AML in benzene-exposed workers and have

identified a dose-response relationship.

Second, Dr. Smith considered evidence concerning the

etiology, or origins, of leukemia indicating that all types of AML

derive from a genetically damaged pluripotent stem cell. Dr. Smith

referred to a recent peer-reviewed article that provided a review

of the current literature and reported numerous studies

demonstrating that both AML and CML are stem cell diseases. He

cited peer-reviewed studies finding that in the APL and Core

Binding Factor (CBF) subtypes of AML, as well as in CML, the stem

cell mutation is often in part caused by a chromosomal

translocation. He also cited evidence that APL and CBF share

common genetic susceptibility factors, common risk factors, and the

same incidence pattern occurring at a constant incidence with age

after age 20. Dr. Smith concluded that the best explanation for

-16-
this evidence is that all AMLs, including APL, have a common

etiology.10

Third, Dr. Smith considered toxicology studies

establishing that metabolites of benzene cause significant

chromosomal damage at the stem cell level in the bone marrow--the

type of damage that is known to cause APL and other types of AML.11

He also cited peer-reviewed work published by his lab showing that

leukemia cases associated with benzene exposure are more likely to

contain clonal chromosome aberrations than leukemias arising in the

general population.

Fourth, Dr. Smith considered two sets of studies

concerning the inhibition of a cellular enzyme known as

topoisomerase II (or "topo II") that is essential for the

maintenance of proper chromosome structure and segregation. One

set of studies--including both test tube and animal studies--has

established that two benzene metabolites are catalytic inhibitors

10
Defendants' experts questioned Dr. Smith's conclusion
that all of the subtypes of AML have a common etiology. However,
on cross examination in the district court Daubert hearing,
defendants' expert Dr. Pyatt agreed with the statement that "there
are a group of reasonable scientists who reasonably believe that
all forms of AML arise from the same progenitor cell" and stated
that Dr. Smith's opinion was "consistent with most of the
evidence." Defendants' expert Dr. Bennett likewise agreed that
"reasonable scientists can and do" agree with Dr. Smith.
11
Defendants' expert Dr. Bennett agreed that "there have
been innumerable studies that demonstrate that benzene actually
works at multiple levels to create damage to the DNA structure of
this hematopoietic stem cell."

-17-
of topo II. A second set of studies has established that a variety

of chemotherapeutic agents that are catalytic inhibitors of topo II

cause APL.12 Dr. Smith explained that taken together, these studies

provided evidence of a known biological mechanism by which exposure

to benzene could cause APL.

Fifth, Dr. Smith considered the small set of

epidemiological studies that provide data on the relationship

between benzene exposure and subtypes of AML.13 He concluded that

the evidence showed an increased risk factor for APL, consistent

with causality, and provided no grounds for concluding otherwise.

Dr. Smith explained that taking into account all of the

evidence described above--the fact that benzene causes AML as a

class, that all subtypes of AML likely have a common etiology, that

12
Dr. Smith cited a long list of peer-reviewed publications
and quoted a recent authoritative paper in a prominent journal
stating that "[t]herapy-related acute promyelocytic leukemia
(t-APL) with the t(15;17) translocation is a well recognized
complication of cancer treatment with agents targeting
topoisomerase II." Syed Khizer Hasan et al., Molecular Analysis of
t(15;17) Genomic Breakpoints in Secondary Acute Promyelocytic
Leukemia Arising After Treatment of Multiple Sclerosis, 112 Blood
3383, 3383 (2008). Defendants' hematopathologist, Dr. Bennett,
acknowledged that chemotherapeutic compounds that inhibit topo II
can cause APL.
13
He considered a multi-center Chinese case-control study
of 1257 cases of leukemia in which there was a 40% increased risk
of APL in benzene-exposed workers; a cohort study of 74,828 workers
exposed to benzene in China in which APL was the most common form
of AML diagnosed; a multi-center Italian case-control study of 38
cases of APL that showed a strong association between APL and
shoe-making, an industry that had for many years used benzene as an
adhesive; and several case reports of APL in benzene-exposed
workers.

-18-
benzene is known to cause the type of chromosomal damage

characteristic of APL, that benzene is known to inhibit an enzyme

whose inhibition is known to cause APL, and that APL has been

reported in benzene-exposed workers in a number of epidemiological

studies--he reached the opinion that the weight of the evidence

supports the conclusion that benzene exposure is capable of causing

APL. Dr. Smith's opinion rests on a scientifically sound and

methodologically reliable foundation, as is required by Daubert.

III.

In finding Dr. Smith's opinion inadmissible under Rule

702, the district court relied on (a) its evaluation of the

mechanistic and epidemiological evidence on which Dr. Smith based

his opinion, and (b) its understanding of the scientific concept of

"biological plausibility" as used by Dr. Smith when he explained

his conclusions. As we explain below, on both of these points, the

district court erred. In the end, the court's exclusion of the

testimony was based on its evaluation of the weight of the

evidence, which is an issue that is the province of the jury, and

on its misperception of the methodology and analysis that provided

the basis for Dr. Smith's opinion.

A. The Evidentiary Basis of Dr. Smith's Opinion

1. Mechanistic Evidence

The district court's exclusion of Dr. Smith's testimony

was based to a significant extent on its rejection of what it took

-19-
to be his three key subsidiary conclusions regarding the weight of

the mechanistic evidence. We briefly summarize the court's

analysis on these points before turning to our discussion of the

ways in which the court erred in its analysis.

First, the court held that there was insufficient

evidence to support Dr. Smith's opinion that all subtypes of AML

likely have a common etiology. The court reasoned that the "clear

differences" among AML subtypes-–in particular, APL's unique

response to certain types of therapy, and the subtypes' different

chromosomal abnormalities--made "a broad extrapolation from AML

generally to APL specifically" inappropriate.14 Milward, 664 F.

Supp. 2d at 144. The court also noted that a series of recent

studies had "led investigators to think that the 'leukemic stem

cell' may exist in more mature, differentiated cell lines," such

that "the 'leukemic stem cell' may not be a stem cell in the usual

sense, but rather a differentiated cell that has somehow acquired

the ability to reproduce itself, as a stem cell can." Id. at 145.

If the various AML subtypes did not arise from the same progenitor

or stem cell, the court reasoned, they might well not share a

14
Dr. Smith's supplemental report makes it clear that this
is something on which reasonable scientific disagreement is
possible. He explained that in his view, defendants' experts erred
in concluding that the fact that APL is therapeutically unique
means that it is also etiologically unique. Identifying the
biological mechanism that made APL therapeutically unique--the
sensitivity of the PML-RARá fusion gene to retinoic acid and
arsenic--he explained that this was "irrelevant" to APL's etiology.

-20-
common etiology.15 Finally, the court emphasized that there was "no

scientific consensus" on this issue, and that the question of when

the key chromosomal translocation occurs was considered by

researchers to be "a question that remains unanswered in the APL

field." Id. (quoting S. Wojiski et al., PML-RARá Initiates

Leukemia by Conferring Properties of Self-Renewal to Committed

Promyelocytic Progenitors, 23 Leukemia 1462, 1469 (2009) (emphasis

added)) (internal quotation marks omitted).

Second, the court held that what was known about the

types of chromosomal translocations caused by benzene did not offer

sufficient support for Dr. Smith's opinion that it is biologically

plausible that benzene causes the characteristic t(15;17)(q22;q12)

translocation seen with APL. The court explained that this opinion

would be warranted if benzene's impact on chromosomes were randomly

experienced, but it noted that a paper co-authored by Dr. Smith

concluded that "benzene can initiate or promote leukemia induction

by a nonrandom selective effect" on specific chromosomes. Id. at

147 (emphasis added). This defeated "the generalization that

because . . . benzene causes damage to some chromosomes, it is

15
In the Daubert hearing, Dr. Smith made it clear that he
had considered the key paper cited by the district court on this
point. He noted that it was based on studies in "mice using a
highly artificial system," and he explained that even if the
mutation could occur at a later point in differentiation as
indicated by this paper, "it doesn't mean that it has to occur only
in that compartment."

-21-
'biologically plausible' that it causes damage to other

chromosomes." Id.

Third, the court held that there was insufficient

evidence to support the inference that benzene metabolites inhibit

topo II in such a way as to cause the chromosomal translocation

seen in cases of APL. The court's conclusion was in part based on

evidence that "[t]here are different classes of topo II inhibitors

and the different classes have been associated with different AML

subtypes." Id. Highlighting one article's finding that leukemias

induced by benzene do not appear to exhibit the defining

characteristics associated with four other classes of topo II

inhibitors, id. at 148, the court held that to "the extent that Dr.

Smith's opinion rests on the proposition that all topo II

inhibitors act similarly to cause a similar effect, then, it does

not appear to be based on reliable scientific knowledge," id. at

147.

In reaching these three conclusions about some of the

evidence on which Dr. Smith based his opinion, the court both

placed undue weight on the lack of general acceptance of Dr.

Smith's conclusions and crossed the boundary between gatekeeper and

trier of fact.

Although general acceptance is still a relevant

consideration under Daubert, the court's demands went too far. Cf.

Smith v. Ford Motor Co., 215 F.3d 713, 721 (7th Cir. 2000)

-22-
(reversing district court that had treated lack of peer review as

dispositive grounds for excluding expert opinion). On the question

of the origins of APL, for example, the court explained that in the

absence of consensus about the target cell for the leukemic

mutation, Dr. Smith's opinion that all forms of AML likely share a

common origin was "at best a plausible hypothesis." Milward, 664

F. Supp. 2d at 146. The court explained that the fact that "other

plausible hypotheses . . . might be true as well, including the

hypothesis that the genetic mutation that leads to APL occurs in

relatively mature cells," meant that Dr. Smith's opinion was not

"based on sufficient facts and data to be accepted as a reliable

scientific conclusion." Id.; see also id. at 148 (focusing on lack

of consensus as to the topo II question). But the fact that

another explanation might be right is not a sufficient basis for

excluding Dr. Smith's testimony. "Lack of certainty is not, for a

qualified expert, the same thing as guesswork." Primiano v. Cook,

No. 06-15563, 2010 WL 1660303, *5 (9th Cir. Apr. 27, 2010).

In addition, the alleged flaws identified by the court go

to the weight of Dr. Smith's opinion, not its admissibility. There

is an important difference between what is unreliable support and

what a trier of fact may conclude is insufficient support for an

expert's conclusion.

The court's analysis repeatedly challenged the factual

underpinnings of Dr. Smith's opinion, and took sides on questions

-23-
that are currently the focus of extensive scientific research and

debate--and on which reasonable scientists can clearly disagree.

In this, the court overstepped the authorized bounds of its role as

gatekeeper. "The soundness of the factual underpinnings of the

expert's analysis and the correctness of the expert's conclusions

based on that analysis are factual matters to be determined by the

trier of fact." Smith, 215 F.3d at 718. "When the factual

underpinning of an expert's opinion is weak, it is a matter

affecting the weight and credibility of the testimony--a question

to be resolved by the jury." Vargas, 471 F.3d at 264 (quoting

Int'l Adhesive Coating Co. v. Bolton Emerson Int'l, 851 F.2d 540,

545 (1st Cir. 1988)) (internal quotation marks omitted); see also

Quiet Tech. DC-8, Inc. v. Hurel-Dubois UK Ltd., 326 F.3d 1333, 1345

(11th Cir. 2003); Amorgianos v. Nat'l R.R. Passenger Corp., 303

F.3d 256, 267 (2d Cir. 2002).

Of course, following Joiner, a "district court properly

may exclude expert testimony if the court concludes too great an

analytical gap exists between the existing data and the expert's

conclusion." Kennedy v. Collagen Corp., 161 F.3d 1226, 1230 (9th

Cir. 1998). Here, however, "the gap was of the district court's

making." Id. Dr. Smith's opinion was based on a reliable

methodology and substantial evidence that he carefully explained.

The questions that the court posed were sensible ones, but ones for

the jury to resolve.

-24-
 At times, the court's error in excluding Dr. Smith's

testimony derived from a mistake in its understanding of the weight

of the evidence methodology employed by Dr. Smith. The court

treated the separate evidentiary components of Dr. Smith's analysis

atomistically, as though his ultimate opinion was independently

supported by each. For example, the court referred to "Dr. Smith's

opinion that because benzene metabolites inhibit topo II and

because some classes of topo II inhibitors appear to have a causal

relationship to APL, therefore benzene has a causal relationship to

APL." Milward, 664 F. Supp. 2d at 148 (emphasis added). This

overstates Dr. Smith's conclusion as to the topo II evidence, and

is indicative of an error in the court's understanding of the

nature of Dr. Smith's analysis.

In Dr. Smith's weight of the evidence approach, no body

of evidence was itself treated as justifying an inference of

causation. Rather, each body of evidence was treated as grounds

for the subsidiary conclusion that it would, if combined with other

evidence, support a causal inference. The district court erred in

reasoning that because no one line of evidence supported a reliable

inference of causation, an inference of causation based on the

totality of the evidence was unreliable. Cf. NutraSweet Co. v. X-L

Eng'g Co., 227 F.3d 776, 789 (7th Cir. 2000) (holding that an

expert's reliance on individual pieces of evidence, insufficient in

themselves to prove a point, "did not render his opinion

-25-
speculative").16 The hallmark of the weight of the evidence

approach is reasoning to the best explanation for all of the

available evidence. Cf. Dalkon Shield, 156 F.3d at 253 (reversing

district court's exclusion of expert testimony as "guesswork" or

without "basis" when testimony was based on differential diagnosis

and there was no showing that any one of the expert's premises was

"so faulty that it could not even be tendered to the jury for its

consideration"); see also Hardyman v. Norfolk & W. Ry. Co., 243

F.3d 255, 261 (6th Cir. 2001).

2. Epidemiological Evidence

As to the epidemiological evidence on which Dr. Smith

based his opinion in part, the court held that the published

articles on which Dr. Smith relied did not support his opinion, and

that in any event, the evidence was not statistically significant.

On these grounds, the court rejected Dr. Smith's conclusion that

the available epidemiological evidence offered some support for an

inference of causation.

In concluding that the papers cited by Dr. Smith did not

support his opinion, the court reasoned that "Dr. Garabrant

convincingly demonstrated, especially with respect to the Golomb

16
As a general evidentiary matter, "individual pieces of
evidence, insufficient in themselves to prove a point, may in
cumulation prove it," and "a piece of evidence, unreliable in
isolation, may become quite probative when corroborated by other
evidence." Bourjaily v. United States, 483 U.S. 171, 179-80
(1987).

-26-
and Travis papers, that Dr. Smith's conclusions that there was a

positive association between exposure to benzene and APL were based

on faulty calculations of odds ratios." Milward, 664 F. Supp. 2d

at 149. An odds ratio represents the difference in the incidence

of a disease between a population that has been exposed to benzene

and one that has not. In Dr. Garabrant's opinion, Dr. Smith should

have used the incidence rate of APL for the general population as

a baseline, rather than the rate for non-benzene-exposed workers.

In the Daubert hearing and in his supplemental report, however, Dr.

Smith explained that he disagreed with Dr. Garabrant on this point,

but that in any event, the odds ratio was still elevated,

consistent with an inference of causation. Where, as here, both

experts' opinions are supported by evidence and sound scientific

reasoning, the question of who is right is a question for the

jury.17

The court explained, however, that even if "some of the

data reported in the various studies could be properly understood

to suggest a positive association, the findings are not

statistically significant," id., and that although "epidemiological

17
The court also rejected Dr. Smith's analysis of the
epidemiological evidence on the grounds that "none of the studies
purports to give direct support to the proposition that benzene
causes APL." Milward v. Acuity Specialty Prods. Grp., Inc., 664 F.
Supp. 2d 137, 148 (D. Mass. 2009). Yet Dr. Smith did not claim
that the studies provided direct support. Rather, his
characterization of his methodology makes clear that he was using
them as indirect support.

-27-
evidence is not always essential," the defendants were "correct

that sound epidemiological studies are ordinarily needed to

confirm, by consistent observation, an hypothesis of causation,"

id. at 148.

In context, the district court read too much into the

paucity of statistically significant epidemiological studies. The

absence of peer-reviewed epidemiological studies does not, as

defendants contend, make it "almost impossible" for Dr. Smith's

opinion to be admissible. Epidemiological studies are not per se

required as a condition of admissibility regardless of context.

See Rider v. Sandoz Pharm. Corp., 295 F.3d 1194, 1198 (11th Cir.

2002) ("It is well-settled that while epidemiological studies may

be powerful evidence of causation, the lack thereof is not fatal to

a plaintiff's case."); Restatement § 28 reporters' note cmt. c(3)

(listing federal circuit cases holding that epidemiological data is

not necessary). Nor are such studies treated as always essential

in the relevant scientific communities.

To be clear, this is not a situation in which the

available epidemiological studies found that there is no causal

link, or even one in which no cases of APL were found among

benzene-exposed workers. Cf. Norris v. Baxter Healthcare Corp.,

397 F.3d 878, 882 (10th Cir. 2005) (holding that epidemiological

studies are not required to prove causation, but that a substantial

body of epidemiological evidence challenging causation cannot be

-28-
ignored); Allen v. Pa. Eng'g Corp., 102 F.3d 194, 197 (5th Cir.

1996) (finding it significant that "numerous reputable

epidemiological studies covering in total thousands of workers"

indicated that there was no causation).

Rather, this is a case in which the few studies that

differentiate between AML and APL do not offer conclusive

statistically significant evidence either way, in part because the

rarity of APL makes it nearly impossible to perform a large enough

study.18 Dr. Smith estimated that in order to obtain statistically

significant results, one would need hundreds of thousands of highly

exposed workers, the same number of controls, and millions of

dollars in funding. The court erred in treating the lack of

statistical significance as a crucial flaw. See Collagen Corp.,

161 F.3d at 1229 (finding that the district court placed too much

emphasis on lack of epidemiological studies where such studies

"would be almost impossible to perform"); see also Primiano, 2010

WL 1660303, at *5-6 (noting that peer-reviewed studies are not

necessary, especially when there are good reasons why such studies

have not been performed). Under these circumstances, the court

18
The difficulty of performing such a study is not
contested by defendants, and it has even been expressly affirmed in
the scientific literature. See Dan Douer, The Epidemiology of
Acute Promyelocytic Leukaemia, 16 Best Prac. & Res. Clinical
Haematology 357, 358 (2003) ("It is difficult to perform
epidemiological studies in AML subtypes classified according to
cytogenetic abnormalities owing to the small number of patients
within each subgroup.").

-29-
erred in holding that "Dr. Smith's attempt to support his

conclusion with data that concededly lacks statistical

significance" was "a deviation from sound practice of the

scientific method" that provided grounds for exclusion. Milward,

664 F. Supp. 2d at 149.

The court's evaluation of the epidemiological evidence is

also in tension with the weight of the evidence methodology. Dr.

Smith explained that his citation to epidemiological data was meant

to challenge the theory that benzene exposure could not cause APL,

and to highlight that the limited data available was consistent

with the conclusions that he had reached on the basis of other

bodies of evidence. He stated that "[i]f epidemiologic studies of

benzene-exposed workers were devoid of workers who developed APL,

one could hypothesize that benzene does not cause this particular

subtype of AML." The fact that, on the contrary, "APL is seen in

studies of workers exposed to benzene where the subtypes of AML

have been separately analyzed and has been found at higher levels

than expected" suggested to him that the limited epidemiological

evidence was at the very least consistent with, and suggestive of,

the conclusion that benzene can cause APL.

The court rejected Dr. Smith's reasoning, stating that a

"'suggestion' may give rise to a plausible hypothesis, but not a

reliable inference." Milward, 664 F. Supp. 2d at 149. But as

noted above, this is inconsistent with the scientifically accepted

-30-
methodology employed by Dr. Smith. Dr. Smith did not infer

causality from this suggestion alone, but rather from the

accumulation of multiple scientifically acceptable inferences from

different bodies of evidence.

B. The Concept of "Biological Plausibility"

The district court also erred in its apprehension of the

scientific concept of biological plausibility and its place in Dr.

Smith's analysis. The concept of biological plausibility, which

numbers among the nine Hill viewpoints, asks whether the

hypothesized causal link is credible in light of what is known from

science and medicine about the human body and the potentially

offending agent. At two places in the court's analysis, it

conflated the scientific question of biological plausibility with

the legal question of probability.

In the court's discussion of the epidemiological

evidence, it stated that even if the evidence "'suggests' a causal

relationship," providing support for Dr. Smith's opinion regarding

biological plausibility, a "plausible hypothesis" is not a

"reliable inference" and is therefore inadmissible. Id. Here, the

court not only misconstrued the concept of biological plausibility

by equating it with a merely plausible or possible hypothesis, but

also misconstrued the concept's role in Dr. Smith's analysis by

assuming that Dr. Smith treated the criteria as sufficient grounds

-31-
for inferring causality (rather than as one consideration that

entered into his weighing of the evidence).

The court made a similar error in its conclusion, where

it stated:

While Dr. Smith's hypotheses are, to use his
term, "plausible," they remain hypotheses,
the validity of which has not been reliably
established. . . . [T]he sum of Dr. Smith's
testimony, fairly understood, is that benzene
might be a cause of APL.

Id. Again, the district court misunderstood Dr. Smith to be saying

that causation is possible rather than probable. The sum of Dr.

Smith's testimony was not merely that it is possible, or even

biologically plausible, that benzene causes APL. Rather, the sum

of his testimony was that a weighing of the Hill factors, including

biological plausibility, supported the inference that the

association between benzene exposure and APL is genuine and causal.

IV.

The record clearly demonstrates that Dr. Smith's opinion

was based on an analysis in which he employed the "same level of

intellectual rigor" that he employs in his academic work. Kumho

Tire, 526 U.S. at 152. In excluding Dr. Smith's testimony, the

district court did not properly apply Daubert and exceeded the

scope of its discretion. We reverse the district court's judgment

-32-
for the defendants and its exclusion of Dr. Smith's testimony, and

we remand for proceedings consistent with this opinion.19

So ordered.

19
We wish to acknowledge the able briefing of the issues by
the parties and amici.

-33-