Recently, two lawyers wrote an article in a legal trade magazine about excluding epidemiologic evidence in civil litigation.[1] The article was wildly wide of the mark, with several conceptual and practical errors.[2] For starters, the authors discussed Rule 702 as excluding epidemiologic studies and evidence, when the rule addresses the admissibility of expert witness opinion testimony. The most egregious recommendation of the authors, however, was their recommendation that counsel urge the classifications of chemicals with respect to carcinogenicity, by the International Agency for Research on Cancer (IARC), and by regulatory agencies, as probative for or against causation.
The project of evaluating the evidence for, or against, carcinogenicity of the myriad natural and synthetic agents to which humans are exposed is certainly important. Certainly, IARC has taken the project seriously. There have, however, been problems with IARC’s classifications of specific chemicals, pharmaceuticals, or exposure circumstances, but a basic problem with the classifications begins with the classes themselves. Classification requires defined classes. I don’t mean to be anti-semantic, but IARC’s definitions and its hierarchy of carcinogenicity are not entirely coherent.
The agency was established in 1965, and by the early 1970s, found itself in the business of preparing “monographs on the evaluation of carcinogenic risk of chemicals to man.” Originally, the IARC set out to classify the carcinogenicity of chemicals, but over the years, its scope increased to include complex mixtures, physical agents such as different forms of radiation, and biological organisms. To date, there have been 134 IARC monographs, addressing 1,045 “agents” (either substances or exposure circumstances).
From its beginnings, the IARC has conducted its classifications through working groups that meet to review and evaluate evidence, and classify the cancer hazards of “agents” under discussion. The breakdown of IARC’s classifications among four groups currently is:
Group 1 – Carcinogenic to humans (127 agents)
Group 2A – Probably carcinogenic to humans (95 agents)
Group 2B – Possibly carcinogenic to humans (323 agents)
Group 3 – Not classifiable as to its carcinogenicity to humans (500 agents)
Previously, the IARC classification included a Group 4 for agents that are probably not carcinogenic for human beings. After decades of review, the IARC placed only a single agent in Group 4, caprolactam, apparently because the agency found everything else in the world to be presumptively a cause of cancer. The IARC could not find sufficiently strong evidence even for water, air, or basic foods to declare that they do not cause cancer in humans. Ultimately, the IARC abandoned Group 4, in favor of a presumption of universal carcinogencity.
The IARC describes its carcinogen classification procedures, requirements, and rationales in a document known as “The Preamble.” Any discussion of IARC classifications, whether in scientific publications or in legal briefs, without reference to this document should be suspect. The Preamble seeks to define many of the words in the classificatory scheme, some in ways that are not intuitive. This document has been amended over time, and the most recent iteration can be found online at the IARC website.[3]
IARC claims to build its classifications upon “consensus” evaluations, based in turn upon considerations of
(a) the strength of evidence of carcinogenicity in humans,
(b) the evidence of carcinogenicity in experimental (non-human) animals, and
(c) the mechanistic evidence of carcinogenicity.
IARC further claims that its evaluations turn on the use of “transparent criteria and descriptive terms.”[4] This last claim is, for some terms, is falsifiable.
The working groups are described as engaged in consensus evaluations, although past evaluations have been reached on simple majority vote of the working group. The working groups are charged with considering the three lines of evidence, described above, for any given agent, and reaching a synthesis in the form of the IARC classificatory scheme. The chart, from the Preamble, below roughly describes how working groups may “mix and match” lines of evidence, of varying degrees of robustness and validity (vel non) to reach a classification.
Agents placed in Category I are thus “carcinogenic to humans.” Interestingly, IARC does not refer to Category I carcinogens as “known” carcinogens, although many commentators are prone to do so. The implication of calling Category I agents “known carcinogens” is to distinguish Category IIA, IIB, and III as agents “not known to cause cancer.” The adjective that IARC uses, rather than “known,” is “sufficient” evidence in humans, but IARC also allows for reaching Category I with “limited,” or even “inadequate” human evidence if the other lines of evidence, in experimental animals or mechanistic evidence in humans, are sufficient.
In describing “sufficient” evidence, the IARC’s Preamble does not refer to epidemiologic evidence as potentially “conclusive” or “definitive”; rather its use of “sufficient” implies, perhaps non-transparently, that its labels of “limited” or “inadequate” evidence in humans refer to insufficient evidence. IARC gives an unscientific, inflated weight and understanding to “limited evidence of carcinogenicity,” by telling us that
“[a] causal interpretation of the positive association observed in the body of evidence on exposure to the agent and cancer is credible, but chance, bias, or confounding could not be ruled out with reasonable confidence.”[5]
Remarkably, for IARC, credible interpretations of causality can be based upon evidentiary displays that are confounded or biased. In other words, non-credible associations may support IARC’s conclusions of causality. Causal interpretations of epidemiologic evidence are “credible” according to IARC, even though Sir Austin’s predicate of a valid association is absent.[6]
The IARC studiously avoids, however, noting that any classification is based upon “insufficient” evidence, even though that evidence may be less than sufficient, as in “limited,” or “inadequate.” A close look at Table 4 reveals that some Category I classifications, and all Category IIA, IIB, and III classifications are based upon insufficient evidence of carcinogenicity in humans.
Non-Probable Probabilities
The classification immediately below Category or Group I is Group 2A, for agents “probably carcinogenic to humans.” The IARC’s use of “probably” is problematic. Group I carcinogens require only “sufficient” evidence of human carcinogenicity, and there is no suggestion that any aspect of a Group I evaluation requires apodictic, conclusive, or even “definitive” evidence. Accordingly, the determination of Group I carcinogens will be based upon evidence that is essentially probabilistic. Group 2A is also defined as having only “limited evidence of carcinogenicity in humans”; in other words, insufficient evidence of carcinogenicity in humans, or epidemiologic studies with uncontrolled confounding and biases.
Importing IARC 2A classifications into legal or regulatory arenas will allow judgments or regulations based upon “limited evidence” in humans, which as we have seen, can be based upon inconsistent observational studies, and studies that fail to measure and adjust for known and potential confounding risk factors and systematic biases. The 2A classification thus requires little substantively or semantically, and many 2A classifications leave juries and judges to determine whether a chemical or medication caused a human being’s cancer, when the basic predicates for Sir Austin Bradford Hill’s factors for causal judgment have not been met.[7]
An IARC evaluation of Group 2A, or “probably carcinogenic to humans,” would seem to satisfy the legal system’s requirement that an exposure to the agent of interest more likely than not causes the harm in question. Appearances and word usage in different contexts, however, can be deceiving. Probability is a continuous quantitative scale from zero to one. In Bayesian analyses, zero and one are unavailable because if either were our starting point, no amount of evidence could ever change our judgment of the probability of causation. (Cromwell’s Rule). The IARC informs us that its use of “probably” is purely idiosyncratic; the probability that a Group 2A agent causes cancer has “no quantitative” meaning. All the IARC intends is that a Group 2A classification “signifies a greater strength of evidence than possibly carcinogenic.”[8] Group 2A classifications are thus consistent with having posterior probabilities less than 0.5 (or 50 percent). A working group could judge the probability of a substance or a process to be carcinogenic to humans to be greater than zero, but no more than say ten percent, and still vote for a 2A classification, in keeping with the IARC Preamble. This low probability threshold for a 2A classification converts the judgment of “probably carcinogenic” into little more than precautionary prescriptions, rendered when the most probable assessment is either ignorance or lack of causality. There is thus a practical certainty, close to 100%, that a 2A classification will confuse judges and juries, as well as the scientific community.
In addition to being based upon limited, that is insufficient, evidence of human carcinogenicity, Group 2A evaluations of “probable human carcinogenicity” connote “sufficient evidence” in experimental animals. An agent can be classified 2A even when the sufficient evidence of carcinogenicity occurs in only one of several non-human animal species, with the other animal species failing to show carcinogenicity. IARC 2A classifications can thus raise the thorny question in court whether a claimant is more like a rat or a mouse.
Courts should, because of the incoherent and diluted criteria for “probably carcinogenic,” exclude expert witness opinions based upon IARC 2A classifications as scientifically insufficient.[9] Given the distortion of ordinary language in its use of defined terms such as “sufficient,” “limited,” and “probable,” any evidentiary value to IARC 2A classifications, and expert witness opinion based thereon, is “substantially outweighed by a danger of … unfair prejudice, confusing the issues, [and] misleading the jury….”[10]
Everything is Possible
Category 2B denotes “possibly carcinogenic.” This year, the IARC announced that a working group had concluded that aspartame, an artificial sugar substitute, was “possibly carcinogenic.”[11] Such an evaluation, however, tells us nothing. If there are no studies at all of an agent, the agent could be said to be possibly carcinogenic. If there are inconsistent studies, even if the better designed studies are exculpatory, scientists could still say that the agent of interest was possibly carcinogenic. The 2B classification does not tell us anything because everything is possible until there is sufficient evidence to inculpate or exculpate it from causing cancer in humans.
It’s a Hazard, Not a Risk
IARC’s classification does not include an assessment of exposure levels. Consequently, there is no consideration of dose or exposure level at which an agent becomes carcinogenic. IARC’s evaluations are limited to whether the agent is or is not carcinogenic. The IARC explicitly concedes that exposure to a carcinogenic agent may carry little risk, but it cannot bring itself to say no risk, or even benefit at low exposures.
As noted, the IARC classification scheme refers to the strength of the evidence that an agent is carcinogenic, and not to the quantitative risk of cancer from exposure at a given level. The Preamble explains the distinction as fundamental:
“A cancer hazard is an agent that is capable of causing cancer, whereas a cancer risk is an estimate of the probability that cancer will occur given some level of exposure to a cancer hazard. The Monographs assess the strength of evidence that an agent is a cancer hazard. The distinction between hazard and risk is fundamental. The Monographs identify cancer hazards even when risks appear to be low in some exposure scenarios. This is because the exposure may be widespread at low levels, and because exposure levels in many populations are not known or documented.”[12]
This attempted explanation reveals important aspects of IARC’s project. First, there is an unproven assumption that there will be cancer hazards regardless of the exposure levels. The IARC contemplates that there may circumstances of low levels of risk from low levels of exposure, but it elides the important issue of thresholds. Second, IARC’s distinction between hazard and risk is obscured by its own classifications. For instance, when IARC evaluated crystalline silica and classified it in Group I, it did so for only “occupational exposures.”[13] And yet, when IARC evaluated the hazard of coal exposure, it placed coal dust in Group 3, even though coal dust contains crystalline silica.[14] Similarly, in 2018, the IARC classified coffee as a Group 3,[15] even though every drop of coffee contains acrylamide, which is, according to IARC, a Group 2A “probable human carcinogen.”[16]
[1] Christian W. Castile & and Stephen J. McConnell, “Excluding Epidemiological Evidence Under FRE 702,” For The Defense 18 (June 2023) [Castile].
[2] “Excluding Epidemiologic Evidence Under Federal Rule of Evidence 702” (Aug. 26, 2023).
[3] IARC Monographs on the Identification of Carcinogenic Hazards to Humans – Preamble (2019).
[4] Jonathan M. Samet , Weihsueh A. Chiu , Vincent Cogliano, Jennifer Jinot, David Kriebel, Ruth M. Lunn, Frederick A. Beland, Lisa Bero, Patience Browne, Lin Fritschi, Jun Kanno , Dirk W. Lachenmeier, Qing Lan, Gerard Lasfargues, Frank Le Curieux, Susan Peters, Pamela Shubat, Hideko Sone, Mary C. White , Jon Williamson, Marianna Yakubovskaya , Jack Siemiatycki, Paul A. White, Kathryn Z. Guyton, Mary K. Schubauer-Berigan, Amy L. Hall, Yann Grosse, Veronique Bouvard, Lamia Benbrahim-Tallaa, Fatiha El Ghissassi, Beatrice Lauby-Secretan, Bruce Armstrong, Rodolfo Saracci, Jiri Zavadil , Kurt Straif, and Christopher P. Wild, “The IARC Monographs: Updated Procedures for Modern and Transparent Evidence Synthesis in Cancer Hazard Identification,” 112 J. Nat’l Cancer Inst. djz169 (2020).
[5] Preamble at 31.
[6] See Austin Bradford Hill, “The Environment and Disease: Association or Causation?” 58 Proc. Royal Soc’y Med. 295 (1965) (noting that only when “[o]ur observations reveal an association between two variables, perfectly clear-cut and beyond what we would care to attribute to the play of chance,” do we move on to consider the nine articulated factors for determining whether an association is causal.
[7] Id.
[8] IARC Monographs on the Identification of Carcinogenic Hazards to Humans – Preamble 31 (2019) (“The terms probably carcinogenic and possibly carcinogenic have no quantitative significance and are used as descriptors of different strengths of evidence of carcinogenicity in humans.”).
[9] See “Is the IARC lost in the weeds” (Nov. 30, 2019); “Good Night Styrene” (Apr. 18, 2019).
[11] Elio Riboli, et al., “Carcinogenicity of aspartame, methyleugenol, and isoeugenol,” 24 The Lancet Oncology P848-850 (2023);
IARC, “Aspartame hazard and risk assessment results released” (2023).
[12] Preamble at 2.
[13] IARC Monograph 68, at 41 (1997) (“For these reasons, the Working Group therefore concluded that overall the epidemiological findings support increased lung cancer risks from inhaled crystalline silica (quartz and cristobalite) resulting from occupational exposure.”).
[14] IARC Monograph 68, at 337 (1997).
[15] IARC Monograph No. 116, Drinking Coffee, Mate, and Very Hot Beverages (2018).
[16] IARC Monograph no. 60, Some Industrial Chemicals (1994).
