Perchloroethylene (PCE) Epidemiology and Neurotoxicity

RoseyFields evaluates the human epidemiological evidence for PCE as follows.

Study Populations and Data Sources

RoseyFields’ epidemiological review of tetrachloroethylene (PCE, also called perchloroethylene) draws on four principal categories of human data, and it is the convergence of findings across these categories—rather than reliance on any single cohort—that anchors RoseyFields’ confidence in the conclusions below.

Occupational cohorts of dry cleaners, laundry workers, and industrial degreasing or aircraft-maintenance workers provide the largest and longest-running body of data. Key cohorts include Nordic dry-cleaning and laundry workers followed for cancer incidence and mortality (Lynge et al., 2006; Seldén and Ahlborg, 2011; Pukkala et al., 2009; Travier et al., 2002; Andersen et al., 1999), a large U.S. dry-cleaner mortality cohort (Blair et al., 2003; Calvert et al., 2011), and aircraft-maintenance and aerospace workers with mixed solvent exposure (Radican et al., 2008; Boice et al., 1999). These cohorts benefit from decades of follow-up and, in the more recent studies, quantitative or semiquantitative exposure reconstruction, but many of the earlier ones classified exposure only by job title, a limitation discussed in Section 5.

Community and contamination-site cohorts capture non-occupational, largely residential exposure. The most influential is the Cape Cod, Massachusetts drinking-water cohort, in which PCE leached from vinyl-lined water-distribution pipe into the public water supply of eight towns between the late 1960s and 1980; this population has been studied repeatedly for cancer (Aschengrau et al., 1993, 1998, 2003; Vieira et al., 2005; Gallagher et al., 2011) and reproductive and developmental outcomes (Aschengrau et al., 2008, 2009, 2019; Ozonoff et al., 2018) using progressively refined water-distribution modeling to reconstruct individual exposure. A second, more recent population type—residents of apartment buildings and children in day-care facilities co-located with dry-cleaning establishments—captures chronic, low-level residential inhalation exposure and has been used specifically to evaluate neurobehavioral and visual effects (Schreiber et al., 2002; Altmann et al., 1995; Storm et al., 2011).

General-population biomonitoring, principally the U.S. National Health and Nutrition Examination Survey (NHANES), measures blood PCE concentration in a nationally representative sample and provides context for how occupational and residential exposure levels compare to typical background body burden from general environmental and consumer-product sources.

Finally, nested and population-based case-control studies—drawing cases and controls from population cancer registries and using either job-exposure matrices or, in the higher-quality studies, semiquantitative cumulative PCE exposure indices—have been essential for examining specific cancer sites, including bladder cancer, non-Hodgkin lymphoma, multiple myeloma, and renal cell carcinoma (Pesch et al., 2000; Seidler et al., 2007; Gold et al., 2010a, 2010b).

RoseyFields’ assessors give particular weight to the fact that PCE’s principal human health signals—neurobehavioral and visual deficits, and a modest excess of bladder cancer and lymphohematopoietic malignancy—have each been observed independently in occupational cohorts exposed primarily by inhalation in the workplace and in residential or community cohorts exposed by a different route or a different (generally lower) intensity. This cross-population, cross-route consistency is considerably more persuasive than replication within a single research group’s occupational cohort, and it is the primary reason RoseyFields’ assessors treat the neurotoxicity and bladder-cancer findings as resting on a genuinely robust epidemiological foundation, while treating other cancer sites—discussed in Section 3—as resting on a thinner and less consistent one.

Health Outcomes by Body System and Strength of Evidence

RoseyFields’ review organizes the full range of human health outcomes investigated for PCE into three evidentiary tiers, consistent with the characterization EPA itself adopted in the 2012 IRIS Toxicological Review and with independent literature (IARC, 2014).

Neurological system. Deficits in color vision, reaction time, cognitive function, and visuospatial processing represent the single most extensively replicated and best-characterized human health effect of PCE, observed with consistency across independent occupational and residential study populations and discussed in detail in Section 4. This is the only outcome for which EPA itself derived quantitative noncancer reference values (the RfD and RfC), based directly on this epidemiological literature rather than on animal data.

Genitourinary/oncologic (bladder cancer). An elevated risk of bladder cancer of modest magnitude (roughly 10–40% increased risk in the higher-quality studies), with an exposure-response gradient demonstrated in at least one large case-control study, represents the strongest cancer signal for PCE and is discussed further in Section 4.

Hematologic/lymphatic malignancies. Non-Hodgkin lymphoma shows a moderately consistent elevated risk across several cohort studies using higher-quality exposure assessment, with some evidence of an exposure-response gradient, and is best characterized as suggestive-to-moderate. Multiple myeloma shows a more inconsistent pattern: studies using broad, job-title-based exposure measures generally show no excess risk, while studies using PCE-specific cumulative exposure measures report an association, including a notably elevated (though imprecise) relative risk in women in one aircraft-maintenance cohort; this outcome is best characterized as suggestive but genuinely inconsistent, and highly sensitive to exposure-assessment methodology.

Other cancer sites. Esophageal, kidney, lung, liver, cervical, and breast cancer have each been investigated in relation to PCE exposure, with a more mixed pattern of risk estimates and, notably, an absence of exposure-response data from the studies using quantitative, PCE-specific cumulative exposure measures. RoseyFields’ assessors characterize the evidence for each of these sites as suggestive at best, and for several sites as inadequate to distinguish a true association from chance or confounding.

Reproductive and developmental. Spontaneous abortion shows a moderately consistent elevated risk across several Scandinavian dry-cleaner cohorts (Olsen et al., 1990) and the Cape Cod drinking-water cohort, and is best characterized as suggestive-to-moderate. Birth defects, specifically oral clefts and neural tube defects, showed elevated risk in the Cape Cod retrospective cohort (Aschengrau et al., 2009) but have not been extensively replicated in other populations and are best characterized as suggestive. Reduced fecundity/increased time to pregnancy has been reported in the Cape Cod cohort (Aschengrau et al., 2019) and is similarly suggestive. Male reproductive effects (reduced sperm quality) were reported in a modestly sized dry-cleaner cohort (Eskenazi et al., 1991) but have not been widely replicated and are best characterized as suggestive-to-limited. Other developmental and reproductive outcomes investigated in the Cape Cod cohort, including polycystic ovary syndrome, endometriosis, ischemic placental disease, and low birth weight, are supported by more limited, single-cohort evidence and are best characterized as suggestive at most.

Hematological and immunological (non-malignant). Changes in standard hematological parameters (hemoglobin, platelet counts, white cell counts) were reported in a single cross-sectional study of Egyptian dry-cleaning workers with elevated blood PCE (Emara et al., 2010); this finding lacks corroborating human studies and is best characterized as suggestive but inadequately corroborated. Immunological effects specific to PCE in humans have not been adequately studied and are best characterized as inadequate.

Hepatic and renal systems. Beyond the renal cell carcinoma and end-stage renal disease signals discussed above, non-malignant hepatic and renal effects specific to PCE have not been well characterized in human epidemiological studies independent of the neurotoxicity, cancer, and reproductive literatures already described; RoseyFields’ assessors regard this as an area of inadequate human data, notwithstanding a more developed animal toxicological literature on both organ systems.

Detailed Evidence for the Strong/Consistent Outcomes

Neurobehavioral and visual deficits. This is the paradigm case of strong, consistent, quantitatively characterized human evidence for PCE, and—unusually among the chemicals RoseyFields has reviewed—it is corroborated across occupational and residential populations using different exposure routes and intensities. Cavalleri et al. (1994) found a 6% average decrement in color-vision performance (Lanthony D-15 Hue desaturated panel) among 35 dry-cleaning and laundry workers compared to 35 matched controls, with a clear exposure-response pattern by job category (time-weighted average exposures of 33–49 mg/m³); Gobba et al. (1998), following 33 of these same workers two years later, found further color-vision decrements among those whose exposure had increased and no improvement among those whose exposure had decreased, providing a within-subject, self-controlled replication of the original finding. Echeverria et al. (1995), studying 65 Detroit-area dry cleaners with a standardized neurobehavioral battery, found dose-related deficits in reaction time and cognitive/visuospatial function (changes of 4–14% from internal referent levels depending on subtest). Critically, these occupational findings are corroborated in residential populations exposed only to the much lower concentrations found in buildings co-located with dry-cleaning establishments: Altmann et al. (1995) found neurobehavioral and neurophysiological deficits among residents of neighborhoods near dry-cleaning shops; Schreiber et al. (2002) found deficits in visual contrast sensitivity among apartment residents and day-care workers in buildings with in-house dry cleaners; and Storm et al. (2011), building on New York State Department of Health data, found visual contrast sensitivity deficits in children residing in such buildings. EPA’s own review found that these effects occurred at residential exposure concentrations three- to one-hundred-fold lower than the occupational exposures producing comparable effects, a pattern RoseyFields’ assessors regard as more consistent with a genuine, low-threshold toxic effect than with confounding, since the confounding structures of an occupational dry-cleaning cohort and a residential apartment-dwelling population are quite different, yet the same class of deficit was observed in both.

Bladder cancer. Among the cancer outcomes, bladder cancer has the most internally consistent epidemiological signal. Effect estimates from the higher-quality exposure-assessment studies ranged from 1.44 to 4.03 (Calvert et al., 2011; Lynge et al., 2006; Blair et al., 2003; Pesch et al., 2000; Aschengrau et al., 1993), and Pesch et al. (2000), using a semiquantitative cumulative PCE exposure metric in a large German case-control study, found an adjusted odds ratio rising from 0.8 at medium exposure to 1.3 at high exposure to 1.8 at substantial exposure relative to low exposure—a genuine exposure-response gradient, a feature notably absent from most of the other cancer-site literatures discussed in Section 3. Relative risk estimates using a broader occupational job-title definition (ever worked as a dry cleaner or laundry worker) in four large Nordic cohort studies were more modest, ranging from 1.01 to 1.44 (Pukkala et al., 2009; Wilson et al., 2008; Ji et al., 2005; Travier et al., 2002), consistent with the expected dilution of a true effect when exposure is measured only by broad job title rather than by a PCE-specific cumulative metric. Confounding by smoking, a well-established independent bladder-cancer risk factor, is an unlikely explanation for these findings because Pesch et al. (2000) and several of the case-control studies explicitly adjusted for smoking without materially changing the association.

Study Quality Considerations

RoseyFields’ assessors evaluate every PCE epidemiological finding against the standard validity threats, and the PCE literature offers clear examples of several of them, most notably exposure misclassification.

Exposure misclassification is the single most consequential issue in the PCE epidemiological literature. Many of the older and larger cohort studies (e.g., the Nordic registry-linkage studies of Pukkala et al., 2009 and Ji and Hemminki, 2006) classify exposure only by broad occupational job title—”dry cleaner” or “laundry worker”—without any PCE-specific exposure concentration or duration data, and such studies necessarily dilute any true underlying association toward the null because job title captures a heterogeneous mix of PCE intensity, other solvent exposures, and, in many cases, workers who never had substantial direct PCE contact. This is precisely why the bladder-cancer and non-Hodgkin-lymphoma associations are strongest and most consistent in the subset of studies using PCE-specific semiquantitative or quantitative cumulative exposure metrics (Pesch et al., 2000; Seidler et al., 2007), and why RoseyFields’ assessors give considerably more evidentiary weight to those studies than to the broader job-title-based registry studies, notwithstanding the latter’s much larger sample sizes.

Confounding by co-exposure to other chlorinated solvents is a related and pervasive concern, since dry-cleaning, laundry, and aircraft-maintenance workers are frequently exposed to trichloroethylene and other structurally related solvents in addition to PCE, and few studies had the exposure resolution to fully separate PCE’s independent contribution; this is particularly relevant to the aircraft-maintenance cohorts (Radican et al., 2008; Boice et al., 1999), where trichloroethylene is often the dominant solvent of concern in the underlying facility, and RoseyFields’ assessors weigh findings from mixed-solvent occupational cohorts less heavily for PCE-specific causation than findings from cohorts (dry cleaners, laundry workers) whose primary chlorinated-solvent exposure is overwhelmingly PCE. Confounding by common lifestyle factors, principally smoking and alcohol use, is a less significant concern for the specific outcomes with the strongest PCE signal, since smoking and alcohol are not established risk factors for non-Hodgkin lymphoma (Morton et al., 2005; Besson et al., 2006), and the leading bladder-cancer studies explicitly adjusted for smoking.

Reverse causation is not a significant concern for the mortality and incidence cohort studies discussed above, given their prospective design, but is more relevant to the cross-sectional hematological findings of Emara et al. (2010), where a health condition affecting standard blood counts could conceivably influence workplace assignment or exposure duration in ways not fully captured by the study design.

The Cape Cod drinking-water cohort deserves particular methodological note because it illustrates both a strength and a limitation characteristic of environmental (as opposed to occupational) PCE epidemiology. Its principal strength is a genuinely novel, validated exposure-reconstruction method: water-distribution modeling software (EPANET) that incorporates pipe network geometry, water flow, and PCE leaching kinetics from vinyl-lined pipe to estimate individual-level exposure intensity and duration, an approach that has been refined across successive reanalyses (Aschengrau et al., 2003; Vieira et al., 2005; Gallagher et al., 2011). Its principal limitation is that, as with any single-site, single-source community cohort, its findings for site-specific outcomes such as birth defects and breast cancer have not yet been extensively replicated in other PCE-contaminated drinking-water communities, meaning RoseyFields’ assessors characterize these findings as an important suggestive signal from a methodologically sophisticated single cohort rather than as an independently replicated, multi-population finding of the kind that supports the neurotoxicity and bladder-cancer conclusions.

Weight-of-Evidence and Agency Determinations

Independent expert panels and regulatory agencies have reached materially convergent, though carefully qualified, conclusions regarding PCE’s principal human health hazards. EPA’s own 2012 IRIS Toxicological Review classifies PCE as “likely to be carcinogenic in humans by all routes of exposure,” a characterization EPA is explicit rests on “suggestive evidence of carcinogenicity in epidemiologic studies” combined with “conclusive evidence” of tumor induction in rats and mice—language that itself signals EPA’s own view that the animal evidence, not the human epidemiological evidence, carries the primary weight for the cancer classification. The International Agency for Research on Cancer, in its 2014 reevaluation (IARC Monograph Volume 106), classified tetrachloroethylene as Group 2A, “probably carcinogenic to humans,” based on sufficient evidence in experimental animals and limited evidence in humans, principally the bladder-cancer signal described in Section 4. The U.S. National Toxicology Program’s Report on Carcinogens lists tetrachloroethylene as reasonably anticipated to be a human carcinogen. Notably, and unusually for a chemical RoseyFields has reviewed, EPA’s own noncancer reference values (the oral RfD and inhalation RfC, both finalized in 2012) are themselves derived directly from human epidemiological data—the occupational neurotoxicity studies of Cavalleri et al. (1994) and Echeverria et al. (1995)—rather than from animal toxicology, a distinction that reflects just how much confidence EPA places in the human neurobehavioral literature relative to the cancer epidemiology. A National Research Council panel (NRC, 2010) that reviewed EPA’s draft assessment likewise concluded that mouse hepatocellular tumor data, rather than the human cancer epidemiology, should anchor the quantitative cancer risk estimates, with only a minority of panel members favoring use of the rat leukemia data instead.

Application to Litigation: General and Specific Causation

RoseyFields’ assessors use this epidemiological evidence, alongside the toxicological and pharmacokinetic methodology described in the companion PCE risk-assessment page, to support general and specific causation opinions, while drawing a clear distinction between the neurotoxicity and bladder-cancer outcomes—where the human evidence is genuinely strong—and the broader cancer and reproductive/developmental literature, where the evidence is more limited, more exposure-assessment-dependent, or drawn from a single cohort.

For general causation, the neurobehavioral/visual outcome described in Section 4 is unusually well supported for litigation purposes precisely because it is the one outcome for which EPA itself relied on human epidemiological data, rather than animal bioassay data, to derive its own quantitative noncancer toxicity values—meaning a general causation opinion for PCE neurotoxicity can be grounded in the same occupational and residential studies EPA itself used, reinforced by the cross-population, cross-route replication described in Section 4. The bladder-cancer general causation opinion is similarly well supported by the exposure-response gradient in Pesch et al. (2000) and the consistency across studies using PCE-specific exposure metrics, though RoseyFields’ assessors are careful to note that EPA and IARC’s carcinogenicity classifications themselves rest primarily on animal data, with the human epidemiology characterized as “suggestive” (EPA) or “limited” (IARC) rather than “sufficient.” For non-Hodgkin lymphoma, multiple myeloma, and the other cancer sites discussed in Section 3, RoseyFields’ assessors offer more cautious general causation opinions that explicitly disclose the exposure-assessment-dependent nature of the findings (stronger in PCE-specific-exposure studies, weaker or absent in job-title-based studies) described in Section 5. For the Cape Cod-derived reproductive and developmental findings (birth defects, breast cancer, reduced fecundity), RoseyFields’ assessors are explicit that these remain single-cohort findings pending broader replication, which affects both the strength of any general causation opinion and the case-specific applicability of the underlying dose-reconstruction methodology to plaintiffs exposed via different sources or routes.

For specific causation, epidemiological effect estimates from the strong-evidence-tier outcomes are combined with individualized exposure and dose reconstruction (as described in RoseyFields’ risk-assessment methodology), latency and exposure-timing analysis appropriate to the outcome, differential diagnosis addressing alternative causes (co-exposure to trichloroethylene and other solvents, smoking, family history, idiopathic disease base rates), and, where available, biomarker or clinical findings (color-vision testing, visual contrast sensitivity, documented urothelial pathology) that corroborate a biologically significant individual exposure. As with every RoseyFields work product, the tier of evidence supporting each specific opinion, and the exposure-assessment and study-quality limitations described in Section 5 as they apply to the individual case, are disclosed explicitly.

Bibliography