Ironing Out Welding Carcinogenicity?
In April 2017, IARC reclassified welding fumes as “known” to be carcinogenic to humans (group 1) from the previous classification as “possibly” carcinogenic in 1990 (group 2B). The full monograph 118 will probably be available next year. For now, you can read a summary in The Lancet.

Welding fume exposure is widespread and diverse. The Bureau of Labor Statistics estimates there are 400,000 welders in the U.S., and there are likely more jobs where workers do some welding. So IARC’s finding expands our view of the extent of exposures that pose a disease risk, both to bystanders at work and for the environment where the welding fume components are present. The working group noted that after exhaustive literature review,
[m]ost studies, including more than 20 case-control studies and nearly 30 occupational or population-based cohort studies, reported increased risks of lung cancer in welders or other workers exposed to welding fumes. Exposure-response associations with indices of longer or greater cumulative exposure to welding fumes were also reported in several studies, some of which were large, high-quality studies.
The monograph also noted consistent increases in kidney cancer, but opined that other non-fume exposures in the welding environment could not be ruled out as causes. Summaries of this literature note that the excess of lung cancer in these studies of welders was in the range of 25 percent, which translates to about 1 percent risk (10 in 1,000) of lung cancer at whatever summary measure of exposure can be derived from this body of studies, compared to the benchmark of 1 in 1,000 the Supreme Court defined as “significant” risk.
This calculation shows that any agent clearly carcinogenic in a study in people is a high priority risk. Practitioners require an occupational exposure limit (OEL) to evaluate the riskiness and need for additional controls in a specific environment. We should also consider how this new authoritative conclusion impacts our view of the protection conferred by OELs promulgated before this finding.
Carcinogen classification communicates risk, stratified by the certainty of the direct observations of the toxic potential of an agent, mixture, or exposure circumstance. The IARC monograph program, the National Toxicology Program's
Report on Carcinogens
, and the California Proposition 65 list are comprehensive resources. The IARC classifications, “known” (group 1), “probable” (group 2A), and “possible” (group 2B), contrast with the NTP classifications of “known” or “reasonably anticipated” (combining IARC 2A and 2B). The NTP terms are set by federal legislation. Most of the IARC 2B carcinogens got there via “sufficient” evidence in laboratory studies combined with “inadequate” evidence (usually no studies) in people. IARC 2A usually means there is “limited” evidence in people. IARC also lists about 500 agents as “not classifiable”; for many of these there is “limited” evidence in laboratory studies and no data in people. Personally, I think “reasonably anticipated” communicates more certainty of risk than “possibly.” NTP does not address “exposure circumstances,” such as welding.
The priorities for IARC reviews are set by a panel of scientific experts (
). The IARC process consists of working groups: scientists in the field without conflicts of interest, who review publications, draft the monograph, meet face to face over the course of 10 days to revise it, and vote on classification. Experts with conflicts may also be invited and can contribute to the discussion but have no vote.

The priorities for listing in the
Report on Carcinogens
are set by NTP staff and by petition. Typically, the ROC lags classifications by IARC, and may not include all the agents classified by IARC. The NTP review process consists of staff (who are scientists) compiling literature and proposing a classification, followed by external peer review by independent scientists and a public meeting at which stakeholders can comment. Following the external peer review, the listing is voted by agency heads and reviewed by the office of the secretary of the Department of Health and Human Services (which is subject to lobbying). The political review explains why the “annual” report on carcinogens became a report without an interval. In my view the IARC process is stronger in epidemiology than NTP’s, and usually precedes consideration by NTP. EPA will also assess carcinogenic potential under its IRIS program, as will ACGIH, but these classifications have little direct relationship to OEL recommendations.
IARC’s assessment of laboratory evidence for carcinogenicity is straightforward. Usually it’s considered “sufficient” if two experiments observe increased tumors. Laboratory evidence was lacking for welding fume. Sufficient evidence in people requires multiple studies, although usually less than the 50 welding fume studies noted in the IARC summary quoted above. The 1990 Monograph 49 on welding fume cited nine cohort studies (two were considered null) and 12 case control studies (again, two considered null). Those data were determined to be “limited.” A detailed comparison of the new evidence that carried welding fume from “possibly” to “known” will have to wait for the full Monograph 118 to be published. This evolution took 18 years, and maybe 30 major studies, which is 18 years’ delay in taking the original evidence into account.
Public health regulation has retreated from a default of banning carcinogens or regulating carcinogen exposure to the lowest feasible level.
is a professor in the CUNY School of Public Health in New York. He can be reached at (212) 396-7782 or
via email
Why does classification of carcinogenic potential have such an impact? Because of its potential role in setting OELs. Public health regulation has retreated from a default of banning carcinogens or regulating carcinogen exposure to the lowest feasible level. But it’s still generally accepted that carcinogen risk rates can be extrapolated from the limit of direct observation, 1 in 10 for a bioassay or maybe 1 in 100 in a study of people, to the 1 in 1,000 or 10,000 benchmark applied by OSHA in defending new PELs. Cancer is likely not the lowest exposure effect of an agent, but it provides leverage.
Industry disputes of classifications usually aim to stop estimates of potency—the fraction of population affected at an exposure level—which are needed for a robust OEL. Welding fume contrasts with most of these disputes, which center on action needed in response to group 2B or even group 3 carcinogens. Welding fume is carcinogenic in people with little data in laboratory studies.
Increased lung cancer is observed in welders of mild steel, where the predominant exposure is iron oxide. While studies of stainless steel welders with exposure to chrome (VI) and nickel (both known human carcinogens) find excesses, exposure to mild steel welding fume is also associated with increased lung cancer. The status of iron oxide as a group 3 carcinogen should be reconsidered. IARC opined that iron oxide and iron ore dust were not classifiable in 1972 and 1987 based on virtually no inhalation data. However, mechanistically the carcinogenicity of nanoparticles with the oxidative potential of iron oxide should be assumed. There are still no published chronic studies of iron oxide particles, although short-term studies indicate oxidative activity. The “trace” iron oxide exposure of urban air pollution may contribute to the observation that particulate air pollution is known to be carcinogenic.
The 1988 NIOSH Criteria Document for welding fume, rarely quoted, cites most of the studies cited by the 1990 IARC Monograph but gives no summary statement on the evidence for carcinogenicity. This document’s low impact demonstrates the importance of a summary statement. However, the
NIOSH Pocket Guide to Chemical Hazards
indicates welding fumes are carcinogenic. NIOSH declined to recommend an exposure limit for welding fume beyond the limits for individual constituents.
Existing exposure limits for welding fume were set without reference to carcinogenicity. There is no OSHA PEL for welding fume, although a 5 mg/m3 PEL was proposed in 1989 and then overturned by an industry lawsuit. ACGIH has deleted a previous TLV of 5 mg/m3, although its TLV of 3 mg/m3 for respirable particulates not otherwise specified (PNOS) would appear to apply.
Without the horsepower to compel a PEL or enforceable OEL in the current environment, practicing IHs should measure exposures down to levels that may plausibly be determined to pose a risk. Since welding fume and iron oxide fume are essentially all nanoparticles, I suggest that air sampling methods be sensitive to exposures as low as 100 µg/m3 respirable for total welding fume and iron oxide, by analogy to my
general recommendation for particles