A systematic review of the epidemiology evidence on talc and cancer

Abstract

Over the past several decades, there have been many epidemiology studies on talc and cancer published in the scientific literature, and several reviews and meta-analyses of talc and respiratory, female reproductive, and stomach cancers, specifically. To help provide a resource for the evaluation of talc as a potential human carcinogen, we applied a consistent set of examination methods and criteria for all epidemiology studies that examined the association between talc exposure (by various routes) and cancers (of various types). We identified 30 cohort, 35 case-control, and 12 pooled studies that evaluated occupational, medicinal, and personal-care product talc exposure and cancers of the respiratory system, the female reproductive tract, the gastrointestinal tract, the urinary system, the lymphohematopoietic system, the prostate, male genital organs, and the central nervous system, as well as skin, eye, bone, connective tissue, peritoneal, and breast cancers. We tabulated study characteristics, quality, and results in a systematic manner, and evaluated all cancer types for which studies of at least three unique populations were available in a narrative review. We focused on study quality aspects most likely to impact the interpretation of results. We found that only one study, of medicinal talc use, evaluated direct exposure measurements for any individuals, though some used semi-quantitative exposure metrics, and few studies adequately assessed potential confounders. The only consistent associations were with ovarian cancer in case-control studies and these associations were likely impacted by recall and potentially other biases. This systematic review indicates that epidemiology studies do not support a causal association between occupational, medicinal, or personal talc exposure and any cancer in humans.

Keywords:

• Talc
• systematic review
• cancer
• study quality
• epidemiology

Introduction

Talc is a naturally occurring hydrous magnesium silicate mineral (US EPA 1992; Fiume et al. 2015). The mineral content, morphology, and purity of talc varies based on the geological conditions under which it is created and its excavation (Drechsel et al. 2018). Talc is classified into two grades. Cosmetic talc is used in cosmetics, personal care products, pharmaceuticals, and food.¹ It is currently > 98% pure and has historically been obtained from mines with > 95-99% platy talc free of other minerals (IARC 2010; Drechsel et al. 2018). Industrial talc can be, but is not always, less pure and may be comprised of a variety of fibrous morphologies and other minerals (Drechsel et al. 2018; IARC 2010).

People are generally exposed to talc in one of three ways: (1) occupational exposure in talc mining or industrial applications of talc; (2) through use of cosmetic or personal hygiene products; or (3) ingestion via food or pharmaceuticals. Occupational exposure can occur directly though the mining and milling of talc, when talc is extracted from the earth and separated from other minerals (IARC 2010; Drechsel et al. 2018). Notably, millers and miners are also exposed to other minerals present in the rock deposit being mined, and the makeup of these other minerals is unique to each location (IARC 2010). Occupational talc exposures can also occur during the manufacturing of paper, plastics, paint, ceramics, rubber, roofing materials, patching compounds, flooring products and fertilizers (IARC 2010). Talc is a common ingredient in cosmetic and personal hygiene products, and has historically been used in body or baby powders, and continues to be used in other cosmetics, creams and lotions, soaps, and pharmaceuticals. Talc is used as an anti-stick or anti-caking agent in foods for both humans and livestock. It is used in gum, candies, and cured meats (IARC 2010), and has been used on rice to delay spoiling and prolong shelf-life (Stemmermann and Kolonel 1978).

Over the past several decades, there have been many epidemiology studies on talc and cancer published in the scientific literature. These studies have reported mixed results. The International Agency for Research on Cancer’s (IARC) most recent review of talc concluded that epidemiological studies provided “limited evidence in humans for the carcinogenicity of perineal use of talc-based body powder,” and classified this use as possibly carcinogenic to human beings (i.e. group 2B) (IARC 2010). In the same review, IARC (2010) concluded that there was “inadequate evidence in humans for the carcinogenicity of inhaled talc not containing asbestos or asbestiform fibers” and that overall “inhaled talc not containing asbestos or asbestiform fibers is not classifiable as to its carcinogenicity (Group 3).” Health Canada identified ovarian cancer as a potential concern for human health from perineal exposure to talc (Environment and Climate Change Canada; Health Canada 2021), and the National Institute of Public Health and the Environment of the Netherlands recently proposed to the European Chemicals Agency that talc should be classified as a Category 2 carcinogen (i.e. suspected human carcinogen) under the Globally Harmonized System of Classification and Labeling of Chemicals (Netherlands RIVM 2023). Regulatory agencies, such as the US EPA and the National Toxicology Program, have not classified the human carcinogenicity of talc. IARC will review talc again in June 2024 at the IARC Monographs Meeting 136 (IARC, 2024).

Several reviews, including meta-analyses, have been conducted that evaluate talc and respiratory, female reproductive, and stomach cancers. The results and conclusions of these reviews vary considerably, with some concluding associations are likely causal (e.g. Taher et al. 2019) and others concluding that they are not (e.g. Goodman et al. 2020; Lynch et al. 2022, Lynch et al. 2023). These reviews and meta-analyses have considered and incorporated individual study quality in various ways (e.g. using Newcastle Ottawa Scale or Institute of Medicine [IOM] criteria); however, many have not fully considered the implications of study quality on the interpretation of the results of individual studies or across studies as whole. For example, Taher et al. (2019) assessed the quality of studies evaluating perineal talc use and ovarian cancer using the Newcastle Ottawa Scale, but only used that quality score in a single analysis where the results were stratified by scores < 7 and ≥ 7, without considering the impact of specific limitations on the interpretation of results. This is problematic because meta-analyses and reviews cannot overcome limitations in the original research studies, particularly with regard to confounding and exposure misclassification. It is possible that a study could have one critical aspect of study quality that trumps all high-quality aspects, and this would be missed by stratifying by a score, without further analysis.

Most reviews of talc and cancer focus on female reproductive and respiratory cancers. We are not aware of any review of talc that systematically evaluates the evidence for all cancer types, and the regulatory process requires a consideration of all potential cancer types. To help provide a resource for the overall evaluation of talc as a potential human carcinogen, we aimed to apply a consistent set of examination methods and criteria for all epidemiology studies that examined the potential connection of talc exposure (by various routes) and cancers (of various types). Given the number of cancer types evaluated, our goal was to tabulate study characteristics, quality, and findings in a systematic and transparent manner, and to provide an overview of the results and their interpretation in the context of study quality.

Methods

The protocol for this systematic review was registered with Open Science Framework (OSF) on January 23, 2024 (https://osf.io/sd46p).

Literature search

The identification of individual studies was guided by a population, exposure, comparator, outcomes, and study design (PECOS) statement, as follows:

In studies of human populations (P), what is the risk of having or dying of cancer (O) for each unit higher exposure to talc (C1/E1) or among those who have ever been exposed to talc (E2) compared to those who have not (C2), as observed in primary observational studies with data at the individual level (S)?

Corresponding inclusion and exclusion criteria are shown in Appendix A Table A1.

Table 1. Bradford Hill analysis of talc and ovarian cancer.

Bradford Hill Consideration	Analysis
Strength	With few exceptions, risk ratios for ovarian cancer and ovarian cancer subtypes reported in cohort studies are < 1.2, and many are < 1. Associations in case-control studies are modest (generally < 1.5).
Consistency	The results of epidemiology studies are not consistent across study designs. Cohort studies, which are not subject to recall bias, consistently report no association for ovarian cancer overall and for most ovarian cancer types. The exception is a small association with serous ovarian cancer in the Nurses’ Health Study (Gates et al. 2010; Gertig et al. 2000). Case-control studies report largely consistent, small, positive findings, but these findings are likely due to similar sources of bias (e.g. recall bias and selection bias). Goodman et al. (2024) demonstrated that even a small amount of recall bias, equal to or less than seen in real world recall of talc use (O'Brien et al. 2023), could have led to statistically significant associations in these studies.
Specificity	The biological processes that result in the development of ovarian cancer are likely complex and not well understood, and there is no evidence to suggest that there is a specific relationship between talc and ovarian cancer.
Temporality	Talc use generally occurred prior to diagnosis.
Dose-Response	There is no consistent exposure-response relationship in the epidemiology studies of talc and ovarian cancer. Case-control studies that evaluated exposure-response and accounted for frequency of application, duration of talc use, or both had mixed results. There were no statistically significant exposure-response relationships found in any of the cohort studies. There was also no increased cancer risk with the use of diaphragms stored in talc, which would be expected to yield greater talc transport to the ovaries (if it occurs) versus external use. Exposure misclassification from both random and systematic error has not been ruled out in the case-control studies that reported exposure-response relationships. None of the studies had any quantitative information on the level of talc exposure per use or any quantitative estimates of the amounts of talc that are could be transported to various locations in the reproductive tract. Biomonitoring studies do not provide evidence that perineal talc use is correlated with the presence of talc particles in ovarian tissues, and there is no relationship between the number of particles found in ovarian tissue and incidence of tumors (Goodman et al. 2020; Prueitt et al. 2024).
Biological Plausibility	There is no biologically plausible mechanism by which talc could cause cancer in humans. Talc is not genotoxic or mutagenic, but can induce some effects that could be events on a possible pathway to carcinogenicity. These effects are mainly at high exposures or in in vitro studies with exposures of unclear relevance in vivo, but these effects are not consistent across studies and cell types. The evidence for retrograde transport of talc from perineal applications to the ovaries is insufficient in both animals and humans, and there is no relationship between particles found in ovaries and talc use or tumor incidence. There is no evidence that talc applied to the perineum causes inflammation in the reproductive tracts of women (Prueitt et al. 2024).
Coherence	The epidemiology evidence for an association between talc and ovarian cancer is not consistent across study types (and recall bias may explain associations in case-control studies). As discussed in Prueitt et al. (2024), talc cancer bioassays evaluate many exposure routes, species, and exposure durations. None of these studies indicate that talc is a carcinogen except in rats under conditions of extremely high exposure that likely resulted in lung particle overload. This is a nonspecific effect of high exposures to poorly soluble particles, and not from any carcinogenic properties of talc, and has not been demonstrated in humans. Mechanistic studies do not support biological plausibility. Collectively, the evidence supporting causation does not fit together in a coherent manner.
Experiment	There is no experimental evidence on talc and ovarian cancer in humans.
Analogy	There is no evidence that any substance analogous to talc causes ovarian cancer.

Note: Underlined text is the ultimate conclusion for that Bradford Hill consideration.

Literature searches were conducted using Scopus and PubMed to identify relevant studies published through January 23, 2024. Search terms are listed in Appendix A Table A2. We also reviewed the reference lists of identified studies and relevant reviews and meta-analyses we identified.

Table 2. Bradford hill analysis of talc and lung cancer.

Bradford Hill Consideration	Analysis
Strength	Many risk ratios were small in magnitude (<1.3). Some studies reported risk ratios around 3, but these were not consistent with weaker estimates reported in the same studies. This is particularly the case for studies of miners and millers, who likely had the highest exposures. Risk estimates were weak and close to 1 in the studies in Val Chisone and Altermark and Knarrevik. Estimates were slightly higher in GTC, but stratified analyses demonstrated that risks were highest among those who had other known prior risks for lung cancer. Katsnelson and Mokronosova (1979) reported higher risks for men but did not provide CIs. Some higher risk ratios were reported, but these were all based on a small number of exposed cases (<10). None of these studies adjusted for smoking.
Consistency	Results were not consistent within and across studies of talc and lung cancer risk. Among mining and milling populations, which may have had the highest potential exposures, there were inconsistencies in the strength and significance of the risk estimates, with some studies reporting weak risk estimates and no statistically significant associations, and others reporting statistically significant increased risks, but generally in specific subgroups (e.g. those who had other known prior risks for lung cancer).
Specificity	Most lung cancers are caused by smoking, and there are several other known causes (e.g. arsenic). There is no evidence to suggest a specific relationship between talc and lung cancer.
Temporality	Talc exposures generally occurred prior to diagnosis in case-control and cohort studies.
Dose-Response	No trends were found in most exposure-response analyses in epidemiology studies of talc and lung cancer, including in the studies of miners and millers. Exceptions were increasing risk with increasing duration of exposure in one ceramic plumbing casting and maintenance plant worker cohort and possibly in one of two rubber plants. These analyses were not adjusted for smoking.
Biological Plausibility	There is no biologically plausible mechanism by which talc could cause cancer in humans. Inhaled talc is rapidly cleared from the lung. Talc is not genotoxic or mutagenic, but can induce some effects that could be events on a possible pathway to carcinogenicity, mainly at high exposures or in in vitro studies with exposures of unclear relevance in vivo, but these effects are not consistent across studies and cell types (Prueitt et al. 2024).
Coherence	The epidemiology evidence does not provide evidence of a consistent, significant increased association between talc and lung cancer. As discussed in Prueitt et al. (2024), talc cancer bioassays evaluate many exposure routes, species, and exposure durations. None of these studies indicates that talc is a carcinogen except in rats under conditions of extremely high exposure that likely resulted in lung particle overload. In this study, the micronized talc likely deposited deeper in the lung and had a longer retention time, higher lung burden, and greater potential to elicit adverse effects in the lungs. Thus, this outcome is a nonspecific effect of high exposures to poorly soluble particles, and not from any carcinogenic properties of talc itself. These observations have not been demonstrated in humans. Mechanistic studies do not support biological plausibility. Collectively, the evidence supporting causation does not fit together in a coherent manner.
Experiment	There is no experimental evidence on talc and lung cancer in humans.
Analogy	There is no evidence that any substance analogous to talc causes lung cancer.

Note: Underlined text is the ultimate conclusion for that Bradford Hill consideration.

When we found more than one study of the same population and outcome, we extracted information from the most recent study with the longest follow-up or the study or studies (even if older) reporting the most informative data (e.g. greater population coverage, more reliable exposure, confounder, and outcome estimates, and/or longer duration of follow-up). In these instances, we extracted results relevant to the same population from multiple studies and have included footnotes in the tables to indicate which results were extracted from which studies. Older studies that reported results for similar analyses as newer studies are also referenced in the tables for completeness, but results were not extracted from these studies.

Study Selection and data collection

Titles, abstracts, and full article texts, as appropriate, of the relevant studies identified from the systematic literature search were independently screened by two reviewers (LME and KJC) and any discrepancies were resolved by a third reviewer (DNB). Eligible studies were selected based on the PECOS statement. The study screening process was managed using the systematic review software tool, Covidence (https://www.covidence.org/), which captured the results of the study screening and reasons for study exclusion.

For each included study, one reviewer independently extracted data (e.g. study characteristics, study results) and evaluated study quality, this was checked for accuracy by another reviewer and any discrepancies were resolved through discussion. Extracted data were stored and organized in Excel or Word tables.

If a study reported multiple exposure-outcome pairs of interest, each exposure-outcome pair was recorded as a separate record. If multiple effect estimates were reported for a single exposure-outcome pair, only the most adjusted one was extracted, unless the purpose of the most adjusted model was to evaluate potential mediation, effect modification, or sensitivity of the main study result, in which case a less adjusted one was extracted. In addition, when subgroup effect estimates were available, we also extracted those data.

For each individual study included in the review and for each exposure-outcome pair, the following study characteristics were extracted: citation (author, year), study design (cohort/case-control), study population (country, age, sex, preexisting conditions, cohort or sources of cases and controls, sample size, exposure (period, source, ascertainment), and outcome (cancer type, incidence or mortality, ascertainment, follow-up period [cohort], year[s] of case identification [case-control]). In addition, the following information related to study results were extracted: citation, study design, incidence or mortality, risk metric, referent group, exposure group, number of exposed cases, number of expected cases or exposed non-cases, risk estimates and confidence intervals (CI), p-trend, and covariates controlled for.

Study quality evaluation

Prior to evidence synthesis, we evaluated study quality across the domains most likely to impact the interpretation of study results to identify study strengths and any threats to validity, in order to determine how accurate and reliable the results of each study are for addressing the research question. For each individual cohort and case-control study included in the review, specific aspects of study quality were evaluated as higher (+) or lower (−) quality according to the criteria described below and summarized briefly in Appendix A Table A3. We based these criteria on those described in several frameworks, including Newcastle Ottawa Scale the ROBINS-I tool, the National Toxicology Program (NTP), Office of Health Assessment and Translation (OHAT) Risk of Bias Rating Tool for Human and Animal Studies, and the United States Environmental Protection Agency (US EPA) Office of Research and Development (ORD) Staff Handbook for Developing Integrated Risk Information System (IRIS) Assessments (US EPA 2022). If a talc cohort or case-control study met the criteria for higher quality as defined in Table A3, we recorded a “+”; otherwise, where studies were limited by lower quality we recorded a “−” in the quality assessment tables. These tables are intended to demonstrate study quality at a very high level, in a dichotomous manner. The impact of different study quality aspects can vary within each individual study, and we discussed those aspects with the largest impacts in the text, below.

Table 3. Bradford hill analysis of talc and upper gastrointestinal tract cancer.

Bradford Hill Consideration	Analysis
Strength	Risk ratios have a wide range of magnitudes, including among mining and milling populations (∼0-10). Many of the extreme estimates (<0.5, and >2) were based on very few exposed cases or were in analyses that didn’t adjust for smoking or alcohol consumption.
Consistency	Risk estimates are not consistent across all studies or within occupational groups, such as miners and millers. There were no reported associations that adequately adjusted for smoking or alcohol consumption.
Specificity	Smoking and alcohol consumption can cause upper gastrointestinal cancers (Scherübl 2022). There is no specific relationship between talc and upper gastrointestinal system cancer.
Temporality	Talc exposure generally occurred prior to diagnosis in case-control and cohort studies.
Dose-Response	There are no consistent exposure-response relationships in epidemiology studies of talc and upper gastrointestinal tract cancers. The only statistically significant dose-response trend reported was an inverse trend for esophageal cancer in the Val Chisone millers and miners in analyses of years of employment. Most of the dose-response analyses, including in Val Chisone workers, were limited by small numbers of exposed cases.
Biological Plausibility	There is no biologically plausible mechanism by which talc could cause cancer in humans. Inhaled talc is rapidly cleared from the lung and there is no evidence that talc translocates from the site of exposure to other organs. Talc is not genotoxic or mutagenic, but can induce some effects that could be events on a possible pathway to carcinogenicity, mainly at high exposures or in in vitro studies with exposures of unclear relevance in vivo, but these effects are not consistent across studies and cell types (Prueitt et al. 2024).
Coherence	The epidemiology evidence for an association between talc and upper gastrointestinal cancer is not consistent. As discussed in Prueitt et al. (2024), talc cancer bioassays evaluate many exposure routes, species, and exposure durations. None of these studies indicates that talc is a carcinogen except in rats under conditions of extremely high exposure that likely resulted in lung particle overload. This outcome is a nonspecific effect of high exposures to poorly soluble particles, and not from any carcinogenic properties of talc, and has not been demonstrated in humans, and is likely not relevant to the gastrointestinal tract. Mechanistic studies do not support biological plausibility. Collectively, the evidence supporting causation does not fit together in a coherent manner.
Experiment	There is no experimental evidence on talc and upper gastrointestinal tract cancer in humans.
Analogy	There is no evidence that any substance analogous to talc causes upper gastrointestinal tract cancer.

Note: Underlined text is the ultimate conclusion for that Bradford Hill consideration.

Exposure assessment

While direct measurements of talc concentrations to which study participants were exposed over time would be ideal, as they provide actual quantitative measurements of an individual’s exposure, these data are not available for personal care product use and are uncommon in occupational studies. In the absence of directly measured exposures, most studies rely on indirect measurements to estimate talc exposures, the accuracy and reliability of which varies both by the method of measurement and the expertise underlying the method (e.g. job histories can be self-reported or obtained from historical records and the linkage of exposures associated with a job history can be self-reported or linked to a job exposure matrix (JEM) by an industrial hygienist with specific expertise on potential exposures. For occupational exposures, in the absence of directly measured exposures, we considered a JEM to be more reliable than some other means of estimating exposure (e.g. categorization based solely on occupational industry) in occupational settings, but the quality of an exposure assessment based on a JEM varies based on the expertise of the person(s) linking job titles to exposures, and whether the assessment is semi-quantitative or not. Exposure estimates based on job titles or industries, including those used in JEMs, are subject to potential misclassification due to possible differences in tasks and exposure conditions for specific jobs. Using these (or other) broad categories for exposure assessments also does not account for individual job characteristics that could modify an individual’s exposure potential. They also usually do not account for differences in exposure duration, frequency, or intensity.

We considered studies that estimate duration of exposure to talc by length of employment, or by using information from self-reported work history, census data, or company records as proxies for exposure dose to be of higher quality than those that do not, but these exposure estimates may still be subject to misclassification and recall inaccuracies. Similarly, we considered studies that evaluated duration, intensity, or the time-varying nature of personal care product exposure to be stronger than those that do not.

Self-reported exposure information (for both occupational, personal care, or ingestion exposures) can be subject to potential recall inaccuracy due to the long time periods between talc exposure or use and interviews. Reliability and accuracy of recalled information can vary considerably and can be a function of how frequently information on exposure or use is collected. Prospective collection of exposure information is less likely to be subject to recall bias than retrospectively collected information. Recall bias can be of particular concern in case-control studies, where participants are selected into the study based on their disease status. Also, studies that use exposure estimates based solely on the presence of talc in air or personal care products do not account for individual exposure variations due to behavior or environmental differences, and these studies do not provide exposure estimates for individuals that are as reliable as those studies that do account for these exposure variations. Finally, exposures estimated for a short period of time (e.g. use of personal care products in past 12 months, occupational talc miner for at least 1 year) may not reflect long term exposures.

Outcome assessment

Self-reported outcomes may be inaccurate or incomplete, while those confirmed by medical professionals or in medical records, death certificates, or registries are less likely to be subject to misclassification and more likely to be reliable and complete. A sufficient duration of follow-up or time between exposure and outcome must be considered with respect to the latency of the disease being examined (Gordis 2014). We considered studies that considered a minimum latency period for solid tumors of 4 years and 6 months for hematologic or lymphatic cancers to be of higher quality (CDC 2014). We chose shorter latency periods to err on the side of overestimating quality. We also considered studies that evaluate aggregated cancer types (e.g. all cancers combined) to generally not be appropriate when the cancers have different underlying etiologies (Foster et al. 2022).

Confounding/covariate consideration

A confounder is associated with an exposure, but is not caused by that exposure, and can cause the disease of interest. Confounding bias can occur when a confounder is not fully accounted for in an analysis. For example, smoking is associated with alcohol consumption and can cause lung cancer. One might find an association between alcohol consumption and lung cancer because people who drink more alcohol smoke more, and not because drinking alcohol causes lung cancer. We concluded that no observational study is truly high quality with respect to confounders, because residual confounding can occur (Sterne et al. 2016). Also, quality among studies can vary based on what potential confounders or covariates are considered. Sex, age, height/weight/obesity, genetic factors, prior or family history of cancer, dietary factors, smoking, and alcohol consumption are important factors to consider in most cancer epidemiology studies because they have the potential to bias study results if they are not appropriately addressed (Arem & Loftfield 2018). Confounders specific to a cancer type are also important to consider (e.g. work place exposures to other minerals or metals in mining occupations can be risk factors for lung cancer, [ACS 2023a]); therefore, studies of talc miners and lung cancer would need to adequately control for these other exposures. Other risk factors for specific disease outcomes that may not be confounders per se (e.g. hysterectomy for ovarian and uterine cancers or menopausal status, hormone replacement therapy, or BRCA1 mutations for breast cancer) should also be considered as covariates in studies of those specific diseases. Common risk factors by cancer type are listed in Appendix A Table A4.

Very few studies have information on all potentially important confounders. Because we were concerned with relative quality, we concluded that studies that controlled for at least two key risk factors (at least one of which was smoking for lung cancer and bladder cancer), and age and sex were deemed higher quality, or lower risk of serious confounding. We considered cohort studies that assessed covariates that may vary over time (e.g. smoking) and consider changes over time in their analyses to be stronger than studies that only assessed covariates that were measured or estimated at a single time-point.

Most cancers have different etiologies and risk factors; this is also true for some subtypes of certain cancers. As a result, we concluded that studies in which groups of cancers that have different etiologies are combined as the outcome are likely to have uncontrolled confounding in these aggregated analyses, as it is unlikely that all relevant confounders for all of the exposure-outcome relationships are controlled for.

Sample Selection

We determined that stronger studies include participants that represent the larger population that researchers are evaluating. Studies that have higher rates of enrollment or retention (>80%) and in which the rates are non-differential by exposure or outcome are less likely to have selection bias and are also considered stronger (Kristman et al. 2004). If a study recruited or retained participants that systematically differ from the population of interest, or if the exposed and unexposed (cohort study) or diseased and non-diseased (case-control study) differ on important factors, risk estimates may be impacted (Gordis 2014).

Evidence synthesis

Evidence was synthesized for each cancer outcome. The consistency of results within and across individual studies and study quality is discussed in the text for cancers for which we identified at least three studies conducted in independent populations. Guided by the Bradford Hill considerations (Hill 1965), we determined the overall plausibility of causality with respect to talc and each cancer outcome for which there was sufficient evidence to evaluate (i.e. for which findings were positive or mixed among at least three studies), and for which there were at least 10 exposed cases in the studies with positive or mixed findings.² We did this through the evaluation of strength of association, consistency, specificity, temporality, dose-response, biological plausibility, coherence, experiment, and analogy, keeping study quality and relevance into account. We also considered the conclusions from the companion study on experimental animal and mechanistic evidence to assess biological plausibility and coherence (Prueitt et al. 2024). The reporting of this systematic review was guided by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) checklist (Appendix B) (PRISMA 2022).

Results

Study Selection

As shown in Appendix C, we identified 488 records from PubMed and 424 from Scopus, 346 of which were duplicates. After screening titles and abstracts of the remaining 566 records, we excluded 477 studies. The primary reasons for exclusion at this stage were that the study did not examine talc as an exposure, report an exposure contrast, examine any cancer outcome, or that the study was a design we decided a priori to exclude (e.g. a case-series, review, in vitro study, animal study, letter to the editor, or commentary). We reviewed the full text of the 89 remaining records. We identified an additional 10 studies from the reference lists of reviews. Ultimately, 77 studies met our inclusion criteria. Of these, 30 were cohort studies, 35 were case-control studies, 11 were pooled studies, and one study reported results from both cohort and a pooled nested-case-control analysis. These studies evaluated talc and cancers of the respiratory system, including mesothelioma, lung, and laryngeal cancer; the female reproductive tract, including ovarian, cervical and uterine/endometrial cancers; the gastrointestinal tract, including oral/pharyngeal, esophageal, stomach, colorectal, liver, gallbladder, bile duct, and pancreatic cancer; the urinary system, including bladder and kidney cancers; the lymphohematopoietic system, including leukemia, lymphoma, myeloma, and thyroid cancer; male cancers, including prostate and male genital organ cancers; the central nervous system (CNS) and brain; as well as skin, eye, bone, connective tissue, peritoneal, and breast cancers. Several studies also evaluated several or all cancers combined. Table C1 in Appendix C provides a summary of the studies included, what endpoints were examined in each, and which studies overlapped in terms of populations and outcomes.

Study characteristics and quality

Characteristics of cohort and case-control studies are described in Appendix D Tables D1 and D2, while pooled cohort and pooled case-control study characteristics are described in Appendix D Tables D3 and D4. Cohort and case-control study quality is shown in Appendix D Tables D5 and D6. Occupational studies were conducted in populations that included talc millers and miners, workers in the rubber, ceramic plumbing and fixture, paper plant, and fiberglass industries, while other studies were conducted in people who used cosmetic talc. One study evaluated medicinal talc exposure and another evaluated exposure to talc on gloves and surgical coverings.

Observational epidemiology studies encounter many challenges where uncontrolled or unmeasured, but potentially variable, factors limit the clarity and certainty of conclusions that may be drawn. It is necessary to recognize these as they bear on each study examined, to consider the possible impacts on apparent outcomes, and to gauge the limits on study-specific conclusions. Overall conclusions are best reached by relying most heavily on studies relatively less affected by these challenges, and to seek consistent patterns across studies to serve as evidence that the potential challenges have not misled the analysis. As such, we have provided an overview of the studies in the context of their quality, highlighting strengths and weaknesses. This is followed by a discussion of study findings by cancer type organized in organ systems.

Most cohort studies had appropriate comparison groups and a high enrollment or retention rate. All but five studies (Bulbulyan et al. 1999; Houghton et al. 2014; Crawford et al. 2012; Negri et al. 1989; Zhang et al. 1989) assessed talc-specific exposures and only two (Thomas 1990; Thomas and Stewart 1987) did not assess exposures prior to cancer diagnosis. About half assessed exposure using validated questionnaires, detailed occupational histories, or a JEM based on high-quality occupational information (e.g. company records, industrial hygiene data). Some studies collected exposure information at multiple timepoints and most considered the frequency, dose, or duration of exposure. The occupational cohort studies most frequently relied on prospectively collected company records, which may be more reliable and accurate than the self-reported estimates provided in most of the studies that evaluated personal care products containing talc. Three studies included some self-reported cancers that were not medically confirmed (O'Brien et al. 2019, 2021b; Gonzalez et al. 2016); all other studies included cancers that were physician-diagnosed, recorded in a registry, or reported by the individual or a proxy but validated clinically. All studies except for three (Stille and Tabershaw 1982; Lamm et al. 1988; Zhang et al. 1989) included sufficient follow up time. While several studies included results for all cancers or other aggregated groups of cancers, all of the cohort studies reported results for specific (non-aggregate) cancer types or cancers with similar etiologies. However, while most studies used proper statistical models, none of the occupational cohort studies and few of the studies that evaluated personal care products containing talc sufficiently considered key confounders, or collected information on time-varying covariates at multiple time points.

Most case-control studies had appropriate comparison groups but low enrollment, and enrollment often differed between cases and controls. All but five studies (Chen et al. 1992; Ke and Shunzhang 1999; Leung et al. 2023; Nielsen et al. 1994; Schildkraut et al. 2016), assessed talc-specific exposures but only six used data on exposures that were recorded prior to cancer diagnosis/mortality (Chiazze et al. 1993; Gamble 1993; Ke and Shunzhang 1999; Langseth and Kjaerheim 2004; Leung et al. 2023; Nielsen et al. 1994). About one-third of the studies considered the frequency, dose, or duration of exposure. Most case-control studies had high-quality outcome assessments; only one study (Chiazze et al. 1993) did not report on cancers that were physician-diagnosed, recorded in a registry, or reported by the individual or a proxy but validated clinically. All studies reported results that were for specific (non-aggregate) cancer types or cancers with similar etiologies, but only about half of the studies included a sufficient duration of time between the exposure and the outcome. Most case-control studies used appropriate statistical models, but about one-third did not fully consider key confounders.

With respect to the populations being studied, we reviewed 15 occupational studies of talc miners and millers. Of these, 14 were cohort studies (Brown et al. 1990; Ciocan et al. 2022a; Fordyce et al. 2019; Pira et al. 2017; Wergeland et al. 2017; Coggiola et al. 2003; Honda et al. 2002; Brown et al. 1990; Lamm et al. 1988; Stille and Tabershaw 1982; Katsnelson and Mokronosova 1979; Rubino et al. 1979; Rubino et al. 1976; Wild et al. 2002) and one was a case-control study (Gamble 1993). The mining and milling populations worked at the Gouverneur Talc Company (GTC) in upstate New York (Honda et al. 2002; Gamble 1993; Brown et al. 1990; Lamm et al. 1988; Stille and Tabershaw 1982), the Val Chisone region in Northern Italy (Ciocan et al. 2022a; Pira et al. 2017; Coggiola et al. 2003; Rubino et al. 1979; Rubino et al. 1976), Vermont (Fordyce et al. 2019), the former USSR (Katsnelson and Mokronosova 1979), the Altermark mine and Knarrevik mill in Norway (Wergeland et al. 2017), or France and Austria (Wild et al. 2002). All talc mining and milling cohorts were exposed to high levels of talc, especially at the early stages of follow-up, but the talc deposits varied in composition of other minerals. Some deposits had no detectable level of asbestos or asbestiform particles (e.g. Ciocan et al. 2022a; Fordyce et al. 2019; Wergeland et al. 2017), while other deposits naturally co-occur with asbestiform minerals and silica (e.g. Katsnelson and Mokronosova 1979).

We identified four occupational studies of rubber workers. Three of these were cohort studies (Straif et al. 2000; Negri et al. 1989; Zhang et al. 1989) and one was a case-control study (Ke and Shunzhang 1999). Two of these studies were conducted in China (Ke and Shunzhang 1999; Zhang et al. 1989), one in Germany (Straif et al. 2000), and one in Italy (Negri et al. 1989).

Two publications reported cancer risks in the same employees at a ceramic plumbing and fixture plant (Thomas 1990; Thomas and Stewart 1987) in the US. We also identified one cohort study of printing press workers in Russia (Bulbulyan et al. 1999) and three case-control studies: one on fiberglass workers in the US (Chiazze et al. 1993), one on female pulp and paper mill workers in Norway (Langseth and Kjaerheim 2004), and one that evaluated talc exposure via surgical gloves in Denmark (Nielsen et al. 1994). Two case-control studies examined occupational talc exposure (Leung et al. 2023; Hartge and Stewart 1994), and one cohort study evaluated the ingestion of talc via medicinal supplements (Chang et al. 2019).

The largest group of talc studies examining cancer outcomes evaluated women who used cosmetic talc. We identified nine cohort studies (O'Brien et al. 2021b; O'Brien et al. 2019; Gonzalez et al. 2016; Urban et al. 2015; Houghton et al. 2014; Crawford et al. 2012; Karageorgi et al. 2010; Gates et al. 2010; Gertig et al. 2000) and 28 case-control studies (Gabriel et al. 2019; Cramer et al. 2016; Schildkraut et al. 2016; Kurta et al. 2012; Neill et al. 2012; Rosenblatt et al. 2011; Moorman et al. 2009; Wu et al. 2009; Merritt et al. 2008; Jordan et al. 2007; Mills et al. 2004; Ness et al. 2000; Cramer et al. 1999; Wong et al. 1999; Godard et al. 1998; Chang and Risch 1997; Cook et al. 1997; Green et al. 1997; Purdie et al. 1995; Tzonou et al. 1993; Chen et al. 1992; Harlow et al. 1992; Rosenblatt et al. 1992; Booth et al. 1989; Harlow and Weiss 1989; Whittemore et al. 1988; Hartge et al. 1983; Cramer et al. 1982) that examined the association between the use of talc as a feminine hygiene product and endometrial or ovarian cancer.

Finally, we identified 12 pooled studies. Two studies pooled data from five unique cohorts of talc miners and millers (Ierardi et al. 2022; Ierardi and Marsh 2020) and analyzed pleural cancer and mesothelioma. One study pooled data from two cohorts of miners and millers in Austria and France and analyzed all cancers, lung cancer, mesothelioma, and stomach cancer (Wild et al. 2002). O'Brien et al. (2020, 2021a) pooled data from four US cohorts and analyzed ovarian and endometrial cancer. Six studies pooled case-control studies of ovarian cancer (Phung et al. 2022; Davis et al. 2021; Wu et al. 2015; Terry et al. 2013; Gates et al. 2008; Cramer and Xu 1995), and one study pooled case-control studies examining cosmetic talc and industrial talc exposure and lung cancer (Ramanakumar et al. 2008).

Study results

We report the results of all studies organized by organ system and cancer type. We tabulated all study results for completeness but only discuss results reported for specific cancers (e.g. bladder cancer), or cancers presumed to occur via the same underlying mode of action (e.g. oral and pharyngeal cancers) and not results for aggregate cancers (e.g. all cancers). We also only discuss the results of cancers for which there are at least three studies conducted in independent populations.

Female reproductive tract

Ovarian cancer

We identified six cohort studies and 30 case-control studies of 27 unique populations that examined talc exposures and ovarian cancer (Bulbulyan et al. 1999; Gates et al. 2010; Gertig et al. 2000; Gonzalez et al. 2016; Houghton et al. 2014; Urban et al. 2015; Booth et al. 1989; Chang and Risch 1997; Chen et al. 1992; Cook et al. 1997; Cramer et al. 1982; Cramer et al. 1999; Cramer et al. 2016; Gabriel et al. 2019; Godard et al. 1998; Harlow and Weiss 1989; Harlow et al. 1992; Hartge et al. 1983; Hartge and Stewart 1994; Kurta et al. 2012; Langseth and Kjaerheim 2004; Leung et al. 2023; Merritt et al. 2008; Jordan et al. 2007; Mills et al. 2004; Moorman et al. 2009; Ness et al. 2000; Purdie et al. 1995; Green et al. 1997; Rosenblatt et al. 1992; Rosenblatt et al. 2011; Schildkraut et al. 2016; Tzonou et al. 1993; Whittemore et al. 1988; Wong et al. 1999; Wu et al. 2009). Of the identified studies, four studies focused on occupational exposures to talc. The occupational studies examined a cohort of female printing plant workers (Bulbulyan et al. 1999) a case-control study of pulp and paper workers (Langseth and Kjaerheim 2004), and general occupational exposures based on self-reported work histories (Hartge and Stewart 1994; Leung et al. 2023). The other five cohort and 27 case-control studies focused on cosmetic applications of talc and ovarian cancer risk. The results of these studies are summarized in Appendix E Tables E3 to E6, E8 to E13, and E15).

In a cohort study of female workers at two printing plants in Moscow, Russia, Bulbulyan et al. (1999) observed an association among bookbinding workers (SMR = 2.9, 95% CI: 1.5-5.0) but not among all employees (SMR = 1.2, 95% CI: 0.6-2.0), when compared with the Moscow general population. Of note, ovarian cancer rates for the entire follow-up period (1979-1993) were not available, so only rates from 1992 were used for comparison. No other studies that assessed occupational exposures reported increased ovarian cancer risks (Langseth and Kjaerheim 2004; Hartge and Stewart 1994; Leung et al. 2023).

Of the five cohort studies that investigated ovarian cancer risk from cosmetic talc exposures, only Gonzalez et al. (2016) observed a statistically significant association. The authors conducted several analyses based on genital talc use and douching practices, including stratified by patency, hysterectomy, tubal ligation, parity, and menopausal status, and only found a statistically significant increase in risk of incident ovarian cancer among subjects who both used talc and douched (HR = 1.9, 95% CI: 1.2-2.9), when compared to subjects who used neither. In analyses conducted based on talc use only, no significant associations were observed. Of the 27 case-control studies that investigated ovarian cancer risk from cosmetic talc exposures, two-thirds reported some statistically significant relationship. These associations were typically weak but greater than one.

We identified three cohort studies and 11 case-control studies in 11 unique populations that evaluated serous ovarian cancer (Gates et al. 2010; Gertig et al. 2000; Houghton et al. 2014; Chang and Risch 1997; Cramer et al. 1999; Cramer et al. 2016; Gabriel et al. 2019; Harlow et al. 1992; Merritt et al. 2008; Jordan et al. 2007; Mills et al. 2004; Rosenblatt et al. 2011; Schildkraut et al. 2016; Wong et al. 1999). Most studies reported statistically significant and positive results for various categories of genital talc exposure (Gates et al. 2010; Gertig et al. 2000; Cramer et al. 1999; Cramer et al. 2016; Gabriel et al. 2019; Merritt et al. 2008; Rosenblatt et al. 2011; Schildkraut et al. 2016).

We identified three cohort studies and 10 case-control studies in 10 unique populations that analyzed mucinous ovarian cancer (Gates et al. 2010; Gertig et al. 2000; Houghton et al. 2014; Chang and Risch 1997; Cramer et al. 1999; Cramer et al. 2016; Gabriel et al. 2019; Harlow et al. 1992; Merritt et al. 2008; Jordan et al. 2007; Mills et al. 2004; Rosenblatt et al. 2011; Wong et al. 1999). Cramer et al. (2016) reported a statistically significant negative association among premenopausal women who had used talc regularly for <1 year (OR = 0.13, 95% CI: 0.02-0.99) and a positive association for use >5 years (OR = 2.28, 95% CI: 1.23-4.26), when compared with never users, and this trend for duration of use was statistically significant (p-trend = 0.005). Merritt et al. (2008) only reported a statistically significant association among women post-hysterectomy or tubal ligation who used talc for more than 10 to 25 years (OR = 2.04, 95% CI: 1.09-3.79), and a negative not statistically significant association for users of more than 25 years (OR = 0.91, 95% CI: 0.27-3.05). Rosenblatt et al. (2011) only presented a statistically significant relationship among talc powder users of 10 to <20 years (OR = 3.12, 95% CI: 1.22-7.97) but not for shorter or longer duration exposures. All 10 other studies, including all three cohort studies, reported non-statistically significant results.

We identified three cohort studies and seven case-control studies in eight unique populations of endometrioid ovarian cancer (Gates et al. 2010; Gertig et al. 2000; Houghton et al. 2014; Chang and Risch 1997; Cramer et al. 2016; Gabriel et al. 2019; Harlow et al. 1992; Merritt et al. 2008; Mills et al. 2004; Wong et al. 1999). Cramer et al. (2016) reported statistically significant results for invasive endometrioid ovarian cancer among women with any genital talc use (OR = 1.38, 95% CI: 1.06-1.80) and women who regularly used talc for at least 25 years (OR = 1.67, 95% CI: 1.06-2.63). Harlow et al. (1992) also reported a statistically significant increase in endometrioid cancer (OR = 2.8, 95% CI: 1.2-6.4) among women with any perineal talc use. Other studies did not report any associations.

We identified five case-control studies in four unique populations of clear cell ovarian cancer (Cramer et al. 2016; Gabriel et al. 2019; Merritt et al. 2008; Mills et al. 2004; Wong et al. 1999). None of these studies observed any significant associations.

With respect to all ovarian cancer studies, it is likely that recall bias has had a large impact on risk estimates. All cosmetic use studies relied on self-reported talc use, sometimes many years or decades in the past. Schildkraut et al. (2016), in their US-based study, hypothesized that the publicity of two high-profile lawsuits in 2014 may have influenced participants recall of talc use, and the authors noted large changes in risk based on subjects interviewed prior to and after 2014. The only study that conducted a recall bias analysis on perineal talc exposure was Cramer et al. (2016). This study calculated that recall sensitivity in controls would have to have been 83% for a risk estimate of 1 if recall sensitivity and specificity were both 99% in cases and recall sensitivity was 99% in controls. Real-world recall data have since been reported from a cohort study (O'Brien et al. 2023), and showed differential recall between cases and controls, with < 99% sensitivities and specificities in cases. In fact, the sensitivity in controls was estimated to be 63% – far below that calculated by Cramer et al. (2016). When Goodman et al. (2024) replicated analyses in Cramer et al. (2016) using sensitivities and specificities reported in O'Brien et al. (2023), a substantial attenuation of ORs were observed, with a reversed direction of the association in one scenario.

Overall, cohort studies mostly reported non-statistically significant associations and no exposure-response relationships. Studies of occupational exposures, in particular, do not show any consistent association between talc and ovarian cancer. Cosmetic talc use case-control studies consistently showed weak associations, though they were based solely on self-reported exposure estimates and recall bias may have had a large impact on risk estimates (Goodman et al. 2024). All studies that evaluated personal care products containing talc were based on semi-quantitative or qualitative measures of exposure, which are weaker than direct quantitative exposure measurements. In addition, some studies did not control for appropriate confounding factors, such as pregnancy history, hormone therapy, endometriosis, and other occupational co-exposures. Therefore, epidemiology studies do not provide evidence supporting a causal association between talc exposures and ovarian cancer risk.

Endometrial/uterine cancer

Four cohort studies (Bulbulyan et al. 1999; Crawford et al. 2012; Karageorgi et al. 2010; O'Brien et al. 2019), one case-control study (Neill et al. 2012), and one pooled analysis (O'Brien et al. 2021a) evaluated talc use and endometrial/uterine cancer risk. The results of these studies are summarized in Appendix E Tables E2, E7, and E14.

Bulbulyan et al. (1999) did not observe any increased risks among a cohort of female workers at two printing plants in Moscow, Russia. The other three cohort studies evaluated perineal talc use in cohorts that were later pooled in O'Brien et al. (2021a). With few exceptions, none of these studies found associations between talc use and endometrial or uterine cancer overall or in any sub-analyses. In their pooled analysis of the Nurses’ Health Studies 1 and 2, the Sister Study, and the Women’s Health Initiative, O'Brien et al. (2021a) found no association between talc use and uterine cancer overall (HR = 1.01, 95% CI: 0.94-1.09), for specific uterine cancer subtypes, or in other sub-analyses (e.g. by long term use or by ≥ one use per week).

Neill et al. (2012) evaluated Type 1 and Type 2 endometrial cancer in 1,399 cases and 740 controls from Australia. They observed no increased risks overall, and reported both increased and decreased non-statistically significant and statistically significant associations in some analyses. They reported a decreased trend of endometrial cancer risk with an increased duration of perineal (p < 0.001) or upper body (p = 0.001) powder use.

The epidemiology evidence does not support a causal association between talc and endometrial/uterine cancer.

Respiratory tract

We identified 25 studies that examined talc exposures and respiratory tract cancers in 15 unique populations (Brown et al. 1990; Bulbulyan et al. 1999; Chiazze et al. 1993; Ciocan et al. 2022a; Coggiola et al. 2003; Fordyce et al. 2019; Gamble 1993; Honda et al. 2002; Ierardi et al. 2022; Ierardi and Marsh 2020; Katsnelson and Mokronosova 1979; Lamm et al. 1988; Negri et al. 1989; Nielsen et al. 1994; Pira et al. 2017; Ramanakumar et al. 2008; Rubino et al. 1979; Rubino et al. 1976; Straif et al. 2000; Thomas 1990; Thomas and Stewart 1987; Wergeland et al. 2017; Wergeland et al. 1990; Wild et al. 2002; Zhang et al. 1989). These studies examined cancers of the larynx, lung, and pleura, or mesothelioma; and cancers of the respiratory organs in aggregate. The results of these studies are summarized in Appendix F Tables F1 to F8.

Laryngeal cancer

We identified nine cohort studies that examined talc exposures and laryngeal cancer in six unique populations. All of the identified studies examined occupational exposures to talc. Six studies of three cohorts were conducted in miners and millers (Ciocan et al. 2022a; Coggiola et al. 2003; Pira et al. 2017; Rubino et al. 1976; Honda et al. 2002; Fordyce et al. 2019), two studies examined workers exposed to talc in tire and rubber industries (Negri et al. 1989; Straif et al. 2000), and one study examined talc exposures in printing plant employees (Bulbulyan et al. 1999). Importantly, none of these studies adjusted for smoking.

There were no statistically significant associations observed in any of the three miner and miller cohorts. Honda et al. (2002) reported an SMR of 3.16 (95% CI: 0.38-11.42) based on two cases. Fordyce et al. (2019) did not observe any cases, and Ciocan et al. (2022a) reported results based on eight exposed cases among all workers (SMR = 0.92, 95% CI: 0.4-1.81), seven exposed miners (SMR = 1.541, 95% CI: 0.617-3.176), and one exposed miller (SMR = 0.414, 95% CI: 0.005-2.306). No significant trends were reported.

Negri et al. (1989) and Straif et al. (2000) reported similar risk estimates for laryngeal cancer in all workers at a rubber tire factory (SMR = 1.26, 95% CI: 0.67-2.16) and a rubber plant (SMR = 1.17, 95% CI: 0.50-2.30), respectively. Straif et al. (2000) reported an increase in risk with increasing exposure, with a statistically significant increased risk among the highest exposure category (medium exposure HR = 2.8, 95% CI: 0.5-16.7; high exposure HR = 5.4, 95% CI: 1.1-27.0) compared to the lowest exposed workers, but these estimates were based on small numbers of cases (range 2-3).

Bulbulyan et al. (1999) only reported one exposed case, and did not report any results for associations between job rolls in the printing industry and laryngeal cancer in Moscow printing plants compared to the general population.

Among all studies, most risk estimates were not statistically significant, with point estimates both above and below 1, and there was only a very small number of exposed cases (most ≤ 8) in most analyses. Only one study reported statistically significant results (Straif et al. 2000), but this estimate was based on only three exposed cases. None of these studies had direct talc exposure measurements, and only the Val Chisone mining or milling studies and the rubber plant workers had semi-quantitative exposure estimates. No study adjusted for smoking. Overall, the epidemiology evidence does not support a causal association between talc exposure and laryngeal cancer.

Lung cancer

We identified 17 cohort studies (Bulbulyan et al. 1999; Ciocan et al. 2022a; Coggiola et al. 2003; Honda et al. 2002; Katsnelson and Mokronosova 1979; Lamm et al. 1988; Negri et al. 1989; Pira et al. 2017; Rubino et al. 1979; Rubino et al. 1976; Thomas 1990; Thomas and Stewart 1987; Straif et al. 2000; Wergeland et al. 2017; Wergeland et al. 1990; Wild et al. 2002; Zhang et al. 1989), two case-control studies (Chiazze et al. 1993; Gamble 1993), and one pooled study (Ramanakumar et al. 2008) that assessed talc exposures and lung cancer risks in 13 unique populations. All of the identified studies examined occupational exposures to talc. Among talc miners and millers, five studies were conducted in male Italian talc miners and millers in Val Chisone, Italy (Ciocan et al. 2022a; Coggiola et al. 2003; Pira et al. 2017; Rubino et al. 1979; Rubino et al. 1976), three studies were conducted among talc miners and millers at the GTC in New York, US (Gamble 1993; Honda et al. 2002; Lamm et al. 1988), two studies were conducted in a cohort of talc miners and millers at the Altermark mining and Knarrevik milling operations in Norway (Wergeland et al. 2017; Wergeland et al. 1990), one study was conducted among a cohort of miners and millers in the former USSR (Katsnelson and Mokronosova 1979), and one study was conducted Austrian and French talc mining and milling operation employees (Wild et al. 2002). Three studies were conducted in cohorts on rubber or tire industry workers (Negri et al. 1989; Straif et al. 2000; Zhang et al. 1989). Two studies were conducted in cohort of workers at ceramic plumbing fixture factories in the US (Thomas 1990; Thomas and Stewart 1987). One study each was conducted among female printing press employees in Moscow, Russia (Bulbulyan et al. 1999) and among employees of the Owens-Corning Fiberglass Corporation’s plant in Newark, US (Chiazze et al. 1993).

Several analyses were conducted in miners and millers at the GTC (Gamble 1993; Honda et al. 2002; Lamm et al. 1988). Gamble (1993) conducted a nested case-control study and found “[w]hen stratified by smoking status, risk of lung cancer decreased with talc tenure and remained negative when excluding cases with < 20 years’ latency and short-term workers.” They concluded that smoking was a confounder and that results didn’t support talc as a causal factor. Almost 10 years later, Honda et al. (2002) reported an association with lung cancer mortality among all miners and millers at the GTC (SMR = 2.32, 95% CI: 1.57-3.29), but the excess was in miners, and not millers. Honda et al. (2002) noted that they “lacked comprehensive information on potential confounders such as cigarette smoking and other occupational exposures” and concluded that associations were “unlikely to be related to respirable talc ore dust per se.”

In a cohort study of employees of a talc mining and milling plant in the former USSR, Katsnelson and Mokronosova (1979) reported a statistically significant higher risk lung cancer among men (SMR = 4.5), but not among women (SMR = 9.3) when compared to town residents. However, CIs, the number of cases, and smoking information were not provided, making it impossible to judge the reliability of these effect estimate. No associations or trends were found in talc miners and millers in Val Chisone, Italy (Ciocan et al. 2022a; Coggiola et al. 2003; Pira et al. 2017; Rubino et al. 1979; Rubino et al. 1976), Altermark mining and Knarrevik milling operations in Norway (Wergeland et al. 2017; Wergeland et al. 1990), or in miners and millers in France and Austria (Wild et al. 2002). None of these studies adjusted for smoking.

In a cohort study of workers in ceramic plumbing fixture plants in the US, Thomas (1990) and Thomas and Stewart (1987) reported several statistically significant associations, but did not report CIs or adjust for smoking in any analyses. Non-statistically significant results were reported in a case-control study of workers at the Owens-Corning Fiberglas Corporation’s Newark plant in the in US (Chiazze et al. 1993), in the cohort of Moscow printing plant workers (Bulbulyan et al. 1999), and in a cohort of rubber tire factory workers (Negri et al. 1989). In another cohort study of rubber tire workers, Straif et al. (2000) reported a statistically significant increased risk of mortality from lung cancer among workers with the highest exposure (HR = 1.6, 95% CI: 1.1-2.4) compared to workers with the lowest exposure, but did not adjust for smoking.

In a cohort of workers in the rubber industry, Zhang et al. (1989) did not report any associations between employment in the industry and lung cancer overall, but reported statistically significant increased risk for employees who smoked ≥ 10 cigarettes per day, suggesting that smoking, not talc exposures, are driving lung cancer mortality in these workers. No confidence intervals were provided.

In a pooled analysis of two case-control studies from hospitals in the Montreal metropolitan area, Ramanakumar et al. (2008) did not report any associations between industrial or cosmetic talc and lung cancer. Almost all the reported effect estimates were either close to or below one (Ramanakumar et al. 2008). In a nested case-control analysis that pooled the Austrian and French talc operation employee cohorts, Wild et al. (2002), did not report an association when comparing exposed employees to unexposed employees (OR = 0.98, 95% CI: 0.88-1.10)

Overall, the epidemiology evidence does not support a causal association between talc exposure and lung cancer. Most associations were not statistically significant and very few studies controlled for smoking. A number of studies also failed to report confidence intervals.

Mesothelioma

We identified six cohort studies and one case-control study of five unique populations that examined talc exposures and mesothelioma (Ciocan et al. 2022a; Pira et al. 2017; Coggiola et al. 2003; Rubino et al. 1976; Negri et al. 1989; Nielsen et al. 1994; Wild et al. 2002). All of the identified cohort studies evaluated occupational exposures to talc, with five studies conducted among three cohorts of miners and millers (Ciocan et al. 2022a; Pira et al. 2017; Coggiola et al. 2003; Rubino et al. 1976; Wild et al. 2002) and one study of rubber workers (Negri et al. 1989). The case-control study focused on hospital patients exposed to talc from surgical gloves and coverings during abdominal surgeries (Nielsen et al. 1994).

No mesotheliomas were reported among Val Chisone, French, or Austrian miners or millers (Ciocan et al. 2022a; Pira et al. 2017; Coggiola et al. 2003; Rubino et al. 1976; Wild et al. 2002), and no association was found in hospital patients exposed during abdominal surgeries (Nielsen et al. 1994). Significant findings were only reported in one study of Italian rubber tire factory workers (Negri et al. 1989). Only nine deaths were identified, and the authors noted that workers may have been heavily exposed to asbestos and other fibers from machinery in the factory. They also reported no pleural cancer deaths among tire building workers, the job category they reported most likely to be exposed to talc.

We think it worth noting that Fordyce et al. (2019) attempted to examine mesothelioma in the cohort of Vermont talc miners, but were limited by nonspecific International Classification of Diseases (ICD) codes, so we do not to report those results here. Based on a secondary review by a nosologist, there was one case of mesothelioma in their cohort, but this case had previously been exposed to asbestos. Similarly, Wergeland et al. (1990, 2017) stated that there were no cases of mesothelioma in their cohort, but did not report risk estimates.

There were two pooled analyses of mesothelioma in talc miners and millers (Ierardi et al. 2022; Ierardi and Marsh 2020), both examining risk in four studies of five cohorts (Ciocan et al. 2022a; Wergeland et al. 2017; Wild et al. 2002; Fordyce et al. 2019). Ierardi et al. (2022) provided the most recent update and reported no association between talc mining and milling and mesothelioma, based on a single reported case in the Vermont mining and milling population reported in Fordyce et al. (2019). Marsh et al. (2019) and Finley et al. (2017) also examined mesothelioma through pooling of four of the five cohorts examined in Ierardi et al. (2022) and Ierardi and Marsh (2020), but they did not report risk estimates and are, therefore, not included in our review.

The epidemiology evidence does not support a causal association between talc exposure and pleural cancers or mesothelioma.

Gastrointestinal system

We identified 16 cohort studies and one case-control study in 11 unique populations that examined talc exposures and gastrointestinal system cancers (Bulbulyan et al. 1999; Ciocan et al. 2022a; Coggiola et al. 2003; Pira et al. 2017; Fordyce et al. 2019; Brown et al. 1990; Stille and Tabershaw 1982; Rubino et al. 1976; Negri et al. 1989; Zhang et al. 1989; Chang et al. 2019; Honda et al. 2002; Katsnelson and Mokronosova 1979; Straif et al. 2000; Wergeland et al. 2017; Wergeland et al. 1990; Ke and Shunzhang 1999). These studies examined cancers of the mouth and pharynx, esophagus, stomach, colon and rectum, liver, pancreas, and gastrointestinal organs in aggregate. The results of these studies are summarized in Appendix G Tables G1 to G8.

Oral and pharyngeal cancer

We identified five cohort studies that examined oral and pharyngeal cancer mortality in three unique populations (Bulbulyan et al. 1999; Ciocan et al. 2022a; Coggiola et al. 2003; Pira et al. 2017; Fordyce et al. 2019). All of the identified studies focused on occupational exposures to talc, with four studies conducted in two cohorts of miners and millers (Ciocan et al. 2022a; Coggiola et al. 2003; Pira et al. 2017; Fordyce et al. 2019) and one study in a cohort of employees at printing plants (Bulbulyan et al. 1999).

The studies examining miners and millers reported mixed results. Fordyce et al. (2019) reported no cases of buccal cavity or pharyngeal cancers in Vermont miners and millers. Ciocan et al. (2022a); Coggiola et al. (2003); and Pira et al. (2017) reported increased risks of oral/pharyngeal cancers overall (SMR = 3.65, 95% CI: 2.53-5.10), and in miners (SMR = 4.06, 95% CI: 2.62-5.99) and among millers (SMR = 2.86, 95% CI = 1.31-5.43) separately in Val Chisone, Italy. They conducted several analyses examining associations by duration of exposure and, while several estimates were statistically significant, the highest risk estimates in all analyses were not in the longest duration category, and there were no statistically significant p-trends reported.

Bulbulyan et al. (1999) reported only a single case of buccal cavity or pharyngeal cancer in female employees from two printing plants in Moscow, Russia which was less than expected compared to the Moscow general population.

While the epidemiology studies of a cohort of miners and millers in Val Chisone reported positive and statistically significant SMRs for oral and pharyngeal cancer, these studies had no information on any direct exposure measurements, and results did not demonstrate any pattern of increased risk with increasing exposure. No other studies reported increased risks. Potential confounders, including tobacco use or alcohol consumption, were not considered in any study. Therefore, epidemiology studies do not provide evidence of a causal association between talc and oral/pharyngeal cancer.

Esophageal cancer

Ten cohort studies assessed whether talc exposures were associated with esophageal cancer mortality in six unique populations (Brown et al. 1990; Stille and Tabershaw 1982; Bulbulyan et al. 1999; Ciocan et al. 2022a; Coggiola et al. 2003; Pira et al. 2017; Rubino et al. 1976; Fordyce et al. 2019; Negri et al. 1989; Zhang et al. 1989). All of the identified studies focused on occupational exposures to talc, with seven studies conducted in three unique populations of miners and millers (Brown et al. 1990; Stille and Tabershaw 1982; Ciocan et al. 2022a; Coggiola et al. 2003; Pira et al. 2017; Rubino et al. 1976; Fordyce et al. 2019), two studies conducted in two unique populations of rubber workers (Negri et al. 1989; Zhang et al. 1989), and one study in female printing plant workers (Bulbulyan et al. 1999).

An increased risk of esophageal cancer was reported in male miners and millers in Val Chisone, Italy overall (SMR = 1.92, 95% CI: 1.05-3.22) and in miners (SMR = 2.30, 95% CI: 1.14-4.11), but not millers (SMR = 1.20, 95% CI: 0.25-3.49) (Ciocan et al. 2022a; Coggiola et al. 2003; Pira et al. 2017; Rubino et al. 1976). Ciocan et al. (2022a) conducted several analyses examining associations by exposure duration and dose. Most of the results of these analyses were not statistically significant, and risks did not increase with increasing exposure. There was a statistically significant increased risk among those with the shortest duration of employment (SMR = 3.13, 95% CI: 1.35-6.18), and the risk decreased with longer duration of employment (p-trend = 0.044).

There were very few esophageal cancer deaths in other studies of male miners and millers, and there were no associations with talc in these studies (Brown et al. 1990; Stille and Tabershaw 1982; Fordyce et al. 2019). The two studies of rubber factory workers (Negri et al. 1989; Zhang et al. 1989) also did not report any associations between employment and esophageal cancer mortality.

Bulbulyan et al. (1999) investigated esophageal cancer mortality among female employees of two printing plants in Moscow, Russia, and observed increased risks among all printing plant employees overall (SMR = 2.6, 95% CI: 1.1-5.4), bookbinders (SMR = 4.1, 95% CI: 1.0-10.4), and employees hired prior to 1958 (SMR = 7.1, 95% CI: 1.9-18.3). This analysis was based on a small number of cases (≤ 7), and there were other occupational exposures in this population that were not controlled for. There were also inconsistencies between the results reported in the text and tables.

Epidemiology studies of male miners and millers in Val Chisone and female printing plant workers in Moscow report some statistically significant positive associations with esophageal cancer mortality, but these studies did not use direct exposure measurements, and studies in two other miner and miller populations and studies of workers in the rubber industry did not report increased risks, although these studies also did not have direct exposure measurements. In addition, smoking and drinking alcohol, which are risk factors for esophageal cancer, were not considered in any studies. Therefore, the epidemiology studies do not provide evidence of a causal association between talc and esophageal cancer.

Stomach cancer

We identified 15 cohort studies and one case-control study in 10 unique populations that assessed talc and stomach cancer (Bulbulyan et al. 1999; Chang et al. 2019; Ciocan et al. 2022a; Coggiola et al. 2003; Pira et al. 2017; Rubino et al. 1976; Fordyce et al. 2019; Honda et al. 2002; Stille and Tabershaw 1982; Katsnelson and Mokronosova 1979; Negri et al. 1989; Straif et al. 2000; Wergeland et al. 2017; Wergeland et al. 1990; Zhang et al. 1989; Ke and Shunzhang 1999). Of the identified studies, 14 focused on occupational exposures to talc, while one study evaluated oral intake of talc powder as a medical treatment (Chang et al. 2019).

Five populations of miners and millers were evaluated. A significant association with stomach cancer was only reported in a single study of employees at a talc mining, grinding, and processing company in the former USSR, (RR_men = 3.7; RR_women = 6.3), but the total number of exposed cases and CIs were not reported (Katsnelson and Mokronosova 1979). There were no other statistically significant results reported overall or in evaluations of exposure frequency, dose or duration in any of the other mining and milling studies (Ciocan et al. 2022a; Coggiola et al. 2003; Pira et al. 2017; Rubino et al. 1976; Fordyce et al. 2019; Honda et al. 2002; Stille and Tabershaw 1982; Wergeland et al. 2017; Wergeland et al. 1990).

In studies of rubber workers, Negri et al. (1989), Zhang et al. (1989), and Ke and Shunzhang (1999) observed non-statistically significant associations, around or <1. Straif et al. (2000) reported no statistically significant increased risk of stomach cancer overall in rubber workers in western Germany, but reported a statistically significant increased risk of stomach cancers among those considered to have had high exposure to talc (i.e. using talc as filler material or heavy use as antitacking material) compared to workers with low exposure (i.e. wet application of talc as antitacking material or no use of talc) (RR = 2.4, 95% CI: 1.2-4.9).

Bulbulyan et al. (1999) investigated mortality from stomach cancer among female employees of two printing plants in Moscow, Russia, and observed an association in press-operators (SMR = 2.2, 95% CI: 1.0-4.2) but no other occupational groups.

Chang et al. (2019) evaluated the association between stomach cancer and medicinal oral intake of talc powder using data from Taiwan’s health insurance database. The authors reported an increased risk of stomach cancer overall (HR = 2.13, 95% CI: 1.54-2.94). They also reported statistically significant increased risk among those with low (HR = 2.40, 95% CI: 1.54-3.73) or high (HR = 1.89, 95% CI: 1.19-3.01) cumulative exposure (≤ or >10.5 g, respectively) when compared to those who were unexposed. However, when the authors restricted cases of stomach cancers to those with 5 or more years between their first exposure and stomach cancer, these relationships were not statistically significant.

Twelve studies of seven unique populations did not observe any associations between talc exposure and stomach cancer. Only one study relied on directly measured exposure measurements (medicinal oral intake) (Chang et al. 2019), and the results of this study were not statistically significant when a latency period between first exposure and stomach cancer was accounted for. Only Chang et al. (2019) considered potential confounders beyond age and sex. The epidemiology evidence does not support a causal association between talc exposure and stomach cancer.

Colon and rectal cancer

We identified 10 cohort studies in six unique populations that assessed whether talc exposures were associated with colon or rectal cancers (Bulbulyan et al. 1999; Ciocan et al. 2022a; Pira et al. 2017; Coggiola et al. 2003; Rubino et al. 1976; Fordyce et al. 2019; Honda et al. 2002; Wergeland et al. 2017; Wergeland et al. 1990; Zhang et al. 1989). All of the identified studies focused on occupational exposures to talc, with eight studies conducted among four unique populations of miners and millers (Ciocan et al. 2022a; Pira et al. 2017; Coggiola et al. 2003; Rubino et al. 1976; Fordyce et al. 2019; Honda et al. 2002; Wergeland et al. 2017; Wergeland et al. 1990), one study conducted among rubber workers (Zhang et al. 1989), and one study of female printing plant employees (Bulbulyan et al. 1999).

Most studies reported small risk estimates around 1 that were not statistically significant. Only a single study of Altermark mine and Knarrevik mill workers in Norway reported a statistically significant association between incident colorectal cancer risk among all workers (SIR = 1.59, 95% CI: 1.07-2.26) and millers (SIR = 1.62, 95% CI: 1.04-2.41) but not miners (SIR = 1.47, 95% CI: 0.54-3.20) compared with the Norwegian general population (Wergeland et al. 2017).

The studies of Norwegian mining and milling workers were the only studies to report statistically significant results for colon and rectal cancers, however, these findings were based on employment ≤1 (miners) or 2 years (millers), and not on direct talc exposure measurements (Wergeland et al. 2017). None of the studies reporting colorectal cancer results considered potential confounders except for age and sex, which may have biased results. The epidemiology evidence does not support a causal association between talc exposure and colon or rectal cancers.

Liver, gallbladder and bile duct cancer

We identified 10 cohort studies that assessed whether talc exposures were associated with liver cancer in six unique populations (Brown et al. 1990; Stille and Tabershaw 1982; Bulbulyan et al. 1999; Ciocan et al. 2022a; Pira et al. 2017; Coggiola et al. 2003; Rubino et al. 1976; Fordyce et al. 2019; Negri et al. 1989; Zhang et al. 1989). All of the identified studies focused on occupational exposures to talc and mortality, with seven studies conducted among three unique populations of miners and millers (Brown et al. 1990; Stille and Tabershaw 1982; Ciocan et al. 2022a; Pira et al. 2017; Coggiola et al. 2003; Rubino et al. 1976; Fordyce et al. 2019), two studies in rubber workers (Negri et al. 1989; Zhang et al. 1989), and one study of female printing plant employees (Bulbulyan et al. 1999).

There were no significant increased risks of liver or gall bladder cancer reported in populations of miners and millers except in the GTC cohort (Brown et al. 1990; Stille and Tabershaw 1982) where two deaths from liver and gallbladder cancer were reported. Among GTC workers with known prior employment a statistically significant increase in mortality and liver and gallbladder cancer was reported (SMR = 10.13, CI not provided) (Stille and Tabershaw 1982).

The studies of rubber workers (Negri et al. 1989; Zhang et al. 1989) and female printing plant employees (Bulbulyan et al. 1999) did not report any associations with liver cancer.

While Stille and Tabershaw (1982) observed a statistically significant increase in mortality risk from liver and gallbladder cancer among their workers with known prior employment, this was based on only two exposed deaths, there was no information on what the prior employment was, and there were no CIs provided to help gauge the confidence in the result. The other nine studies did not report any associations between talc exposure and liver, gallbladder, or bile duct cancer. None of the studies considered key potential confounders beyond age and sex. The epidemiology evidence does not support a causal association between talc exposure and liver cancer.

Pancreatic cancer

We identified eight cohort studies in five unique populations that assessed whether talc exposures were associated with pancreatic cancer (Brown et al. 1990; Stille and Tabershaw 1982; Bulbulyan et al. 1999; Ciocan et al. 2022a; Coggiola et al. 2003; Pira et al. 2017; Fordyce et al. 2019; Negri et al. 1989). All of the identified studies focused on occupational exposures to talc and pancreatic cancer mortality. Six studies were conducted in three unique populations of miners and millers (Brown et al. 199; Stille and Tabershaw 1982; Ciocan et al. 2022a; Coggiola et al. 2003; Pira et al. 2017; Fordyce et al. 2019), one in rubber workers (Negri et al. 1989), and one in female printing plant employees (Bulbulyan et al. 1999).

In a study of miners and millers in Vermont, Fordyce et al. (2019) reported two pancreatic cancer deaths, and did not report an increased risk among all workers (SMR = 0.563, 95% CI: 0.068-2.035) compared with the US general population. When analyses were stratified by categories of latency, they reported a statistically significant increased risk only among the shortest latency period (0-14 years) (SMR = 8.452, 95% CI: 1.023-30.534), but this risk estimate was based on only two deaths, and there were no cases observed in longer latency categories. Brown et al. (1990), also observed only two exposed cases in the GTC cohort of miners and millers and did not observe an increased risk in this population (SMR = 1.54, 95% CI: 0.18-5.44).

Ciocan et al. (2022a), Coggiola et al. (2003), and Pira et al. (2017) did not observe any associations in miners and millers in Val Chisone. Overall, they reported 11 exposed cases (SMR = 0.93, 95% CI: 0.46-1.66) and other analyses examining associations by duration of exposure were based on a small number of cases within each strata (≤4).

Negri et al. (1989) reported a statistically significant decrease in risk among all rubber factory workers (SMR = 0.26, 95% CI: 0.03-0.96), but this was based on only two deaths. Bulbulyan et al. (1999) did not report any statistically significant associations in female printing plant workers.

The results from epidemiology studies looking at talc and pancreatic cancer were mixed, but both studies reporting statistically significant results, Fordyce et al. (2019) positive and Negri et al. (1989) negative, were based on only two deaths. None of the other eight studies observed any statistically significant associations with pancreatic cancer. In addition, none of the studies controlled for potential confounders beyond age and sex. The epidemiology evidence does not support a causal association between talc exposure and pancreatic cancer.

Urinary system

We identified 10 studies that examined talc exposures and urinary system cancers in six unique populations (Brown et al. 1990; Bulbulyan et al. 1999; Ciocan et al. 2022a; Coggiola et al. 2003; Pira et al. 2017; Negri et al. 1989; Wergeland et al. 2017; Wergeland et al. 1990; Stille and Tabershaw 1982; Fordyce et al. 2019). These studies examined cancers of the bladder, kidney, or urinary organs in aggregate. The results of these studies are summarized in Appendix H Tables H1 to H3.

Bladder cancer

We identified eight cohort studies that examined talc exposures and bladder cancer specifically in five unique populations (Brown et al. 1990; Bulbulyan et al. 1999; Ciocan et al. 2022a; Coggiola et al. 2003; Pira et al. 2017; Negri et al. 1989; Wergeland et al. 2017; Wergeland et al. 1990). All of the identified studies examined occupational exposures to talc. Three studies were conducted in miners and millers (Brown et al. 1990; Ciocan et al. 2022a; Coggiola et al. 2003; Pira et al. 2017; Wergeland et al. 2017; Wergeland et al. 1990), and one each in printing plant (Bulbulyan et al. 1999) and rubber tire factory (Negri et al. 1989) employees.

The three studies that evaluated talc exposures and bladder cancer in miners and millers reported mixed results. Brown et al. (1990) and Wergeland et al. (2017) reported no statistically significant associations among all workers at a mine and mill in the US (SMR = 1.43, 95% CI: 0.04-8.64) and in Norway (SMR = 1.35, 95% CI: 0.72-2.30), respectively, compared to the general populations in these countries. Coggiola et al. (2003) reported a statistically significant decreased risk of bladder cancer among all workers (SMR = 0.23, 95% CI: 0.05-0.66) and among miners (SMR = 0, 95% CI: 0-0.659), and no association among millers (SMR = 0.319, 95% CI: 0.004-1.775), in Italy compared to the general population. Wergeland et al. (2017) reported a statistically significant increase in millers first employed between 1960-1964, but not among the millers, and not among miners overall or during that same time period. Only Ciocan et al. (2022a) evaluated an exposure-response relationship, and reported no significant increase in bladder cancer mortality by increasing duration of exposure among the same cohort of Italian miners and millers, but noted a statistically significant decreased risk among those with >30 years since their first exposure (SMR = 0.157, 95% CI: 0.002-0.872) based on a single exposed case. All of the mining and milling results were based on a very small number of exposed cases (n = 1-13).

Bulbulyan et al. (1999) and Negri et al. (1989) evaluated risks of occupations with talc exposures, but did not have any information on individual exposures to talc or other substances. Bulbulyan et al. (1999) reported a statistically significant increased risk of bladder cancer mortality among press operators (SMR = 12.5, 95% CI: 1.5-45.1), but not among all workers (SMR = 2.2, 95% CI: 0.5-6.3), in Moscow printing plants compared to the general population. Similar to the milling and mining studies, there was a small number of exposed cases (n = 3) in this study, resulting in wide CIs. Negri et al. (1989) reported an increased risk of bladder cancer mortality among all workers at an Italian rubber tire factory compared to the Italian general population (SMR = 1.83, 95% CI: 1.05-2.96). Negri et al. (1989) reported point estimates and whether results were statistically significant only for each subanalysis; they did not report CIs, which limits the ability to interpret the confidence in the reported results.

Most epidemiology studies of talc and bladder cancer reported either no statistically significant or conflicting associations, and all results were based on a very small number of exposed cases. Bladder cancer risk factors (e.g. smoking) were not accounted for in any study, and most studies assessed broad categories of workers whose talc exposures may have differed. Co-exposures to other agents were also not considered. Overall, the epidemiology evidence does not support a causal association between talc exposure and bladder cancer.

Kidney cancer

We identified nine cohort studies that examined talc exposures and kidney cancer in five unique populations (Brown et al. 1990; Bulbulyan et al. 1999; Ciocan et al. 2022a; Coggiola et al. 2003; Pira et al. 2017; Wergeland et al. 2017; Wergeland et al. 1990; Stille and Tabershaw 1982; Fordyce et al. 2019). All of the identified studies examined occupational exposures to talc. Four studies were conducted in miners and millers (Brown et al. 1990; Ciocan et al. 2022a; Coggiola et al. 2003; Pira et al. 2017; Wergeland et al. 2017; Wergeland et al. 1990; Stille and Tabershaw 1982; Fordyce et al. 2019), and one in printing plant employees (Bulbulyan et al. 1999).

None of the four studies that evaluated talc exposures and kidney cancer in miners and millers reported statistically significant increased or decreased risks. Brown et al. (1990) and Fordyce et al. (2019) reported SMRs >1 in all analyses (range: 1.67-3.25), Coggiola et al. (2003) and Ciocan et al. (2022a) reported SMRs <1 in all analyses (range: 0-0.913), and Wergeland et al. (2017) reported risks >1 for millers (SMR = 1.11, 95% CI: 0.30-2.85), and <1 for miners and for the cohort as a whole (range: 0-0.87). All of the milling and mining studies were limited by the very small number of exposed cases (n = 1-5).

Bulbulyan et al. (1999) reported SMRs >1 (range: 1.4-4.4) among all printing plant employees, compositors, and bookbinders, but none of these results was statistically significant. Similar to the milling and mining studies, there was a small number of exposed cases (n = 6) in this study, resulting in wide CIs.

The epidemiology evidence does not support a causal association between talc exposure and kidney cancer. All of the studies reported non-statistically significant results with point estimates that spanned both above and below 1, and all results were based on a very small number of exposed cases.

Lymphohematopoietic system

We identified 10 cohort studies of six populations that examined talc exposures and lymphohematopoietic cancers, including leukemia, NHL, Hodgkin’s Disease, myelomas, thyroid and endocrine gland cancer, and hematopoietic and lymphatic system cancers in aggregate (Brown et al. 1990; Bulbulyan et al. 1999; Ciocan et al. 2022a; Coggiola et al. 2003; Fordyce et al. 2019; Honda et al. 20028; Negri et al. 1989; Pira et al. 2017; Stille and Tabershaw 1982; Thomas and Stewart 1987). The results of these studies are summarized in Appendix I Tables I1 to I4.

We identified eight cohort studies of five unique populations that examined leukemia mortality (Brown et al. 1990; Bulbulyan et al. 1999; Ciocan et al. 2022a; Coggiola et al. 2003; Fordyce et al. 2019; Negri et al. 1989; Pira et al. 2017; Stille and Tabershaw 1982). All of the identified studies focused on occupational exposures to talc. Six of the studies were conducted among three cohorts of talc miners and millers (Brown et al. 1990; Ciocan et al. 2022a; Coggiola et al. 2003; Fordyce et al. 2019; Pira et al. 2017; Stille and Tabershaw 1982). One study analyzed female printing plant workers in Moscow (Bulbulyan et al. 1999), and another a cohort of rubber tire factory employees (Negri et al. 1989). Among these studies, four also evaluated mortality from lymphoma (Brown et al. 1990; Coggiola et al. 2003; Fordyce et al. 2019; Stille and Tabershaw 1982).

None of the studies reported statistically significant results or any trends for any lymphohematopoietic cancer. Risk estimates were both below and above 1, and all analyses were based on ≤9 exposed cases, and most were based on ≤4 exposed cases, and very few studies examined the same specific cancer outcomes. None of the studies relied on direct exposure measurements. Overall, the epidemiology evidence does not support a causal association between talc exposure and lymphohematopoietic cancers.

Male cancers

We identified eight cohort studies in four unique populations examining talc exposures and male cancers (Brown et al. 1990; Ciocan et al. 2022a; Coggiola et al. 2003; Fordyce et al. 2019; Pira et al. 2017; Stille and Tabershaw 1982; Wergeland et al. 2017; Wergeland et al. 1990). These studies examined prostate cancer and aggregate testicular and other male genital organ cancers. The results of these studies have been summarized in Appendix J Tables J1 and J2, and prostate cancer studies are discussed below.

Prostate cancer

We identified five cohort studies that examined prostate cancer (Brown et al. 1990; Ciocan et al. 2022a; Coggiola et al. 2003; Fordyce et al. 2019; Pira et al. 2017; Stille and Tabershaw 1982; Wergeland et al. 2017; Wergeland et al. 1990) (Appendix J Table J1). All of the identified studies focused on occupational exposures among talc miners and millers. Two studies assessed prostate cancer mortality in a cohort of talc miners and millers at GTC (Brown et al. 1990; Stille and Tabershaw 1982), three studies assessed prostate cancer mortality in miners and millers in Val Chisone, Italy (Ciocan et al. 2022a; Coggiola et al. 2003; Pira et al. 2017), one study assessed prostate cancer mortality in a cohort of miners and millers in Vermont, US (Fordyce et al. 2019), and two studies assessed prostate cancer incidence in miners and millers at the Altermark Mine and Knarrevik Talc Mill in Norway (Wergeland et al. 2017; Wergeland et al. 1990).

None of these studies reported statistically significant associations with prostate cancer or trends with increased exposure. No study relied on any direct exposure measurements, and potential confounders were not fully considered in any study. Overall, the epidemiology evidence does not support a causal association between talc exposure and prostate cancer.

Brain and other CNS tumors

We identified seven occupational cohort studies of five unique populations that examined talc exposures and brain and other CNS cancers (Brown et al. 1990; Bulbulyan et al. 1999; Ciocan et al. 2022a; Fordyce et al. 2019; Negri et al. 1989; Pira et al. 2017; Stille and Tabershaw 1982). The results of these studies are summarized in Appendix K Table K1.

Five of these were studies of talc miners and millers (Brown et al. 1990; Ciocan et al. 2022a; Fordyce et al. 2019; Pira et al. 2017; Stille and Tabershaw 1982), one was a study of rubber tire workers in Italy (Negri et al. 1989), and one was as study of female printing plants workers in Russia (Bulbulyan et al. 1999). Most analyses involved ≤ 6 exposed cases. Risk estimates were both below and above 1, and none were statistically significant. None of these studies relied on direct exposure measurements. Collectively, these studies do not support causal associations between talc and brain or other CNS cancers.

Skin cancer

We identified four cohort studies of in four unique populations that examined talc exposures and skin cancer mortality (Bulbulyan et al. 1999; Coggiola et al. 2003; Fordyce et al. 2019; Negri et al. 1989). Two studies were conducted among miners and millers (Coggiola et al. 2003; Fordyce et al. 2019), one study among rubber workers (Negri et al. 1989), and one study focused on female printing plant workers (Bulbulyan et al. 1999). The results of these studies are summarized in Appendix L Table L1.

Risk estimates were both above and below one, with several risk estimates equal to zero, likely due to the small number of cases, resulting unstable risk estimates. All confidence intervals were wide because of the small number of exposed cases (≤ 3 in all analyses). Most results were not statistically significant, except Bulbulyan et al. (1999) reported an increased risk of melanoma mortality among press operators based on two cases (SMR = 8.7, 95% CI: 1.1-31.3) compared with the Moscow general population, but the risk among all printing plant employees, also based on these two cases, was not statistically significant (SMR = 1.3, 95% CI: 0.2-4.8). None of these studies relied on direct exposure measurements or controlled for important risk factors for skin cancer (e.g. sun exposure). Collectively, these studies do not support a causal association between talc and skin cancer.

Other cancers

We identified 17 cohort studies that evaluated cancers in aggregate (e.g. all cancers) in 11 unique populations (Brown et al. 1990; Stille and Tabershaw 1982; Honda et al. 2002; Bulbulyan et al. 1999; Ciocan et al. 2022a; Coggiola et al. 2003; Pira et al. 2017; Rubino et al. 1976; Fordyce et al. 2019; Selevan et al. 1979; Katsnelson and Mokronosova 1979; Straif et al. 2000; Thomas and Stewart 1987; Wergeland et al. 2017; Wergeland et al. 1990; Wild et al. 2002; Zhang et al. 1989). We also identified studies that examined the following cancers in only one population: bone cancer (Fordyce et al. 2019), connectival sarcoma (Coggiola et al. 2003), eye cancer (Fordyce et al. 2019) and peritoneal cancer (Ciocan et al. 2022a; Coggiola et al. 2003; Pira et al. 2017). There were also two cohort studies that examined talc exposures and breast cancer (Bulbulyan et al. 1999; Fordyce et al. 2019). The results of these studies are summarized in Appendix M Tables M1 to M6. The evidence for each cancer type is too limited to address causation.

Discussion

This is the first systematic review of talc and all cancer types of which we are aware. For many cancers, there were only a small number of studies available. In the results section, we provided an overview of the results for cancer types for which at least three studies in independent populations were available; the characteristics, quality, and results of all studies are tabulated in the Appendices. The evidence for most cancers either clearly demonstrated no association or was not sufficient to draw conclusions regarding causality (e.g. because of a small number of exposed cases across studies). Studies of ovarian cancer, lung cancer, and upper gastrointestinal tract cancer had positive or mixed findings and at least 10 exposed cases in most studies, so we discuss the overall plausibility of causality with respect to talc and each of these cancers, guided by Bradford Hill’s considerations, below.

Ovarian cancer

Several epidemiology and experimental studies have evaluated talc and ovarian cancer (Appendix E and Prueitt et al. 2024). No epidemiology study had direct individual talc measurements, though some used semi-quantitative metrics. Most case-control studies were likely impacted by recall and selection bias, and few studies adequately considered confounding by controlling for age, sex, and at least two of the following covariates: endometriosis, menstrual history, pregnancy history, hormone therapy, personal or family history of breast or colorectal cancer, and fertility treatment. Although limited in number and quality, experimental studies do not provide evidence for causation (Prueitt et al. 2024). Keeping these study strengths and limitations in mind, we evaluated the available evidence as a whole in the context of Bradford Hill’s considerations to determine whether it supports a causal association between talc and ovarian cancer in Table 1. Text underlined in the table is the conclusion for each consideration. We also discuss other reviews and meta-analyses of this topic.

In 2006, IARC reviewed all published research relevant to the potential carcinogenic effects of talc, including risks associated with inhalation in occupational settings as well as perineal use (IARC 2010). The review covered epidemiology evidence published up through 2006, and IARC stated that the evidence for ovarian cancer risk associated with perineal talc use was “limited” in human studies. However, some IARC Working Group members concluded that the epidemiology evidence was “inadequate” IARC (2010). The IARC Working Group indicated that recall bias “could not be ruled out” as an explanation for positive findings and that, if present, this bias would tend to inflate observed risk estimates (IARC 2010).

Wentzensen and Wacholder (2014) reviewed the available epidemiology evidence relevant to talc exposure and ovarian cancer. They concluded that the overall evidence was inconclusive, and they echoed the IARC Working Group’s concerns that case-control studies of talc use and ovarian cancer, while generally positive, were likely affected by recall bias. Wentzensen and Wacholder (2014) recommended that new techniques in exposure assessment be employed in future epidemiology studies of talc use and ovarian cancer to help resolve the issue of recall bias.

That same year, the United States Food and Drug Administration (US FDA) indicated that it did not find conclusive evidence of a causal association between perineal talc use and ovarian cancer. Specifically, US FDA noted that exposure to talc is not well characterized in epidemiology studies, such that it is not known whether there was contamination by asbestos or other asbestiform minerals, or other structurally similar compounds, and that various consumer brands or lots of talc were not identified (US FDA 2014). US FDA also noted that several studies acknowledged biases in study design and that no single study considered all the factors that could potentially contribute to ovarian cancer, including selection bias and/or confounding that could result in spurious positive associations (US FDA 2014). US FDA reported that case-control studies do not demonstrate a consistent positive association across studies; lower confidence limits are often close to 1 and exposure-response is lacking. There was also mention of the lack of specificity (i.e. exposure does not account for all cases of ovarian cancer) and the lack of a cogent biological mechanism by which talc could cause ovarian cancer. Finally, US FDA (2014) referenced the IARC (2010) review but noted that the US-based Nurses’ Health Study revealed no overall association between ever using talc and ovarian cancer.

Several meta-analyses of talc use and ovarian cancer have been published, each reporting small increased risks in case-control studies or all studies combined, but not in cohort studies. For example, Berge et al. (2018) reported a summary RR of 1.22 (95% CI: 1.13-1.30) for having ever used perineal talc and ovarian cancer. When the studies were analyzed by study type, the RR for case-control studies was 1.26 (95% CI: 1.17-1.35), while no association was found for cohort studies (RR = 1.02; 95% CI: 0.85-1.20). Weak trends were reported for duration and frequency of talc use. When stratified by cancer subtype, an association was detected for serous carcinoma, on the basis of 13 case-control studies only (RR = 1.24; 95% CI: 1.15-1.34). The authors also reported significant heterogeneity by study design (case-control vs. cohort) and noted that this result does not support a causal association between talc exposure and ovarian cancer. In addition, the authors proposed that a significant association for combined case-control, but not cohort, studies suggests that the case-control results could be explained by recall bias.

Another meta-analysis was conducted by Penninkilampi and Eslick (2018). These authors also reported a significant association between perineal use of talc and ovarian cancer in case-control studies (OR = 1.35; 95% CI: 1.27-1.43 for having ever used talc perineally) but not in cohort studies (OR = 1.06; 95% CI: 0.90-1.25), confirming the difference in outcomes between the two study types. Penninkilampi and Eslick (2018) also reported a significant association between talc use and invasive serous ovarian cancer in cohort studies. However, the authors did not use the most recent publications for two of the cohort studies (Houghton et al. [2014] was superseded by Urban et al. [2015] for the Women’s Health Initiative population, and Gertig et al. [2000] was superseded by Gates et al. [2010] for the Nurses’ Health Study population). The use of the older cohort studies did not affect the results for overall ovarian cancer risk, but it did affect the results by histological types. Urban et al. (2015) did not report results by histological type (Houghton et al. [2014] reported no association), but Gates et al. (2010) reported a small, non-statistically significant association between talc use and serous invasive ovarian cancer (HR = 1.06; 95% CI: 0.84-1.35). It is likely that if Penninkilampi and Eslick (2018) had used the most recent results from the same population as Gates et al. (2010), the association between serous invasive ovarian cancer and talc use would have been closer to 1 and not statistically significant.

Taher et al. (2019) conducted the most recent meta-analysis of talc and ovarian cancer and reported a similar association (OR = 1.32, 95% CI: 1.24-1.40) in case-control studies and a no association among cohort studies (RR = 1.06, 95% CI: 0.90-1.25). These authors evaluated study quality using the Newcastle Ottawa Scale (Wells et al. 1999), but did not fully consider the impact of study quality on the interpretation of study findings.

Important limitations inherent in meta-analysis should be considered when interpreting results. Even though they can be a powerful tool for leveraging statistical power across a number of studies, producing high precision in meta-results, a risk is that this precision can result in “over-conclusiveness.” As described by Greenland and O'Rourke (2008), the CIs and p-values associated with meta-analysis results can give the impression that results are more precise and conclusive than they really are. When interpreting meta-analysis results, it is essential to remember that CIs and p-values reflect random error only and do not reflect sources of bias; accurate CIs that account for uncertainty about bias are likely to be much wider (Greenland and O'Rourke 2008). Bias in the underlying case-control studies, such as recall bias, would bias the meta-analysis results in a similar manner, which is a particular concern in meta-analyses of talc and ovarian cancer, given that recall bias is likely to operate in a consistent manner across all case-control studies. In other words, bias in individual studies becomes compounded, in effect, in meta-analysis. Standard meta-analysis methods do not account for biases in underlying studies in any way. Greenland and O'Rourke (2008) recommended adjusting results from individual studies to account for biases, if possible, before combining results in meta-analysis and stated that “failure to fully and properly emphasize the nonrandom sources of uncertainty in a meta-analysis may encourage and even support faulty conclusions and bad policy decisions.”

A very recent systematic review of talc and ovarian cancer was published by Lynch et al. (2023). They used the IOM (2008) framework to evaluate epidemiology, animal, and mechanistic studies, and concluded that this “framework yielded classifications of suggestive evidence of no association between perineal application of talcum powders and risk of ovarian cancer at human-relevant exposure levels.” They noted that the quality of case-control studies was mixed, with most studies having issues with exposure characterization and risk of bias. They also stated that most case-control studies reported positive associations, but there were no consistent exposure-response relationships, and that cohort studies reported no associations with epithelial ovarian cancer overall.

Goodman et al. (2024) recently conducted a quantitative bias analysis on recall in talc and ovarian cancer case-control studies, using Cramer et al. (2016) as a case-study. Assumptions on sensitivity and specificity of talc use recall were based on information provided by O'Brien et al. (2023) in a study of participants of the Sister Study, a large US cohort of women with sisters with a history of breast cancer. In the Sister Study, recall sensitivity and specificity overall were 63% (95% CI: 62-64%) and 95% (95% CI: 95–95%), respectively, and they were 83% (95% CI: 66–93%) and 87% (95% CI: 78–93%), respectively, for cases. Goodman et al. (2024) found that the observed association between talc use and ovarian cancer was attenuated, with simulation intervals that included 1 or were statistically significantly lower (OR = 0.62, 95% CI: 0.36-0.95) based on the Sister Study data. In two of three scenarios modeling higher sensitivity and specificity, results were also attenuated. The Goodman et al. (2024) analysis clearly demonstrates how a modest, yet realistic, amount of recall bias can lead to false positive results.

In summary, relying on Bradford Hill’s considerations, we found evidence did not support an association between talc exposure and ovarian cancer. This is consistent with recent reviews, many of which have concluded that evidence is limited and may be affected by recall bias and other uncertainties. Meta-analysis results from prospective cohort studies consistently show no association. In contrast, the meta-analyses indicate that case-control studies collectively report small positive associations between talc and ovarian cancer, but it is difficult to interpret these results without accounting for bias in underlying studies. Narrow CIs and small p-values associated with meta-analyses of case-control studies should be interpreted with caution to avoid overconfidence in the results. Given critical limitations, the results of meta-analyses of talc use and ovarian cancer showing highly precise, modestly positive associations are likely impacted by recall and other biases, and despite being statistically significant, they do not support a causal association between talc and ovarian cancer.

Respiratory cancers

Several epidemiology and experimental studies have evaluated talc and respiratory cancers (Appendix F and Prueitt et al. 2024). No epidemiology study had direct individual talc measurements, though some used semi-quantitative metrics, and very few adequately considered smoking. Experimental studies do not provide evidence for lung cancer causation except in rats exposed to doses high enough to cause lung particle overload; these studies are not relevant to human exposures (Prueitt et al. 2024). No studies found increased risks of mesothelioma. We evaluated the available evidence as a whole in the context of Bradford Hill’s considerations in Table 2 to determine whether it supports a causal association between talc and lung cancer. We also discuss other reviews and meta-analyses of this topic.

Wild (2006) conducted the first review of talc and lung cancer of which we are aware. Wild (2006) concluded that exposure information was generally lacking, but there was no increase in lung cancer mortality among talc millers, who had high exposures. They found that lung cancer risks were only increased when talc was associated with other possible carcinogens. Similarly, IARC (2010) concluded that there was “inadequate evidence from epidemiological studies to assess whether inhaled talc not containing asbestos or asbestiform fibers causes cancer in humans.” The IARC Working Group noted there were no consistent patterns in studies of talc millers and miners, which it “considered to provide the best source of evidence,” and that the one study that observed an increased lung cancer risk among miners may have been confounded by exposures to other carcinogens” IARC (2010).

Several years later, Chang et al. (2017) conducted a meta-analysis of talc exposure and lung cancer. They reported a significant association (SMR = 1.45, 95% CI: 1.22–1.72, p < 0.0001), but the studies had a high degree of heterogeneity (I-squared = 72.9%), and talc exposures were likely minimal in several of the meta-analyzed studies. Fewer than half of the studies had any information on smoking, and none adjusted for smoking.

Ciocan et al. (2022b) conducted a systematic review and meta-analysis of malignant and nonmalignant respiratory diseases in talc millers and miners. They reported an increased lung cancer risk overall (SMR = 1.42, 95% CI: 1.07-1.89), but concluded, “The small excess in lung cancer mortality may be, in part, explained by the high prevalence of the smokers in some of the analyzed cohorts or by the exposure to other carcinogens like radon decay products and diesel engine exhaust” Ciocan et al. (2022b). They did not meta-analyze mesothelioma risks due to an insufficient number of studies.

Lynch et al. (2023) conducted a systematic review of industrial and cosmetic talc and respiratory cancers. They found that lung cancer risk was generally not elevated across studies, even in highly exposed millers and miners, and that smoking and previous employment was not adequately considered. Lynch et al. (2023) also noted the small number of mesothelioma cases in people exposed to talc.

Finally, while we excluded studies of pleurodesis because pleurodesis is used to treat mesothelioma (among other conditions), Finley et al. (2017a) reviewed retrospective clinical studies of over 300 patients who received pleurodesis treatments and reported that none developed mesothelioma.

We concur with the conclusions of several recent reviews that epidemiology evidence does not support a causal association between talc exposure and respiratory cancers.

Upper gastrointestinal tract cancers

We identified several epidemiology studies of oral and pharyngeal, esophageal, and stomach cancer that reported mixed results (Appendix G). No epidemiology study had direct individual talc measurements except for Chang et al. (2019) which evaluated oral intake of talc powder as a medical treatment, though some studies used semi-quantitative metrics. Experimental studies do not provide evidence for causation (Prueitt et al. 2024). We evaluated the available evidence for upper gastrointestinal tract cancers as a whole in the context of Bradford Hill’s considerations in Table 3 to determine whether it supports a causal association between talc and upper gastrointestinal (GI) tract cancers. We also discuss other reviews of this topic.

Chang et al. (2018) conducted a meta-analysis of talc exposure and stomach cancer. They included several studies written in Chinese and a proportionate mortality (PMR) study (Stern et al. 2001). PMR studies did not meet our inclusion criteria because they have several limitations that make them difficult to interpret. The authors reported that all workers exposed to talc had an increased risk of stomach cancer (meta-RR of 1.21, 95% CI: 1.03-1.42), though risks in talc-using industries were not significant (meta-RR = 1.18, 95% CI: 0.93-1.50). When Stern et al. (2001) was excluded from the analysis the results were no longer significant (meta-RR = 1.11, 95% CI: 0.96-1.29). We are not aware of any other reviews or meta-analyses of upper gastrointestinal cancers.

Given the lack of direct exposure measurements, inconsistent results, lack of control for smoking and alcohol intake, we conclude the evidence does not support talc as a potential cause of upper gastrointestinal cancer.

Conclusions

This is the first systematic review of all types of talc exposure (i.e. occupational, medicinal, and personal care product use) and all cancer types. We reviewed the literature and tabulated study characteristics, quality, and results in a systematic manner, and evaluated all cancer types for which studies of at least three unique populations were available in a narrative review. We focused on study quality aspects most likely to impact the interpretation of results. We found that with only one exception (Chang et al. 2019), no studies had direct exposure measurements for any individuals, though some used semi-quantitative exposure metrics, and few studies adequately assessed potential confounders. The only consistent associations were with ovarian cancer in case-control studies, and these associations were likely impacted by recall bias. This systematic review indicates that epidemiology studies do not support a causal association between occupational or personal talc exposure and any cancer in humans. Future studies using causal methods could complement this analysis and be helpful to address talc and cancer causation.

Abbreviations
BRCA1	=	Breast Cancer Gene 1
CI	=	Confidence Interval
CNS	=	Central Nervous System
EMA	=	Essential Minerals Association
GI	=	Gastrointestinal
GTC	=	Gouverneur Talc Company
HR	=	Hazard Ratio
IARC	=	International Agency for Research on Cancer
ICD	=	International Classification of Diseases
IOM	=	Institute of Medicine
IRIS	=	Integrated Risk Information System
JEM	=	Job Exposure Matrix
NHL	=	Non-Hodgkin’s Lymphoma
NTP	=	National Toxicology Program
OHAT	=	Office of Health Assessment and Translation
OR	=	Odds Ratio
ORD	=	Office of Research and Development
OSF	=	Open Science Framework
PECOS	=	Population, Exposure, Comparator, Outcomes, and Study Design
PMR	=	Proportionate Mortality
PRISMA	=	Preferred Reporting Items for Systematic Reviews and Meta-Analyses
RR	=	Risk Ratio
SIR	=	Standardized Incidence Ratio
SMR	=	Standardized Mortality Ratio
US EPA	=	United States Environmental Protection Agency
US FDA	=	United States Food and Drug Administration

Acknowledgments

The authors thank the Gradient staff members who provided essential assistance in preparing this paper: Rich Gomes for editorial assistance, Eric Fischbach for administrative assistance, and Holly Howes for assistance with the literature searches. We also thank Dr. Lorenz Rhomberg, an Advising Principal at Gradient, and the five external peer reviewers selected by the journal editor, whose identities were anonymous to the authors, for reviewing this paper and providing helpful comments that strengthened the quality of this paper.

Declaration of interest

All authors are employed by Gradient, an environmental sciences consulting firm founded in 1985 (https://gradientcorp.com/). Gradient conducts work for both private and government clients. Funding for this work was provided by Essential Minerals Association (EMA), an organization that represents the interests of companies that mine or process minerals (https://www.essentialminerals.org/). Gradient prepared this analysis with the intent of submitting it to the International Agency for Research on Cancer (IARC) Monographs Meeting 136: Talc and Acrylonitrile upon journal acceptance. The authors reserve the right to submit this accepted manuscript to other scientific or regulatory groups or legal proceedings.

This paper was drafted during the authors’ normal course of employment. The authors have sole responsibility for the writing and content of this paper. This paper represents the professional opinions of the authors and not necessarily those of EMA or its member companies. EMA and its member companies were not involved with the study design, analysis, interpretation of the evidence, or writing of this paper. EMA and its member companies did not have input on the substance of this paper and did not review this paper prior to its acceptance. This paper was not reviewed by anyone except the authors and the individuals listed in the Acknowledgments section prior to its acceptance.

Gradient has previously submitted comments to US EPA, US FDA, and Health Canada regarding issues related to talc measurements and epidemiology. These activities were funded by the Cosmetics Alliance (CA) Canada, the Industrial Minerals Association – North America (IMA-NA, now known as EMA), and the National Stone Sand and Gravel Association (NSSGA). CA Canada and IMA-NA also funded previous Gradient work on recall bias in talc ovarian cancer studies and a critical review of talc and ovarian cancer. Gradient recently published a paper on recall bias that was informed by some of this earlier work, but did not receive funding for the analysis or preparation of the paper for publication (Goodman JE, Espira LM, Zu K, Boon D. 2024. Quantitative recall bias analysis of the talc and ovarian cancer association. Glob. Epidemiol. 7:100140).

Dr. Julie E. Goodman is on the Scientific Advisory Board of NSSGA. Dr. Goodman has also served as an expert witness in litigation involving talc. Drs. Goodman and Robyn L. Prueitt have served as scientific expert witnesses in litigation involving asbestos.

Supplemental material

Supplemental data for this article can be found on the https://doi.org/10.1080/10408444.2024.2351081.

Notes

1 Cosmetic talc refers to talc in cosmetics and personal care products throughout this paper.

2 We are not aware of a standard cutoff for what constitutes a small number of exposed cases. We chose 10 because risk estimates based on <10 exposed cases are generally statistically unstable.