A critical review of talc and ovarian cancer

ABSTRACT

The association between perineal talc use and ovarian cancer has been evaluated in several epidemiology studies. Some case-control studies reported weak positive associations, while other case-control and three large prospective cohort investigations found this association to be null. A weight-of-evidence evaluation was conducted of the epidemiology, toxicity, exposure, transport, in vitro, and mechanistic evidence to determine whether, collectively, these data support a causal association. Our review of the literature indicated that, while both case-control and cohort studies may be impacted by bias, the possibility of recall and other biases from the low participation rates and retrospective self-reporting of talc exposure cannot be ruled out for any of the case-control studies. The hypothesis that talc exposure induces ovarian cancer is only supported if one discounts the null results of the cohort studies and the fact that significant bias and/or confounding are likely reasons for the associations reported in some case-control investigations. In addition, one would need to ignore the evidence from animal experiments that show no marked association with cancer, in vitro and genotoxicity studies that did not indicate a carcinogenic mechanism of action for talc, and mechanistic and transport investigations that did not support the retrograde transport of talc to the ovaries. An alternative hypothesis that talc does not produce ovarian cancer, and that bias and confounding contribute the reported positive associations in case-control studies, is better supported by the evidence across all scientific disciplines. It is concluded that the evidence does not support a causal association between perineal talc use and ovarian cancer.

KEYWORDS:

• Talc
• ovarian cancer
• weight of evidence
• mode of action

Introduction

Many epidemiology studies have been conducted to investigate the potential association between perineal exposure to cosmetic-grade talc and ovarian cancer. There is no known well-established cause for most cases of ovarian cancer (American Cancer Society (ACS), 2019). One hypothesis is that the tissue damage and subsequent repair to the ovary that occurs during each ovulation may increase the risk of cancer due to gene replication errors during the tissue repair process (ACS 2019; Hankinson and Danforth 2006; Purdie et al. 2003). Another hypothesis is that ovarian cancer may be mediated by fluctuating levels of endogenous hormones or higher endogenous levels of certain sex hormones in some individuals (Hankinson and Danforth 2006; Lukanova and Kaaks 2005).

A woman’s risk of developing epithelial ovarian cancer may be affected by a wide range of factors. Approximately 10-20% of ovarian cancers are believed to be initiated by inherited factors The relative contributions of other risk factors are poorly understood, and the remaining 80-90% of cases are attributed to unknown causes (i.e., are idiopathic) (Hankinson and Danforth 2006; Hunn and Rodriguez 2012; Walsh et al. 2011). Factors associated with an elevated risk of ovarian cancer include increasing age; longer duration of menstruation (i.e., starting menstruation before age 12 or undergoing menopause after age 52); obesity; hormone therapy after menopause; family history of ovarian, colorectal, or breast cancer; and personal history of breast cancer (ACS 2019; CDC, 2015; Mayo Clinic 2019; Memorial Sloan Kettering Cancer Center 2019; Booth, Beral, and Smith 1989; La Vecchia 2017). Factors associated with a reduced risk of ovarian cancer include reproductive history (risk decreases with each full-term pregnancy in women under 35), oral contraceptive use, and gynecologic surgery such as tubal ligation or hysterectomy (ACS 2019; Memorial Sloan Kettering Cancer Center 2019; Booth, Beral, and Smith 1989; La Vecchia 2017).

Cosmetic-grade talc is used in baby powders, feminine hygiene products, antiperspirants, deodorants, creams, hair care products, lipsticks, shampoos, shaving products, wound ointments, foot powders, and sun care products (Fiume et al. 2015). Talc is also used in medical applications, including for pleurodesis, which is a medical procedure performed to treat malignant pleural effusions (Rosenman 2012). In 1976, cosmetic-grade talc specifications required that there be no detectable fibrous, asbestos minerals (Fiume et al. 2015).

Women have reported utilizing talc-containing body powder in the genital/rectal area, on the feet and or thighs, on sanitary napkins, and on underwear, as well as to store diaphragms (Cramer et al. 1999; Ness et al. 2000; Wong et al. 1999). Exposure via the genital tract might also occur because many brands of condoms use talc as a surface lubricant (Cook, Kamb, and Weiss 1997).

The association between perineal talc use and ovarian cancer was examined in a number of epidemiology studies in the past 4 decades. Some case-control studies and meta-analyses reported associations, while other case-control investigations, three large prospective cohort studies, and a pooled analysis of cohort investigations found this association to be null (Cramer et al. 1999; Gross and Berg 1995; Huncharek, Geschwind, and Kupelnick 2003; Huncharek and Muscat 2011; Huncharek et al. 2007; IARC, 2010; Langseth et al. 2008; Musser 2014; NCI, 2016; O’Brien et al. 2020; Penninkilampi and Eslick 2018; Taher et al. 2019; Wentzensen and Wacholder 2014).

In light of the questions surrounding talc and ovarian cancer, a weight-of-evidence evaluation was conducted of epidemiology, toxicity, exposure and transport, in vitro, and mechanistic evidence to determine whether, collectively, relevant available evidence supports a causal association. To our knowledge, this is the first review of the body of evidence across scientific disciplines that considers study quality and risk of bias and their impact on the interpretation of results.

Methods

Literature search strategy and study selection

The PubMed database was searched for studies published through January, 2020, that examined talc and ovarian cancer in humans or any cancer in animals, genotoxic or in vitro effects of talc, or exposure and transport of talc in the reproductive tract using several search terms: (talc OR talcum) AND (cancer OR tumor OR tumors OR neoplasm) AND “animal”; (talc OR talcum) AND (ovarian OR ovary OR ovaries OR reproductive OR fallopian OR uterus OR cervix OR cervical OR vagina) AND (epidemiol* OR cohort OR women OR female OR adult OR adults OR population); (talc OR talcum) AND (genotoxic* OR DNA); (talc OR talcum) AND “in vitro”; (talc OR talcum) AND (perineal OR genital); (talc OR talcum) AND (transport OR migration). In addition, references were checked in review articles of talc and cancer to identify any studies that may not have been identified by our literature search.

Weight-of-evidence evaluation

To assess whether talc use might result in an increased risk of ovarian cancer, a weight-of-evidence analysis was conducted across scientific disciplines. This involved reviewing all relevant studies, considering strengths and weaknesses of each, and weighing their points of agreement and contradiction. In a weight-of-evidence evaluation, all relevant evidence must be considered as a collective body to enable a robust and balanced analysis. With this objective, all of the relevant epidemiology, toxicity, exposure and transport, and mechanistic evidence was assessed in the context of the Bradford Hill considerations of strength of association, consistency, specificity, temporality, dose-response, biological plausibility, coherence, experiment, and analogy. Both positive and null studies were evaluated to determine overall plausibility for causality in humans, bearing in mind study quality and relevance, uncertainties and inconsistencies in the evidence, and ad hoc assumptions that may be required to accept a causal association.

Study quality was determined by considering study design, study size, participation rate, exposure assessment, and adjustment for confounders. Investigations were considered to be of higher quality if exposures were assessed prospectively and if these included a validation study to assess the accuracy of self-reported exposures, a larger sample size, a higher participation rate, a longer follow-up period, and adequate adjustment for confounders. When evaluating results across epidemiology studies, more weight was placed on higher quality studies such as those that considered their results to be more reliable and thus informative to the evaluation of whether they support perineal talc use as a cause of ovarian cancer. To determine consistency of results across epidemiology studies, results were tabulated and whether effects were evident across studies, regardless of magnitude or statistical significance, was assessed. If a pattern was observed, it might indicate either a causal association or a consistent bias or confounding factor across investigations.

To evaluate the evidence in animal studies, the route, magnitude, and duration of exposure, and the outcome such as presence of tumors were considered. For in vitro and genotoxicity studies, the magnitude and duration of exposure to the cells, and the cellular effects including cell proliferation, oxidative stress, or DNA effects were taken into account. For experiments of talc migration in the reproductive tract and quantitation of talc in reproductive tissues, the mode of talc delivery and comparative amounts of talc (or particles identified as possible talc) in tissues of women who had differing exposures were considered, as well as the processes by which the particle content of reproductive tissues were evaluated. Experimental study quality was assessed on the basis of exposure characterization, outcome assessment, and control groups. Investigations were given more weight if they included well-characterized exposures that are relevant to humans in terms of route and amount, sufficient time for tumors to develop, appropriate statistical analyses, and appropriate controls, including positive controls for tumor development and negative controls for inert dust exposure.

Results

Epidemiology

Three prospective cohort studies and 27 case-control studies were identified that evaluated the association between perineal talc or body powder use and ovarian cancer. The cohort studies were all conducted in the US and were described in 5 articles published since 2000 and a recent pooled analysis. The characteristics and results of the cohort studies are summarized in Table 1. The case-control studies were conducted in a variety of countries – including the US, Canada, United Kingdom, Norway, Greece, Australia, and China – and described in 33 papers published since the early 1980s. Twenty-seven of these are population-based, 5 are hospital-based studies, and one includes both general and hospital-based populations. The characteristics and results of the case-control studies are summarized in Table 2.

Table 1. Talc and ovarian cancer cohort study characteristics and results.

					Potential Confounders Adjusted								Risk Estimate^a (95% Confidence Interval [CI]) for Mode of Perineal Talc Use				Extent of Perineal Talc Exposure ptrend or Risk Estimate (95% CI)^b		Histological Subtypes of Invasive Ovarian Cancer^c
Study	Cohort	Sample Size	Follow-up (years)	Loss to Follow-up (%)	Age	Obesity	Family History	Parity	OC Use	Tubal Ligation	Hysterectomy	PMH Use	Any	Dusting Powder	Sanitary Napkin	Diaphragm	Frequency	Duration	Serous	Mucinous	Endometrioid	Clear Cell
Gertig et al. (2000)	NHS	78,630	14	2	✓	✓		✓	✓	✓		✓	1.09(0.86, 1.37)		0.89 (0.61, 1.28)		1.12 (0.82, 1.55)^d		1.4 (1.02, 1.91)	0.93 (0.53, 1.66)	0.91 (0.49, 1.87)
Gates et al. (2010)	NHS	108,870	24	4.8–6.4	✓		✓	✓	✓	✓	✓	✓							1.06 (0.84, 1.35)^e	1.50 (0.84, 2.66)^e	1.06 (0.66, 1.69)^e
Houghton et al. (2014)	WHI-OS	61,576	12.4 (average)	NR	✓	✓	✓	✓	✓	✓		✓	1.06 (0.87, 1.28)	1.12 (0.92, 1.36)	0.95 (0.76, 1.20)	0.92 (0.68, 1.23)		0.77	1.13 (0.84, 1.51)	1.03 (0.47, 2.27)	1.29 (0.64, 2.61)
Urban et al. (2015)	WHI-OS	74,786	12.3 (average)	NR	✓													0.97 (0.78, 1.22)^f
Gonzalez et al. (2016)	Sister Study	41,654	6.5 (median)	0.1	✓			✓	✓		✓		0.73 (0.44, 1.2)^g
O’Brien et al. (2020)	Pooled (NHS, NHSII, WHI-OS, Sister Study)	252,745	11.2 (median)	NR		✓		✓	✓	✓	✓	✓	1.08 (0.99, 1.17)				0.20	0.49	1.10 (0.97, 1.25)^h	1.03 (0.69, 1.54)^h	1.15 (0.83, 1.58)^h	1.17 (0.73, 1.89)^h

Notes:

NHS = Nurses’ Health Study; NHSII = Nurses’ Health Study II; OC = Oral Contraceptive; PMH = Postmenopausal Hormones; NR = Not Reported; WHI-OS = Women’s Health Initiative Observational Study.

Bold denotes a statistically significant result.

(a) All studies used hazard ratios as the risk metric, with the exception of the two studies of the NHS cohort, which used relative risks.

(b) P_trend is presented, unless otherwise noted.

(c) Risk estimates for invasive cancers are presented when available.

(d) Relative risk for daily talc use vs. never use.

(e) Relative risk for talc use at least once per week vs. less than once per week.

(f) Hazard ratio for talc use of a duration of more than 10 years vs. less than 10 years.

(g) For any talc use in the 12 months prior to study enrollment.

(h) Hazard ratio for ever use of talc vs. never use.

Table 2. Talc and ovarian cancer case-control study characteristics and results.

			Number		Response Rates		Potential Confounders Adjusted
Study	Location	Control	Cases	Controls	Cases	Controls	Age	Obesity	Family History	Parity	OC Use	Tubal Ligation	Hysterectomy	PMH Use
Cramer et al. (1982)	US	P	215	215	72.4%	48.2%	✓	✓		✓	✓			✓
Hartge et al. (1983)	US	H	135	171	NR	NR	✓
Whittemore et al. (1988)	US	H, P	188	539^d	NR	NR	✓			✓	✓
Booth, Beral, and Smith (1989)	UK	H	235	451	NR	NR	✓
Harlow and Weiss (1989)	US	P	116	158	68.0%	74.0%	✓			✓	✓
Harlow et al. (1992)	US	P	235	239	69.0%	50.5%	✓	✓		✓
Chen et al. (1992)	China	P	112	224	48.7%	93.7%	✓			✓
Rosenblatt, Szklo, and Rosenshein (1992)	US	H	77	46	77.1%	NR		✓		✓		✓
Tzonou et al. (1993)	Greece	H	189	200	90.0%	94.0%	✓	✓		✓
Cramer and Xu (1995)	US	P	450	454	NR	NR
Purdie et al. (1995)	Australia	P	824	860	90.0%	73.0%				✓
Chang and Risch (1997)	Canada	P	450	564	71.3%	64.6%	✓		✓	✓	✓	✓	✓
Cook, Kamb, and Weiss (1997)	US	P	313	422	64.3%	68.0%	✓
Green et al. (1997)	Australia	P	824	855	90.0%	73.0%	✓	✓	✓	✓	✓
Godard et al. (1998)	Canada	P	170	170	87.0%	22.7%	✓					✓
Cramer et al. (1999)	US	P	563	523	64.2%	72.0%	✓	✓	✓	✓	✓	✓
Wong et al. (1999)	US	H	499	755	NR	NR	✓		✓	✓	✓	✓	✓
Ness et al. (2000)	US	P	767	1,367	87.9%	74.0%	✓		✓	✓	✓	✓	✓
Langseth and Kjaerheim (2004)	Norway	N	35	121	76.1%	57.1%	✓		✓	✓
Mills et al. (2004)	US	P	256	1,122	40.3%	57.1%	✓				✓
Jordan et al. (2007)	Australia	P	363	752	84.0%	47.0%	✓			✓	✓		✓
Merritt et al. (2008)	Australia	P	1,576	1,509	74.0%	47.1%	✓			✓	✓
Gates et al. (2008)	US	P	1,175	1,202	71.0%	68.0%	✓	✓		✓	✓	✓		✓
Wu et al. (2009)	US	P	609	688	62.5%	NR	✓		✓	✓	✓	✓
Moorman et al. (2009)^i	US	P	550	533	66.5%	43.9%	✓
Moorman et al. (2009)^j	US	P	83	134			✓
Rosenblatt et al. (2011)	US	P	812	1,313	76.7%	69.0%	✓			✓	✓
Kurta et al. (2012)	US	P	902	1,802	71.0%	46.1%	✓
Terry et al. (2013)	US, Canada, Australia	P	8,525	9,859	NR	NR	✓	✓		✓	✓	✓
Wu et al. (2015)	US	P	1,701	2,391	63.2%	NR	✓	✓	✓	✓	✓	✓		✓
Cramer et al. (2016)	US	P	2,041	2,100	71.5%	56.4%	✓	✓	✓	✓	✓	✓	✓	✓
Schildkraut et al. (2016)	US	P	584	745	50.0%	61.0%	✓	✓	✓	✓	✓	✓
Gabriel et al. (2019)	US	P	2,040	2,100	71.0%	54.0%	✓	✓		✓	✓	✓
	Relative Risk (95% Confidence Interval) for Mode of Perineal Talc Use				Extent of Perineal Talc Exposure ptrend or Risk Estimate (95% CI)^a			Histological Subtypes of Invasive Ovarian Cancer^b
Study	Any	Dusting Powder	Sanitary Napkin or Underwear^c	Diaphragm or Cervical Cap	Frequency	Duration	Total Applications	Serous	Mucinous	Endometrioid	Clear Cell
Cramer et al. (1982)	1.61(1.04, 2.49)
Hartge et al. (1983)	2.5(0.7, 10.0)			0.8(0.4, 1.4)
Whittemore et al. (1988)	1.36(0.91, 2.04)	1.45(0.81, 2.60)	0.62(0.21, 1.80)	1.5(0.63, 3.58)	0.19	0.61
Booth, Beral, and Smith (1989)		1.3^e(0.8, 1.9)		Null	0.05
Harlow and Weiss (1989)	1.1(0.7, 2.1)	1.2(0.6, 2.6)	2.2(0.8, 19.8)	0.5(0.2, 1.4)
Harlow et al. (1992)	1.5(1.0, 2.1)	1.7(1.1, 2.7)	1.1(0.4, 2.8)	1.2(0.6, 2.4)	0.046	0.07	0.015
Chen et al. (1992)		3.9(0.9, 10.6)
Rosenblatt, Szklo, and Rosenshein (1992)	1.0(0.2, 4.0)	1.7(0.7, 3.9)	4.8(1.3, 17.8)	3.0(0.8, 10.8)
Tzonou et al. (1993)		1.05(0.28, 3.98)
Cramer and Xu (1995)	1.6(1.2, 2.1)
Purdie et al. (1995)		1.27(1.04, 1.54)
Chang and Risch (1997)	1.42(1.08, 1.86)	1.31(1.00, 1.73)	1.26(0.81, 1.96)		Null	Null		1.34(0.96, 1.85)	1.59(0.97, 2.58)	1.67(1.00, 2.79)
Cook, Kamb, and Weiss (1997)	1.5(1.1, 2.0)	1.8(1.2, 2.9)	1.5(0.6, 3.6)	0.8(0.4, 1.4)			Null	1.7(1.1, 2.5)	0.7(0.4, 1.4)	1.2(0.6, 2.3)
Green et al. (1997)		1.3(1.1, 1.6)				Null
Godard et al. (1998)		2.49(0.94, 6.58)
Cramer et al. (1999)	1.60(1.18, 2.15)	1.45(0.97, 2.18)	1.45(0.68, 3.09)			0.48	0.68	1.70(1.22, 2.39)	0.79(0.44, 1.40)	1.04 ^f(0.67, 1.61)
Wong et al. (1999)	0.92 (0.24, 3.62)	1.0(0.8, 1.3)	0.9(0.4, 2.0)			Null
Ness et al. (2000)		1.5(1.1, 2.0)	1.7(1.2, 2.4)	0.6(0.3, 1.2)		Null
Langseth and Kjaerheim (2004)	1.15 (0.41, 3.21)
Mills et al. (2004)	1.37(1.02, 1.85)				0.015	0.045	0.051	1.77(1.12, 2.81)	2.56(0.89, 7.39)	1.28(0.62, 2.62)	0.63(0.15, 2.64)
Jordan et al. (2007)		1.10(0.84, 1.45)			0.3 ^g
Merritt et al. (2008)	1.17(1.01, 1.36)					0.021		1.21(1.03, 1.44)	1.10(0.80, 1.52)	1.18(0.81, 1.70)	1.08(0.68, 1.72)
Gates et al. (2008)		1.40(1.15, 1.70)			0.002			1.62(1.26, 2.09)
Wu et al. (2009)		1.53(1.13, 2.09)	1.71(0.99, 2.97)	1.14(0.46, 2.87)	0.032 ^h		0.0004
Moorman et al. (2009)^i	1.04(0.82, 1.33)
Moorman et al. (2009)^j	1.19(0.68, 2.09)
Rosenblatt et al. (2011)		1.27 (0.97, 1.66)	0.82 (0.58,1.16)	0.72 (0.48, 1.10)		Null	Null	1.01 (0.69, 1.47)		1.53 ^f(0.91, 2.57)
Kurta et al. (2012)		1.40(1.16, 1.69)
Terry et al. (2013)		1.24(1.15, 1.33)					0.17	1.20(1.09, 1.32)	1.09(0.84, 1.42)	1.22(1.04, 1.43)	1.24(1.01, 1.52)
Wu et al. (2015)		1.46(1.27, 1.69)				1.14 ^k(1.09, 1.20)
Cramer et al. (2016)	1.32(1.15, 1.53)	1.42(1.04, 1.96)		0.73(0.57, 0.93)	<0.0001	0.002	0.02	1.42(1.19, 1.69)	0.87(0.53, 1.44)	1.38(1.06, 1.80)	1.01(0.65, 1.57)
Schildkraut et al. (2016)	1.44(1.11, 1.86)				<0.01	0.02	0.01
Gabriel et al. (2019)	1.30 (1.13–1.50)				0.0001 ^h			1.39 (1.14–1.69)	0.97 (0.68–1.40)	1.26 (0.94–1.69)	1.08 (0.66–1.78)

Notes:

P = Population-based; H = Hospital-based; N = Nested within the same occupational cohort; NR = Not Reported; OC = Oral Contraceptive; PMH = Postmenopausal Hormones.

Bold denotes a statistically significant result. Results are classified as null if a study reported they were not statistically significant but did not report risk estimates or p-values.

(a) P_trend is presented, unless otherwise noted.

(b) Risk estimates for invasive cancers are presented when available.

(c) If results for sanitary napkins and underwear were reported separately, I present the larger risk estimate of the two.

(d) 280 hospital-based controls, 259 population-based controls.

(e) Daily use.

(f) For endometrioid/clear cell combined.

(g) Frequency for use in the perineal region.

(h) Frequency and duration were evaluated together.

(i) Caucasian women.

(j) African American women.

(k) Relative risk for every five years of talc use.

Study quality

In the cohort studies, perineal exposures to talc were assessed prospectively (i.e., before the diagnosis of ovarian cancer). This eliminates the possibility of differential recall bias between those with and without ovarian cancer. However, exposures were still self-reported, with limited exposure information that was not quantitative and often only assessed frequency or duration of talc use, but not both. In addition, exposure information was not updated after the baseline interview, increasing the potential for non-talc users to become talc users (and vice versa). Each of these factors might lead to potential exposure misclassification.

All three cohorts contained large sample sizes and follow-up times that enabled detection of small increases in ovarian cancer risk, although the study by Gonzalez et al. (2016) included a follow-up period (median 6.5 years) that was relatively short with regard to latency for ovarian cancer, given that information on talc use in this study was only collected for the 12-month period prior to participant enrollment (see below and Table 1). Some of the cohort investigators collected information on talc exposure duration, such as whether talc use occurred for up to 30 years or more prior to participant enrollment (Houghton et al. 2014; O’Brien et al. 2020; Urban et al. 2015). The cohorts also had ample information on demographic, lifestyle, and reproductive factors that enabled robust adjustment for potential confounders.

With regard to case-control studies, some were more robust than others methodologically; for example, some included fewer than 100 cases or controls, whereas others included 1,000 or more cases or controls (see Table 2). As with cohort studies, some collected information on whether talc exposure occurred for up to 30 years or more (see Table 2 for investigations that examined duration of exposure). However, many of the case-control studies did not assess or adjust for several established risk factors for ovarian cancer such as obesity, oral contraceptive or postmenopausal hormone use, and history of gynecological surgeries. In addition, all of the case-control studies were subject to recall bias, given that study participants needed to recall exposure that occurred many years in the past, and none of the investigations included a validation study to assess the accuracy of these self-reports.

Even the higher-quality case-control studies exhibited considerable methodological limitations. For example, the study by Cramer et al. (2016) contained extensive information on talc exposure and other covariates, assessed the extent of talc use and exposure-response relationships, and evaluated different histological subtypes of ovarian cancer; however, this study and several others reported differential response rates between cases and controls, and the response rate was very low (see Table 2), indicating a potential for selection bias.

Cohort studies

Below, results of the three prospective cohort studies that evaluated the association between perineal exposures to talc and ovarian cancer are described: The Nurses’ Health Study (NHS), the Women’s Health Initiatives Observational Study (WHI-OS), and the Sister Study. A pooled analysis is also presented that included results from each of these cohort studies. The results of these studies are summarized in Table 1.

The Nurses’ Health Study

Gertig et al. (2000) and Gates et al. (2010) evaluated the association between perineal talc use and ovarian cancer risk in the NHS, an ongoing large-scale prospective cohort of 121,700 registered female nurses in the US aged 30–55 years in 1976. Questionnaires were mailed to participants at baseline and every two years during follow-up, to obtain information on medical history, lifestyle factors, and health-related issues. Use of talc powder was assessed on the 1982 questionnaire by asking whether the participants had ever commonly applied talc, baby powder, or deodorizing powder to the perineal area or on sanitary napkins. The frequency of talc use was also assessed on that questionnaire. Incident cases of ovarian cancer were self-reported through biennial questionnaires and searches of the National Death Index during follow-up, and confirmed by medical record review or death certificates. Potential confounders including body mass index (BMI), physical activity, smoking, age at menarche, parity, oral contraceptive use, tubal ligation, hysterectomy/oophorectomy, menopausal status, age at menopause, postmenopausal hormone use, and family history of breast and ovarian cancer were assessed in this cohort at baseline and/or on multiple questionnaires during follow-up.

Talc use was determined once at baseline in the NHS prior to the onset of ovarian cancer in any participant. Gertig et al. (2000) identified 307 incident ovarian cancer cases during 14 years of follow-up (1982–1996) and reported that having ever used talc perineally was not associated with an elevated risk of ovarian cancer (relative risk [RR] = 1.09, 95% confidence interval [CI]: 0.86–1.37). Talc use on sanitary napkins was also not associated with ovarian cancer risk (RR = 0.89, 95% CI: 0.61–1.28). Ovarian cancer risk did not change with increased frequency of perineal talc use (< 1 use/week: RR = 1.14, 95% CI: 0.81–1.59; 1–6 uses/week: RR = 0.99, 95% CI: 0.67–1.46; daily use: RR = 1.12, 95% CI: 0.82–1.55). When stratified by histological subtype of ovarian cancer, having ever used talc was associated with a small rise in risk for serous invasive cancers (RR = 1.40, 95% CI: 1.02–1.91), but not for endometrioid cancers (RR = 0.91, 95% CI: 0.49–1.87) or mucinous cancers (RR = 0.93, 95% CI: 0.53–1.66). There was a borderline significant linear trend between frequency of talc use and risk of serous invasive cancers (< 1 use/week: RR = 1.29, 95% CI: 0.81–2.04; 1–6 uses/week: RR = 1.49, 95% CI: 0.77–2.11; daily use: RR = 1.49, 95% CI: 0.98–2.26; p_trend = 0.05).

The most recent analyses of talc use and ovarian cancer in the NHS cohort were reported by Gates et al. (2010). During 24 years of follow-up (1982–2006), a total of 797 ovarian cases were identified with confirmed histological subtypes. A higher frequency of use of talc on the perineum (at least once per week vs. less than once per week) was not associated with an increased risk of ovarian cancer overall (RR = 1.06, 95% CI: 0.89–1.28), nor with any of the subtypes (serous invasive: RR = 1.06, 95% CI: 0.84–1.35; endometrioid: RR = 1.06, 95% CI: 0.66–1.69; mucinous: RR = 1.50, 95% CI: 0.84–2.66). There was no evidence that talc use might have different impacts on different histological types (p_{heterogeneity} = 0.55).

The Women’s Health Initiatives Observational Study

Houghton et al. (2014) and Urban et al. (2015) examined talc use and ovarian cancer in the WHI-OS. The WHI-OS cohort consisted of 93,676 postmenopausal women aged 50–79 years at enrollment (1993–1998). Questionnaires were mailed to participants at baseline and annually during the follow-up to obtain and update information on risk factors and disease outcomes including ovarian cancer. Perineal powder use was assessed at baseline. Information on potential confounders, including alcohol consumption, smoking status, physical activity, BMI, family history of ovarian or breast cancer, age at menarche, age at menopause, age at first birth, age at last birth, parity, breastfeeding duration, history of tubal ligation, history of hysterectomy, history of irregular cycles, history of endometriosis, duration of oral contraceptive use, and duration of postmenopausal hormone use were also obtained at baseline. The participants were followed for an average of 12.4 years, and 429 incident cases of ovarian cancer were identified (Houghton et al. 2014). Having ever been exposed to talc perineally was not associated with an enhanced risk of ovarian cancer (hazard ratio [HR] = 1.06, 95% CI: 0.87–1.28). The risk of ovarian cancer also did not alter with increased duration of perineal talc use (≤ 9 years: HR = 1.09, 95% CI: 0.88–1.36; ≥ 10 years: HR = 1.02, 95% CI: 0.80–1.30; p_trend = 0.77). The results for each mode of talc application were also null (perineal powder use: HR = 1.12, 95% CI: 0.92–1.36; powder use on sanitary napkins: HR = 0.95, 95% CI: 0.76–1.20; powder use on diaphragm: HR = 0.92, 95% CI: 0.68–1.23). There were no exposure-response relationships between ovarian cancer risk and duration of any mode of talc application.

In an updated analysis of the WHI-OS cohort (Urban et al. 2015), perineal talc use was dichotomized by duration (> 10 years vs. < 10 years). Perineal talc use for > 10 years was not associated with an increased risk of ovarian cancer compared to use < 10 years (HR = 0.97, 95% CI: 0.78–1.22).

The Sister Study

Gonzalez et al. (2016) analyzed talc use and ovarian cancer in the Sister Study, which consisted of 50,884 women in the US and Puerto Rico who had sisters diagnosed with breast cancer. The participants were aged 35–74 years at enrollment (2003–2009) and completed a computer-assisted telephone interview to provide information on reproductive history, health conditions, and lifestyle factors. The participants also completed a self-administered questionnaire at baseline to report personal care products used in the 12 months prior to enrollment, including frequency and mode of genital talc use. With a median follow-up of 6.5 years, 154 incident ovarian cancers were identified. Gonzalez et al. (2016) noted that talc use in the 12 months prior to enrollment was not associated with an increased risk of ovarian cancer (HR = 0.73, 95% CI: 0.44–1.2).

In summary, the prospective cohort studies consistently reported a null association between perineal talc use and ovarian cancer, and a lack of exposure-response between ovarian cancer risk and frequency or duration of talc use. In addition, talc exposure was not associated with increased risks of any subtypes of ovarian cancer, and associations did not appear to vary by subtype.

Pooled analysis of cohort studies

O’Brien et al. (2020) conducted a pooled analysis of data from the NHS, WHI-OS, and Sister Study cohorts, as well as the Nurses’ Health Study II (NHSII), a prospective cohort of 116,429 registered female nurses in the US aged 25–42 years in 1989, for whom genital talc use was assessed via questionnaire in 2013. With a pooled sample size of 252,745 women and a median follow-up of 11.2 years, O’Brien et al. (2020) found that having ever used talc in the genital area was not associated with an elevated risk of ovarian cancer (HR = 1.08, 95% CI: 0.99–1.17). Ovarian cancer risk also did not change markedly for frequent (at least once per week) vs. never use (HR = 1.09, 95% CI: 0.97–1.23; p_trend = 0.20), long-term (at least 20 years) vs. never use (HR = 1.01, 95% CI: 0.82–1.25; p_trend = 0.49), or for ever vs. never use among cases with histologic subtypes of serous (HR = 1.10, 95% CI: 0.97–1.25), endometrioid (HR = 1.15, 95% CI: 0.83–1.58), mucinous (HR = 1.03, 95% CI: 0.69–1.54), or clear cell (HR = 1.17, 95% CI: 0.73–1.89) ovarian cancer.

Case-control studies

In general, case-control studies reported a small increase in ovarian cancer risk associated with perineal exposure to talc, as most of the RR estimates fell between 1 and 2 (see Table 2). These investigators did not consistently observe positive exposure-response relationships between extent of talc exposure and ovarian cancer risk (see Exposure-response section below and Table 2). Different modes of perineal talc exposures were not consistently associated with increased cancer risks. Below, the two recent case-control studies with the largest numbers of cases, the New England Case-control (NECC) study (Cramer et al. 1999, 2016; Gabriel et al. 2019) and the Ovarian Cancer Association Consortium (OCAC) study (Terry et al. 2013), are discussed in more detail.

The New England Case-control Study

Cramer et al. (2016) conducted the NECC study, in which cases and controls were recruited from Massachusetts and New Hampshire in three phases spanning over two decades. Combining data from the three phases, 2,041 pathologically confirmed epithelial ovarian cancer cases and 2,100 population-based controls were included in the analyses (Cramer et al. 2016). Personal interviews were conducted to obtain information on potential ovarian cancer risk factors and talc use that occurred more than one year prior to diagnosis for cases and one year before the interview for controls. Specific modes of talc application and extent of talc exposure were assessed, and a number of established ovarian cancer risk factors and potential confounders were adjusted for in the analyses.

Cramer et al. (2016) noted that any genital powder use was associated with an OR of 1.33 (95% CI: 1.16–1.52) for ovarian cancer. Similar results were found in the most recent analysis of the NECC study by Gabriel et al. (2019) (see Table 2), though this investigation focused on the joint effects of genital talc use and douching. Cramer et al. (2016) also reported significant exposure-response relationships between ovarian cancer risk and the frequency, the duration, months per year, and total genital applications of talc use. Women who were only exposed to talc via diaphragm use were at significantly reduced risks for ovarian cancer (odds ratio [OR] = 0.73, 95% CI: 0.57–0.93).

The Ovarian Cancer Association Consortium Study

The OCAC study, founded in 2005, is a consortium of ovarian cancer case-control studies. Terry et al. (2013) conducted a pooled analysis of 8 population-based case-control studies in the OCAC study, including the first two enrollment phases of the NECC study. Information on genital powder use varied across studies, and the OCAC pooled analysis developed harmonized analytic exposure variables by defining genital powder use as any type of powder applied directly or indirectly (by application to sanitary pads, tampons, or underwear) to the genital, perineal, or rectal area. The total number of applications was also estimated from duration and frequency of genital powder use. Information on known and suspected risk factors for ovarian cancer, including oral contraceptive use, parity, tubal ligation history, BMI, race, and ethnicity, was available from each study. Histological subtypes of ovarian cancer were also evaluated. A total of 8,525 ovarian cancer cases and 9,859 controls were included in the pooled analysis.

Terry et al. (2013) reported a pooled OR of 1.24 (95% CI: 1.15–1.33) associated with any genital powder use. Study-specific ORs for genital powder use ranged from 0.99 to 1.37, with most studies demonstrating significant elevated risks. Terry et al. (2013) also observed significant risks for three histological subtypes of ovarian cancer: serous (pooled OR = 1.20, 95% CI: 1.09–1.32), endometrioid (pooled OR = 1.22, 95% CI: 1.04–1.43), and clear cell (pooled OR = 1.20, 95% CI: 1.09–1.32).

Different uses of talc

Several of the epidemiology studies assessed the risks associated with different modes of talc powder use: powder on sanitary napkins, pads, or underwear and talc on the diaphragm or cervical cap. Among the prospective cohort studies, Houghton et al. (2014) examined powder use on the genitals, sanitary napkins, and diaphragms, while Gertig et al. (2000) examined talc use on perineum and on sanitary napkins. Both studies reported null results for all of these specific powder uses.

A number of the case-control studies also evaluated specific uses of talc. For dusting powder use on the perineum, results from most case-control studies suggest increases in ovarian cancer risk (Booth, Beral, and Smith 1989; Chang and Risch 1997; Chen et al. 1992; Cook, Kamb, and Weiss 1997; Cramer et al. 1999, 2016; Gates et al. 2008; Godard et al. 1998; Green et al. 1997; Harlow et al. 1992; Kurta et al. 2012; Ness et al. 2000; Purdie et al. 1995; Rosenblatt, Szklo, and Rosenshein 1992; Rosenblatt et al. 2011; Terry et al. 2013; Whittemore et al. 1988; Wu et al. 2015, 2009), while other investigators noted null associations (Harlow and Weiss 1989; Jordan et al. 2007; Tzonou et al. 1993; Wong et al. 1999). For talc use on sanitary napkins, pads, or underwear, 6 studies reported increased risks (Cook, Kamb, and Weiss 1997; Cramer et al. 1999; Harlow and Weiss 1989; Ness et al. 2000; Rosenblatt, Szklo, and Rosenshein 1992; Wu et al. 2009), but 5 demonstrated null associations (Chang and Risch 1997; Harlow et al. 1992; Rosenblatt et al. 2011; Whittemore et al. 1988; Wong et al. 1999). For talc use on diaphragms or cervical caps, most investigators reported null associations or reduced ovarian cancer risks (Booth, Beral, and Smith 1989; Cook, Kamb, and Weiss 1997; Cramer et al. 2016; Harlow et al. 1992; Harlow and Weiss 1989; Hartge et al. 1983; Ness et al. 2000; Rosenblatt et al. 2011; Wu et al. 2009), and two studies reported numerically higher risks that were not statistically significant (Rosenblatt, Szklo, and Rosenshein 1992; Whittemore et al. 1988). A meta-analysis that included 8 of these studies (those published through 2005) found no elevated risk of ovarian cancer associated with use of talc on diaphragms (summary RR of 1.03, 95% CI: 0.8–1.37) (Huncharek et al. 2007). Huncharek et al. (2007) also conducted sensitivity analyses that explored the effects of specific study characteristics on this summary risk estimate and reported that all resultant summary RRs indicated no marked association between talc use on diaphragms and enhanced risk of ovarian cancer.

Exposure-response

Many epidemiology studies examined the exposure-response relationship between ovarian cancer risk and the extent of perineal talc exposure (as an estimate of dose-response). The most frequently assessed metrics of talc exposure include frequency and duration of talc application and total applications (frequency × duration). Other metrics include age at first use, age at last use, time since first use, and time since last use. To compare results across studies, this review focused on the three most frequently evaluated metrics: frequency, duration, and total number of talc applications. The p_trend values for these metrics are listed in Table 2.

Some case-control investigators found a significant exposure-response association for frequency of talc use (Cramer et al. 2016; Gates et al. 2008; Harlow et al. 1992; Mills et al. 2004; Schildkraut et al. 2016; Wu et al. 2009). In contrast, several case-control studies (Booth, Beral, and Smith 1989; Chang and Risch 1997; Jordan et al. 2007; Whittemore et al. 1988) and the NHS cohort (Gates et al. 2008; Gertig et al. 2000) (Table 2) demonstrated no marked relationship.

Six case-control studies reported a significant linear trend in ovarian cancer risk with an increased duration of talc use (Cramer et al. 2016; Merritt et al. 2008; Mills et al. 2004; Schildkraut et al. 2016; Wu et al. 2015, 2009); while 8 case-control investigations (Chang and Risch 1997; Cramer et al. 1999; Green et al. 1997; Harlow et al. 1992; Ness et al. 2000; Rosenblatt et al. 2011; Whittemore et al. 1988; Wong et al. 1999) and the WHI-OS cohort (Houghton et al. 2014) noted no marked correlation (Table 2).

Several case-control studies evaluated ovarian cancer risk and total applications (frequency × duration). As presented in Table 2, the results from these studies were mixed (Cook, Kamb, and Weiss 1997; Cramer et al. 1999, 2016; Harlow et al. 1992; Mills et al. 2004; Rosenblatt et al. 2011; Schildkraut et al. 2016; Terry et al. 2013; Wu et al. 2009). Our focus was on the NECC (Cramer et al. 2016) and OCAC study (Terry et al. 2013), because these investigations had the largest numbers of cases, and the OCAC study combined data of several previous studies. As discussed above, the NECC study had three enrollment phases. Combining data from three phases, the results showed a significant positive exposure-response relationship between total number of talc applications and ovarian cancer risk. An earlier publication of the NECC study (Cramer et al. 1999), which noted findings from the first enrollment phases, did not demonstrate a significant trend with elevated total talc applications. The OCAC study included data from the first two enrollment phases of the NECC study, as well as data from 7 other population-based case-control studies (Terry et al. 2013). Pooling data from all 8 case-control investigations and using uniform exposure categories and confounder adjustment, Terry et al. (2013) did not observe a significant trend in ovarian cancer risk with increased total talc applications.

Reviews and meta-analyses

A number of systematic reviews and meta-analyses of epidemiology evidence regarding talc exposure and ovarian cancer have been published (Cramer et al. 1999; Gross and Berg 1995; Huncharek, Geschwind, and Kupelnick 2003; Huncharek and Muscat 2011; Huncharek et al. 2007; International Agency for Research on Cancer (IARC) 2010; Langseth et al. 2008; Musser 2014; National Cancer Institute (NCI) 2016; Penninkilampi and Eslick 2018; Taher et al. 2019; Wentzensen and Wacholder 2014). These analyses reported similar findings of modest, positive meta-RRs among relevant case-control studies for ever vs. never talc use and no marked associations among cohort studies. It is noteworthy that several investigators indicated that the influence of recall bias could not be ruled out for case-control studies (Langseth et al. 2008; Penninkilampi and Eslick 2018; Taher et al. 2019).

Animal studies

Three studies of talc effects in animals after direct application to reproductive tissues or exposure in the perineal area were identified. Because the data on talc carcinogenicity in the reproductive tract were limited, our scope was expanded to include studies of the carcinogenicity of talc in other tissues. Several animal studies were identified that examined tissues for tumor development following talc exposure by various routes (intraperitoneal (ip) injection, capsule implantation, intratracheal (IT) instillation, ingestion, subcutaneous (sc) injection, or intrapleural injection). These studies are described below.

Reproductive tract tissues

Two animal studies were identified that determined the effects of talc application to reproductive tract tissues specifically. Hamilton et al. (1984) injected 10 mg of asbestos-free talc per ovary directly into the surrounding bursal space in rats and examined histopathology at various time points up to 18 months after injection. Hamilton et al. (1984) found focal areas of papillary change in the surface epithelium of the ovaries in 4 of 10 treated animals but detected no atypical cellular features that would indicate a preneoplastic condition (e.g., no mitotic figures) and no carcinogenicity. Hamilton et al. (1984) hypothesized that the papillary changes may have been produced not by talc directly, but by high concentrations of steroid hormones that likely accumulated in the bursal space. Considering that a relatively large amount of talc (10 mg) was applied directly into the bursa sac that surrounds the rat ovary, this study is not representative of potential exposure of the ovaries to talc following perineal application. The investigation also lacked a non-talc particle control to determine whether the histological changes may be attributed to properties specific to talc, or merely to the presence of a large amount of foreign inert particulate matter. The presence of granulomas in the treated rat ovaries suggests that the ovarian tissue was responding to the presence of an inert, nontoxic agent (de Brito and Franco 1994).

Keskin et al. (2009) applied 100 mg of talc (type unspecified) in a saline solution either intravaginally or perineally to rats daily for three months. At the end of the three months, genital and reproductive tissues were evaluated for histopathological alterations. There was no evidence of carcinogenicity in any of the tissues. All of the rats that received talc intravaginally or perineally (7 rats per treatment group) developed reproductive tract infections. It is unclear, however, whether the infections were talc related, as 2 of 7 untreated (no injection or application) animals also developed reproductive tract infections (including one ovary infection), and one of 7 saline-injected rats developed endometritis. No information was provided as to sterility of the test materials or living conditions. This study is limited by the small number of animals, a test period (three months) that is not sufficient to study tumor development, and a possible infection in the animal colony unrelated to talc.

In an investigation of indirect talc application, Boorman and Seely (1995) reported findings from the NTP (1993) study described below in which rats and mice of both genders were exposed to high-purity micronized talc at aerosol concentrations of 6 or 18 mg/m³ via whole body inhalation (discussed in more detail below). Although animals were exposed by inhalation, the air concentrations were so high that dermal (perineal) exposure was also present. As described by Boorman and Seely (1995), “animals were exposed for 6 hours per day with talc covering the fur and the cage bars, and there was ample opportunity for perineal as well as oral and respiratory exposure”. Boorman and Seely (1995) found no talc-related lesions of the ovaries in either rats or mice. In addition, no talc particles were identified in the ovarian tissues of exposed rats. This investigation had several strengths, including large numbers of animals (49–50 of each gender per treatment group), lifetime exposure, and histological examination of both lung (as a positive control) and ovarian tissue for the presence of talc particles from randomly selected animals in each treatment group.

Non-reproductive tract tissues

Several experiments evaluated the influence on rodents of non-reproductive tract talc exposure via various exposure routes. Many of these studies have methodologies that limit their interpretation in terms of relevance to human exposures, including small numbers of test animals, use of only one dose level, a single administration, and/or a dose that was much higher than any relevant human exposures. In some investigations, the routes of exposure (e.g., ip, intrapleural, or sc injection) were not relevant to typical human exposures. Some of the reseachers did not report the source or grade of the talc used, and many did not include a particulate matter control to assess whether effects were specific to talc. In addition, most of these experiments did not include an assessment of talc effects on the reproductive tract. Nevertheless, these are informative as to talc’s potential carcinogenicity in general.

Ozesmi et al. (1985) exposed mice to 20 mg of UV-sterilized commercial talc in physiological saline via ip injection and followed animals until spontaneous death or obvious tumor formation. Incidence of tumor formation in exposed mice was similar to controls. Pott, Huth, and Friedrichs (1974) exposed rats to 4 ip injections of 25 mg talc in saline and noted that the exposure “did not lead to the development of tumors except for a few cases.” Stanton et al. (1981) tested 7 different samples of refined commercial talc in the form of a gelatin capsule implanted in the pleural surface of female rats for at least one year. The samples were each from a different source and meant to represent potential range in dimension of talc. Stanton et al. (1981) found no significant change in pleural sarcomas in any of the talc-exposed rats compared to controls. No local tumors were detected in mice observed for life following a single sc injection of 80 g talc (type unspecified) (Neukomm and de Trey, 1961, as cited in International Agency for Research on Cancer (IARC) 2010).

Wagner et al. (1977) reported no clear evidence of carcinogenicity in rats following an intrapleural inoculation with 20 mg Italian 00000 grade talc or dietary exposure to 100 mg/day Italian talc on 101 days over a 5-month period. This Italian talc (mean particle size 25 µm) was selected for the experiments due to its 50-year history of use and because it served as the source of over 40% of cosmetic grade talc in Great Britain at the time. In the rats exposed via intrapleural inoculation, injection site granulomas were reportedly common and one small pulmonary adenoma was identified as a possibly incidental finding. In the dietary portion of the experiment, a single leiomyosarcoma of the stomach was detected in a talc-fed rat, but Wagner et al. (1977) were unable to determine whether it was related to exposure. In addition, two sarcomas of the uterus were found in treated rats but were not considered treatment related due to the location of the tumors and occurrence of this tumor type in historical control animals.

Studies in hamsters (Stenback and Rowland 1978), rats (Friemann et al. 1999), and mice (Sahu, Shanker, and Zaidi 1978) investigated pulmonary effects following IT instillation of talc and noted no marked alteration in tumors of the respiratory system. Non-respiratory tumors were either not investigated or not reported in the studies. Stenback and Rowland (1978) treated hamsters with 18 IT instillations of talc (United States Pharmacopoeia grade; 93.3% of particle less than 25 µm). The hamsters showed “signs of minor respiratory disorders” during the treatment period but did not develop any respiratory or non-respiratory tumors attributed to the talc exposure. Friemann et al. (1999) exposed rats to a single IT instillation of 25 g of asbestos-free talc and examined histological effects of the respiratory system to identify potential pre-neoplastic lesions. Talc-exposed rats exhibited reversible proliferation of bronchiolar epithelium but no respiratory neoplasms or significant changes in alveolar bronchiolization, an effect that was considered a pre-neoplastic lesion. In a study of effects related to pulmonary silicosis, Sahu, Shanker, and Zaidi (1978) exposed mice to 5 mg talc dust sourced from India (particle size < 5 µm) via IT instillation and observed animals for a period of 210 days. No gross pathological effects of the respiratory system resembling neoplasms were detected.

Similarly, 3 of 4 inhalation studies reported no excess tumors in rats (Wagner et al. 1977), hamsters (Wehner, Zwicker, and Cannon 1977), or mice (NTP, 1993). Wagner et al. (1977) exposed rats to a mean concentration of 10.8 mg/m³ Italian talc (as previously described) as respirable dust for 3, 6, or 12 months and sacrificed rats following exposure or after a one-year recovery period. A single lung adenoma was observed in the 12-month exposure group, which was noted as a possibly incidental finding, although no statistical analysis was provided. Wehner, Zwicker, and Cannon (1977) exposed hamsters to a respirable fraction of approximately 8 to 10 mg/m³ aerosolized cosmetic grade talc for various exposure durations, including a maximum of 300 days, and found no treatment-related tumors in any of the exposed animals. National Toxicology Program (NTP) (1993) conducted a standard cancer bioassay in which mice of both genders were exposed to aerosols containing 0, 6, or 18 mg/m³ talc for up to 104 weeks. The high-purity talc (MP 10–52 grade) was sourced from a mine in Montana, contained no tremolite or asbestiform minerals, and had a maximum particle size of 10 µm with 75% of particles between 1 and 3 µm. No significant excess in tumor incidence was found, and NTP concluded that there was no evidence of carcinogenic activity of talc based on the findings in mice.

Also in the NTP (1993) study, female rats, but not male rats, developed a significant excess of alveolar/bronchiolar adenomas, carcinomas, and combined adenomas and carcinomas at the highest dose only (18 mg/m³ for more than two years). NTP (1993) also observed an excess of pheochromocytomas in the high-dose male and female rats, but IARC (2010) did not consider this increase to be talc-related because pheochromocytomas may result from stress and hypoxia, and background incidence of these tumors is quite high in the F344/N rat strain that NTP studied. NTP (1993) concluded that there was “clear evidence of carcinogenicity” in female rats based upon the findings of alveolar/bronchiolar adenomas, carcinomas, and combined adenomas and carcinomas or benign or malignant pheochromocytomas of the adrenal gland.

Several investigators disagreed with NTP’s conclusion, including a panel of experts at a 1994 International Society of Regulatory Toxicology & Pharmacology (ISRTP)/US Food and Drug Administration (FDA) workshop (Goodman 1995; Musser 2014; Oberdorster 1995; Wehner 2002), because of weaknesses in the study design and its relevance to human risk. A talc concentration of 18 mg/m³, 6 hr/day, 5 days/week, is much higher than any relevant occupational or consumer exposures and would have resulted in lung overload to the animals, which might result in tumor formation due to an inflammatory reaction to even non-carcinogenic particles (Oberdorster 1995; Wehner 2002). It is notable that the study did not include positive or negative dust controls, which would have permitted an assessment of talc’s carcinogenicity relative to other control dusts. The rats in the highest exposure group also exhibited marked chronic lung toxicity (chronic granulomatous inflammation, non-neoplastic changes, and interstitial fibrosis), indicating that they were exposed at a level that produced general toxicity (Goodman 1995). Experts at the 1994 ISRTP/FDA workshop concluded that this 1993 NTP study had “no relevance to human risk” (Musser 2014).

Summary

Several studies, described above, assessed whether talc might induce ovarian cancer in experimental animals, including two that evaluated talc directly applied to reproductive tissues. Although the two studies of direct application to reproductive tract tissues have some methodological limitations, both reported no marked association between talc exposure and ovarian cancer, even at doses much higher than might reasonably be expected to result from consumer exposure to talc. In addition, of all the studies examining other routes of exposure, only one reported lung cancer in female rats at the highest dose, but this finding has no relevance to human risk (Musser 2014). No other cancer in either gender or other species was found.

Genotoxicity

Endo-Capron et al. (1993) did not observe any marked alterations in sister chromatid exchange, DNA repair, or aneuploidy in rat pleural mesothelial cells that were exposed to three different talc samples at concentrations ranging from 2 to 50 μg/cm², for 24 or 48 hr. Litton Bionetics, Inc. (1974, as cited in Fiume et al. 2015) found no significant talc-initiated genotoxic effects in host-mediated, cytogenetic, or dominant lethal assays. In the host-mediated assay, male mice were administered talc either by a single gavage, or once daily for 5 days with 30, 300, 3,000, or 5,000 mg/kg talc. The indicator organisms Salmonella typhimurium TA1530 and G46 and Saccharomyces cerevisiae D3 (injected into the host mouse abdominal cavity and then withdrawn) did not exhibit any mutations. In the cytogenetics assay, 2, 20, or 200 μg/ml talc were added to human embryonic lung culture cells. Talc produced no significant aberrations and was not genotoxic. In the dominant lethal assay, no dominant lethal mutations were detected following exposure of rats to talc at 30, 300, 3,000, or 5,000 mg/kg.

In vitro studies

Three studies evaluated the influence of talc on ovarian cells in vitro. Buz’Zard and Lau (2007) noted neoplastic transformation and a time-dependent increase in reactive oxygen species (ROS) in human ovarian cell line cultures following addition of 0.5–500 μg/ml talc for 24–72 hr. This report is misleading, as the generation of ROS actually decreased with increasing talc concentration in the ovarian cell lines. Small increases were observed at 72 hr compared to 24 hr after talc addition at some, but not all, concentrations of talc. While the generation of ROS in cells may be one step in a pathway leading to tumor formation, it is not an indication of carcinogenesis per se. Many different exposures and cell processes lead to enhanced oxidation but do not result in cancer development (Goodman and Lynch 2017). Buz’Zard and Lau (2007) did not demonstrate that the generation of ROS following talc exposure led to cell damage or genetic alterations. In addition, there was no recovery period, and thus it remains unknown whether the concentration of ROS would return to baseline levels after removal of talc.

The reported neoplastic transformation was measured by an elevation in the ability of cells to grow in soft agar, but findings were inconsistent. In one ovarian epithelial cell line (OSE2a), the increased growth occurred in a reverse exposure-response relationship, with the greatest effect occurring at 5 μg/ml talc, and a reduction in growth at 100 μg/ml talc. In the stromal ovarian cell line (GC1a), increasing talc concentrations were associated with enhanced growth. In addition to these inconsistencies, there was no discussion of the relevance of the in vitro concentrations used in the cell cultures in relation to any plausible dose in vivo to the ovaries of women who use talc. This study has not been replicated by other investigators, does not exhibit consistent talc effects in a clear exposure-response-related manner, and is not consistent with other genotoxicity findings.

Fletcher et al. (2019) exposed normal human primary ovarian epithelial cells and several ovarian cancer cell lines to 5, 20, or 100 μg/ml talc for 72 hr and demonstrated concentration-dependent elevation in various markers of oxidative stress, such as nitric oxide and inducible nitric oxide synthase, and diminished expression of antioxidant enzyme activities including catalase, superoxide dismutase, glutathione peroxidase, and glutathione reductase. Fletcher et al. (2019) also reported a rise in expression of CA-125 protein. CA-125 is a biomarker of ovarian cancer but is not specific to ovarian cancer, and may also be elevated in patients with other, noncancerous conditions of the reproductive tract and in patients with some other types of cancer (Jelovac and Armstrong 2011). Increased cell proliferation and decreased apoptosis with talc treatment of both normal and ovarian cancer cells were noted, as well as an induction of specific mutations in key oxidant and antioxidant enzymes that correlate with alterations in their activities. As with the study by Buz’Zard and Lau (2007), Fletcher et al. (2019) did not discuss the relevance of the in vitro concentrations employed in their study to any plausible in vivo dose to the ovaries of women who use talc. This is important because, as Fletcher et al. (2019) also did not include any inert particulate matter controls, they could not rule out effects due to the presence of an overload of inert particulate matter rather than talc specifically. Further, any parameters after a recovery period were not measured to determine whether the changes were permanent, or if these might return to baseline over time.

Mandarino et al. (2020) examined whether talc exerted an effect on the ability of macrophages to inhibit tumor growth in the presence of estrogen. Mouse macrophage cell lines that were exposed to talc, with or without estradiol, were co-cultured with an ovarian cell line, and the cell line was monitored for growth. There was more growth in the ovarian cells cultured in the presence of macrophages exposed to talc plus estradiol than in the presence of macrophages exposed to either substance alone. Mandarino et al. (2020) concluded that the combination of talc plus estradiol may inhibit the anti-tumor activity of macrophages in reproductive tissues. However, the effects of talc and estradiol appear to be additive in this study, which does not indicate a novel mechanism of immune suppression that occurs only when both agents are present. Mandarino et al. (2020) did not specify the concentration of talc used in most of the trials and did not indicate the relevance of the amount of talc to human exposures.

Taken together, these in vitro studies provide inconclusive evidence regarding whether talc induces pre-carcinogenic effects in the ovary at human relevant doses. This is consistent with longer-term studies of animals that do not indicate carcinogenicity from talc exposure.

Exposure and transport evaluations

There are several types of exposure studies that may be considered in evaluating whether perineal talc application (either direct or by application to underwear or sanitary pads) and subsequent retrograde transport into the reproductive tract comprise a plausible exposure pathway that substantially contributes to development of ovarian cancer. This type of exposure information includes the following:

• Biological monitoring data, including results from studies that explored the presence of talc in samples of ovarian cancer and healthy ovarian tissues;

• Mechanistic transport studies, which investigated the movement of talc or other types of particles in the female reproductive tract; and

• Modeling evaluations that assessed the magnitude of exposures that might occur through this pathway.

The exposure information collected in epidemiology investigations also provides insights into this transport pathway.

Biological monitoring and measurements in ovarian tissues

The scientific literature includes a number of individual case studies and small-scale studies that explored the presence of talc particles in ovarian tissues. These investigations examined tissue samples from surgical patients, including patients with ovarian cancer, and others judged to have healthy ovarian tissues.

Based on results from an extraction-replication technique that their lab developed to examine the presence of foreign particles in tissue samples, Henderson et al. (1971) reported that talc was present in tissue samples from 10 of 13 ovarian tumors, 12 of 21 cervical tumors, and 5 of 12 samples judged to reflect normal ovarian tissues from patients with breast cancer. There was some question regarding whether the particles detected in the samples may have been due to sample contamination such as from particles on the gloves of individuals obtaining or analyzing the samples. In a subsequent investigation, Henderson, Hamilton, and Griffiths (1979) noted that talc was found in all 9 additional samples: three samples from normal ovaries, three from cystic ovaries, and three from adenocarcinomas. These investigators noted they were particularly careful to avoid external talc contamination sources. Using several microscopic methods, Cramer et al. (2007) in a case study indicated the presence of talc particles in samples collected from the pelvic lymph nodes of a woman with ovarian cancer. These investigators stated that the use of polarized light microscopy identified “diffuse areas of birefringence compatible with talc,” and that examination of the samples using scanning electron microscopy and X-ray spectroscopy confirmed the presence of talc.

In a study of ovarian samples from 100 women with “grossly normal” ovaries undergoing surgery for pelvic disease, Mostafa et al. (1985) observed “crystalline foreign particles” in histological evaluations of the samples from 9% of subjects. Subsequently 4 samples containing foreign particles were analyzed using a scanning electron microscope and computer-assisted microscopic X-ray analysis to evaluate elemental composition of the particles. Based upon these analyses, Mostafa et al. (1985) concluded that the particles were composed largely of magnesium and silicon but did not definitively confirm the observed particles as being talc; however, talc and asbestos were identified as the “most common compounds containing magnesium silicates.” In addition, these investigators did not report any precautions taken to ensure that the particles were not derived from contamination (from surgical gloves, for example) during surgery and/or the processing of the tissues for examination.

Heller et al. (1996) examined the presence of talc particles in normal ovarian tissue samples collected from 24 women undergoing incidental oophorectomy (i.e., surgical removal of one or both ovaries) to address benign ovarian neoplasms. Talc particles were identified in samples from all 24 women by either polarized light microscopy or analytic electron microscopy. Heller et al. (1996) also collected information regarding the study subjects’ perineal talc use and approximated talc exposure levels by estimating the subjects’ “lifetime talc applications.” Data indicated that talc particle counts from both types of microscopy showed no quantitative relationship with estimated level of talc use. For example, in unexposed women, particle counts observed using light microscopy ranged from 0 to 2,200/g, while counts in exposed women ranged from 26-464/g. The lowest particle count in the exposed group was reported for the woman with the highest estimated lifetime talc applications. In the electron microscopy results, talc particle counts in exposed and unexposed groups spanned a similar range, and the counts for approximately one-half of the women in each group were not-detectable (zero).

Two recent studies (McDonald et al. 2019b, 2019a) examined the ovaries, pelvic lymph nodes, and other pelvic tissues of ovarian cancer patients for the presence of talc particles. McDonald et al. (2019a) measured talc particles in the pelvic lymph nodes of 22 ovarian cancer patients. Results demonstrated that the mean concentration of talc particles in the pelvic lymph nodes of women who used perineal talc was higher than that in women who did not use perineal talc. However, of the 10 women who reported perineal talc use, 9 also indicated regular use of talc on other parts of the body. McDonald et al. (2019a) did not conduct an analysis of talc in tissues of women who used talc as a body powder vs. women who did not, thus it is not possible to conclude that talc in the pelvic area lymph nodes was specifically derived from perineal application.

In another study, McDonald et al. (2019b) measured talc in lymph nodes, cervix, uterine corpus, Fallopian tubes, and ovaries of 5 women with ovarian cancer who had a history of perineal talc use. These tissue concentrations were compared to those in 6 ovarian cancer patients who did not have a history of talc use. McDonald et al. (2019b) indicated that talc concentrations were higher in tissues from talc users than non-talc users. The ages of the patients were provided but no other characteristics were included.

Mechanistic transport studies

A number of investigations examined particle transport in the female reproductive tract. Almost all of the identified human studies involved hospitalized populations and particles other than talc. Animal experimentation investigated the transport of talc as well as other particle types. No identified studies of humans or animals specifically examined particle transport following external application to the perineal area. Instead, the studied particles were typically in a solution and placed within the reproductive tract (e.g., intravaginally). In one case, starch particles from surgical gloves were introduced into the reproductive tract during gynecological examinations.

In one study of hospitalized patients, Egli and Newton (1961) deposited inert carbon particles in solution in the vaginas of patients undergoing elective abdominal hysterectomy. In two of the three patients, particles were observed in the patients’ Fallopian tubes within approximately 30 min. Wehner, Weller, and Lepel (1986) noted that this study was not quantitative and did not include any blanks or negative controls.

De Boer (1972) placed India ink (a suspension of carbon particles) at various locations within the reproductive tracts of 178 women undergoing abdominal surgery. Data demonstrated that placement location was associated with frequency of observing particles in the Fallopian tubes. In particular, noting the role of the cervix as a barrier to transport, De Boer (1972) found that particles placed in the vagina were seen in the Fallopian tubes in only 1 of 37 patients.

Venter and Iturralde (1979) placed a solution of albumin microspheres labeled with radionuclides intravaginally in 24 patients undergoing gynecological surgeries. In 9 out of 14 cases in which radioactivity measurements in the uterus were counted separately from those in the ovaries and Fallopian tubes, radioactivity levels were detected in either the ovaries or Fallopian tubes. The 5 negative results were all from patients with tubal damage from previous infections. Venter and Iturralde (1979) concluded that these findings indicated that particle transport from the vagina to Fallopian tubes is possible. Kunz et al. (1996) also investigated particle transport in the female reproductive tract using radiolabeled albumen microspheres. In this study of 64 women, a solution containing the particles was placed in the vagina just outside the entrance to the cervix. Some microspheres entered the uterus and reached the Fallopian tubes within minutes.

In a study of approximately 60 surgical patients, Sjosten, Ellis, and Edelstam (2004) examined retrograde migration of starch particles from powdered gloves. Either one or 4 days prior to elective hysterectomies, study subjects had an intravaginal gynecological examination where powdered gloves (exposed group) or powder-free gloves (control group) were employed. For individuals undergoing the exam one day prior to the surgery, there were significant differences in the numbers of particles between exposed and control groups in the cervix, uterus, and Fallopian tubes. Ovaries were not examined. For individuals undergoing the exam 4 days prior to the surgery, significant differences were observed for small and large particles in the cervix and uterus.

Potential talc transport in the reproductive tract was also studied in rodents, rabbits, and primates. Seventy-two hr following a single intravaginal administration of a radiolabeled talc suspension in three rabbits, radioactivity was detected only at the site of administration (Phillips et al. 1978). Similar readings taken 72 hr after 6 daily intravaginal doses found no migration of talc to the animals’ ovaries; a small amount of radioactivity was noted to be associated with the cervix and Fallopian tubes. In a study using 26 rabbits, Edelstam, Sjosten, and Ellis (1997) reported finding starch particles in peritoneal cavity rinsate collected on multiple days following intravaginal administration of a single dose. Edelstam, Sjosten, and Ellis (1997) stated that this finding indicates “the possibility of retrograde migration,” but also noted that no overall significant differences were seen in the numbers of particles in the rinsates between control and exposed groups. In a study of 8 rats some evidence of talc transport to the ovaries occurred following intrauterine administration of a single dose of a talc-containing solution (Henderson et al. 1986). In 6 rats administered a single intravaginal dose, evidence of talc transport to the ovaries was found for two rats examined 4 days after dosing but not in those examined 24 or 48 hr post-dosing.

Studies in monkeys, where reproductive tracts most closely resemble that of humans (Wehner et al. 1985), provided no evidence of retrograde talc transport. In one study, 6 Cynomolgus monkeys received 30 intravaginal applications of a radioactive talc suspension over a 45-day period (Wehner, Weller, and Lepel 1986). Despite the extensive dosing, radioactivity readings provided no evidence of talc transport to the uterus or beyond. A similar study by Wehner et al. (1985) using bone black detected no evidence of particle transport to the uterus or beyond. In this investigation, 5 Cynomolgus monkeys were administered a single intravaginal dose of the test solution, and evidence of translocation was collected one hr (three monkeys) or 72 hr (two monkeys) after administration. These animals also received an injection of oxytocin to mimic certain experimental procedures used in previous human studies.

No animal studies directly examined potential talc transport following external perineal administration of talc. However, potential transport of talc particles to the ovaries was explored as one element of a lifetime whole body exposure toxicity study in rats undertaken by NTP (Boorman and Seely 1995). In this study, rats were exposed to a talc aerosol at concentrations sufficient to cover their fur and cage bars, providing “ample opportunity for perineal as well as oral and respiratory exposure.” Examination of the ovaries and ovarian bursa showed no evidence of material consistent with talc.

Modeling

Virtually no information is available that reflects the use of exposure modeling techniques to estimate the degree of talc exposure associated with or needed to potentially induce ovarian cancer. Based upon information collected in epidemiology studies regarding talc use patterns, some investigators developed semi-quantitative exposure measures (e.g., estimates of “lifetime talc applications”) (Heller et al. 1996) or “talc-years” (Cramer et al. 2016) generated from information regarding subjects’ frequency and duration of perineal talc use. None of the exposure evaluations included in the epidemiology or talc biomonitoring studies attempted to quantify the amount of talc exposure associated with ovarian cancer based upon such measures as the amount of talc applied during each perineal application event, or the number of particles expected to reach biological target locations (e.g., the ovaries). Thus, more rigorous quantitative evaluation of this exposure pathway is lacking.

Epidemiology studies

Comparative results from epidemiology studies also provide insights regarding the potential role of perineal talc use in ovarian cancer. For example, storage of diaphragms used for contraception in talc was identified as a mechanism whereby talc might be introduced into the vagina, in proximity to the cervix (Fiume et al. 2015; Muscat and Huncharek 2008). Relative to the perineal external exposure pathway, the use of a talc-stored diaphragm would be expected to transport talc particles further into the reproductive tract and in a more deliberate and direct manner. In contrast, the exposure pathway for externally applied talc particles is more attenuated, requires transport over a greater distance, and involves physical barriers that need to be crossed. A number of epidemiology investigations of ovarian cancer examined this potential exposure source. Table 2 shows that, despite this greater exposure potential, none of the 12 studies that evaluated this potential talc exposure source reported a significant effect on ovarian cancer risk (Cramer et al. 2016 reported a significant decrease in ovarian cancer risk in women who applied talc to their diaphragms).

Discussion

Our analysis of the epidemiology studies indicates that there is a small, increased association observed between perineal talc use and ovarian cancer in some case-control studies. In contrast, in meta-analyses, despite the above observations, there is no consistent finding of a positive exposure-response relationship. Further, prospective cohort studies consistently reported a null association between perineal talc use and ovarian cancer with lack of exposure-response between ovarian cancer risk and frequency or duration of talc use. Animal studies indicate that talc is not carcinogenic in the ovaries or in any tissue. In vitro and genotoxicity data demonstrate that talc is not genotoxic, and no other potential mechanism of carcinogenicity has been reliably noted. Transport studies have not demonstrated that dry talc, applied to the perineal region, can reach the ovaries. In exposure studies, particles detected in reproductive tissue were not related to perineal use of talc or tumor formation. In the following discussion, the evidence from these different types of studies was assessed to determine whether data support the hypothesis that talc produces ovarian cancer.

Epidemiology studies

A striking feature of the body of epidemiology evidence regarding talc and ovarian cancer is the difference in results reported between case-control and cohort studies. As discussed above, small positive associations are generally observed across case-control studies, while results of cohort studies are consistently null. This feature of the epidemiology evidence is important because, while both case-control and cohort studies can be impacted by non-differential misclassification, case-control studies are generally more susceptible than cohort studies to certain types of systematic errors, including various sources of bias and confounding (Rothman, Greenland, and Lash 2008; Rothman, Pastides, and Samet 2000). If these sources of bias cannot be ruled out with certainty in individual case-control studies, the most likely explanation for the discrepancy is that the small positive associations observed in case-control studies are caused by biases that consistently skew case-control results toward false positive associations.

A difference in results between cohort and case-control studies for the same causal question is not unique to studies of talc and ovarian cancer. For example, early case-control studies of associations between coffee consumption and pancreatic cancer reported positive associations, whereas cohort studies were generally null (Boffetta et al. 2008; IARC, 1991). More recent evaluations of the literature on coffee consumption and pancreatic cancer indicate that there is no association (IARC, 2018). The false positive result in early case-control studies was likely attributable to selection bias, as the control populations excluded individuals with a history of diseases related to cigarette smoking and alcohol consumption (which are highly correlated with coffee consumption), resulting in a lower percentage of coffee consumers in the control groups (Boffetta et al. 2008).

Below, the specific sources of bias that may have contributed to the small positive associations observed in talc case-control studies are discussed. These biases have not been directly ruled out for any case-control study of talc and ovarian cancer, despite existing methods for addressing and mitigating these biases. It is possible that these biases influenced results of several studies in a consistent manner and that compounding of bias across individual studies explains meta-analyses results showing modest significant elevations of risk (as discussed further below).

Recall bias

In every case-control study of talc, exposure was determined based solely upon subjects’ self-reports about whether or not they used talc in the past; for some studies, subjects were additionally asked details on how talc was applied (e.g., on genital areas only or all over the body; on sanitary pads, diaphragms, or condoms) and/or the frequency and duration of use (Table 2). In these studies, subjects needed to recall exposure that occurred many years in the past – in some cases, several decades prior. None of the studies of talc and ovarian cancer included a validation study to assess the accuracy of these self-reports. Validation studies assess the reliability of self-reported data by comparing self-reported exposure measurements to known past exposures, typically in a subset of the overall study population. For example, validation studies have been conducted to evaluate self-reported mobile phone use (Vrijheid, Cardis, and Armstrong et al. 2006), cigarette smoking (Coultas et al. 1988), and pesticide use (Engel et al. 2001). Results of these studies may provide information on the type (i.e., systematic or random) and degree of error associated with the self-reported data, giving researchers a better understanding of study limitations and the opportunity to adjust for error in future analyses.

Besides conducting a validation study, other methods for mitigating recall bias in case-control studies include using objective measures of exposure instead of reported exposure, but none of the case-control studies of talc and ovarian cancer utilized an objective measure of talc exposure. Another approach to reducing recall bias is to design surveys such that subjects are asked about a variety of chemical exposures, in addition to the exposure of interest (i.e., talc in this case), such that subjects are unaware of the main hypothesis of the study and do not become fixated on one environmental exposure in particular. In some studies, questionnaires employed are focused on talc use, and no other environmental exposures (besides smoking) are assessed (see for example, Cramer et al. 2016). These questionnaires are not designed to mitigate recall bias.

In 2000, one group of leading academic researchers reviewed the epidemiology evidence linking talc with ovarian cancer and concluded that recall bias “can readily introduce enough bias to produce the modestly-sized overall effect (RR = 1.3)” observed in a pooled analysis of available studies (Rothman, Pastides, and Samet 2000). One approach to evaluating the actual magnitude of recall bias is to conduct a quantitative analysis of bias (Greenland and Lash 2008; Lash, Fox, and Fink 2009). This was not done in the vast majority of epidemiology studies; the only epidemiology investigation of talc and ovarian cancer to conduct such an analysis is Cramer et al. (2016). However, the analysis of potential recall bias conducted by Cramer et al. (2016) is overly simplistic and is not informative regarding the variable nature of recall errors; their analysis did not meet standards suggested by researchers for “good practices” in quantitative bias analyses (Lash et al. 2014). A more thorough analysis of recall bias indicates that differential reporting errors are a plausible cause of positive associations in Cramer et al. (2016).

It is noteworthy that a recent case-control study of talc and ovarian cancer explored temporal patterns in risk estimates to assess the potential for recall bias (Schildkraut et al. 2016). It was hypothesized that publicity of two high-profile lawsuits in 2014 may have influenced study participants’ recollections of talc use, and stratified analyses were conducted according to whether subjects were interviewed before or after 2014. The OR for ovarian cancer associated with genital talc use was found to be 1.19 (95% CI: 0.87–1.63) for women interviewed prior to 2014 and 2.91 (95% CI: 1.70–4.97) for women interviewed in 2014 or later. Schildkraut et al. (2016) acknowledged that recall bias may have caused inflation of ORs. If recall bias led to false positive findings for years following high-profile lawsuits in this study, it has also likely affected studies conducted over the past several decades, because media coverage of possible relationships between talc and cancer started as early as the 1980s, when the first epidemiology studies were published.

In summary, recall bias cannot be ruled out with certainty as an explanation for positive findings in individual case-control studies of talc and ovarian cancer. Based upon recent analyses discussed above, recall bias is likely to have biased associations away from the null, and this bias was likely consistent across all published case-control studies, leading to biased meta-analysis results, as described further below.

Selection bias

Selection bias is most likely to exert an impact on study findings when participation rates are relatively low (e.g., 70% or lower) and especially when participation rates differ between cases and controls (Lash, Fox, and Fink 2009). These factors increase the chance that cases and controls are not representative of the source population in terms of exposure status. While many investigators did not include details on participation rates, several epidemiology investigations of talc have participation rates well below 70% (Table 2). If participation rates are low, investigators can assess the potential for selection bias by conducting a validation study. None of the available case-control studies of talc and ovarian cancer conducted a validation analysis to compare the characteristics of those who participated to those who refused. Therefore, selection bias may have contributed to positive associations reported in these studies.

Confounding

Several risk factors for ovarian cancer might have confounded measured associations in both the case-control and cohort studies, including age, obesity, parity, oral contraceptive use, tubal ligation, hysterectomy, and postmenopausal hormone use. Some of the case-control studies of talc and ovarian cancer measured and attempted to control for one or more of these factors, but many did not (Table 2), whereas all of the cohort investigations addressed the majority of these factors. Several of these potential confounders may have been mismeasured leading to residual confounding. It is important to note that because the causes of most ovarian cancer cases are unknown, there are likely many potential confounders that are not measured or considered in the cohort and case-control studies. According to a review of epidemiology evidence of talc and ovarian cancer conducted by leading academic researchers, it is possible that unidentified risk factors for ovarian cancer that are also correlated with talc use might confound observed associations, and the compounded effect of several unidentified confounders might easily have caused the weak associations observed between talc and ovarian cancer in the case-control studies (Rothman, Pastides, and Samet 2000).

Exposure misclassification

In addition to the systematic errors discussed above, a major uncertainty affecting epidemiology evidence is random misclassification errors. All talc epidemiology studies, both cohort and case-control, are likely to be affected by substantial amounts of random exposure misclassification, especially considering that reported use (e.g., ever vs. never, frequency, duration, cumulative) is likely a poor representation of true dose.

A common misconception held by many epidemiologists is that random exposure misclassification will always bias results toward the null, attenuating measured associations and potentially obscuring true associations (Jurek et al. 2005). In fact, random exposure misclassification can bias in either direction (Jurek et al. 2006), and evaluations of epidemiology evidence need to account for this possibility. Further, random exposure misclassification cannot bias observed associations toward the null in the event that no true causal association exists. As noted by Rothman, Pastides, and Samet (2000), “one cannot draw the conclusion that the overall slight positive relation between talc exposure and ovarian cancer must be an underestimate of a larger effect because of nondifferential misclassification.” Rothman, Pastides, and Samet (2000) further assert that the possibility of this misclassification “does not provide any help by itself” in evaluating evidence regarding talc use and ovarian cancer. Study designs for evaluating associations between perineal talc use and ovarian cancer have remained the same in the two decades since this was published, so the conclusions of Rothman, Pastides, and Samet (2000) are still valid.

All of the epidemiology studies relied on self-reported talc use, and none of the studies, even if they evaluated cumulative exposure, had any quantitative information on the level of talc exposure per use. This indicates that exposure misclassification was likely in all of these studies and likely impacted results. That is, total exposure is not known in any study, even those that evaluated cumulative exposure, so the actual dose cannot be known. In addition, results from ovarian biomonitoring studies did not provide evidence that perineal talc use was correlated with the presence of talc particles in ovarian tissues. Recall bias was likely consistently present in case-control studies and may have impacted all metrics of talc exposure. Thus, any observed associations or exposure-response relationships in case-control studies may be due, at least partially, to recall bias. Cohort studies with prospective exposure assessments were not impacted by recall bias and consistently reported no overall association and no exposure-response relationship between perineal talc use and ovarian cancer. The null findings from the cohort studies further support that recall bias likely contributed substantially to the positive findings in the case-control studies.

Meta-analyses

Although several meta-analyses of case-control studies (Gross and Berg 1995; Huncharek, Geschwind, and Kupelnick 2003; Huncharek et al. 2007; Langseth et al. 2008; Penninkilampi and Eslick 2018; Taher et al. 2019) reported modest, positive associations between talc and ovarian cancer for case-control studies and null results for prospective cohort studies, meta-analyses are susceptible to the same biases as the underlying studies. Unless biases are resolved, meta-analysis results also will be biased due to compounded bias from individual studies (Greenland and O’Rourke 2008). For example, recent reviews concluded that evidence for associations between perineal talc use and ovarian cancer is limited and may be affected by recall bias and other uncertainties (International Agency for Research on Cancer (IARC) 2010; Huncharek and Muscat 2011; Wentzensen and Wacholder 2014; Musser 2014; National Cancer Institute (NCI) 2016). Therefore, despite being modestly positive and significant, results of meta-analyses of talc use and ovarian cancer do not conclusively support a causal relationship between talc and ovarian cancer.

Animal studies

Of the several experiments that evaluated whether talc may induce ovarian cancer in animals, including two that evaluated talc directly applied to reproductive tissues, none reported that ovarian carcinogenicity is produced by talc exposure, even at doses much higher than might reasonably be expected to result from consumer exposure to talc. In addition, of all the studies evaluating other routes of exposure, only one (NTP 1993) reported lung cancer in female rats at the highest dose, but this study has no relevance to human risk (Goodman 1995; Musser 2014; Oberdorster 1995; Wehner 2002). NTP (1993) evaluated talc concentrations that are much higher than any relevant occupational or consumer exposures and resulted in lung overload to the animals (Oberdorster 1995; Wehner 2002). No other cancer in either gender or other species was found. Although study quality varied, no investigator reported increased tumor formation in response to talc except in the case of lung overload. Taken together, the animal evidence does not indicate talc is a carcinogen, much less an ovarian carcinogen.

Genotoxicity and in vitro studies

There is very little information regarding the potential for talc to initiate changes in DNA, either from standard genotoxicity assays or from studies of ovarian cells grown in culture. Nonetheless, the limited information that is available is mostly negative and does not indicate that talc is genotoxic (Endo-Capron et al. 1993; Litton Bionetics, Inc., 1974, as cited in Fiume et al. 2015).

Exposure and transport evaluations

Certain aspects of the studies of talc particle counts in reproductive tissues need to be considered when evaluating their significance. Some methods used in these experiments did not definitively identify talc as the particle type detected in the samples. For example, Mostafa et al. (1985) characterized the particles as “crystalline foreign particles” containing magnesium and silicon based upon histological evaluation and scanning electron microscopy. As these investigators noted, asbestos is another magnesium silicate that is relatively common. In addition, the particles detected in the samples might have been due to sample contamination (e.g., from particles on the gloves of individuals obtaining or analyzing the samples). For example, because such questions had been raised regarding their earlier study (Henderson et al. 1971), Henderson, Hamilton, and Griffiths (1979) noted that they were particularly careful to avoid contamination sources in their 1979 study.

Of particular note, there is no evidence of a difference in ovarian particle content between healthy study subjects and those with ovarian cancer (i.e., no dose-response relationship). Henderson, Hamilton, and Griffiths (1979) measured the highest amount of talc particles in a sample from normal ovaries.

With regard to the transport studies in women, conditions in these investigations may not apply to healthy women. For example, the health status of the women might have affected clearance function (IARC 2010). Study conditions in which women were under anesthesia and had restricted movements during surgical procedures might also have impacted typical transport functions. In some instances, the researchers deliberately instituted measures to encourage mobility of the instilled particles. For example, Egli and Newton (1961) scheduled women for observation/surgery at or near ovulation (when they expected transport to be optimal), had the women lying down during the observation period, and administered oxytocin as a potential agent to facilitate particle transport. A concern specific to the Venter and Iturralde (1979) study using radionuclides is whether the radioactivity counts reflected actual particle transport or leaching of the radioactivity from the particles (e.g., as discussed in Wehner, Weller, and Lepel 1986). Further, it has not been demonstrated that mobility of these particle types other than talc is representative of talc mobility.

Based upon its review of the available literature, IARC (2010) characterized the evidence for retrograde transport of talc to the ovaries in normal women as “weak.” Studies of talc biomonitoring and transport published since 2010 confirm this characterization. IARC’s conclusion is consistent with the observation of an earlier review at a workshop sponsored by ISRTP/FDA that, while results for other types of particles were mixed, the “available histological and physiologic studies provide no basis to conclude that talc can migrate to the ovaries from the perineal region” (Carr 1995).

Overall, studies of talc exposure and transport in the reproductive tract do not support the hypothesis that perineal use of talc plays a causal role in ovarian cancer. This is in agreement with the conclusions of IARC (2010), the Cosmetic Ingredient Review (Fiume et al. 2015), and FDA (Musser 2014).

Mechanisms

Approximately 10-20% of ovarian cancers are caused by inherited factors. The causes of the remaining 80-90% are unknown (Hankinson and Danforth 2006; Hunn and Rodriguez 2012; Walsh et al. 2011), but the literature describes several hypotheses. Although none of these hypotheses can fully explain the causes of all ovarian cancers, it is likely that several factors may play a role (Hankinson and Danforth 2006; Hunn and Rodriguez 2012; Sueblinvong and Carney 2009).

The “incessant ovulation” hypothesis suggests that the process of ovulation, which entails repeated minor trauma to the epithelial surface of the ovary and subsequent cell proliferation to repair the tissue damage, may propagate mutations or promote carcinogenesis (Hankinson and Danforth 2006; Purdie et al. 2003). This is supported by several findings that noted a higher number of lifetime ovulations, a longer duration of ovulation, and increased age are associated with enhanced ovarian cancer risk (Hankinson and Danforth 2006; Hunn and Rodriguez 2012; Purdie et al. 2003).

Another hypothesis involves hormonal effects. Several investigators proposed that fluctuating hormone levels that occur naturally before and during ovulation may stimulate the growth of abnormal cells. Women with consistently higher levels of androgens, estrogens, and prolactin, and decreased progesterone may also be at greater risk. In addition, other hormones may also be involved in the development of ovarian cancer (Cramer et al. 2016; Hankinson and Danforth 2006; Lukanova and Kaaks 2005). This proposed mechanism and the “incessant ovulation” mechanism are not mutually exclusive, as elevated numbers of ovulations correspond with increased episodes of hormonal fluctuation. Further support for a hormonal mechanism is provided by the enhanced risk of ovarian cancer in women who have never taken oral contraceptives, who have taken hormone replacement therapy, or who are obese, as each of these factors affects the levels of certain hormones in the body (Hankinson and Danforth 2006).

A third hypothesis for causation involves inflammation and effects on the immune system. Inflammation is known to be associated with certain types of cancer (e.g., mesothelioma) (Murphy et al. 2012). Endometriosis (another risk factor for ovarian cancer) and ovulation are both inflammatory events and may serve to precipitate development of ovarian cancer (Cramer and Finn 2011; Hankinson and Danforth 2006; Ness and Cottreau 1999). Again, this hypothesis is in concordance with the “incessant ovulation” hypothesis of ovarian cancer.

Some investigators proposed that talc acts as an agent of chronic inflammation that might act to diminish the immune response to ovarian tumors, thus inducing ovaries to be more susceptible to cancer (Cramer 2012; Mandarino et al. 2020; Williams et al. 2014). This hypothesis would negate the need for talc to physically reach the ovaries in order to exert an effect. Various lines of evidence have been put forth in support of this hypothesis.

Williams et al. (2014) claimed to present evidence that talc produces inflammation in women who apply talc perineally, based upon the neutrophil-to-lymphocyte (NLR) ratio in peripheral blood as a general measure of inflammation in women with ovarian cancer. Williams et al. (2014) found that talc use was associated with greater NLR. In fact, there was no exposure-response relationship, and more importantly there was a significant decrease in NLR among all women who used talc for less than 20 years compared to women who did not use talc. Among women who used talc for more than 20 years, the NLR was not significantly greater than NLR for women who did not use talc (geometric mean = 4.1, 95% CI: 3.4–4.9, for > 20 years of use, compared to 3.7, 95% CI: 3.4–4.0, for women who did not use talc). NLRs for premenopausal women followed the same pattern, except the reduction in women with less than 20 years of talc use was not significant.

There is no evidence of talc carcinogenicity in any other tissues, including non-reproductive tissues. If talc were carcinogenic to the ovaries via an inflammatory mechanism, it would be reasonable to expect that it might also exhibit carcinogenicity by the same mechanism in other tissues. There is no evidence, however, of respiratory cancers in humans from occupational or consumer inhalation exposure to asbestos-free talc in settings where the talc dose to lungs is likely far greater than any amount of talc that might reach the human reproductive tract after perineal application (Wehner 2002). The Occupational Safety and Health Administration’s permissible exposure limit for cosmetic talc in air is 20 million particles per cubic foot (Centers for Disease Control and Prevention (CDC), 2016), the recommended exposure limit set by the National Institute for Occupational Safety and Health is 2 mg/m³ (CDC 2016), and the American Conference of Governmental Industrial Hygienists threshold limit value is also 2 mg/m³ (ACGIH, 2010). These values all represent the upper limit of continuous exposure deemed safe by these agencies for workers who are exposed for a 40-hour work week and would result in much higher exposures than perineal talc use. Further, talc has been utilized for decades in a medical procedure, pleurodesis, whereby it is surgically deposited between the parietal and the visceral pleura of the chest to produce an adhesion between them, in order to treat recurrent pneumothorax and other conditions. No increased incidence of lung cancer or mesothelioma has occurred in pleurodesis patients (Baiu, Yevudza, and Shrager 2019; Wehner 2002).

Taken together, the mechanistic evidence does not support a chronic inflammatory role for talc in the causation of ovarian cancer and does not support a viable proposed mechanism for perineal talc use leading to ovarian cancer.

Weight of evidence regarding a causal relationship between talc and ovarian cancer

To determine whether perineal application of talc is causally associated with ovarian cancer, it is critical to evaluate the weight of evidence (WoE) across all the scientific disciplines described in this report. The results of all relevant studies were reviewed, strengths and weaknesses of each considered, and their points of agreement and contradiction weighed. This review was conducted in the context of the Bradford Hill considerations described in the Methods section.

Strength of association

As described in this and other reviews and meta-analyses including IARC (2010) and Penninkilampi and Eslick (2018), cohort studies consistently reported that perineal talc use was not associated with increased ovarian cancer risk, while case-control studies generally show small, positive associations.

Consistency

The results of epidemiology investigations are not consistent across study designs. Cohort studies are consistently null. Case-control investigations generally report small, positive associations. A number of epidemiology studies assessed different modes of talc powder use that contribute to women’s overall exposures and found that observed risks of ovarian cancer with regard to these specific uses are not consistent or are not what would be expected if the association between perineal talc exposure and ovarian cancer were causal.

Specificity

Ovarian cancer is not specific to one particular cause, and most causes are unknown. There are known risk factors and several hypotheses regarding causation such as incessant ovulation, hormone effects or inflammation. Talc and ovarian cancer are not specifically associated with each other, and consequently the conditions of specificity are not met.

Temporality

For women who utilized talc regularly, the use of talc typically preceded the diagnosis of ovarian cancer.

Exposure-response

There is no consistent exposure-response relationship in the epidemiology investigations of talc and ovarian cancer. Case-control studies that evaluated exposure-response and took into account frequency of application, duration of talc use, or both had mixed results. There was no exposure-response relationship found in any of the cohort studies. There was also no enhanced 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. Importantly, none of the investigations had any quantitative information on the level of talc exposure per use or any quantitative estimates of the amounts of talc that are claimed to be transported to various locations in the reproductive tract. Biomonitoring experiments on the whole 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 number of particles found in ovarian tissue and incidence of tumors.

Taken together, the talc-ovarian cancer epidemiology investigations, including cohort studies and several case-control studies, have not consistently reported exposure-response relationships.

Biological plausibility

The evidence regarding the direct genotoxicity of talc and neoplasticity in talc-exposed cultured cells is mostly negative and does not indicate that talc is genotoxic.

Evidence regarding the retrograde transport of talc applied perineally to the ovaries in humans and animals does not support the hypothesis that perineal use of talc plays a causal role in ovarian cancer. Almost all of the identified human studies exploring particle transport in the female reproductive tract involved hospitalized populations and may not be generalizable. Studies that evaluated the interaction between talc use and tubal ligation/hysterectomy noted mixed results (Cramer et al. 2016; Cramer and Xu 1995; Wong et al. 1999) and, thus, do not consistently support the retrograde transport hypothesis. Further, the degree to which the mobility of the non-talc particles employed in many human and animal studies is representative of talc mobility, even in unhealthy patients, has not been established. The exposure setting of primary interest – i.e., talc transport following external application of talc powder on or near the perineum – has not been directly studied in humans or animals.

In addition, there is no consistent evidence of a dose-response relationship between talc use and the number of talc particles found in ovarian tissue. Similarly, there is no marked relationship between number of particles found in ovarian tissue and incidence of tumors. The presence of talc particles in ovarian cancer tissue does not prove that they played any causal role in development of cancer.

There is a proposed mechanism of carcinogenicity involving talc effects on the immune system, including inflammation and immune system suppression, but this has not been demonstrated. There is no evidence in humans that talc applied perineally produces general inflammation or inflammation of the reproductive tract. There is also no solid evidence in animal experiments of talc-induced inflammation of the reproductive tract to support this mechanism for perineally applied talc. Thus, there is insufficient evidence for any proposed mechanism by which talc might plausibly cause ovarian cancer.

Coherence

The epidemiologic evidence for an association between talc and ovarian cancer is contradictory, and there is no evidence of an association in animal studies. The evidence for retrograde transport of talc from perineal applications to the ovaries is insufficient in both animals and humans, and there is no consistent relationship between particles found in ovaries and talc use or tumor incidence. There is no evidence that perineally applied talc initiates inflammation in the reproductive tracts of women. Taken together, the evidence is not coherent or consistent with the hypothesis that talc causes ovarian cancer.

Experiment

There is no relevant experimental evidence on talc and ovarian cancer in humans.

Analogy

There is no evidence of ovarian cancer being induced by a substance analogous to talc. There is no evidence that any other substance applied perineally might enter the reproductive tract and either reach the ovaries and initiate cancer or cause an inflammatory effect that results in ovarian cancer. Although talc is chemically similar to some forms of asbestos, an analogy with asbestos is not appropriate because the fiber size, shape, solubility, durability, and surface chemistry of asbestos determines its carcinogenicity (Boffetta, Mundt, and Thompson 2018; Celsi et al. 2019: Chatfield 2018; Crane 2018; Dodson 2016; Garabrant and Pastula 2018; Gunter 2018; Lemen 2016; NIOSH, 2011), and talc is not analogous to asbestos based on these properties.

Evaluation of alternative hypotheses

Whether the WoE best supports the hypothesis that talc induces ovarian cancer or the alternative hypothesis that talc does not produce ovarian cancer, and the positive associations observed in case-control studies are attributable to bias and confounding rather than any carcinogenic action of talc, was examined.

The small positive findings in case-control studies (and meta-analyses based upon these investigations) have not been confirmed in any of the high-quality prospective cohort studies that have been published. Our review of the literature indicates that, while both case-control and cohort studies may be impacted by bias, the possibility of recall and other biases from the low participation rates and retrospective self-reporting of talc exposure cannot be ruled out for any of the published case-control investigations; in fact, some specific aspects of case-control studies strongly suggest that biases and/or confounding affected the outcomes. The hypothesis that talc exposure induces ovarian cancer is only supported if one ignores the null results of the cohort studies and the fact that bias and/or confounding are the likely reasons for the small, positive associations reported in the case-control investigations. In addition, one would need to ignore the evidence from animal experiments that do not show a relationship with cancer, the in vitro and genotoxicity findings that do not indicate a carcinogenic mechanism for talc, and the mechanistic and transport data that do not support the ability of talc to undergo retrograde transport to reach the ovaries.

The alternative hypothesis that talc does not cause ovarian cancer and that bias and confounding yielded positive associations in case-control studies is better supported by the totality of the available evidence. It is only the case-control studies that may support a causal association, whereas the consistent null results in cohort investigations (which are less susceptible to bias than case-control studies), null results in animal carcinogenicity experiments, the inadequate evidence for retrograde transport of talc, and lack of evidence for genotoxicity or any other plausible mechanism of action support a lack of an association between talc use and ovarian cancer.

The controversy surrounding talc and ovarian cancer began after reports of positive associations in case-control studies. However, when this evidence is balanced with other lines of evidence, namely cohort investigations, animal data, genotoxicity, in vitro, exposure, and transport studies, the hypothesis that talc produces ovarian cancer is not adequately supported by the evidence as a whole. The alternative hypothesis of no causation best fits the evidence across all scientific disciplines, and only requires an explanation for positive associations in the case-control studies, that recall and other biases likely impacted their observations. This explanation is well-supported; thus, it is concluded that the available evidence does not support a causal association between talc and ovarian cancer.

Acknowledgments

The authors thank Drs. Ke Zu and Christine Loftus for technical support, and Kathryn Landoe and Eric Fischbach for administrative support. This manuscript was written with funding from the Cosmetics Alliance Canada (CA Canada) and Industrial Minerals Association-North America (IMA-NA).

Disclosure statement

The authors are employed by Gradient, a private environmental consulting firm. The work reported in this paper was conducted during the normal course of employment, with financial support provided by CA Canada and IMA-NA. Gradient has conducted work on talc in general, and this topic specifically, in the context of litigation and regulatory comments to Health Canada. That work informed this manuscript; however, it did not influence the work presented here. CA Canada and IMA-NA received a draft of this manuscript before it was submitted for publication, but CA Canada and IMA-NA were not involved with the conception or drafting of the manuscript. The authors have the sole responsibility for the writing, content, and conclusions in this article.

Additional information

Funding

This work was supported by the Cosmetics Alliance Canada and Industrial Minerals Association-North America.