Many readers may be aware of large payments made by such companies as Johnson & Johnson to compensate women with a history of ovarian cancer who have claimed that perineal application of talc played a causative role in their cancer development. This column serves to review the purported role of perineal talc use in the development of ovarian cancer, and explore some of the pitfalls of observational science.
ThitareeSarmkasat/iStock/Getty Images Plus
Talc, a hydrated magnesium silicate, is the softest mineral on earth, and has been sold as a personal hygiene product for many decades. Perineal application of talc to sanitary pads, perineal skin, undergarments, and diapers has been a common practice to decrease friction, moisture build-up, and as a deodorant. Talc is chemically similar, although not identical, to asbestos and is geologically located in close proximity to the known carcinogen. In the 1970s, there were concerns raised regarding the possible contamination of cosmetic-grade talc with asbestos, which led to the development of asbestos-free forms of the substance. Given that a strong causal relationship had been established between asbestos exposure and lung and pleural cancers, there was concern that exposure to perineal talc might increase cancer risk.
In the 1980s, an association between perineal talc exposure and ovarian cancer was observed in a case-control study.1 Since that time, multiple other observational studies, predominately case-control studies, have observed an increased ovarian cancer risk among users of perineal talc including the findings of a meta-analysis which estimated a 24%-39% increased risk for ovarian cancer among users.2 Does this establish a causal relationship? For the purposes of legal cases, these associations are adequate. However, science demands a different standard when determining cause and effect.
It is not unusual to rely on observational studies to establish a causal relationship between exposure and disease when it is unethical to randomize subjects in a clinical trial to exposure of the potential harmful agent. This was the necessary methodology behind establishing that smoking causes lung cancer. Several factors must be present when relying on observational studies to establish plausible causation including an observable biologic mechanism, dose-effect response, temporal relationship, consistent effect observed in multiple study populations, and statistical strength of response. These elements should be present in a consistent and powerful enough way to balance the pitfalls of observational studies, namely biases.
A particularly problematic bias is one of recall bias, which plagues case-control studies. Case-control studies are a popular tool to measure a relationship between an exposure and a rare disease, because they are more feasible than the prospective, observational cohort studies that require very large study populations observed over very long periods of time to capture enough events of interest (in this case, cases of ovarian cancer). In case-control studies, researchers identify a cohort of patients with the outcome of interest (ovarian cancer) and compare this population to a control group of similar demographic features. They then survey directly or indirectly (through medical records) for the exposure of interest (perineal talc use).
Recall bias occurs when subjects who have the disease are more likely to have memory of exposure than do control subjects because of the natural instincts individuals have toward attribution. This is emphasized when there is public commentary, justified or not, about the potential risks of that exposure. Given the significant publicity that these lawsuits have had with companies that produced cosmetic talc, it is plausible that ovarian cancer survivors are more likely to remember and negatively attribute their talc exposure to their cancer than are subjects without cancer. Additionally, their memory of volume and duration of exposure generally is enhanced by the same pressures. The potential for this bias is eliminated in prospective, cohort observational studies such as the Women’s Health Initiative Observational Study which, among 61,576 women, half of whom reported perineal talc exposure, did not measure a difference in the development of ovarian cancers during their 12 years of mean follow-up.3
Given these inherent biases, The biologic mechanism of talc carcinogenesis is largely theoretical. As mentioned earlier, prior to the 1970s, there was some observed contamination of talc with asbestos likely caused by the geologic proximity of these minerals. Asbestos is a known carcinogen, and therefore possibly could be harmful if a contaminant of talc. However, it is not known if this level of contamination was enough to be achieve ovarian carcinogenesis. Most theories of talc carcinogenesis are based on foreign body inflammatory reaction via talc particle ascent through the genital tract. This is proposed to induce an inflammatory release of prostaglandins and cytokines, which could cause a mutagenic effect promoting carcinogenesis. The foreign body inflammatory mechanism is further supported by the observation of a decreased incidence of ovarian cancer after hysterectomy or tubal ligation.4 However, inconsistently, a protective effect of NSAIDs has not been observed in ovarian cancer.5
A recent meta-analysis, which reviewed 27 of the largest, best-quality observational studies, identified a dose-effect response with an increased risk for ovarian cancer with greater than 3,600 lifetime applications, compared with less than 3,600 applications.2 The observed association between perineal talc exposure and increased risk of ovarian cancer appears to be consistent across a number of observational studies, including both case-control studies and prospective cohort studies (although somewhat mitigated in the latter). Additionally, there appears to be consistency in the finding that the risk is present for the epithelial subtypes of serous and endometrioid, but not mucinous or clear cell cancer. However, when considering the magnitude of effect, this remains somewhat small (odds ratio, 1.31; 95% confidence interval, 1.24-1.39) when compared with other better established carcinogenic relationships such as smoking and lung cancer where the hazard ratio is 12.12 (95% CI, 6.94-21.17).2,6
If talc does not cause ovarian cancer, why would this association be observed at all? One explanation could be that talc use is a confounder for the true causative mechanism. A theoretical example of this would be if the genital microbiome (a subject we have reviewed previously in this column) was the true culprit. If a particular microbiome profile promotes both oncogenic change in the ovary while also causing vaginal discharge and odor, it might increase the likelihood that perineal talc use is reported in the history of these cancer patients. This is purely speculative, but it always is important to consider the potential for confounding variables when utilizing observational studies to attribute cause and effect.
Dr. Emma C. Rossi
Therefore, there is a consistently observed association between perineal talc application and ovarian cancer, however, the relationship does not appear to be strong enough, associated with a proven carcinogenic mechanism, or free from interfering recall bias such to definitively state that perineal talc exposure causes ovarian cancer. Given these findings, it is reasonable to recommend patients avoid the use of perineal talc application until further definitive safety evidence is provided. In the meantime, it should be noted that even though talc-containing products are not commercially labeled as carcinogens, many pharmaceutical and cosmetic companies have replaced the mineral talc with corn starch in their powders.
Dr. Rossi is assistant professor in the division of gynecologic oncology at the University of North Carolina at Chapel Hill. She had no relevant financial disclosures. Email her at obnews@mdedge.com.
References
1. Cancer. 1982 Jul 15;50(2):372-6.
2. Epidemiology. 2018 Jan;29(1):41-9.
3. J Natl Cancer Inst. 2014 Sep 10;106(9). pii: dju208.
4. Am J Epidemiol. 1991 Aug 15;134(4):362-9.