Association Between Polycystic Ovary Syndrome and Cancer Risk

Inconsistent results have been reported for the association between polycystic ovary syndrome (PCOS) and risks of ovarian and breast cancers.¹ Few studies had sufficient sample sizes, well-defined diagnostic criteria, or reasonable confounding adjustment.^1,2 Even fewer studies have addressed the risks of other cancers. Thus, we conducted a large, population-based cohort study with a long follow-up period and rather sufficient confounding adjustment.

Using Swedish register data, we established a cohort including all females aged 15 to 50 years between 1985 and 2009 (N = 3 511 357). Those with cancers before entry (n = 17 750) and PCOS diagnosed when they were younger than 9 years (n = 3) were excluded. All eligible participants (n = 3 493 604) were followed up until cancer, emigration, death, or December 31, 2009, whichever came first. The present study was conducted from October 1, 2014, to March 21, 2018. Ethical approval for study was granted by the Regional Ethics Vetting Board in Stockholm, Sweden, with waiver of informed consent.

Exposure of interest was a diagnosis of PCOS. The outcome measure was histologically verified primary cancer. Cancer diagnosed within 1 year from exposure was considered occurring in the unexposed period. The cohort was cross-linked to other registers to identify information on exposure, outcome, and covariates (including the highest achieved educational level, parity, parental cancer history, use of medications, and diabetes). Hazard ratios (HRs) for cancers among PCOS vs non-PCOS individuals were derived from Cox proportional hazards regression models with 2-tailed 95% CIs. The significance level was set at α = .05. Proportionality assumption was tested and no obvious deviation was found. To address influence of menopause, HRs over attained age younger than 51 and 51 or older years (median menopause age in Sweden³) were estimated separately.

We further analyzed the interaction between PCOS and covariates and performed sensitivity analyses by excluding individuals with missing educational level data or by additionally adjusting for use of medications (metformin, oral contraceptives, and hormone replacement therapy) with follow-up constrained to 2005-2009 (the Prescribed Drug Register became available July 1, 2005). SAS, version 9.4 (SAS Institute) and Stata, version 14.0 (StataCorp LP) were used for all statistical analyses.

Among 14 764 women with diagnosed PCOS (mean [SD] age at study entry, 28.5 [6.9] years), 182 had primary cancers. Polycystic ovary syndrome was positively associated with an increased overall cancer risk (fully adjusted HR, 1.15; 95% CI, 1.00-1.33) (Table 1). Excess cancer risks observed at specific sites determined from fully adjusted models were as follows: endometrium (HR, 2.62; 95% CI, 1.58-4.35), ovary (HR, 2.16; 95% CI, 1.30-3.59), endocrine gland (HR, 1.92; 95% CI, 1.21-3.06), pancreas (HR, 3.40; 95% CI, 1.41-8.20), kidney (HR, 3.07; 95% CI, 1.27-7.39), and skeletal and hematopoietic system (HR, 1.69; 95% CI, 1.05-2.72).

We further performed analysis by menopause status (Table 2). The excess risks were prominent only before menopause. The HR for overall cancer was 1.22 (95% CI, 1.03-1.44); endometrial cancer, 6.45 (95% CI, 3.65-11.40); ovarian cancer, 2.55 (95% CI, 1.45-4.50); pancreatic cancer, 6.68 (95% CI, 2.49-17.91); kidney cancer, 4.57 (95% CI, 1.71-12.22); and endocrine gland cancers (except thyroid cancer), 2.90 (95% CI, 1.56-5.40).

No significant interaction between PCOS and covariates was observed. In sensitivity analyses, excluding patients with missing educational level data or further adjusting for use of medications did not change the results materially.

Several carcinogenic processes are associated with PCOS, including dyslipidemia, hyperinsulinemia, and chronic inflammation.⁴ However, studies on risk of all types of cancer are scarce (except for endometrial, ovarian, and breast cancers). To our knowledge, only Gottschau et al⁵ reported a higher risk of kidney, colon, and brain cancer without adjusting for confounding factors. Our study indicates that cancer may need to be added to the spectrum of long-term health consequences of PCOS and warrants increased surveillance among these patients.

The limitations of the study include underestimated PCOS prevalence from the National Patient Register (patients in primary care, private clinics, or outpatient care before 2001 could not be identified) and the variation of PCOS morbidity owing to differing diagnostic criteria. Further studies are needed to define the cancer risk pattern among patients with PCOS more clearly.

Accepted for Publication: September 6, 2018.

Corresponding Authors: Weimin Ye, MD, PhD, Department of Medical Epidemiology and Biostatistics, Karolinska Institutet; PO Box 281, 171 77 Stockholm, Sweden (weimin.ye@ki.se); Liangzhi Xu, MD, PhD, Department of Reproductive Endocrinology, West China Second University Hospital, Sichuan University, Chengdu, China (xuliangzhi_art@126.com).

Published Online: November 29, 2018. doi:10.1001/jamaoncol.2018.5188

Author Contributions: Drs W. Yin and Ye had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: W. Yin, Falconer, L. Yin, Ye.

Acquisition, analysis, or interpretation of data: W. Yin, L. Yin, Xu, Ye.

Drafting of the manuscript: W. Yin.

Critical revision of the manuscript for important intellectual content: Falconer, L. Yin, Xu, Ye.

Statistical analysis: W. Yin, L. Yin.

Obtained funding: Xu, Ye.

Administrative, technical, or material support: Falconer, Ye.

Supervision: Xu, Ye.

Conflict of Interest Disclosures: None reported.

Funding/Support: The creation of the database used in this work was supported by grant 80748301 from the Swedish Research Council. The project was supported by National Natural Science Foundation of China grant 81270665.

Role of the Funder/Sponsor: The funding organizations had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.