Relation Between Hysterectomy, Oophorectomy and the Risk of Incident Differentiated Thyroid Cancer

The E3N Cohort

Agathe Guenego; Sylvie Mesrine; Laureen Dartois; Laurence Leenhardt; Françoise Clavel-Chapelon; Marina Kvaskoff; Marie-Christine Boutron-Ruault; Fabrice Bonnet


Clin Endocrinol. 2019;90(2):360-368. 

In This Article

Materials and Methods

The E3N Cohort

TheE3N cohort study (Etude Epidemiologique de Femmes de la Mutuelle Générale de l'Education Nationale) is a prospective cohort of 98 995 French women born between 1925 and 1950 and insured by a national health system primarily covering teachers.18Participants have received 11 follow-up questionnaires (mailed every 2–3 years) and responded to each at a rate of ~80%.[18] The women were enrolled in 1990 after they returned a baseline self-administered questionnaire on their lifestyle and medical history. All women signed an informed consent, in compliance with the rules of the French National Commission for Data Protection and Individual Freedom from which approval was obtained.

Thyroid Disease Assessment

Each questionnaire inquired about cancer occurrence, and requested information and permission to contact participants' physicians. Cases were confirmed by pathology reports.[18] Here, we considered only histologically confirmed first incident primary differentiated thyroid cancer cases (ie, papillary (International Classification of Diseases for Oncology, ICD-O codes: 8050, 8052, 8130, 8260, 8263, 8340- 8344, 8350, 8450) and follicular (8290, 8330–8335, 8480, 8490)), excluding 21 cases of anaplastic or medullary thyroid cancer. Micro- and macro-carcinomas were defined as tumours sized <10 mm and ≥10 mm, respectively.

Each questionnaire also inquired about diagnoses of dysthyroidism (hyperthyroidism or hypothyroidism) and of benign morphological thyroid conditions.

Assessment of Hysterectomy and Benign Gynaecological Diseases

Hysterectomy and oophorectomy status (none, or uni- or bilateral), and age at surgery were recorded in each questionnaire. Benign gynaecological diseases considered in this analysis were endometriosis, ovarian cysts, fibroids and uterine polyps recorded in each questionnaire. Ovarian cysts reported by women diagnosed with endometriosis were not considered, to avoid potentially misdiagnosed ovarian endometrioid cyst cases. For all diseases, diagnostic procedures as laparoscopy, biopsy, hysterography, hysteroscopy or ultrasonography were recorded in the 1992, 1993 and 1994 questionnaires and surgical treatments were available in each questionnaire except the baseline questionnaire. For each gynaecological disease, we separately considered those confirmed by surgery or laparoscopy, confirmed by at least one treatment or diagnosis examination, and only self-reported gynaecological diseases.

Population for Analysis

Follow-up started at the date of return of the 1990 questionnaire. Participants contributed person-years of follow-up until the date of diagnosis of any cancer (except basal cell carcinoma and in situ colorectal cancer), the date of the last completed questionnaire, or December 2011 (date of mailing of the 10th questionnaire), whichever occurred first. Women studied were censored at the date of a first thyroid cancer and we did not consider a hysterectomy or other benign gynaecological diseases which occurred following a first thyroid cancer. Among the 98 995 women included in the study, we excluded those who reported a prevalent cancer at baseline other than a basal cell cancer or in situ colorectal cancer (n = 4844), and women with no follow-up data (n = 2073), with primary amenorrhoea or unknown age at menarche (n = 2209), with missing information on age at hysterectomy or oophorectomy (n = 418), or with unknown date of cancer diagnosis (n = 111), ending up with 89 340 women for analysis.

Statistical Analyses

Cox proportional hazard models with age as timescale were used to estimate Hazard ratios (HR) and 95% confidence intervals (CI) of first differentiated thyroid cancer associated with history of hysterectomy or benign gynaecological disease. Women diagnosed with non-differentiated thyroid cancer were censored at the date of diagnosis. Hysterectomy status or benign gynaecological disease variables were analyzed as time-dependent variables. When the variables were not available at a given questionnaire, the preceding value was considered until the next known value. The proportional hazard hypothesis was verified for all time-independent variables of interest using log-log survivor plots. Covariates included in the models used are listed in Table 2 and Table 3. Information on time-dependent variables was updated at each questionnaire. Missing values for all adjustment variables were replaced by the modal value, as they were missing for less than 5% of women.

We performed stratified analyses to explore associations according to tumour size (micro- (<10 mm) or macro- (≥10 mm) carcinoma) using competing-risk models. Cases with missing values on tumour size were excluded from these analyses, and analyses were performed in each strata by censoring the cases belonged to the other strata at date of diagnosis.

We used homogeneity tests to compare risk estimates across strata using the Wald chi-square statistic.

Smoking status and a history of dysthyroidism were evaluated as potential effect modifiers by adding an interaction term in the final model and testing statistical significance. All tests were two-sided, and statistical significance (P-value) was set at the 0.05 level. All analyses were performed using Statistical Analysis Systems (SAS) software, version 9.3 (SAS Institute, Inc, Cary, North Carolina).