Reproductive Factors and Risk of Papillary Thyroid Cancer in Women

Transcription

1 American Journal of Epidemiology Copyright O 2 by The Johns Hopkins University School of Hygiene and Public Health All rights reserved Vol. 5, Printed In USA. Reproductive Factors and Risk of Papillary Thyroid Cancer in Women Mary Anne Rossing, * Lynda F. Voigt, ' 2 Kristine G. Wicklund, and Janet R. Daling - 2 The authors conducted a population-based case-control study of 4 women residing in three counties in western Washington State who were aged -64 years when diagnosed with papillary thyroid cancer in and 54 controls to assess the effects of pregnancy history and other aspects of reproductive life on risk of this disease. Among women aged 45-64, the authors observed no associations with number of live births, age at first live birth, or age at last live birth. Risk was somewhat increased in women <45 years who had given birth within the previous 5 years; this association was most evident among women who reported that cancer symptoms had led to diagnosis. Among women who had given birth within the last 5 years, risk was greatest among those with two or more births during that time period (relative risk (RR) = 4.2, 95% confidence interval (Cl): 2.,.9, relative to parous women whose last birth was >5 years before the reference date). Risk of thyroid cancer was also associated with lactation during the previous 5 years (e.g., RR = 2.9, 95% Cl:.5, 5.5, among parous women who had breastfed 2 months, vs. - months, during that interval). Our results suggest that thyroid stimulation during both pregnancy and lactation may result in a transient increase in risk of papillary thyroid cancer. Am J Epidemiol 2;5:-2. lactation; menopause; pregnancy; thyroid neoplasms While a number of previous studies have examined the relation of reproductive factors with risk of female thyroid cancer, most of these have been relatively small, and consistent findings have not emerged. In various studies, increased thyroid cancer risk has been associated with late age at menarche, increasing number of births, increasing age at first or last birth, and recency of last birth. Recently, larger studies of thyroid cancer have been conducted (-4) and suggest that, while the overall effect of increasing parity on thyroid cancer risk is weak, a relatively greater increased risk may be present for some time interval after each live birth. However, these latter studies generally had little ability to explore the effects of other possible correlates of recent childbearing, such as lactation or increased surveillance associated with medical attention during the postpartum months. The purpose of this study was to examine, in a study of 4 women with papillary thyroid cancer and 54 population-based controls, the effects of pregnancy history and other related aspects of reproductive life on risk of this disease. Received for publication February 4,999, and accepted for publication June, 999. Abbreviations: Cl, confidence interval; RR, relative risk; SIR, standardized incidence ratio. Program in Epidemiology, Division of Public Health Sciences, Fred Hutchinson Cancer Research Center, Seattle, WA. 2 Department of Epidemiology, School of Public Health and Community Medicine, University of Washington, Seattle, WA. MATERIALS AND METHODS The study population and methods have been previously described (5). Briefly, cases were identified through the Cancer Surveillance System, a populationbased cancer registry operating as part of the National Cancer Institute's Surveillance, Epidemiology, and End Results Program. Eligible cases included women residing in three counties in western Washington State who were aged -64 years and were diagnosed with primary thyroid cancer of the follicular epithelium between January, 9, and December 3, 994. A total of 55 eligible cases were identified, of whom 46 (3.9 percent) were interviewed. The histologic type of thyroid cancer was categorized using data collected by the Cancer Surveillance System. Thyroid cancers classified as papillary (n = 4) for purposes of the present analysis included the following Cancer Surveillance System categories: papillary carcinoma, papillary adenocarcinoma, mixed papillary and follicular adenocarcinoma, and cribriform carcinoma. Because of the small number of women with follicular cancers (n = 5) and the suggestion in previous reports that the risk factors for these two types of thyroid cancer may differ (6, ), these analyses were limited to women with papillary cancer. Control women were identified through random digit telephone dialing (-) and were selected to be similar in age and county of residence to those of the cases. Complete household census information

2 66 Rossing et al. was obtained for 94.5 percent of the residences contacted via telephone screening. Of 3 women who met the study's residence and age requirements, 54 (.9 percent) completed the interview, for a final overall response proportion among controls of 3.6 percent. Controls were assigned a reference date (month and year) that approximated the distribution of diagnosis dates among the cases (5). Information was obtained during an in-person interview using a structured questionnaire that pertained to events before diagnosis (cases) or reference date (controls). The interview included questions on menstrual and reproductive histories, medical history, use of exogenous hormones, family history of cancer, body size, diet, alcohol consumption, smoking habits, and demographic characteristics. Each case was asked to describe the reason her cancer came to be diagnosed, that is, whether the physician visit(s) leading to diagnosis were prompted by a palpable lump or other symptoms related to thyroid cancer, symptoms of an unrelated medical condition (with no symptoms of thyroid cancer), thyroid screening related to childhood irradiation (with no cancer symptoms), or routine medical care with no symptoms of disease. For most analyses, unconditional logistic regression was used to compute odds ratios as estimates of the relative risk (and related 95 percent confidence intervals) of papillary thyroid cancer associated with aspects of reproductive history while controlling for the confounding effects of other variables (). Age at menopause was modeled as a hazard function of case-control status using a Cox proportional hazards model to estimate the relative risk incorporating information from women who had not yet experienced menopause at the reference date (2, 3). (This relative risk can be interpreted as the risk of becoming a case, given menopause occurs at any given age, relative to the risk if menopause occurs later). The reported age at natural menopause or premenopausal bilateral oophorectomy was considered the endpoint of interest, with censoring occurring at the age at which a premenopausal hysterectomy occurred (if some ovarian tissue was retained) or if the reference age was reached before menopause. Characteristics examined as potentially confounding or modifying the associations of interest included age, county of residence, race, marital status, cigarette smoking (never, former, current), alcohol consumption, history of radiation treatment to the head or neck as a child or adolescent (<2 years of age), family history of thyroid cancer, use of oral contraceptives, and history of benign thyroid disease. Analyses shown are adjusted only for the frequency matching variables, age (5-year strata) and county of residence, and for characteristics that appreciably changed the risk estimates. RESULTS Women with papillary thyroid cancer were slightly younger than were controls, and a lesser proportion of cases than controls were cigarette smokers at the reference date (table ). Cases were more likely than controls to report a personal history of benign thyroid disease, while similar proportions of cases and controls reported a prior diagnosis of endometriosis or uterine fibroids. Approximately one half (5. percent) of women with papillary thyroid cancer indicated that self-perceived symptoms of the cancer had led to its diagnosis, while for 4.6 percent of women, an abnormality leading to diagnosis of thyroid cancer was first noted by a care provider either during a medical visit for an unrelated problem or in the absence of any symptoms of illness. Detection of thyroid cancer during screening prompted by history of childhood irradiation was very uncommon (. percent). Weight (assessed year before the reference date) was weakly associated with the risk of papillary thyroid cancer (table 2); the association was somewhat more evident when women with benign hyperplastic thyroid disease (defined as goiter or benign nodule diagnosed 2 or more years before the reference date) were excluded from the analysis (e.g., women >5 pounds vs. <5 pounds, relative risk (RR) =., 95 percent confidence interval (CI):, 2.5). Weight at age was also associated with risk of disease. There was no clear trend in risk with increasing height; however, risk was slightly increased in women >66 inches ( inch = 2.54 cm). We observed no association of body mass index and thyroid cancer risk in analyses based on weight year before the reference date, weight at age, or maximum adult weight (data not shown). The risk increased with increasing age at menarche and was also slightly elevated in women who reported that their periods never became regular. Analyses of various aspects of childbearing were conducted separately in women of reproductive age (<45 years) and older women (45-64 years). Among women aged 45-64, we observed no associations of risk of papillary thyroid cancer with number of live births, age at first live birth, age at last live birth, or breastfeeding (data not shown). Among women,under age 45, we again observed no association of thyroid cancer risk with increasing number of live births (table 3). In analyses adjusted for age, county of residence, and cigarette smoking, parous women within the oldest category of age at first or last birth were at Am J Epidemiol Vol. 5,, 2

3 Thyroid Cancer in Women 6 TABLE. Characteristics of women with papillary thyroid cancer and controls, western Washington State, Cases (n = 4) % Controls (n=54) % Age (years) at reference ^ Self-described race White/Hispanic Black Asian Other Refused Cigarette smoking Never Former Current X-ray treatment to head or neck when <2 years of age Personal history of benign thyroid disease History of endometriosls History of uterine fibroids increased risk of papillary thyroid cancer (RR =.9, 95 percent CI:, 3.9, and RR = 2., 95 percent CI:.9, 4., respectively); however, these associations were either reduced or eliminated upon further adjustment for number of births within the last 5 years (table 3). Risk was somewhat increased among parous women who had given birth within the last 5 years. This association was most evident among women who reported that symptoms of thyroid cancer had led to its diagnosis (e.g., among women with symptoms who had delivered within the last year, RR = 2.5, adjusted for age, county, and smoking). Moreover, among women with a recent (i.e., within the last 5 years) birth, risk was greater among women with two or more births during that time period; relative to parous women whose last birth was more than 5 years before the reference date, the risk among women with two or more births in the last 5 years was 4.2 (95 percent CI: 2.,.9). Among women under age 45, the risk of thyroid cancer was also associated with increasing lifetime duration of lactation and with decreasing time since last lactation. However, similar to our findings regarding the relation of thyroid cancer risk to number of live births, the association with duration of lactation within the 5 years before the reference date was more marked than was the association with lifetime duration of breastfeeding (table 3). The associa- Am J Epidemiol Vol. 5,, 2

4 6 Rossing et al. TABLE 2. Relation of measures of body size and menstrual characteristics with risk of papillary thyroid cancer, western Washington State, Cases % Controls % OR* vn oca/ pi* JJOTO \jl Weight ((lb)t year before reference date)*, ,.6.9,..9,.9 Weight ((Ib) year before reference date)t, < ,2.2 Weight at years of aget Much less than other children Somewhat less than other children About the same as other children Somewhat more than other children Much more than other children ,.4.,.4.,.., 3.2 Height (in)t.h < > ,.,.9 Age (years) at menarche,# ,.2.,.5,3. Ever had regular menstrual periods,# , 2.3 OR, odds ratio; Cl, confidence interval, t One pound (Ib) =.45 kg. t Adjusted for age and county of residence. Adult weight unknown for one control; age at menarche unknown for one case and one control; regular menstrual periods unknown for one case and two controls. H One inch (in) = 2.54 cm. # Adjusted for age, county, race, and relative weight at age. tion of recent lactation and thyroid cancer risk was observed in subgroups of women with and without symptoms leading to cancer diagnosis, although greatest among women without presenting symptoms (data not shown). Of the cases and controls who had breastfed within the 5 years before the reference date, all but four cases (4.6 percent) and four controls (5. percent) had also delivered a live birth during that time. In analyses restricted to women who had had one or more births within the last 5 years, the associations of recent number of live births and recent lactation with risk of papillary thyroid cancer remained apparent after adjustment for each of these factors by the other (table 4). A history of hysterectomy was associated with an increased risk of thyroid cancer in women 45 years of age (table 5), while no association was evident in women under age 45 (RR =, 95 percent CI:.6,.9). Among women aged years, we observed little evidence that risk of thyroid cancer varied according to the time since or age at which hysterectomy had occurred; also, the association was not attributable to concurrent bilateral oophorectomy or to a particular medical indication. Risk estimates were similar for women with and without thyroid cancer symptoms leading to diagnosis. Cases were somewhat more likely to experience menopause (defined as natural menopause or premenopausal bilateral oophorectomy) Am J Epidemiol Vol. 5,, 2

5 Thyroid Cancer in Women 69 TABLE 3. Birth and lactation history and risk of papillary thyroid cancer in vramen under 45 years of age, western Washington State, 9-994* nt VI cases nf controls OR* Parous women 95% Cl ORt 95%Clf Parity ,.., 2.4.5, ,.5.6,.4., 2..5, 2.3 Age (years) at first birth ^M) ,.9.6, 2.., 3.3 Age (years) at last birth ^ ,.3.4,.3.4, 2.6 Time (years) since last birth > < Currently pregnant ,.3,3.2., 4..6, 3. of births in last 5 years ,2. 2.,.9 Lifetime duration (months) of lactation Never or < , 2..5,.., 2..9, 3.3 Years since last lactation > < ,.2., 2..,5.5 Duration of lactation (months) in the last 5 years -< ,3..5,5.5 * Currently pregnant women are excluded from all analyses other than time since last birth, t OR, odds ratio; Cl, confidence interval. X Age at first and last birth adjusted for age, county of residence, smoking, and number of births in the last 5 years. All other analyses adjusted for age, county of residence, and smoking. excluded women who had ever used hormone replace- ment therapy (of any type) or combined estrogen and progestogen therapy. at an earlier age than were controls (RR =.3, 95 percent Cl:.9,.9, adjusted for age, county of residence, and smoking). These results were similar when we Am J Epidemiol Vol. 5,, 2

6 Rossing et al. TABLE 4. Birth and lactation history and risk of papillary thyroid cancer among women under 45 years of age who reported one or more births in the last 5 years, western Washington State, 9-994* of cases of controls ORt 95%Clf of births in the last 5 years}: , 4.6 Duration of lactation (months) in the last 5 years -< , 5.,. Time (months) since last lactation Never lactated > Currently pregnant women are excluded. t OR, odds ratio; Cl, confidence interval. $ Adjusted for age, county of residence, smoking, and duration of lactation in the last 5 years. Adjusted for age, country of residence, smoking, and number erf births in the last 5 years , 5.,9. TABLE 5. Hysterectomy and risk of papillary thyroid cancer In women years of age, western Washington State, of cases of controls OR*,t 95% Cl«Ever had hysterectomy With bilateral oophorectomy Without bilateral oophorectomy Oophorectomy status unknown ,3..6, 2.6.3,4. Years since hysterectomy Never <5 5- > , 5.2.4, 2.9.2,3. Age (years) at hysterectomy Never <3 3-4 > , 4.9.,3.9., 3.2 Reason for hysterectomy Never Fibroids Dysmenorrhea Endometrlosis Cervical cancer/cin* Prolapsed uterus/other uterine problem Other OR, odds ratio; Cl, confidence interval; CIN, cervical Intraepithelial neoplasia. t Adjusted for age and county..5., 4..3, 2..5, 6.., 5..5,3.4.6, 5. Am J Epidemiol Vol. 5,, 2

7 Thyroid Cancer in Women DISCUSSION Differences in characteristics of women who did or did not choose to participate in this study may influence our results as may errors in recall of various aspects of reproductive life. In addition, nearly half of the women with thyroid cancer had their tumors diagnosed in the absence of disease symptoms; thus, it is possible that sociodemographic and other characteristics that influence access to and use of medical care may affect the likelihood or timing of diagnosis. Conceivably, more frequent contact with physicians among women with a recent pregnancy may have prompted detection of thyroid cancer, particularly asymptomatic disease. However, the association between recent pregnancy and risk of papillary thyroid cancer remained apparent in an analysis restricted to women who reported that symptoms of their tumors had led to diagnosis. Also, increased cancer detection among women with a recent pregnancy is not Likely to completely explain the associations we observed with increasing number of recent (i.e., within the 5 years before reference date) pregnancies and duration of recent lactation. Although thyroid cancers are diagnosed in life 2-3 times more frequently in women than in men (4), similar proportions of occult papillary carcinomas were detected in men and women in a consecutive series of autopsies of individuals with no known thyroid disease (5). Based in part on these observations, others (, 6) have suggested that hormone-related factors in women may act to promote the growth of subclinical tumors to the point that they become clinically evident. Stimulation of thyroid activity during pregnancy is evidenced by increases in thyroid volume as well as increased uptake of radioactive iodine; both thyroidstimulating hormone and human chorionic gonadotropin may be involved. Levels of the latter are increased during pregnancy, and human chorionic gonadotropin is structurally similar to thyroid-stimulating hormone and can interact with the thyroid-stimulating hormone receptor (). Little information is available regarding the extent of thyroid stimulation in lactating women. Iodine and thyroid hormones are secreted in milk, and serum thyroid hormone concentrations are positively correlated with milk production (-2). Moreover, in two studies, thyroid-stimulating hormone levels at 3 months postpartum were higher among women who were lactating (, 2). Lambe and Ekbom (2), using linked data from the Swedish national cancer and fertility registries, reported a reduced risk of thyroid cancer during pregnancy (standardized incidence ratio (SIR) =.34, 95 percent CI:.,.5) and an increased risk during the 2 months after delivery (SIR =., 95 percent CI:.44, 2.). They hypothesized that their findings could be due to a delay of cancer diagnosis until the postpartum period, as well as to the stimulation of malignant cell growth by hormonal and/or immunologic changes during childbearing. Galanti et al. () conducted a case-control study of,49 cases of thyroid cancer under 6 years of age and,9 agematched controls. They reported that, among women aged 5-49 years, the relative risk of thyroid cancer was.5 (95 percent CI:.3,.) in the year after conception of a pregnancy that ended in a live birth (i.e., the gestation and immediate postpartum interval) and.5 (95 percent CI:.,.9) 2-4 years after conception, relative to women whose most recent live birth was conceived or more years earlier; this association was most evident in uniparous women. Similar findings were observed for both papillary and follicular cancers, when these subtypes were examined separately. Kravdal et al. (3) reported an increased risk among women who conceived a pregnancy ending in a live birth within the 45 months before diagnosis for follicular, but not papillary, thyroid cancer. Most recently, Negri et al. (4) conducted a pooled analysis of data from 4 case-control studies of thyroid cancer that included 2,24 women with thyroid cancer ( percent of which was papillary) and 3,699 control women. Similar to the current study, risk increased with increasing age at menarche (RR per year increase = 4, 95 percent CI:,.). They observed no clear association of disease risk with parity; the relative risk among parous versus nulliparous women was.2 (95 percent CI:,.4), with no trend of increasing risk with increasing number of births. In women below age 45, the risk of thyroid cancer somewhat decreased with longer time since last birth; the relative risk estimate was.9 (95 percent CI:., ) for a 5-year increase in time since last birth. While risk was slightly increased among women with later age at first or last birth, these associations were generally stronger at younger ages; for example, the relative risk for a 5-year increase in age at first or last birth, respectively, was.3 or.4 at age 35 and younger and above age 55. These authors observed no association of thyroid cancer risk with ever having breastfed; the relative risk for ever versus never breastfeeding was, and the relative risk per 2- month increase in lifetime duration of breastfeeding was. (95 percent CI:,.2). Other aspects of breastfeeding, such as recency, were not examined. Similar to the current study, Negri et al. (4) reported that women with thyroid cancer were more likely than similarly aged controls to have undergone either natural or artificial menopause. A few other studies have also reported an increased risk of thyroid cancer among women who had undergone hysterectomy (6, Am J Epidemiol Vol. 5,, 2

8 2 Rossing et al. 2). Some authors (4, 6) have suggested that this association might reflect either increased detection of thyroid cancers at the time of hysterectomy or confounding by indication for hysterectomy. In the current study, we observed a relation of hysterectomy and risk of thyroid cancer only among women 45 years of age; among those women, there was no clear relation with time since hysterectomy, the age at which hysterectomy had occurred, or a particular indication for this procedure. Thus, the basis for the observed association (whether biologic or artifactual) remains unclear. We observed no relation of risk of papillary thyroid cancer to aspects of pregnancy history (including number of live births and age at first or last live birth) or breastfeeding among women over 45 years of age. Among younger women, increases in risk observed with increasing age at first and last live birth were less evident after adjustment for number of births within the last 5 years. Taken together, these results suggest that associations of thyroid cancer risk with older age at first and last birth may primarily reflect the correlation of these characteristics (in younger women) with having had a recent birth. The risk of thyroid cancer was slightly elevated among women who were currently pregnant; thus, we observed no evidence that increased risk among women who had recently delivered could be attributable to delay of cancer diagnosis to the postpartum period. The increased risk of papillary thyroid cancer that we observed among women who had delivered a live birth within the 5 years before the reference date was particularly evident among women who, during this time interval, had had more than one birth and had breastfed. While our observations are consistent with those of other studies that observed an association of thyroid cancer risk with recent childbearing, they further suggest that hormonal, metabolic, or other effects of both pregnancy and lactation may exert a transient influence on risk of this disease. ACKNOWLEDGMENTS This work was supported by grant RO CA526 and by the Cancer Surveillance System of the Fred Hutchinson Cancer Research Center, which is funded by contract N- CN-523 from the Surveillance, Epidemiology, and End Results (SEER) Program of the National Cancer Institute with additional support from the Fred Hutchinson Cancer Research Center. The authors thank Drs. el Weiss and Mats Lambe for their helpful comments. REFERENCES. Galanti MR, Lambe M, Ekbom A, et al. Parity and risk of thyroid cancer: a nested case-control study of a nationwide Swedish cohort. Cancer Causes Control 995;6: Lambe M, Ekbom A. Cancers coinciding with childbearing: delayed diagnosis during pregnancy? BMJ 995;3: Kravdal O, Glattre E, Haldorsen T. Positive correlation between parity and incidence of thyroid cancer new evidence based on complete rwegian birth cohorts. Int J Cancer 99;49: Negri E, Dal Maso L, Ron E, et al. A pooled analysis of casecontrol studies of thyroid cancer. II. Menstrual and reproductive factors. Cancer Causes Control 999; : Rossing MA, Voigt LF, Wicklund K, et al. Use of exogenous hormones and risk of papillary thyroid cancer. Cancer Causes Control 99;9: Wingren G, Hatschek T, Axelson O. Determinants of papillary cancer of the thyroid. Am J Epidemiol 993;3: Franceschi S, Boyle P, LaVecchia C, et al. The epidemiology of thyroid carcinoma. Crit Rev Oncog 993;4: Waksbcrg J. Sampling methods for random digit dialing. J Am Stat Assoc 9;3:4O Hartge P, Brinton LA, Rosenthal JF, et al. Random digit dialing in selecting a population-based control group. Am J Epidemiol 94;2: Harlow BL, Davis S. Two one-step methods for household screening and interviewing using random digit dialing. Am J Epidemiol 9; 2: Breslow NE, Day NE, eds. Statistical methods in cancer research. Vol I. The analysis of case-control studies. Lyon, France: International Agency for Research on Cancer, 9. (IARC scientific publication no. ). 2. Breslow NE, Day NE, eds. Statistical methods in cancer research. Vol n. The design and analysis of cohort studies. Lyon, France: International Agency for Research on Cancer, 9. (IARC scientific publication no. 2). 3. Stanford JL, Weiss NS, Voigt LF, et al. Combined estrogen and progestin hormone replacement therapy in relation to risk of breast cancer in middle-aged women. JAMA 995;24: 3^2. 4. Henderson BE, Ross RK, Pike MC, et al. Endogenous hormones as a major factor in human cancer. Cancer Res 92;42: Martinez-Tello FJ, Martinez-Cabruja R, Fernandez-Martin J, et al. Occult carcinoma of the thyroid: a systematic autopsy study from Spain of two series performed with two different methods. Cancer 993;: Galanti MR, Hansson L, Lund E, et al. Reproductive history and cigarette smoking as risk factors for thyroid cancer in women: a population-based case-control study. Cancer Epidemiol Biomarkers Prev 996;5: Kennedy RL, Dame J. The role of hcg in regulation of the thyroid gland in normal and abnormal pregnancy. Obstet Gynecol 99;: Strfcak V, Macho L, Uhercik D, et al. The effect of lactation on thyroid activity of women. Endokrinologie 9;2: Motil KJ, Thotathuchery M, Montandon CM, et al. Insulin, cortisol and thyroid hormones modulate maternal protein status and milk production and composition in humans. J Nutr 994;24: Iwatani Y, Amino N, Tanizawa O, et al. Decrease of free thyroxine in serum of lactating women. Clin Chem 9;33: Luotta RA, Auvinen A, Pukkala E, et al. Hysterectomy and subsequent risk of cancer. Int J Epidemiol 99;26:46-3. Am J Epidemiol Vol. 5,, 2