Diabetes Volume 66, March 2017 577

Taulant Muka,1,2 Jana Nano,1 Loes Jaspers,1 Cindy Meun,3 Wichor M. Bramer,4 Albert Hofman,1,2 Abbas Dehghan,1 Maryam Kavousi,1 Joop S.E. Laven,2 and Oscar H. Franco1

Associations of Steroid Sex and Sex –Binding Globulin With the Risk of Type 2 Diabetes in Women: A Population-Based Cohort Study and Meta-analysis

Diabetes 2017;66:577–586 | DOI: 10.2337/db16-0473

It remains unclear whether endogenous sex hormones for being accompanied by an increased risk of cardiovas- (ESH) are associated with risk of type 2 diabetes (T2D) in cular disease and type 2 diabetes (T2D) (1,2). Changes in women. Data of 3,117 postmenopausal women partici- hormonal patterns in menopause, including the decline in pants of the Rotterdam Study were analyzed to examine endogenous (E) levels and the relative – whether ESH and binding globulin (SHBG) excess, contribute to an increase in visceral adiposity that were associated with the risk of incident T2D. Addition- is associated with glycemic traits and therefore may influ- ally, we performed a systematic review and meta-analysis ence the risk of T2D (3,4). Furthermore, polycystic of studies assessing the prospective association of ESH syndrome, a common disorder among women character- and SHBG with T2D in women. During a median follow-up ized by , has been identified as a signif- of 11.1 years, we identified 384 incident cases of T2D in icant nonmodifiable risk factor associated with T2D (5). the Rotterdam Study. No association was observed be- Although the relation between sex hormone–binding tween total (TT) or bioavailable testosterone globulin (SHBG) and T2D has long been recognized (6,7), (BT) with T2D. SHBG was inversely associated with the risk of T2D, whereas total estradiol (TE) was associated literature on the associations of steroid sex hormones, such with increased risk of T2D. Similarly, in the meta-analysis as endogenous E and testosterone (T), with T2D is scarce. of 13 population-based prospective studies involving SHBG, T, and E have been associated with glucose me- – more than 1,912 incident T2D cases, low levels of SHBG tabolism and development of insulin resistance (6 9). and high levels of TE were associated with increased risk Few epidemiological studies investigating the relation of T2D, whereas no associations were found for other between sex hormones and T2D have yielded conflicting hormones. The association of SHBG with T2D did not results (10–12). These studies were limited by their cross- change by menopause status, whereas the associations sectional design, selected samples, or insufficient adjust- of ESH and T2D were based only in postmenopausal ment for diabetes risk factors. To date, no large prospective women. SHBG and TE are independent risk factors for cohort study has examined the association of T2D with the development of T2D in women. SHBG, T, and E in healthy postmenopausal women. Thus, we aimed to investigate the association between SHBG, sex hormones, and T2D in postmenopausal women. Menopause is an important transition in a woman’s life, Furthermore, to clarify the contradictory results, we sys- not only for marking the end of reproductive life but also tematically reviewed and meta-analyzed studies evaluating

1Department of Epidemiology, Erasmus MC, Rotterdam, the Netherlands This article contains Supplementary Data online at http://diabetes 2Department of Epidemiology, Harvard T.H. Chan School of Public Health, Boston, .diabetesjournals.org/lookup/suppl/doi:10.2337/db16-0473/-/DC1. MA T.M., J.N., M.K., and J.S.E.L. contributed equally to this work. 3Department of Obstetrics and Gynecology, Erasmus MC, Rotterdam, the © 2017 by the American Diabetes Association. Readers may use this article as Netherlands long as the work is properly cited, the use is educational and not for profit, and the 4Medical Library, Erasmus MC, Rotterdam, the Netherlands work is not altered. More information is available at http://www.diabetesjournals Corresponding author: Taulant Muka, [email protected]. .org/content/license. Received 13 April 2016 and accepted 4 October 2016. See accompanying article, p. 568. 578 Steroid Sex Hormones, SHBG, and T2D Diabetes Volume 66, March 2017 the association between SHBG, sex hormones, and T2D and ,5%. The free androgen index (FAI), calculated as in women. (T/SHBG)*100,isusedasasurrogatemeasureofbio- available testosterone (BT) (16). RESEARCH DESIGN AND METHODS Population of Analysis The Rotterdam Study The current study used data from the third visit of the The Rotterdam Study is a prospective cohort study which first cohort (RSI-3) and the baseline examinations of the started since 1990 in the Ommoord district, in the city of second (RSII-1) cohort. Overall, there were 3,683 post- Rotterdam, the Netherlands. Details regarding the design, menopausal women eligible for blood measurements. objectives, and methods of the Rotterdam Study have Among them, 122 women did not come for a blood been described in detail elsewhere (13). In brief, in 1990, measurement at the research center and 32 did not have all inhabitants of a well-defined district of Rotterdam T2D follow-up data and were excluded from the analysis. were invited, of whom 7,983 agreed (78.1%). In 2000, Furthermore, 412 women with prevalent T2D were an additional 3,011 participants were enrolled (RS-II), excluded, leaving 3,117 for our final analysis. Potential consisting of all people living in the study district who confounding variables are described in detail in Supple- had become 55 years of age. Follow-up examinations were mentary Appendix 1. performed periodically, approximately every 3–5 years (13). There were no eligibility criteria to enter the Rotterdam Statistical Analysis Study cohorts except the minimum age and residential Person-years of follow-up were calculated from study area based on ZIP codes. The Rotterdam Study has been entrance (March 1997 to December 1999 for RSI-3 and approved by the medical ethics committee according to the February 2000 to December 2001 for RSII-1) to the date Population Screening Act: Rotterdam Study, executed by of diagnosis of T2D, death, or the censor date (date of the Ministry of Health, Welfare and Sport of the Nether- last contact of the living), whichever occurred first. lands. All participants in the present analysis provided Follow-up was until 1 January 2012. Cox proportional written informed consent to participate and to obtain hazards modeling was used to evaluate whether SHBG, TT, information from their treating physicians. TE, and BT were associated with T2D. Relative risks (RRs) and 95% CIs were reported. All sex hormone variables were Ascertainment of T2D assessed in separate models, continuously and in tertiles. The participants were followed from the date of baseline For E, first tertile included all women with levels of E center visit onwards. At baseline and during follow-up, cases lower than the detection limit (n = 992). To study the of T2D were ascertained through active follow-up using relations across increasing tertiles, trend tests were general practitioners’ records, glucose hospital discharge let- computed by entering the categorical variables as con- ters, and glucose measurements from Rotterdam Study vis- tinuous variables in multivariable Cox proportional hazards its, which take place approximately every 4 years (14). T2D fi models. To achieve approximately normal distribution, was de ned according to recent World Health Organization skewed variables (SHBG, TT, BT, plasma , guidelines, as a fasting blood glucose $7.0 mmol/L, a non- LDL [LDL-C], C-reactive [CRP], fasting blood glucose $11.1 mmol/L (when fasting samples thyroid-stimulating hormone [TSH], and insulin) were were absent), or the use of blood glucose–lowering medica- natural log transformed. In the base model (model 1), tion (15). Information regarding the use of blood glucose– we adjusted for age, cohort (1,2), and fasting status (fast- lowering was derived from both structured home ing sample vs. nonfasting sample). To examine whether interviews and linkage to pharmacy records (14). At base- the relations of sex hormones and SHBG with risk of T2D line, .95% of the Rotterdam Study population was covered were independent of established risk factors for T2D, by the pharmacies in the study area. All potential events model 2 included the terms of model 1, BMI (continuous), of T2D were independently adjudicated by two study glucose (continuous), and insulin (continuous). BMI and physicians. In case of disagreement, consensus was sought waist circumference were highly correlated (Pearson cor- with an endocrinologist. Follow-up data were complete relation coefficient = 0.81, P , 0.001), so only BMI was until 1 January 2012. used as a measure of adiposity, consistent with previous Sex Steroid Measurements studies (10,12). Model 3 included all covariates in model All blood samples were drawn in the morning (#11:00 A.M.) 2 and further potential intermediate factors, including and were fasting. Total estradiol (TE) levels were mea- metabolic risk factors (total cholesterol, systolic blood sured with a radioimmunoassay and SHBG with the Im- pressure [continuous], indication for hypertension [yes mulite platform (Diagnostic Products Corporation, vs. no], and use of -lowering [yes vs. Breda, the Netherlands). The minimum detection limit no]), lifestyle factors (alcohol intake [continuous] and for E was 18.35 pmol/L. Undetectable E was scored smoking status [current vs. former/never]), prevalent cor- as 18.35. Serum levels of total testosterone (TT) were onary heart disease (yes vs. no), age of menopause, hor- measured with liquid –tandem mass mone replacement therapy (yes vs. no), CRP (continuous), spectrometry. The corresponding interassay coefficients and sex hormones for each other. Effect modifications of of variations for TE, SHBG, and TT are ,7%, ,5%, sex hormones by BMI and years since menopause were diabetes.diabetesjournals.org Muka and Associates 579

fi tested in the nal multivariable model in addition to per- — fi Table 1 Selected characteristics of study participants, the forming strati ed analysis. We also performed a series of Rotterdam Study sensitivity analyses. Since waist circumference is a better % measure of visceral adiposity, an important risk factor for Women missing diabetes and of sex hormone levels after menopause, we (n = 3,117) values performed the analysis substituting it with BMI. To ac- Age (years) 69.7 6 8.7 0 count for the specific effects of lipid particles on diabetes, Years since menopause (years) 20.9 6 10.0 4.4 we substituted total cholesterol with HDL cholesterol, Age of menopause (years) 48.9 6 5.2 4.4 , and LDL-C. TSH, physical activity, number Number of of at least of pregnancies, and type of menopause (nonnatural vs. 6 months 2.3 6 2 12.4 natural) are associated with sex hormone levels and/or Natural menopause, n (%) 2,433 (78.1) 0 risk of T2D; therefore, the models were further adjusted Current smokers, n (%) 218 (9.2) 1.8 for these factors. To explore potential reverse causation, a we reran the analysis by excluding the first 3 years of Alcohol intake (g/day) 1.3 (10) 26.5 2 follow-up. Multiple imputation procedure was used (n = BMI (kg/m ) 27.0 6 4.3 2.3 5 imputations) to adjust for potential bias associated with Waist circumference (cm) 89.4 6 11.6 5.8 missing data. Rubin method was used for the pooled Prevalent coronary heart regression coefficients (b) and 95% CI (17). A P value disease, n (%) 86 (2.8) 0.06 of ,0.05 was considered statistically significant. All anal- E (pmol/L) 34.2 (41.62)a 0 yses were done using SPSS statistical software (SPSS, ver- TT (nmol/L) 0.8 (0.56)a 0 sion 21.0; SPSS, Inc., Chicago, IL). SHBG (nmol/L) 69.6 6 33.0 0 a Systematic Review and Meta-analysis FAI 1.3 (1.1) 0 Data Sources and Search Strategy. The review was TSH (mU/L) 1.95 (1.7)a 0.03 conducted using a predefined protocol and in accordance Hormone replacement therapy, n (%) 159 (5.3) 4.8 with the Preferred Reporting Items for Systematic Reviews Insulin (pmol/L) 67 (47)a 0.26 and Meta-analyses (PRISMA) (18) and Meta-analysis Of Glucose (mmol/L) 5.5 6 0.6 1.3 Observational Studies in Epidemiology (MOOSE) (19) CRP (mg/mL) 1.7 (2.93)a 3.7 guidelines (Supplementary Appendices 2 and 3). Medline, Total cholesterol (mmol/L) 6.0 6 1.0 1.3 Embase.com, Web of Science, the Cochrane Library, a PubMed, and Google Scholar were searched from inception LDL-C (mmol/L) 4.2 (1.22) 2.5 until 2 November 2015 (date last searched) with the assis- HDL cholesterol (mmol/L) 1.5 6 0.4 2.3 tance of an experienced biomedical information specialist. use, n (%) 681 (14) 4.8 The computer-based searches combined terms related to Triglycerides (mmol/L) 1.27 (0.74)a 0.26 the exposure (e.g., sex hormone binding globulin, T, and Systolic (mmHg) 142.0 6 21.1 1.03 E) with outcomes (e.g., T2D), without any language re- Indication for hypertension, n (%) 794 (25.5) 1.03 striction. Details on the search strategy are provided in Incident T2D, n (%) 384 (12.3) 0 Supplementary Appendix 4. 6 a Study Selection and Eligibility Criteria. Studies were Plus/minus values are mean SD. Median (interquartile range). included if they 1) were observational cohort, case-cohort, or prospective nested case-control studies; 2) had reported on at least one of the sex hormones as exposures (SHBG, Sex Hormones and the Risk of Developing T2D 3 TT, BT, TE, and bioavailable estradiol [BE]); and ) In models adjusted for age, cohort effect, and fasting had assessed associations with risk of T2D in women status, lower SHBG levels (third vs. first tertile: RR 0.33 (pre- and postmenopausal). Two independent reviewers [95% CI 0.25–0.43], P trend ,0.001) and higher levels fi screened the titles and abstracts of all initially identi ed of BT (third vs. first tertile: RR 2.01 [95% CI 1.55–2.60], studies according to the selection criteria. Full texts were P trend ,0.001) and TE (third vs. first tertile: RR 2.02 fi retrieved from studies that satis ed all selection criteria. [95% CI 1.50–2.70], P trend ,0.001) were associated with Data extraction, quality assessment, and data synthesis an increased risk of T2D (Table 2). Further adjustments and analysis are described in detail in Supplementary for BMI, insulin, and glucose attenuated but did not abol- Appendix 5. ish the association between SHBG (third vs. first tertile: RR 0.56 [95% CI 0.41–0.77], P trend ,0.001) or TE and RESULTS incident T2D (third vs. first tertile: RR 1.39 [95% CI Table 1 summarizes the baseline characteristics of the par- 1.004–1.93], P trend = 0.07). On the other hand, adjust- ticipants included in the analysis. Of the 3,117 postmeno- ment for obesity and glycemic traits weakened the associ- pausal women without diabetes at baseline, 384 women ations of BT with T2D such that they were no longer developed diabetes over a median follow-up of 11.1 years. statistically significant (Table 2). Controlling for metabolic 580 Steroid Sex Hormones, SHBG, and T2D Diabetes Volume 66, March 2017

Table 2—Associations of SHBG, TT, FAI, and TE with the risk of T2D in postmenopausal women, the Rotterdam Study (n =3,117) SHBG Tertile 1 Tertile 2 Tertile 3 Continuous P trend Case subjects 191 119 74 Model 1, HR (95% CI) 1.00 0.56 (0.45–0.71) 0.33 (0.25–0.43) 0.37 (0.30–0.46) <0.001 Model 2, HR (95% CI) 1.00 0.82 (0.64–1.04) 0.56 (0.41–0.77) 0.63 (0.49–0.81) <0.001 Model 3, HR (95% CI) 1.00 0.82 (0.64–1.05) 0.56 (0.40–0.79) 0.66 (0.51–0.86) 0.001 TT Tertile 1 Tertile 2 Tertile 3 Continuous P trend Case subjects 126 139 119 Model 1, HR (95% CI) 1.00 1.04 (0.82–1.32) 0.90 (0.69–1.16) 0.91 (0.75–1.10) 0.40 Model 2, HR (95% CI) 1.00 0.94 (0.74–1.20) 0.82 (0.63–1.07) 0.87 (0.71–1.07) 0.15 Model 3, HR (95% CI) 1.00 0.96 (0.75–1.24) 0.88 (0.67–1.16) 0.93 (0.76–1.14) 0.36 FAI Tertile 1 Tertile 2 Tertile 3 Continuous P trend Case subjects 87 124 173 Model 1, HR (95% CI) 1.00 1.39 (1.05–1.82) 2.01 (1.55–2.60) 1.54 (1.32–1.79) <0.001 Model 2, HR (95% CI) 1.00 1.06 (0.79–1.42) 1.17 (0.87–1.57) 1.13 (0.94–1.36) 0.28 Model 3, HR (95% CI) 1.00 1.05 (0.78–1.42) 1.15 (0.85–1.54) 1.10 (0.92–1.32) 0.34 TE Tertile 1 Tertile 2 Tertile 3 Continuous P trend Case subjects 109 132 143 Model 1, HR (95% CI) 1.00 1.28 (0.99–1.65) 2.02 (1.50–2.70) 1.003 (1.001–1.004) <0.001 Model 2, HR (95% CI) 1.00 1.00 (0.74–1.34) 1.39 (1.004–1.93) 1.003 (1.001–1.004) 0.07 Model 3, HR (95% CI) 1.00 1.05 (0.78–1.41) 1.42 (1.01–2.00) 1.002 (1.001–1.004) 0.055 Significant association (P , 0.05) indicated by boldface type. Model 1: adjusted for age, cohort, fasting status; model 2: model 1 + insulin, glucose, and BMI; model 3: model 2 + alcohol intake, smoking status, coronary heart disease, serum total cholesterol, statin use, systolic blood pressure, treatment for hypertension, hormone replacement therapy, age of menopause, CRP, and sex hormones for each other. risk factors, lifestyle factors, inflammatory markers, and prev- Table 1). Also, no effect modification by BMI was found alent coronary heart disease did not materially affect these for TT and BT (Supplementary Table 1). associations (Table 2). No association was found between TT and incident T2D in any of the models (Table 2). Systematic Review and Meta-analysis Because associations of continuous hormone variables fi Literature Search, Characteristics, and Quality of with T2D in model 1 appeared linear, RRs strati ed Eligible Studies and sensitivity analyses were expressed per unit log or The initial search identified 3,209 potentially relevant unit increase in hormone biomarkers. In the sensitivity citations. After screening and detailed assessment, 15 ar- analyses, substituting BMI with waist circumference as a ticles based on 12 unique studies were included (Supple- measure of adiposity, substituting total cholesterol for mentary Fig. 1 and Supplementary Appendix 5). Therefore, other blood , adjusting further for serum TSH, we meta-analyzed estimates from 13 studies (including the physical activity, number of pregnancies of at least current study) involving a total of 14,902 pre- and post- fi 6 months, or menopause type, and excluding the rst menopausal women with 1,912 incident T2D cases, report- 3 years of follow-up did not affect any of the associations ing on the association between sex hormones and T2D risk. fi (Supplementary Table 1). Also, in the strati cation anal- Detailed characteristics of these studies and quality assess- fi ysis, no signi cant interactions were found for SHBG ment have been summarized in Supplementary Table 2. All and TE with BMI or years since menopause (Supplemen- studies were medium to high quality except one. tary Table 1). Significant interaction terms were found for TT (P interaction = 0.019) and FAI (P interaction = Sex Hormones and T2D in Pooled Analysis 0.03) with years since menopause. However, no associa- The meta-analyses for BT, TE, and BE are based only on tion was found between these hormones and T2D after studies examining postmenopausal women; the meta-analysis stratification for time since menopause (Supplementary for TT is based on four studies including postmenopausal diabetes.diabetesjournals.org Muka and Associates 581 women and one study including pre- and postmenopausal the mid SHBG tertile and 2.5% in the high SHBG tertile. women, whereas the findings for SHBG derive from studies Okubo et al. (23) reported lower levels of SHBG in T2D including premenopausal women (n = 2), postmenopausal converters (59.7 6 8.4 nmol/L) than nonconverters women (n =4),andcombined(n = 3). The pooled RRs for (69.5 6 2.5 nmol/L) during 3 years of follow-up, but that T2D adjusted for several metabolic risk factors comparing was not significantly different after adjusting for age, BMI, third versus first tertile of SHBG, TT, BT, TE, and BE were and waist-to-hip ratio. Sex steroids and SHBG were not 0.44 (95% CI 0.30–0.66, I2 = 77.9%, P , 0.001), 1.32 (95% associated with diabetes outcomes in pre- and postmeno- CI 0.79–2.21, I2 =53.8%,P = 0.07), 1.75 (95% CI 0.92– pausal women in the study of Mather et al. (24). 3.33, I2 = 80.7%, P = 0.001), 1.99 (95% CI 1.21–3.27, I2 = 55.1%, P = 0.06), and 3.58 (95% CI 0.86–14.84, I2 =81.0%, Publication Bias P = 0.02) (Figs. 1–3). There was evidence of between-study The appearance of funnel plots was asymmetrical for the heterogeneity for all these analyses, with the possible ex- analysis on SHBG and T2D, and Egger test results were fi P ception of the meta-analysis on the association between TE signi cant ( = 0.014) (Supplementary Fig. 2). This sug- and the risk of T2D (Figs. 1–3). For SHBG, heterogeneity gested that publication bias may be present. After exclu- was not explained by any of the study-level characteristics sion of the four studies that included 50 or fewer case fi fi assessed, such as menopause status, location, and number subjects with T2D, ndings were not statistically signi - P of participants (Supplementary Table 4). For TT, the level cant (Egger test, = 0.93, data not shown). No evidence of heterogeneity was largely explained by location (Supple- of publication bias was observed for the analysis of TT or mentary Table 4). Five studies could not be included in the TE and T2D (Supplementary Fig. 2). meta-analyses. Soriguer found that in pre- and postmeno- pausal women, per one unit log increase in SHBG, TT, and DISCUSSION BT, the corresponding RRs were 0.23 (95% CI 0.1–0.53), In this large population–based study of postmenopausal 1.04 (0.59–1.83), and 1.12 (0.59–2.13), respectively (20). women free of T2D at baseline, we showed that that lower Boyd-Woschinko et al. (21) reported a fivefold increase in T2D levels of SHBG and higher levels of TE were associated incidence in the lowest quintile of SHBG. Similarly, Lindstedt with the risk of T2D, independent of established risk et al. (22) found that among patients in the low SHBG factors for T2D, including BMI, glucose, and insulin. In tertile, 18% converted to T2D as compared with 5% in contrast, the association between T and the risk of T2D

Figure 1—RRs of T2D comparing top vs. bottom thirds of baseline plasma SHBG. The summary estimates presented were calculated using random-effects models (D+L) and fixed effects (I-V). The sizes of the data markers are proportional to the inverse of the variance of the odds ratio; the CIs are represented by the bars. X2 = 36.2. I2 = 77.9%. P < 0.001. 582 Steroid Sex Hormones, SHBG, and T2D Diabetes Volume 66, March 2017

Figure 2—RRs of T2D comparing top vs. bottom thirds of baseline plasma TT and FT levels. The summary estimates presented were calculated using random-effects models (D+L) and fixed effects (I-V). The sizes of the data markers are proportional to the inverse of the variance of the odds ratio; the CIs are represented by the bars. A: X2 = 8.6. I2 = 53.8%. P = 0.07. B: X2 = 15.5. I2 = 80.7%. P = 0.001. diabetes.diabetesjournals.org Muka and Associates 583

Figure 3—RRs of T2D comparing top vs. bottom thirds of baseline plasma TE and free E levels. The summary estimates presented were calculated using random-effects models (D+L) and fixed effects (I-V). The sizes of the data markers are proportional to the inverse of the variance of the odds ratio; the CIs are represented by the bars. A: X2 = 8.91. I2 = 55.1%. P = 0.06. B: X2 = 5.26. I2 = 81.0%. P = 0.02. 584 Steroid Sex Hormones, SHBG, and T2D Diabetes Volume 66, March 2017 was explained by BMI, glucose, and insulin. Pooled results randomized control trials of oral therapy, which from the systematic meta-analysis of 13 studies reinforce showed a lower risk of T2D among postmenopausal the validity and generalizability of our findings, suggesting women assigned to estrogen treatment (35–37). However, that SHBG and TE are robust risk markers of T2D in due to the observational design, our study does not pro- women. vide causality. Mendelian randomization experiments are Unlike the previous meta-analysis by Ding et al. (25), warranted to investigate the potential causal implications which was based mainly on studies with cross-sectional de- of E on T2D. Exogenous estrogen may have different sign and examined only mean differences between case sub- physiological effects depending on type, route, duration, jects with T2D and control subjects without T2D, our current and dose of estrogen therapy (38–41). For example, op- pooled analysis is based on the findings from 13 prospective posing effects of oral estrogen on fasting glucose versus studies (only 2 studies included in the previous review were glucose tolerance have been reported (38,39). Also, in a eligible), including 14,902 participants with 1,912 case sub- randomized trial of postmenopausal women, oral estro- jectswithT2D.Therefore,ourmeta-analysis provides a more gen elevated CRP levels up to 12 months of treatment but detailed assessment of the nature and magnitude of the not transdermal E (40). Moreover, a bimodal relationship association between sex hormones and T2D in women. of estrogen dose may exist. In a clinical trial of postmen- SHBG levels have been associated with metabolic opausal women, a lower dose of estrogen therapy increased syndrome, glucose, and insulin levels, established risk insulin sensitivity whereas a higher dose had the opposite factors for T2D (7,8,26). Also, women with polycystic effect (42). ovary syndrome, a condition of anovulation and hyper- In postmenopausal women, endogenous E may be androgenism, are at increased risk of T2D, and levels of associated with diabetes risk through its relation to SHBG are decreased in these women (27). The complex glucose, insulin, obesity, and inflammation. Indeed, pre- biological mechanisms that explain the association be- vious cross-sectional studies have linked both BE and TE tween circulating SHBG levels and the risk for T2D are with higher glucose and insulin resistance levels in post- not fully understood. Classically, the primary function of menopausal women, independent of obesity (6,9,43). Also, SHBG was thought to be the binding of circulating hor- whereas studies suggestthatEregulatesbody mones in order to regulate free sex hormone bioavailabil- composition, many studies in postmenopausal women ity to target tissues. Therefore, it has been hypothesized have failed to show a consistent beneficial role of E in that the relation between SHBG and T2D may result from weight loss and in the distribution of body fat (44). How- the indirect influence of alterations in SHBG on sex hor- ever, in our study, the association between TE and T2D, mone bioavailability. However, in our study, the associa- although attenuated, remained significant after adjust- tion between SHBG and T2D risk remains significant after ment for plasma levels of glucose and insulin, BMI, and adjustment for TT, BE, and TE, implicating SHBG levels as CRP, suggesting that E may play a direct role in the a risk factor for T2D independent of serum androgen pathophysiology of T2D in postmenopausal women. Fur- levels. Additional evidence in support of an independent thermore, additional adjustment for TT did not affect effect of SHBG on T2D comes from recent studies that this association, suggesting that E may be more than have found several polymorphisms in the SHBG to asso- just a marker of increased conversion. Ex- ciate with insulin resistance and T2D, suggesting that plicit mechanisms of estrogen in relation to T2D require altered SHBG physiology may be a primary defect in the further study. pathogenesis of disease (28–31). Furthermore, a growing Our study showed no association between TT and the body of evidence shows that SHBG may directly mediate risk of T2D, whereas a suggestive positive association was cell-surface signaling, cellular delivery, and biological ac- observed between BT and T2D. The lack of association tion of sex hormones via activation of a specific plasma between free testosterone (FT) and the risk of T2D in our receptor (32–34). At the target tissue level, the fraction study might be due to the lack of a direct measure of BT of SHBG that is not bound to sex steroid has the ability in the blood, which could have biased our results toward to bind plasma membrane high-affinity receptors (RSHBG) the null. These findings are in line with previous studies (32). Sex steroids of variable biological potency can activate reporting higher levels of insulin resistance with in- the anchored SHBG-RSHBG complex, and the activated creasing levels of BT in postmenopausal women, whereas complex can have either an agonist or antagonist effect. no association has been observed between TT and insulin For example, SHBG-RSHBG complex can have direct cellu- resistance (6,14). Similarly, BT has been related to in- lar antagonistic properties against estrogen; SHBG may creased odds of having impaired fasting glucose (14). interact with cellular estrogen receptors, which can trigger The strengths of our study include its prospective a biological antiestrogenic response (32). Specificdown- design, the long follow-up, and adequate adjustments for stream effects of the SHBG-receptor complex merit fur- a broad range of possible confounders. We also performed ther investigation since they may help to clarify the several sensitivity analyses, such as excluding the first underlying mechanisms linking SHBG to T2D. 3 years of follow-up to avoid potential bias of undiagnosed Our result for a positive relation between E and T2D disease at baseline. Furthermore, our study included, in is in contrast with the results from previous large addition to an analysis of primary data, a systematic review diabetes.diabetesjournals.org Muka and Associates 585 of all available published prospective cohorts, which is the approval of the manuscript. The funder/sponsor did not have the ability to veto first-ever quantitative synthesis of these associations thus the publication of study results. far in women. Also, most of the studies included in our Duality of Interest. T.M., L.J., and O.H.F. work at ErasmusAGE, a center meta-analysis adjusted for potential confounding. How- for aging research across the life course funded by Nestlé Nutrition (Nestec Ltd.), Metagenics Inc., and AXA. T.M. and L.J. reported receiving research support from ever, there are several limitations that need to be taken Metagenics Inc. J.N. has been financially supported by Erasmus Mundus Western into account. First, we did not have measures of BE in Balkans (ERAWEB), a project funded by the European Commission. M.K. is the Rotterdam Study, whichcouldhavestrengthened supported by the AXA Research Fund. O.H.F. reported receiving grants or re- our results. Also, TE was measured using an immuno- search support from Metagenics Inc. No other potential conflicts of interest assay with a detection limit of 18.35 pmol/L, which is relevant to this article were reported. considered suboptimal, particularly in postmenopausal These funding sources had no role in the design or conduct of the study; women. However, the observed effect remained the collection, management, analysis, or interpretation of the data; or preparation, same while analyzing TE continuously and categorically. review, or approval of the manuscript. Second, FT levels were not measured directly in the Author Contributions. T.M. conceived and designed the study, ran the blood and therefore have to be interpreted with caution. analysis, screened the title and abstract, obtained the full text, determined the Nevertheless, FT levels in this study were derived from eligibility of articles, participated in data extraction, participated in data synthesis, analysis, and interpretation, drafted the final manuscript, and contributed to the the ratio of T to SHBG, which is considered a precise critical revision of the manuscript and approved the final version. J.N. screened proxy for BT (45). Third, we observed a moderate to high the title and abstract, obtained the full text, determined the eligibility of articles, level of heterogeneity across the included studies. Dif- participated in data extraction, and contributed to the critical revision of the man- ferent assays (Supplementary Table 3) used to assess the uscript and approved the final version. L.J., C.M., A.H., A.D., M.K., and J.S.E.L. levels of sex hormones and SHBG contributed to the contributed to the critical revision of the manuscript and approved the final version. observed heterogeneity. However, since the number of W.M.B. designed and executed the search strategies and contributed to the critical available studies included in each meta-analysis was gen- revision of the manuscript and approved the final version. O.H.F. conceived and erally small, it precluded our ability to conduct subgroup designed the study, drafted the final manuscript, and contributed to the critical analyses involving various study-level characteristics revision of the manuscript and approved the final version. T.M. is the guarantor of (such as age). Fourth, there was evidence of publication this work and, as such, had full access to all the data in the study and takes bias for the association between SHBG and the risk of responsibility for the integrity of the data and the accuracy of the data analysis. T2D, so it is possible that our results constitute an over- References estimation of the performance of the test. However, 1. Atsma F, Bartelink ML, Grobbee DE, van der Schouw YT. Postmenopausal when we excluded small studies, differences were not fi status and early menopause as independent risk factors for cardiovascular statistically signi cant, and therefore, the effect of pub- disease: a meta-analysis. Menopause 2006;13:265–279 lication bias may be only minor. Fifth, except for SHBG, 2. Gambacciani M, Ciaponi M, Cappagli B, et al. Body weight, body fat dis- fi the other ndings come from studies conducted mainly tribution, and hormonal replacement therapy in early postmenopausal women. in postmenopausal women, and thus, these results can- J Clin Endocrinol Metab 1997;82:414–417 not be extended to pre- or perimenopausal women. Fi- 3. Liu Y, Ding J, Bush TL, et al. Relative androgen excess and increased nally, contrary to the results of random-effects models, cardiovascular risk after menopause: a hypothesized relation. Am J Epidemiol the fixed-effects models showed a significant association 2001;154:489–494 of both BE and BT with the risk of T2D. The differences 4. Brand JS, van der Schouw YT, Onland-Moret NC, et al. Age at menopause, in random- versus fixed-effects models might be explained reproductive life span, and type 2 diabetes risk: results from the EPIC-InterAct – by the substantial heterogeneity observed between studies study. Diabetes Care 2013;36:1012 1019 5. Gambineri A, Patton L, Altieri P, et al. Polycystic ovary syndrome is a risk (for example, in the association of BT), which could be factor for type 2 diabetes: results from a long-term prospective study. Diabetes better captured under the random-effects model (46). For 2012;61:2369–2374 BE, the small size of the studies might undermine the 6. Kalish GM, Barrett-Connor E, Laughlin GA, Gulanski BI; Postmenopausal fi precision of the estimate under a xed-effects model. How- Estrogen/Progestin Intervention Trial. Association of endogenous sex hormones ever, in light of these observations, the overall results of and insulin resistance among postmenopausal women: results from the Post- this study should be interpreted with caution. menopausal Estrogen/Progestin Intervention Trial. J Clin Endocrinol Metab 2003; In conclusion, lower levels of SHBG and higher levels 88:1646–1652 of TE are independently associated with risk of T2D in 7. Brand JS, van der Schouw YT. Testosterone, SHBG and cardiovascular postmenopausal women. Further studies are needed to health in postmenopausal women. Int J Impot Res 2010;22:91–104 establish hormone thresholds at which diabetes risk is 8. Brand JS, van der Tweel I, Grobbee DE, Emmelot-Vonk MH, van der Schouw increased, because this may aid in identifying high-risk YT. Testosterone, sex hormone-binding globulin and the metabolic syndrome: a postmenopausal women in the clinical setting. systematic review and meta-analysis of observational studies. Int J Epidemiol 2011;40:189–207 9. Golden SH, Dobs AS, Vaidya D, et al. Endogenous sex hormones and Acknowledgments. The authors thank Dr. Wanes Kazanjian (Erasmus glucose tolerance status in postmenopausal women. J Clin Endocrinol Metab MC) for reviewing the abstract. 2007;92:1289–1295 Funding. This study was sponsored and funded by Metagenics Inc. 10. Ding EL, Song Y, Manson JE, Rifai N, Buring JE, Liu S. Plasma sex steroid Metagenics Inc. had no role in the design or conduct of the study; collection, hormones and risk of developing type 2 diabetes in women: a prospective study. management, analysis, or interpretation of the data; or preparation, review, or Diabetologia 2007;50:2076–2084 586 Steroid Sex Hormones, SHBG, and T2D Diabetes Volume 66, March 2017

11. Oh JY, Barrett-Connor E, Wedick NM, Wingard DL; Rancho Bernardo Study. 30. Zhao JL, Chen ZJ, Zhao YR, et al. Study on the (TAAAA)n repeat poly- Endogenous sex hormones and the development of type 2 diabetes in older men morphism in sex hormone-binding globulin gene and the SHBG serum levels in and women: the Rancho Bernardo study. Diabetes Care 2002;25:55–60 putative association with the glucose metabolic status of Chinese patients suf- 12. Kalyani RR, Franco M, Dobs AS, et al. The association of endogenous sex fering from polycystic ovarian syndrome in Shandong province. Zhonghua Yi Xue hormones, adiposity, and insulin resistance with incident diabetes in post- Yi Chuan Xue Za Zhi 2005;22:644–647 [in Chinese] menopausal women. J Clin Endocrinol Metab 2009;94:4127–4135 31. Perry JR, Weedon MN, Langenberg C, et al.; MAGIC. Genetic evidence that 13. Hofman A, Darwish Murad S, van Duijn CM, et al. The Rotterdam Study: raised sex hormone binding globulin (SHBG) levels reduce the risk of type 2 di- – 2014 objectives and design update. Eur J Epidemiol 2013;28:889 926 abetes. Hum Mol Genet 2010;19:535–544 14. Leening MJG, Kavousi M, Heeringa J, et al. Methods of data collection and 32. Rosner W, Hryb DJ, Khan MS, Nakhla AM, Romas NA. Sex hormone-binding definitions of cardiac outcomes in the Rotterdam Study. Eur J Epidemiol 2012;27: globulin mediates at the plasma membrane. 173–185 J Steroid Biochem Mol Biol 1999;69:481–485 15. World Health Organization. Definition and Diagnosis of Diabetes Mellitus and 33. Catalano MG, Frairia R, Boccuzzi G, Fortunati N. Sex hormone-binding Intermediate Hyperglycemia: Report of a WHO/IDF Consultation. Geneva, World globulin antagonizes the anti-apoptotic effect of estradiol in breast cancer cells. Health Org., 2006, p. 1–50 Mol Cell Endocrinol 2005;230:31–37 16. Rosner W, Auchus RJ, Azziz R, Sluss PM, Raff H. Position statement: utility, 34. Fortunati N, Catalano MG, Boccuzzi G, Frairia R. Sex hormone-binding globulin limitations, and pitfalls in measuring testosterone: an Endocrine Society position (SHBG), estradiol and breast cancer. Mol Cell Endocrinol 2010;316:86–92 statement. J Clin Endocrinol Metab 2007;92:405–413 35. Kanaya AM, Herrington D, Vittinghoff E, et al.; Heart and Estrogen/progestin 17. Rubin BD. Multiple imputation for nonresponse in surveys. Invest Radiol 1987 Replacement Study. Glycemic effects of postmenopausal hormone therapy: the 18. Rubin RB. Multiple Imputation for Nonresponse in Surveys. New York, John Heart and Estrogen/Progestin Replacement Study. A randomized, double-blind, – Wiley & Sons, 1987 placebo-controlled trial. Ann Intern Med 2003;138:1 9 ’ 19. Stroup DF, Berlin JA, Morton SC, et al. Meta-analysis of observational 36. Margolis KL, Bonds DE, Rodabough RJ, et al.; Women s Health Initiative studies in epidemiology: a proposal for reporting. Meta-analysis Of Observational Investigators. Effect of oestrogen plus progestin on the incidence of diabetes in Studies in Epidemiology (MOOSE) group. JAMA 2000;283:2008–2012 postmenopausal women: results from the Women’s Health Initiative Hormone 20. Soriguer F, Rubio-Martin E, Fernandez D, Valdes S, et al. Testosterone, Trial. Diabetologia 2004;47:1175–1187 SHBG and risk of type 2 diabetes in the second evaluation of the Pizarra cohort 37. Bonds DE, Lasser N, Qi L, et al. The effect of conjugated equine oestrogen study. Eur J Invest 2012;42:79–85 on diabetes incidence: the Women’s Health Initiative randomised trial. Dia- 21. Boyd-Woschinko G, Kushner H, Falkner B. Androgen excess is associated betologia 2006;49:459–468 with insulin resistance and the development of diabetes in African American 38. Espeland MA, Hogan PE, Fineberg SE, et al. Effect of postmenopausal women. J Cardiometab Syndr 2007;2:254–259 hormone therapy on glucose and insulin concentrations. PEPI Investigators. 22. Lindstedt G, Lundberg PA, Lapidus L, Lundgren H, Bengtsson C, Bjorntorp P. Postmenopausal Estrogen/Progestin Interventions. Diabetes Care 1998;21:1589– Low sex-hormone-binding globulin concentration as independent risk factor for 1595 development of NIDDM. 12-yr follow-up of population study of women in 39. Zhang Y, Howard BV, Cowan LD, et al. The effect of estrogen use on levels – Gothenburg, Sweden. Diabetes 1991;40:123 128 of glucose and insulin and the risk of type 2 diabetes in American Indian post- 23. Okubo M, Tokui M, Egusa G, Yamakido M. Association of sex hormone- menopausal women: the Strong Heart Study. Diabetes Care 2002;25:500–504 binding globulin and insulin resistance among Japanese-American subjects. 40. Strandberg TE, Ylikorkala O, Tikkanen MJ. Differing effects of oral and – Diabetes Res Clin Pract 2000;47:71 75 transdermal hormone replacement therapy on cardiovascular risk factors in 24. Mather KJKC, Kim C, Christophi CA, et al.; Diabetes Prevention Program. healthy postmenopausal women. Am J Cardiol 2003;92:212–214 Steroid sex hormones, sex hormone-binding globulin, and diabetes incidence 41. Decensi A, Omodei U, Robertson C, et al. Effect of transdermal estradiol and in the Diabetes Prevention Program. J Clin Endocrinol Metab 2015;100:3778– oral conjugated estrogen on C-reactive protein in -placebo trial in healthy 3786 women. Circulation 2002;106:1224–1228 25. Ding EL, Song Y, Malik VS, Liu S. Sex differences of endogenous sex 42. Lindheim SR, Presser SC, Ditkoff EC, Vijod MA, Stanczyk FZ, Lobo RA. A hormones and risk of type 2 diabetes: a systematic review and meta-analysis. possible bimodal effect of estrogen on insulin sensitivity in postmenopausal women JAMA 2006;295:1288–1299 and the attenuating effect of added progestin. Fertil Steril 1993;60:664–667 26. Weinberg ME, Manson JE, Buring JE, et al. Low sex hormone-binding 43. Goodman-Gruen D, Barrett-Connor E. Sex differences in the association of globulin is associated with the metabolic syndrome in postmenopausal women. Metabolism 2006;55:1473–1480 endogenous sex hormone levels and glucose tolerance status in older men and – 27. Nestler JE, Powers LP, Matt DW, et al. A direct effect of hyperinsulinemia on women. Diabetes Care 2000;23:912 918 serum sex hormone-binding globulin levels in obese women with the polycystic 44. Davis SR, Castelo-Branco C, Chedraui P, et al.; Writing Group of the In- ovary syndrome. J Clin Endocrinol Metab 1991;72:83–89 ternational Menopause Society for World Menopause Day 2012. Understanding 28. Ding EL, Song Y, Manson JE, et al. Sex hormone-binding globulin and risk weight gain at menopause. Climacteric 2012;15:419–429 of type 2 diabetes in women and men. N Engl J Med 2009;361:1152–1163 45. Vermeulen A, Verdonck L, Kaufman JM. A critical evaluation of simple 29. Wang Q, Kangas AJ, Soininen P, et al. Sex hormone-binding globulin methods for the estimation of free testosterone in serum. J Clin Endocrinol Metab associations with circulating lipids and metabolites and the risk for type 2 di- 1999;84:3666–3672 abetes: observational and causal effect estimates. Int J Epidemiol 2015;44: 46. Borenstein M, Hedges LV, Higgins JPT, Rothstein HR. Introduction to Meta- 623–637 Analysis. West Sussex, U.K., Wiley, 2009