Sex hormone-binding globulin, testosterone and type 2 diabetes risk in middle-aged African women: exploring the impact of HIV and menopause

Sex hormo ne-binding globuli n, androgens , to tal t estost eron e, fre e tes toste rone , oes trogen , 48 HIV, postmeno pausal, p remeno pausal, A frica 49 50 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 29, 2024. ; https://doi.org/10.1101/2024.12.25.24319619doi: medRxiv preprint 4

51

Sex hormone-binding globuli n (SHBG) and testos te rone a re differen ti ally associated with 52 type 2 diabe tes (T2D) risk. We investigat ed whethe r th ese associa tions differ by HIV and menopausal 53 status in Black Sou th African women livin g with (WLWH) and without HIV (WLWOH). 54 Design: Cross-sectional obse rvation al. 55

Eighty one prem enopausa l (57 WLWOH, 24 WLWH) and 280 postmen o pausal (236 WLWOH, 56 44 WLWH) women from the Middle-Age d Soweto Cohor t (MASC) completed th e following measures: 57 circulating SHBG and sex ho rmones, bod y composition (dual energy x-ray ab sor pt iometry), oral glucose 58 toler ance t est t o estima te insulin sensi tivity (Matsuda inde x), secre tion (insulinog enic index , IG I) and 59 clearance , and be ta-cell function (disposi tion inde x, DI). Dysglycaemia was defined as eithe r impaired 60 fasting or postpr andial glucose o r T2D. 61

SHB G was higher and tot al and free t estos ter one wer e lower in pos tmen opausal WLWH t han 62 WLWOH (all p< 0.023). I rresp ective of HI V serostat us, SHBG was posi tively associated with M atsuda 63 index, insulin cle aranc e and DI and invers ely with HOMA-IR (all p<0 .011). The asso ciation be tween SH BG 64 and Matsud a inde x was stronger in p rem enopausal than pos tmenop ausal women (p=0.043 for 65 inter action). Fr ee t estos ter one (and no t t otal t estos ter one) was only negatively as sociated with bas al 66 insulin clearanc e (p=0.021), and positivel y associated with HO MA-IR in pr emenop ausal and not p ost-67 menopausal women (p=0.015 for in ter ac tion). 68

We show for the first time t hat midlife African WLWH have high er S HBG and lowe r to tal 69 and free t estos ter one th an WLW OH, which correspond ed to their high er be ta-cell function, suggesting a 70 putative p rot ective effect of SHB G on T2D risk in WLWH. 71 72 73 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 29, 2024. ; https://doi.org/10.1101/2024.12.25.24319619doi: medRxiv preprint 5 Significance statement: 74 This study in midlife Black African women suggest that high er sex h ormone bin di ng protein (SHB G) and 75 lower free t estos te rone in women living with HIV (WLWH) may be associated wit h reduced risk of type 2 76 diabet es (T2D) compared to women living without HIV. Fur the r, this study pr ovides a puta tive 77 mechanism underlying the low er preval e nce of T2D in WLWH and obesity compar ed to women living 78 with obesity but with out HIV . However, l ongitudinal stu dies ar e requi red to unde r stand th e clinical 79 implications of thes e findings. 80 81 82 83 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 29, 2024. ; https://doi.org/10.1101/2024.12.25.24319619doi: medRxiv preprint 6

84 Type 2 diabetes (T2D) has reached epid e mic proportio ns globally and is associate d with obesity and 85 ageing 1 . Sub-Sahar an Africa (SSA) is the r egion with the high est projec ted r elat ive increase in T2D (129% 86 by 2045) 1 , with South Africa (SA) 2 having the highest number of peopl e living with T2D in SSA 3 . Midlife 87 SA women are dispr opor tiona tely affecte d by T2D 4 , with putative drive rs being HI V, the menopa usal 88 transi tion, and the high pr evalence of obesity 5 6 . Hormonal changes em erge as a k ey factor connecti ng 89 these h ealth chal lenges, with s tudies sug gesting tha t sex ho rmones play a critica l role in th e 90 developmen t and progr ession of T2D 7-11 . 91 92 The menopausal transi tion is charac teris ed by an increase in the and rogen t o E2 ratio, which reflec ts the 93 substanti al decline in o estr ogen (E2) production and a slowe r rat e of decline in a ndrogen biosynth esis 9 . 94 Higher tes tost eron e in women has bee n associated with inc reas ed risk for T2D 7, 11-13 . However, the re is 95 evidence to suggest tha t the associ ation betwee n tes tost eron e and T2D risk may be driven (directly or 96 indirectly) by the effects of sex ho rmone binding globulin (SHBG) 7, 14 . Indeed, rec e nt Mend elian 97 randomisati on studi es showed an indep e ndent associa tion betwee n SHBG a nd T2D 7, 8, 15 . 98 99 Given the impac t of SHBG and androge ns on the risk of T2D, it is important to inve stigate their p oten tial 100 independ ent r oles in th e preval ence of d ysglycaemia among two high-risk groups; women living with 101 HIV (WLWH) and postmenopausal wome n. On aver age, SHB G decr eases over the menopausal p eriod 16 , 102 with SHBG being significantly lower in po stmenopaus al compared to pr emenopa u sal women 17 . Results 103 from the Women ’s Int erag ency HIV study (WIHS) have shown that WLWH have lo wer E2 and tot al 104 testos te rone, but higher SH BG compa red to women living without HIV (WLWOH) 18-21 . However, this 105 study did not e xplor e the r ole of SHB G a nd androgens on insulin dynamics (i.e . insulin sensitivity, insulin 106 response a nd bet a-cell function), which is integral t o the d evelopmen t of T2D 22 . We rec ently showed 107 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 29, 2024. ; https://doi.org/10.1101/2024.12.25.24319619doi: medRxiv preprint 7 that SHB G was associat ed with inciden t T2D, insulin sensitivity and bet a-cell functi on in a cohort of 108 middle-aged Black SA men withou t HIV 23 . This may be because the alr eady el evate d SHBG levels in men 109 living with HIV (MLWH) may have reache d a satura tion poin t, limiting furt her effe cts 23 . Since women 110 have higher SHB G levels tha n men 24 , it is importan t to und erst and how eleva ted S HBG affects insulin 111 dynamics and T2D risk in WLWH. This is especially rel evant for African women w ho presen t with a 112 phenotyp e of low insulin sensitivity and hyperinsulina emia 25 . 113 114 We hypoth esised th at highe r SHBG an d lower to tal and fre e tes tost eron e in SA W LWH before and aft er 115 the menop ause will be associa ted with f avourable insulin dynamics th at may pro tect agains t th e 116 developmen t of T2D. Accordingly, the ai m of the study was to investigat e associa tions betw een 117 androgen ho rmones, SH BG and T2D risk in pre- and post-meno pausal Black SA wo men living with and 118 without HIV. 119 120

121 Design, study population and setting. 122 Recruitm ent a nd da ta collec tion we re c onducted betwe en Janua ry 2017 and Au gust 2018 at the S out h 123 African Medical Res earch Council/Wits Developmental Pat hways for Health Res earch Unit a t th e Chris 124 Hani Baragwana th Hospit al in Soweto , J ohannesbu rg, South Africa . The sta rting sample for this cross-125 sectional s tudy included women (n=50 1) selected from t he Middl e-Aged So weto Cohor t (MASC) 26 . 126 Exclusion crit eria i ncluded using m enop ausal hormon e ther apy (n=7), hormon al contrac eptives (n=30), 127 had a hyster ectomy (n=47), had no HIV d ata (n=1) or blood sampl es (n=1), were p erimenop ausal (n=54), 128 resulting in a final sampl e of 361 pa rtici pants. Th e final sampl e included 81 pre menopausal wome n (57 129 WLWOH, 24 WLWH), and 280 postm eno pausal women (236 WLWOH, 44 WLWH). 130 131 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 29, 2024. ; https://doi.org/10.1101/2024.12.25.24319619doi: medRxiv preprint 8 The study, conducte d in accordance wit h the decla rati on of Helsinki, was approve d by the Human 132 Research E thics Committee (HREC) (Medical) of the University of the Wi twat ersra nd (clearance 133 certificat e No . M160604). Informed writ t en consent was obtaine d from all partici pants. 134 135 Testing procedures 136 Administered questionnaires 137 Inte rviewer-administ er ed qu estionn aire s were cap tur ed on to REDCap (Version 14.6.7 , Vande rbil t 138 University, 2024). Data collected includ e d age, housing density (number of people per room) and asset 139 index (percen tage from a possible 1 2 assets), current med ication us e, smoking status (curren t 140 smoker/non-smoker) and alcohol consu mption (curr ently consum es/does n ot c onsume). M enopaus al 141 status was classified using the d at e of final menstr ual pe riod . Pre-menopa use w as defined as cu rren tly 142 having a regular menstrual cycle and post-menopause was defined as cessation of the menstrual cycle 143 for > 12 months 27 . 144 145 HIV testing and CD4 count 146 Women wit hout a previous HIV posi tive diagnosis complet ed a HIV antibody t est (On e S tep HIV–1/2 , 147 Guangzhou Wondfo Biot ech, China). In WLWH, venous blood was analysed for CD4 count using flow 148 cytometry (Beckman Coulte r, Be rea , CA, USA). The number of years since HIV dia gnosis, number of years 149 on HIV tre atmen t and th e medica tion us ed were r ecorde d. 150 151 Body composition 152 Weight and h eight wer e measu red using standar d techniq ues. Waist circumfer enc e was measured in the 153 mid–axilla ry line at the midpoin t betwe en the lower margin of the last palpabl e rib and the top of th e 154 iliac cres t a t the end of no rmal expi r ation, and hip ci rcumference was me asured as the grea tes t 155 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 29, 2024. ; https://doi.org/10.1101/2024.12.25.24319619doi: medRxiv preprint 9 protrusi on of the but tocks 28 . Subto tal fat mass (FM) and regional fat distribu tio n were m easur ed usin g 156 dual en ergy x-ray abso rpti omet ry (DXA; Hologic Discovery-A (S/N 86254), Be dford, M A, USA, APEX 157 software versi on 13.4 .2:3). Regional fat distribut ion includ ed trunk a nd leg FM r eport ed rela tive (%) t o 158 FM 29 . Abdominal visceral (VAT) and subcutane ous adipose tissue (SAT) were esti mated 30 . Fa t mass inde x 159 (FMI) was calculated as FM (kg) divided by height squared (m 2 ). 160 161 Fasting venous blood and oral glucose tolerance test (OGTT) 162 At ~0800 h and following an overnigh t fast (10–12 h), baseline venous blood was drawn for the 163 dete rminati on of HbA1c, glucose, insulin, C-peptide, SHB G, follicle stimula ting hor mone (FSH), and 164 luteinizing ho rmone (LH), E2 and testos t erone conc ent ratio ns. Participan ts th en completed a s tanda rd 165 120-minute OGTT. Af ter inges tion of glucose (75 g), venous blood samples (~5 mL) were drawn at 30, 60, 166 90 and 120 minutes for the d et erminat io n of glucose, and insulin, and C-peptid e concentr atio ns. 167 Participants with known diabet es (on medication) did not comple te the O GTT. Ho meostasis model 168 assessment of insulin resis tance (HOM A-IR) was calculated 31 . Basal insulin cle aran ce was estimate d 169 using the ra tio of fasting C-peptide t o insulin 32 and peripher al insulin sensitivi ty was estimat ed using the 170 Matsuda Inde x 33 . Insulin resp onse and se cretion we re es timate d using the insulin ogenic index (I GI, 171 (∆Insulin 0-30 /∆Glucose 0-30 )) 34 and C-peptide index (∆C-pep tide 0-30 /∆Glucose 0-30 ) 35 , respectively . 172 Disposition index (DI), which is an estima te of bet a-cell function, was calculat ed (I GI/Ma tsuda inde x) 34 . 173 World He alth O rganiza tion (WHO) crit eri a were for th e classification of glucose to lerance s tatus 36 . The 174 participan ts were classified as n ormal glu cose tole rance (N GT), impaired glucose metabolism (IG M) 175 including those with impair ed fasting glu cose (IFG) and/or impaired glucos e tole r ance (IGT), or T2D 36 . 176 Dysgly caemia was defined as the combin ation of I GM and T2D. 177 178 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 29, 2024. ; https://doi.org/10.1101/2024.12.25.24319619doi: medRxiv preprint 10 Biochemical analysis 179 Plasma glucose was analysed on the Ra ndox RX Daytona Chemistry An alyzer us ing enzymatic method s 180 (Randox Labora tori es Ltd. , Londo n, UK). Serum i nsulin an d C-peptid e was analy sed on the Immulit e® 181 1000 Immunoassay Syst em (Siemens Chemiluminescent He althca re GmbH, Henk estr, G ermany). Hb A1c 182 levels were me asured on whole blo o d samples using the D-10™ Hemoglo bin Analyzer (Bio-R ad 183 Laborat ories, Inc. , CA, USA). S erum foll icle stimula ting hormo ne (FSH), lu teiniz ing hormone (LH) and 184 SHBG we re a nalysed using ch emilumin escent micr opar ticle immuno assays (Ar chitect assays, Abb ott 185 Laborat ories, IL, USA) and albumin was analysed using colorimetric (Bromcresol Gre en) assay (Alinity c, 186 Abbot t Labora tori es, IL, US A). Endogeno us steroid ho rmones (E2 and to tal t estos teron e) were quan tifie d 187 by ultra-high perform ance liquid chr o matography t andem mass spect rome t ry (LCMS) 37, 38 . Free 188 testos te rone was calcula ted 39 . 189 190 Statistical Analysis 191 Statis tical analysis was performe d using STATA version 18 (StatCorp , College Sta ti on, Texas). The 192 Shapiro –Wilk tes t was used to ass ess the distributi on of continuous varia bles. Un adjusted 193 mean6i08±6i08st andard d eviation (SD), media n and (25 th to 75 th perc entil e), or count ( %) are present ed for 194 normally distribu ted, skewe d or cat egori cal data, r espec tively. Withi n menopaus a l groups, differences 195 betwee n WLWOH an d WLWH were asse ssed using unpaired t- tests , two-sample Wilcoxon r ank sum test 196 or Fisher’s e xac t tes t for normally distri buted, skewed a nd catego rical da ta, r espe ctively. Logistic (or 197 multinomial logistic), linea r or quan tile r egressions wer e used to explo re th e over all effects of HIV and 198 menopause in the combined sampl e, incl uding an HIV x menopausa l inte ractio n t erm. Ove rall group 199 differences in sex ho rmones as well as gl ycaemia and insulin param ete rs were adj usted for age . 200 Associations betwee n androg ens and SH BG, and p revalen t dysglycaemia were e x plored using logistic 201 regression , adjusting for age and FM I. Da ta rep ort ed as odds ra tio (OR) (95% confi dence int erval (CI). In 202 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 29, 2024. ; https://doi.org/10.1101/2024.12.25.24319619doi: medRxiv preprint 11 women living without T2D associations between SH BG, and to tal and fre e t estost erone wit h continuo us 203 glycaemic and insulin paramet ers, wer e explo red using quan tile r egression a t the 50 th percen tile, 204 adjusting for age and FMI . Data r epor ted as beta coefficient (β) (95% CI). In all logistic and quantil e 205 regressions we e xplor ed wheth er HIV or menopausal st atus al ter ed th ese re latio n ships by including 206 inter action t erms with se x hormon es in t he models sepa rat ely. Age was included in all models to 207 explo re th e effect of menopaus al sta tus independ ent of chron ological age. Fo r bo th the logis tic and 208 quantile regressi on models, ro bust Z-scores, calculat ed from th e m edian and scal ed median absolu te 209 deviation (MAD), were de rived for SHB G, total an d free t estos ter one and the glycaemic and insulin 210 measures t o facilitat e comparisons of th e magnitude of th e associati ons using a standa rdised measu re. 211 The robust Z-scores rep rese nt th e numb er of MADs that a d ata p oint lies from th e median. The M ADs 212 were scaled by a constan t facto r of 1.4826 so that th e robus t Z-scores are mor e di rectly comparabl e to a 213 tradi tional Z-score, with 1 r obust Z-score indicating a value ~1 SD from the medi an . As missing data was 214 minimal (<10%) and missing at random (no differences be tween grou ps), pairwise dele tion was used 215 when handling missing data. 216 217

218 Participant characteristics 219 Characte ristics of particip ants st ratified by HIV and menopausal sta tus ar e described in Tab le 1. 220 Postmenopausal women we re old er tha n premenop ausal women, and wit hin th e postmeno pausal 221 group, WLWH wer e younger th an WLWO H. Socioeconomic st atus, cha ract erised by housing density and 222 asset inde x, did no t differ by HIV or men opausal group . Postmenopa usal women were less likely to drink 223 alcohol and smoke cigare tt es than p rem enopausal women . Dieta ry intake and p h ysical activity did not 224 differ between H IV and menopa usal gro ups (data not shown). A gre ate r prop orti on of postmenopa usal 225 women repor ted taking anti-hyper tensiv e medication comp ared to prem enopaus al women, while 226 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 29, 2024. ; https://doi.org/10.1101/2024.12.25.24319619doi: medRxiv preprint 12 diabet es and lipid loweri ng medications did not differ by menopause o r HIV. In W LWH, 81% of women 227 were taking an tire trovir al the rapy (ART), with the majori ty (73%) using non-nucleoside revers e 228 transcrip tase inhibito rs (NNRTIs) and this , togeth er with CD4 count, did no t differ b y menopausal group. 229 230 Differences in body composition 231 Total body fatness (BMI o r body fat %) and body fat distribu tion (waist circumfer ence, t runk and leg fat 232 mass, VAT and SAT areas) betwe en th e menopausal and H IV groups are repo rte d in Table 1. The only 233 difference be tween the groups was body fat percen tage which was higher in post menopausal compa red 234 to prem enopausa l women. Wh en excludi ng participan ts with T2D, weight and FMI were significantly 235 lower in WLWH th an WLWOH (p=0 .037 a nd 0.028, respec tively, dat a not shown). 236 237 Differences in SHBG and sex hormones 238 In age-adjusted a nalyses LH and FSH wer e higher and E2 was lower in postm enop ausal compared to 239 premenop ausal women (Table 2). While LH and FSH did not differ by HIV status, E 2 was lower in 240 postmenop ausal WLWH t han WLW OH (HIV x menopaus e inte ractio n; p<0 .001), while SHBG was higher 241 in postmenop ausal WLWH th an WLW OH . In contr ast, tot al and free tes toste ron e were lower in pr e- and 242 postmenop ausal women WLWH compa r ed to WLW OH. Tot al tes tost eron e was lo wer in postmenop ausal 243 compared t o premen opausal women , bu t free t estos ter one and SH BG did no t differ by menopausal 244 status . The findings did not differ when e xcluding particip ants with T2D. 245 246 Differences in glycaemic and insulin parameters 247 Overall, 70.7% of par ticipan ts prese nted with NGT, 14.8% with I GM and 14.5% wit h T2D, and did not 248 differ by HIV or menopausal sta tus (Table 2). The overall preval ence of dysglycaemia was 29.2 (95% CI: 249 24.6-34.3)%, with 32% of WLWOH prese nting with dysglycaemia compared to 20 % of WLWH. 250 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 29, 2024. ; https://doi.org/10.1101/2024.12.25.24319619doi: medRxiv preprint 13 251 Differences in fasting parame ters wer e o nly evident when exclu ding those with T 2D (n=309). In 252 postmenop ausal women withou t T2D, th e IG I, C-peptide in dex a nd DI were highe r in WLWH compared 253 to WLWOH . 254 255 The association between SHBG, and total and free testosterone with glycaemic and insulin parameters 256 In the tot al sample each ~1SD increase in SHBG was associate d with 40% lower od ds of dysgly caemia 257 (Table 3), with the associatio n being significant in WLWOH (OR (95% CI): 0.50 (0.3 5–0.71), p<0.001) but 258 not in WLWH (OR (95%CI): 0.90 (0.62–1.31), p=0.592)(SHBG x HIV inte ractio n p=0 .055; Figure 1). The 259 association betwee n SHBG a nd dysglycaemia did not differ b y menopausal st atus . 260 261 SHBG was positively associat ed with Ma t suda index , DI and basal insulin clea ranc e, and nega tively 262 associated with fas ting glucose, insulin, a nd HOMA-IR , irr espective of HIV sta tus, a ge, and FMI (Table 3). 263 Notably, the associa tion betwe en SHBG and Matsud a inde x differed b y menopau sal status (p=0.043 for 264 SHBG x men opause , Figure 2A), with th e association st ronger in p remen opausal (standa rdised ß (95%CI): 265 0.46 (0.22-0.70) p<0.001) compared to p ostmenopa usal women (standa rdised ß (95%CI): 0.29 (0.16-266 0.43), p<0.001). 267 268 Total tes tost eron e was not associa ted wi th dysglycaemia or any glycaemic or insulin parame ters (Table 269 3). The association be tween fr ee t estos t erone and HOM A-IR was only signficiant in the pr emenopa usal 270 women (standardised ß (95%CI): 0.40 (0. 13-0.67), p=0.005) and not th e postmen opausal women 271 (standardised ß (95%CI):0.05 (-0.02 – 0.12), p=0.200). Free test oste rone was nega tively associated with 272 basal insulin clear ance, ind epend ent of a ge, HIV status , and FMI . 273 274 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 29, 2024. ; https://doi.org/10.1101/2024.12.25.24319619doi: medRxiv preprint 14

275 This is the first study to r epor t sex ho rmones in Black African women at differ ent stages of the 276 menopausal t ransi tion, e xplor e differenc es by HIV serosta tus, and t o investigat e h ow SHBG and 277 testos te rone leve ls rela te t o T2D risk. We showed tha t SHBG was highe r in postm enopausal WLWH 278 compared t o WLWOH . No tably, higher S HBG was associat ed with a T2D-protec tive pheno type, including 279 higher insulin sensitivity, beta-cell functi on, and b asal insulin cle aranc e, as well as lower fasting glucose, 280 insulin, and HOM A-IR, r egardl ess of HIV status . In cont rast , to tal and fre e tes tost e rone wer e lower in 281 both pr e- and postmen opausal WLWH th an WLWOH . Lower free t estos te rone wa s linked to higher basa l 282 insulin clearanc e, and lower H OMA- IR in premenop ausal, bu t not pos t-menopa usal women. These 283 findings support our hypot hesis tha t high er SHBG and lower fre e tes tost eron e in WLWH compared to 284 WLWOH may be associat ed with r educe d risk of T2D. This study also provides a putative mech anism 285 underlying the lowe r preval ence of T2D in WLWH and obesi ty compared t o wome n living with obesity 286 but withou t HIV 40 . However, longit udinal studies are requi red t o unders tand t he c linical implications of 287 the findings, par ticularly as prem enopau sal women transi tion int o post-menop au se. 288 289 Higher SHBG l evels have bee n consisten t ly and causally associated with lowe r risk for T2D in both men 290 and women 7, 8, 12, 14-16, 23, 41, 42 . Accordingly, one may hypothesise tha t the high er SH BG in post-291 menopausal WLWH may confer r educed risk for T2D compared to WLWOH . Inde e d, higher SHB G levels 292 in postmenop ausal WLWH wer e linked t o great er be ta-cell function compa red t o WLWOH. Fur the r, 293 SHBG was also associat ed with lowe r risk of T2D-related trai ts, such as lower fasti ng glucose and insulin, 294 and higher insulin sensi tivity, with th ese associations bei ng independ ent of fat ma ss and unaffected by 295 HIV serosta tus. Simila r alt era tions in SHB G and andr ogens have be en obse rved in middle-aged African 296 men, along with significant associat ions betwee n SHBG , insulin dynamics and incident T2D. However, 297 these associa tions wer e only significant in men without HIV and no t th ose living with HIV 23 . The reasons 298 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 29, 2024. ; https://doi.org/10.1101/2024.12.25.24319619doi: medRxiv preprint 15 for these differ ences ar e unclear , but t he association be tween SH BG and T2D risk is stronger in women, 299 suggesting sexually dimorphic effects 13 . F urthe r, rec ent r esults from th e KORA st u dy report ed higher 300 SHGB levels in women compar ed to me n contribut ed to s ex differenc es in fasting glucose and incident 301 T2D 24 . 302 303 Higher SHBG l evels in postmen opausal WLWH than WLW OH is consisten t with p revious studies 18-20 . 304 Higher SHBG h as previously been link ed t o chronic viral infections including HIV, e specially in women 305 and those with H IV RNA >400 copi es/mL 20 . The regulati on of SHBG is comple x, but proinflammatory 306 cytokines like inte rleukin-1 and tumou r n ecrosis factor alph a reduc e SHBG mR NA expr ession by 307 downregulati ng its main transc ription fac tor, h epat ocyte nucle ar factor 4 (reviewe d previously 43, 44 ). We 308 can only speculate that elevat ed SHB G le vels in WLWH may be a compensato ry mechanism to pro tec t 309 against systemic inflammation associa te d with HIV 20 . SHBG is also regula ted by h epatic de novo 310 lipogenesis, adip onecti n and sex h ormon es 43-45 . Mendelian ran domisatio n studies have shown that 311 higher SHBG l evels are associ ate d with lo wer free t estos ter one, bu t not tot al tes t oster one levels 7 . 312 Indee d, our stu dy shows that higher SH B G in the pos tmenop ausal WLWH was acc ompanied with lowe r 313 free tes tost eron e concen tra tions, which corrobo rat es the findings of WIHS 18 . Whi le E2 is positively 314 associated with SH BG 44 , we showed lowe r E2 levels in postmenopa usal WLWH co mpared to WLWOH, 315 suggesting that H IV and/or ARVs may directly increase SH BG . Since most of th e women were bei ng 316 trea ted with NNRT I’s it is difficult to diffe renti ate the effect of HIV from thos e of A RVs on endocrine 317 function. Fur ther rese arch is need ed to bett er und ersta nd the m echanisms underl ying this endocrine 318 dysregulation in WLW H over th e menop ausal tra nsition . 319 320 None thel ess, our study showed that the association betwee n SHBG a nd preval ent dysglycaemia was 321 only significant in WLWOH. W e postul ate that th e lower tot al and free test oste ron e concentr atio ns in 322 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 29, 2024. ; https://doi.org/10.1101/2024.12.25.24319619doi: medRxiv preprint 16 WLWH may confer reduced risk for T2D, with higher levels of SHBG h aving no additional be nefit . In 323 contras t, in WLWOH who p resen t with hi gher to tal and fre e tes tost eron e levels co mpared to WLWH, 324 higher SHBG may re duce T2D risk by low ering the bi oavailabili ty of testos ter one. This is supported by 325 our findings, which showed that the asso ciation be tween SH BG and insulin s ensiti vity was stronger in 326 premenop ausal women, who have highe r tot al tes tost eron e levels compar ed to p ostmenopa usal 327 women (Figure 2A). A recen t Mend elian randomisati on study showed a causal lin k between t he 328 bioavailability of tes tost eron e and incr ea sed risk of T2D in women 7 . Similar to our study and oth ers 14, 16, 329 21, 42, 46 , the Mendeli an rand omisation s tu dy showed no association be twee n tot al testos te rone an d T2D 330 risk, suggesting that the associa tions be t ween SHBG a nd bioavailabl e tes tost eron e is likely driven by 331 SHBG, ei the r direc tly or in combinatio n with free t estost eron e 7 . The e xact mech anisms by which SHBG 332 may influence T2D risk are not fully understood . However, SH BG has been shown to act as a he pat okine, 333 mediating th e link betwe en int rahep atic lipids, insulin sensitivity and T2D status, with stronge r effects in 334 women than men 47, 48 . Further , SHBG bin ds to GPRC6A, stimula ting insulin rel ease in a dose depend ent 335 manner 49 , providing a pot enti al mechani sm for the observed r ela tionships be twe en SHBG an d bet a-cell 336 function. 337 338 Our study showed th at fre e tes tost eron e was associated with lower basal insulin clearanc e and higher 339 HOMA-IR , with this associa tion only signi ficant in premeno pausal women . This could be due to the 340 lower to tal andr ogen load af ter men opa use. These r esults a re suppor ted by ot he r studies showing tha t 341 higher free test oste rone in pr emeno pau sal women is associated with r educed h e patic insulin clear ance 342 and lower insulin sensi tivity, but no t wit h obesity-rela ted insulin s ecre tion 50 . Ot h er studi es also suggest 343 that the r elati onship be tween fre e tes tos teron e and incide nt T2D is mediated by a diposity and insulin 344 resistanc e 42 . Taken toge the r, our r esults i ndicated that high er SHB G, couple d with lower free 345 testos te rone leve ls may confer reduced r isk for T2D in WLWH than WLWOH. 346 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 29, 2024. ; https://doi.org/10.1101/2024.12.25.24319619doi: medRxiv preprint 17 347 Despite significantly lower E2 in the pos t menopausal women, n o differences in body composition or T2D 348 risk were observed b etwe en th e menopa usal groups, findings contra ry to oth ers 51 . However, o thers 349 have shown that Bl ack African women may not gain as much abdominal adiposi ty across the menop ause 350 transi tion due to both high er abd ominal adiposity and smalle r fluctuati ons in sex steroid h ormones in 351 the years l eading up to men opause 52 . I n our study, th e majority (67.1%) of the women were living with 352 obesity, limiting th e pot enti al for furth er increases in body fat and rela ted chang e s in glycaemic and 353 insulin parame ters . A compar ative study conducted in SSA reveal ed no significant differences in 354 anthro pomet ric and cardiom eta bolic risk factors betwe en pre- and p ost-menop au sal women from South 355 and East Africa who pres ent ed with a hig h prevalence of obesity (32-66%) 5, 53 . In contras t, women from 356 West Africa , with a lower pr evalence of obesity (1-5%), showed distinct variations in cardiomet abolic 357 risk between men opausal s tages 5 . 358 359 Strengths and limitations 360 This study comprehensively pheno types Black African women, including OGTT-de rived measures of 361 glycaemic and insulin paramet ers, as wel l as detaile d charact erisa tion of sex h orm ones measur ed using 362 LCMS. The study is however limited by th e cross-sectional d esign tha t preclud es inferences abou t 363 causality. Fr ee t estos ter one was not me a sured, and we r elied on calcul ate d free t estost eron e 39 . In 364 addition , the r ela tively small sample of premenop ausal women and WLWH limi ts our analysis and 365 interp re tati on of the r esults. Fu rth er, se x hormone measu remen ts in preme nopau sal women were not 366 taken a t a set time during th e menst rual cycle, however we do not bel ieve th at th e phase of menstr ual 367 cycle influenced the and rogen r esults as differences in thes e hormon es by HIV ser ostatus we re 368 consistent a t each me nopausal s tage. 369 370 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 29, 2024. ; https://doi.org/10.1101/2024.12.25.24319619doi: medRxiv preprint 18

371 For the firs t time, we show th at midlife B lack African WLWH have higher SHB G, an d lower to tal and fre e 372 testos te rone conce ntr ations compa red t o WLWOH, which may confer reduc ed ris k for T2D. These 373 findings offer valuable insights into th e complex r elati onships betw een and rogeni c hormones, HIV, 374 menopause , and T2D risk. However, long itudinal stud ies ar e requi red t o unders ta nd the clinical 375 implications of thes e associatio ns. 376 377 Conflict of interest: The authors decla re no conflict of interes t. 378 379 Funding: 380 This study was jointly funded by the S outh African Medical Rese arch Council (SAMRC) via the Sout h 381 African Na tion al Depar tmen t of He alth, t he UK Me dical Res earch Council (via the Newton Fund) and th e 382 GSK Africa Non-Communicable Disease Open Lab (grant projec t number ES/N01 3891/1) and the South 383 African Nati onal Rese arch Founda tion (grant numbe r UID:99108). The AWI-G en Collaborative Cen tre is 384 funded by the Na tional H uman G en ome Rese arch I nstitu te (NH GR I), the Nati onal I nstitu te of 385 Environmental H ealt h Sciences (NIEHS), the Office of AIDS r esea rch (OAR) and t he Na tional Insti tut e of 386 Diabetes and Digestive and Kidn ey Dise ases (NIDDK), of th e Nat ional Inst itu tes of He alth (NIH) und er 387 award numbe r U54H G006938, as p art o f the H3Afric a Consor tium, and by th e Departmen t of Sci ence 388 and Innovati on, Sout h Africa, award nu mber DST/CON 0056/2014. IDS is also s upport ed by the Unit e d 389 Stat es Na tion al Ins titu tes of He alth (N I H)/Fogarty Inte rna tional Cen tr e (FIC) grant N o. D43TW010937 390 (University of Pittsbu rgh HIV Como rbidi ties R esea rch Training Program in Sou th Africa – Pitt-HRTP-SA, 391 PIs: Prof. Jean B. Nachega and Prof. S oraya Seed at) and th e Departm ent of Science and Innova tio n 392 (DSI)/National Res earch Found ation (NR F) Professional Developme nt Pr ogramme. JHG , APL and IDS ar e 393 support ed by the SAMRC. The con ten ts of this publication are sol ely the resp o nsibility of the autho rs 394 and do not r epr esent the official views of the sponsors . 395 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 29, 2024. ; https://doi.org/10.1101/2024.12.25.24319619doi: medRxiv preprint 19 396

397 We a re gr at eful to th e pa rticipan ts of th e Middl e-Aged S oweto Cohort and the r esearch staff who wer e 398 involved in this study. 399 400 Data availability : Data is available from a uthors upon reason able r eques t 401 402 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 29, 2024. ; https://doi.org/10.1101/2024.12.25.24319619doi: medRxiv preprint 20

1. Sa eedi P , P et ersohn I , Sa lpe a P , Malanda B, Karurang a S , Unw in N, C olag iuri S, G ua r ig uata L, Mo tala A A, Ogurtsova K, S haw JE, B r i ght D , Willia ms R & C ommittee IDF DA. G l o bal an d re gion al diabe tes prev al enc e es timat es f or 2019 a nd projec t i on s f or 203 0 and 2 045: Res ults f r om t h e Int erna ti onal Dia bet es Fede rati on D iabetes A tlas , 9( th) editi on. Di ab etes Re s Clin P r act 2019 157 107 843. 2. P ee r N, S t e yn K, L ombard C , Lamb er t E V , V ythiling um B & Lev itt N S . Ris ing di abet e s prevale nce among urban- dw elling blac k S out h Afric ans . PL oS On e 201 2 7 e4 3336. 3. Inte rna tiona l D iabe tes Fed e r a t i on . ID F D iabe te s Atlas . Brus sle s , 20 21. 4. Gr u ndlingh N, Z ewotir TT , Rob erts D J & Manda S. As s es sm ent of pr eval ence and ris k f ac t o r s of dia bet es and pre- diabe t e s i n Sou th Af rica . J H e a lt h Pop ul N ut r 20 22 41 7 . 5. C hik wati RP , Mahyoode en NG , Jaf f N G, R a msay M, Mick les f ield LK, Wade A N , Ag o ngo G, As iki G, C ho ma SS R, Boua P R, Ge orge J A & C rowther N J . C ardi ometabo l ic di sea se r is k f a cto rs in pre- an d pos tmenop au sal women f rom f our sub -S aharan A f ric an c oun tr i e s: A c ross - sec tional study . Ma t u r i t a s 2023 172 6 0- 68 . 6. C hik wati RP , Jaf f N G, Mah yood een NG, Mic kles f ield LK, Ra msa y M, Gomez- O li ve FX , Mohamed SF, C homa SS R, Ge orge J A & C r o wther N J . The a ss oc ia t io n of menop a use with c ardi ometa bol i c dis eas e ris k f ac tor s in wom en liv ing w ith and w ithout HIV in s ub- Saha r a n Af r i ca : Re sul ts f rom the A WI- G e n 1 s t ud y. Ma tu r i t a s 2024 187 108069. 7. Ruth KS , D ay F R, Tyrrell J , Thomps on D J, Wood AR, Maha jan A , Be aumo nt RN , Wit t e mans L, Ma rtin S , Bus ch AS, Er zurumluog lu A M, Hollis B, O' M ara TA, Endometri al C a ncer As soc ia tion C , M cC arthy MI, Lang enb erg C , Eas ton D F, War eha m N J , Bur g es s S , M ur r ay A, Ong KK, Frayli ng TM & P erry JRB. U si ng huma n gene t i cs t o unders t and th e di sea se i mpac ts of tes tos teron e in men and women . Na t M e d 202 0 26 252- 258 . 8. D ing E L, S ong Y, Manso n JE, Hunter D J , L ee C C , Rif ai N , B ur in g JE, Gazian o JM & Li u S. S ex ho r mo ne- binding glo bulin a nd risk of type 2 diabe tes in women and men. N En gl J Me d 2009 361 1152 -11 63. 9. La mbrinoudak i I, P a sc hou SA , A rmeni E & Goulis DG. T he int erpl ay betwe e n diab e te s me llitus and menop ause : cli nical implic ations . Na tur e Revi ews Endo crinol og y 202 2 18 60 8- 622. 10. P ark SK, Harlow SD , Zhe ng H, Karvo nen- Gut i e r r ez C , T hur s ton RC , Rup p ert K , J ans s en I & Ra ndolph JF , Jr. Ass oc iation b etw e en c hange s in oes tradi o l and f ol licle -s timula t i ng ho rmone l evel s during t h e menop a usa l transi t i on and ris k of d iabe tes . D iabet M ed 2017 34 531- 538 . 11. D ing E L, S ong Y, Manso n JE, Rif ai N , Bur i n g JE & Li u S. P l asma se x s teroid hormon e s and r is k o f devel oping type 2 diabe t e s in women : a pr os pectiv e study . D iabe tolo g ia 200 7 50 207 6- 2084. 12. O' R eilly MW, Glisi c M, Ku mar endr an B, S ubramania n A, Manol opo ulos KN , Tah ra ni AA, K ee rt h y D , Muka T, Toul is K A , H a nif W, Thoma s GN, F ranc o OH , A rl t W & N i r a n t ha ra kumar K. Se rum tes tos te rone, s ex ho rmone- bindi ng gl obulin a nd se x - s pecif ic risk of inc ident type 2 dia bet es i n a re trosp e ctiv e pr i mary c are coh o r t . Cl in End oc ri nol (O x f ) 20 19 90 145- 154. 13. D ing E L, S ong Y, Malik V S & Liu S. Se x di ff erenc es of end og enous se x hormon es an d r is k o f t y pe 2 d iabetes : a s ystem atic rev iew a nd me ta- an alys is . JAMA 2 006 295 1288- 129 9. 14. Muka T, Nano J , Ja sp ers L, M e un C , Bram er WM, Hof man A , Deh ghan A, Kavous i M, L ave n JS & Franc o O H . Ass oc iations of St eroid S ex Hormo nes an d Sex H o r mon e -Bi nding Glo bu lin Wit h th e Ris k of T ype 2 Di abetes in Women: A P opul ation- B as ed C oh ort S t u dy and M e t a - an aly sis . D iabe te s 201 7 66 5 77- 586. 15. P erry JR, W ee don M N , Lang enb erg C , J a c ks on AU, L yss enk o V , Spa r s o T, Thorleif s son G , Grall e r t H, F e rr uc ci L, Magg io M, P a olis so G, W alk er M, P alm e r C N , P ayne F, You ng E, Herder C , N a ris u N, Morke n MA, Bo nnyca stle LL , Owen KR, Shie lds B, Knig ht B, B enne tt A, Groves C J, Ru ok onen A , Ja rve lin MR, P ea r s on E , P as coe L, Fe r ran ni ni E, Bo r ns tein S R, S tringham HM, S cott LJ, Kuusi sto J, Nils son P , Nep tin M, Gjes ing AP , P isi nger C , La uritz en T, Sa n dbaek A , Samps on M, Ma gic, Zegg ini E, Lindg r e n C M, Stei nt h o r s d ottir V, Thors tei nsd otti r U, Hans en T, Schw arz P , Illig T, Laak so M, S t ef anss on K, Mor ris AD , Gr oop L , P e de rse n O , Bo ehnk e M , Ba rr os o I , Wareh am NJ, H at tersl e y AT, Mc C arthy M I & Frayl ing TM. G e ne tic ev idenc e tha t ra is ed s ex hor m o ne bindi ng glo bulin ( SHBG) lev els r e duce t he r i sk of type 2 diabe tes . Hu m M o l G ene t 2010 19 535- 54 4. 16. Hedders o n MM, C apra A, Lee C , Habe l L A, Le e J, G old EB, Bad on S E, Mitr o SD & El Khou dary SR. Long itudin al C hang es in Se x Hormon e Bind ing Gl obuli n ( SHBG ) a nd Ris k of In cident D ia bet es: T he Study of Women ' s Health Ac r o ss the Nati on ( SWAN). Di a bet es Care 202 4 47 67 6- 682. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 29, 2024. ; https://doi.org/10.1101/2024.12.25.24319619doi: medRxiv preprint 21 17. Sof tic D , Mesa lic L & Sof tic D. The in fl uence of anthrop ometr ic p arame ters o n the con centr ation of g lobulin that bind s s e x hor m on es dur i ng th e me n opaus al tr ans itio n. Me d Glas ( Ze nica) 2 024 21 344- 348. 18. Karim R , Mack WJ , Kono N, Tie n P C , Anas tos K, La z ar J, Y oung M, Cohe n M , Go l u b E , G r een b l at t R M , K a p l a n RC & Hodis H N. Gonado t rop in and s ex s t eroid le vels in HIV - inf ec te d prem enopa u sal women and thei r as soc iation with s ubcl inica l a the ros cler o s is i n HIV -i nf ected a nd - uninf ec ted wome n in the women's inter age ncy H IV study (WIHS) . J C lin End o c rinol Me ta b 2013 98 E61 0- 618. 19. C oburn S B, D ionne- Od om J, Alc aid e ML, Moran C A, Ra h angda le L, G o lub ET, Mass ad LS, S eidman D, Michel KG, Mink off H, Murphy K, Brown TT, V is v a natha n K, Lau B & Althof f KN. The Ass oc ia tion B etwe e n HIV Sta tus , Es tr a diol, a nd Sex H or m one Bi ndin g Glob ulin A m ong P remeno paus a l Women i n th e Women ' s Int erage ncy HIV Study . J W ome ns Healt h ( La r chm t) 2022 31 183- 19 3. 20. P ena D ias J, P igg ot t D A, S un J, W ehbeh L, Garz a J, Abr aham A, A s tembo rsk i J, Mos eley KF, Ba saria S , V aradha n R & Br o wn TT . SHBG, bone mi neral de ns ity a nd phys ic al f unc t i on amon g injec tion drug us ers with and without H IV and H C V . J Cl i n E n doc r i no l M e t ab 2022 . 21. Abelma n RA , Sc hneid e r MF , C ox C , Mes s erlia n G, C ohen M, Gus taf son D, P lanke y M, S harma A , P ric e J, Gr u nf eld C & Tien P C . As soc iation of And rog en Hor m on es, Se x H o r mon e Binding G l obulin, a nd the Menopa us al Trans iti on wi t h Incide n t Dia betes Me llitus in W omen w ith and with o ut Hiv. J Ac quir I mmu ne De fi c S ynd r 2024. 22. Kahn SE. T he rela tiv e co nt r ibutio n s of in suli n resis tanc e and b e t a - cell dys f unction to the p atho phys iolog y of Ty pe 2 di abetes . D iabe tolo g ia 2003 46 3- 19. 23. Se ipone ID , Mend ham AE, S tor b e ck KH , O es tlun d I, Kuf e C N, C hiko wore T, M a se mo la M, C rowthe r NJ, Keng ne AP , Nor r i s S , Olss on T, Br ow n T, Mic kle sf iel d LK & Goedec ke J H. S HBG, F ree Tes tos ter one, a nd Ty pe 2 D iabetes Ris k i n Middle- ag ed Af rican M e n: A Lo ngitudinal S tudy. J En do cr Soc 2024 8 bvae 129. 24. Rae isi- D ehk or di H, Amiri M, Rat hman n W , Zell er T, Ad ams ki J , Bano A , v an der S chouw YT, T horand B, Muk a T & Nano J. S e x h or mon e- binding globuli n may ex pla in s ex dif fe re nces f or gluc ose homeos tasi s a nd inc idenc e of type 2 diab e tes : the KORA s t udy. E ur J Epide miol 202 4. 25. Goed ec ke JH & Mendh am AE . P athophy s iol ogy of ty pe 2 di abetes in s ub- Sah a r a n Af r ic ans . D iabe tol o gia 2022 . 26. Goed ec ke JH, Nguye n K, Kuf e C , Mas emola M , C hikow ore T, Mend h am AE, Nor ris S A, C rowthe r NJ , Karpe F , Olss on T, Kengne AP & M i ckle sf iel d LK. W a ist c ircumf ere nc e th resh olds pre dicti ng incid ent dys glyc aemi a and type 2 diabe t e s in Blac k Af rican men an d women. D iabe tes O bes Met ab 2022. 27. Harlow SD , Gass M, Hall JE , Lob o R, Mak i P , Reba r RW , Sh erman S , Sl uss P M, de Vi l liers TJ & G roup S C . Executiv e s ummary of the S tages of Repr oduc tive Ag ing Works hop + 1 0: addr es si n g the unf ini shed agen d a of s t a ging r e pr o du ctive aging . J Cli n Endo cr i n ol Met ab 20 12 97 1159- 1168. 28. World He alth O rganis a tion. Wai st c ir c um f erenc e and w ais t-hi p rati o: re por t of a WHO e x per t C ons ulat ation . Genev a , 2011 . 29. Goed ec ke JH, Mic kles f iel d LK, Levitt NS, L ambert EV , Wes t S, Ma arten s G & Da ve J A. E ff e ct of dif f erent antir etr oviral dr ug regime ns on b ody f at di stribut ion of HIV - inf ec t e d Sou th Af rica n wome n. AID S Res Hum Retrovi rus es 2013 29 557- 56 3. 30. Mic kles f ield LK, Goedec ke JH, P un yanity a M, Wils on KE & Kell y TL. D ual- ene r g y X- ray perfo r ms as we ll as cl inica l co mputed tom ography f or the m eas urem ent of v isc eral f at . O bes it y ( S ilve r Spring) 201 2 20 11 09- 1114 . 31. Matt hews D R, Hosk er JP , Rud ens ki AS, N aylo r BA, Tre ache r DF & T ur n er RC . Hom eos t a sis model as ses sm ent : ins ulin res is t a nc e and be ta- c ell f unc tion f rom f as t i ng pla sma gluc ose and ins ulin c onc entra t i o ns i n man. D iabetol o gi a 198 5 28 41 2- 419. 32. P olon sky KS, P ugh W, Jas pan JB, C ohen D M, Karris on T, T ager HS & Ruben s tein AH . C - peptide and i nsuli n s ecreti on. R el atio ns hip betw een pe riphe r a l con cent rati ons of C - peptide and ins uli n and their s ecre tion r a tes in the dog. J Cl in In v est 19 84 74 1821- 1829. 33. Mats uda M & D eFron z o RA . Ins ulin s ens i t i vity i ndice s obtai n ed f rom oral g lucos e toler an ce t es ting: co mparison with th e e uglyc emic ins ulin c lamp. Di abe tes Care 199 9 22 14 62- 1470. 34. Ut zs chnei d er KM, P rigeon RL, F aul enb ac h MV , T ong J, C a r r D B, Boy ko EJ, Leone tti D L , McNeely MJ, Fujim ot o WY & Kahn S E. Oral dis pos itio n inde x p re dic ts the devel op men t of f utur e diab e tes abo ve a nd beyo nd f as t in g and 2- h gluc ose leve ls. Di abe tes Care 2 009 32 33 5- 341. 35. Tura A, Kautz k y- Wille r A & P ac ini G. Ins ul inog enic indic es from i nsul in and C -pe pti de: c ompariso n of be ta- c ell f unctio n f r om O GTT and IV GTT. Dia b etes Res Clin P ract 2006 72 298- 301 . . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 29, 2024. ; https://doi.org/10.1101/2024.12.25.24319619doi: medRxiv preprint 22 36. World He alth O rganis a tion. C las sif ic ation of di abetes melli tus . Genev a , 2019 . 37. Sc hif f er L, Shahee n F, Gill igan LC , S t orbe c k KH, Haw ley J M, Keevi l BG, A r l t W & Ta ylor AE. Mul ti- s ter oid prof iling by UHP LC- MS/ MS w ith pos t - c olumn in fus ion o f ammonium fl uoride. J C hroma to gr B A n alyt Te c h nol Biome d Li fe Sci 2022 1209 1 23413. 38. Owen LJ, Wu FC & Kee vil BG. A r api d dir e ct a ss ay f or the r ou tine me as urem ent of oes tra diol a nd o es trone by liqu id chroma t og ra phy tand e m mas s s pec tr ome t ry. An n Clin Bi oc h em 2014 51 360 -3 67. 39. V ermeule n A , V erdonck L & Kau fma n JM. A critic al e valua t io n of si mple me t h od s f or the es timati on of free tes t os te rone in se rum. J Cl i n E ndoc r i n o l M et ab 1999 84 3666 -3 672. 40. Mago dor o IM, C as t l e AC , Ts huma N , Go e de cke JH , Sewp a ul R, Ma n asa J, M anne- Goehl e r J , Ntus i NA B, Nyirenda M J & Si edne r MJ . Ass oc iati ons of HIV and prev alen t type 2 dia b etes mel litus in the c on te x t of obes ity i n South Af rica . J our nal of Mul ti m or b id ity and Co mor bidit y 2024 14 . 41. P ers ky V , A b asi lim C , T sints if as K, D ay T, Sa r g is RM , Da vigl us ML, C ai J , Freels S, Unterman T, C hav ez N, Kapl an R, Is as i C R, P ir zada A , Meye r ML, Tal avera G A, Th yaga r a jan B, P e ters B A, Mad r ig al J M, Gri eco A & Turyk ME. S ex Hormon es and Dia bet es in 45- 74- Ye ar- O ld Men and P ostmenop au sa l Women: the His pani c C ommuni t y Health S t ud y . J Clin En docri n ol Meta b 2023. 42. Kal yani RR, Franc o M, D obs AS , Ouyang P , V aid ya D , Bertoni A, G a ps tur SM & Gold en S H . The a ss ocia t i on of endog enous se x hormon es , a dip osi t y , an d i nsul in resi stanc e wi t h incid en t diabe te s in p ostmenopa us a l wom en. J Cli n En docri nol Me ta b 2009 94 4127- 413 5. 43. Boureb ab a N , N g o T, S mies z ek A, B oure b aba L & Maryc z K . Sex hormon e bi nding g lobulin as a pote ntial d rug ca ndidat e f or li ver- rela ted me t a bol ic disorde r s tr e a tmen t. Bi ome d P harma co ther 2022 153 113 261. 44. Si mons P , V alke nburg O, Steh ouw er C D A & Brouw ers M. S e x h or mon e - binding glo buli n: biomarke r and hepa tok ine? Tren ds Endo crino l Met ab 20 21 32 544- 553. 45. Si mons P , V alke nburg O, Telg enka mp I, v a n der Wa ai j KM, de Gr oot D M, V ee raia h P , Bons JAP , Tas ki nen MR, Boren J, S chrauw en P , Ru tt en JHW , C as si man D , Sc halk wijk C G, Stehouw e r C D A , Sch r a uwen- Hinder ling V B, Hodso n L & Brouwe r s M. Rel a t i o nshi p be twee n de n ov o lipog enes is and s e rum se x hormone bin ding globuli n in h umans . Clin E nd ocr in ol ( Oxf ) 2021 95 101- 10 6. 46. Fe nsk e B, Kis che H, Gros s S, Walla sc h of sk i H, V olzke H, D orr M, N a uck M, Ke ev il BG, Bra bant G & Haring R. Endog enous Andr ogens and Se x H ormon e -Bi nding Globul in in Women and Ris k of M e t a b olic Synd r ome and Ty pe 2 D iab etes. J C l in E nd oc r i no l M e ta b 2015 100 45 95- 4603. 47. P eter A , Ka nta r tzis K, Mac han n J, S chic k F , Staig er H, Machic ao F , Sc hl eic her E, F r i t s che A, Ha ring HU & S tef an N. Rel a t io nship s of c ircula t i ng s ex ho rmo ne -b inding globuli n with me t a bolic t ra i t s in hu mans . D iabe tes 2010 59 3 167- 3173. 48. Si mons P , V alke nburg O, van de Waa ren burg MP H , van Greev e nbro ek MMJ, Kooi ME, Jans en J FA, Scha lkwi jk C G, Steho u wer C D A & Brouw ers M. S e ru m se x hormone -b indi ng g lobulin is a media tor of t h e as soc iati on betwe e n intr ahepa t i c lip id con ten t an d t ype 2 diabe te s: th e M a as tric ht S tudy. D iab et olo gia 2022. 49. D e Toni L, Guid olin D , D e F ilippis V , P e t e rle D, Rocc a MS, Di N is io A, De Rocc o P once M & Foresta C . SHBG141- 161 Dom ain- P eptide Stimul ate s GP RC 6A- Mediat ed Respo ns e i n Le ydig and bet a- L angerha ns c ell line s. Sc i R e p 201 9 9 194 32. 50. P eiris AN, Mu elle r R A, St r uv e MF, Smith GA & Kiss ebah AH . Re lati ons hip of a ndr o g enic activ ity to s planc hnic ins ulin m etabolis m and p eriph eral g lucos e utili z a tion in pr emenop a usa l wome n. J Clin Endo cr i nol M et ab 1987 64 162 -1 69. 51. Gr e endal e G A , H a n W, Fink els t e in JS, Bur nett- B owie SM, Huang MH, M a rtin D & K arla mangla AS . C ha nges in Reg ional F a t D istributi on a nd An throp om etr i c Me as ures Across the Meno p ause Tr ans ition . J Cli n En docri nol Meta b 20 21. 52. Marla tt KL , Redman LM, B ey l RA, S mith S R, C ha mpagne C M, Yi F & Love joy JC . Ra c ia l dif f erenc es in body co mpositio n and c ar diomet abol ic ris k du ring the meno p ause t r a nsiti on: a pr ospe c tive , obs er v a tional coh ort st u dy . A m J O b s te t Gyn e c o l 2019. 53. Jaf f NG , Norris SA , Sn yman T , Toman M & C row t he r N J . Body com position in the S t udy of Women E nt e r i ng and in Endoc rine Tran si tion ( S W E ET) : A pers pectiv e of Af rica n women who ha ve a high p r ev alen ce of ob esi ty and H IV i nf ection. M e ta bolis m 2015 64 103 1- 1041. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 29, 2024. ; https://doi.org/10.1101/2024.12.25.24319619doi: medRxiv preprint 23 Table 1. Pa r t ic ipant c ha ract eri stic s Pr e men opausa l P ostm eno pau s a l P V al u e WLW OH (n = 5 7 ) WLWH (n= 24 ) P valu e WL WOH (n=2 36) WL WH (n=4 4 ) P V alue HI V M e n o H I V x Me n o So ci od em ogr aph i c and l i fe s ty l e f ac t o r s A ge ( y ears) 48 ( 4 6 - 5 1) 46 ( 45- 4 9) 0 . 0 65 58 ( 5 4 - 6 1) 5 4 ( 5 0- 59) <0 . 001 0. 2 0 8 <0 .00 1 0. 4 8 4 Ho usi ng de nsit y ( pe ople per roo m ) 1 .2 (0 .9 -1 .5 ) 1 .0 (0 .8 -1 .3 ) 0 .271 1 .2 (0 .8 -1 .6 ) 1 . 1 (0 . 6 -1 .6 ) 0 . 44 7 0 .277 0 .7 6 7 0 .36 8 Ass e t i ndex ( % of 1 2 ) 7 5 . 0 ( 66. 7-83. 3) 75. 0 ( 5 8. 3 -9 1. 6 ) 0 . 6 64 75 . 0 ( 58. 3-83. 3) 75.0 ( 5 8 . 3-8 3 . 3) 0.8 7 1 >0 . 9 99 > 0. 9 99 > 0. 9 99 Current sm o k e r , n ( %) 7 (12 .3 ) 0 0 .07 9 1 0 (4 .2 ) 2 (4 .5 ) 0 . 93 1 0 .93 1 0 .026 Current alco hol co ns ume r , n (% ) 24 ( 4 2 . 1) 1 3 ( 54. 2) 0 . 3 20 53 ( 2 2 . 5) 1 1 ( 2 5.0) 0.712 0 . 3 21 0 .003 0. 5 78 Me d ic atio n us e , n(% ) Diab ete s m e dicat i o n s 4 ( 7.0) 1 ( 4 . 2) 25 ( 1 0 . 6) 6 ( 1 3.6) 0 . 6 30 0 . 4 21 0 . 5 01 Hy p ert e nsi on medicat ion s 10 ( 1 7. 5) 5 ( 21. 7 ) 99 ( 4 1 . 9) 2 2 ( 5 0.0) 0. 6 6 4 0 .001 0. 9 4 2 L i p id me d i c at io n s 3 (5 . 3 ) 1 (4 .4 ) 2 1 (8 .9 ) 2 (4 .5 ) 0 .86 5 0 .371 0 .7 1 0 HIV i n f or m at ion Es ta b li sh e d on A R T, n (%) 1 8 ( 75. 0) 3 7 ( 8 4.1) 0 . 2 21 PIs, n (%) 0 1 ( 2. 3 ) 0 . 4 66 NNR TIs, n ( %) 1 3 ( 7 2. 2) 2 7 ( 73.0) 0 . 721 CD 4 c oun t (cel l / mm 3 ) 4 74 ( 36 1- 619) 6 1 1 ( 4 1 6 - 85 8) 0 . 1 28 CD 4 WC C (ce l l/m m 3 ) 5 ( 4 - 6) 5 ( 4- 7) 0 . 8 62 C D4 ly m poh cy te s ( ce l l / mm 3 ) 33 ( 21- 4 0) 3 4 ( 2 3- 40) 0 . 4 63 B ody co m pos itio n H e i gh t ( c m ) 158 . 1 (15 4 . 1 - 162 . 5 ) 157. 1 (1 53 . 0 -160 .2 ) 0 .26 7 158 . 2 ( 1 54.4-1 6 2.0) 15 6.3 ( 1 53.3- 16 0. 4 ) 0.1 08 0 . 5 06 0 . 9 15 0 . 8 44 Weig ht ( k g ) 8 1 . 8 ( 77. 0-97. 4) 75. 6 ( 6 3. 7 -8 8. 3 ) 0 . 0 65 81. 3 ( 7 2. 4 – 9 5 .1 ) 79.9 ( 6 4 . 3-9 2 . 8) 0.1 06 0 . 755 0 . 8 7 3 0 . 8 46 BMI ( k g / m 2 ) 3 3 . 3 ( 29. 4-37. 7) 31. 6 ( 2 6. 8 -3 5. 7 ) 0 . 1 21 3 3 . 1 ( 29. 1-37. 8) 3 3.1 ( 2 6 . 6-3 6 . 5) 0.1 84 0 . 1 73 0 . 8 68 0 . 2 71 Wais t circu mfe re nce (c m) 9 5.5 ( 8 9 . 6-10 3. 0 ) 9 5 . 5 ( 77. 5-10 6.0) 0 . 4 32 9 6.5 ( 8 9 . 6-10 5. 6 ) 9 6. 6 ( 8 5.1-1 03. 8) 0.3 84 0 . 8 06 0 . 6 86 0 . 9 73 Sub to tal bo dy f a t m a s s (%) 4 3 . 8 ( 41. 8-46. 7) 43. 8 ( 3 9. 0 -4 5 . 6 ) 0 . 3 64 4 6 . 2 ( 42. 2-49. 0) 4 4.3 ( 3 8 . 6-4 9 . 3) 0.1 70 0 . 9 90 0 .014 0. 3 3 5 FMI (kg/m 2 ) 1 4 . 2±3. 5 12. 5±3. 7 0 . 0 59 1 4 . 5±4. 3 1 3.5± 4 . 8 0 . 15 1 0. 1 0 2 0. 6 8 3 0. 5 70 Leg f at mass (% ) 4 4 . 7±7. 1 43. 9±7. 2 0 . 6 81 4 4 . 2±6. 4 4 4.1± 7 . 2 0 . 972 0. 8 0 3 0. 77 5 0. 9 5 9 Tru nk f a t m a s s (%) 4 2 . 7±6. 3 42. 8±6. 1 0 . 9 24 4 3 . 2±6. 0 4 3.2± 6 . 3 0 . 973 0. 8 6 9 0. 9 9 7 0. 70 3 VA T (c m 2 ) 10 2 (60 -12 8 ) 77 (59 -108 ) 0 .22 3 10 8 (7 8 -14 0 ) 109 (64 -1 3 2) 0 . 38 5 0 .12 8 0 .51 8 0 .19 6 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 29, 2024. ; https://doi.org/10.1101/2024.12.25.24319619doi: medRxiv preprint 24 SAT (cm 2 ) 4 6 4 ( 3 72- 51 5 ) 4 13 ( 30 9- 535) 0 . 2 74 4 8 3 ( 3 92- 58 2 ) 4 4 3 ( 2 9 6 - 58 0) 0.1 58 0 . 2 22 0 . 5 29 0 . 7 23 VA T / SA T 0. 21 ( 0 . 1 6-0 . 2 6) 0 . 2 0 ( 0. 1 5 -0. 2 5 ) 0 . 6 28 0. 22 ( 0 . 1 8-0 . 2 7) 0. 22 ( 0.1 8-0.3 0) 0.8 14 0 . 9 12 0 . 2 77 0 . 8 49 Value s a r e medi an (inter q uar til e rang e) or mean ± s t anda r d devia ti on. WLWO H, women li ving without HIV; WLWH, wome n livi ng wi th HIV; PI, pr ot ea se inhi bi tors , NN RT I s , non-nuc l eoside r ev erse transcripta s e inhibitors; WCC, white ce ll count. BMI, body mass i ndex, W HR, wai st-hi p-ratio; FFS TM, fa t - fre e s oft ti s s ue ma ss; F MI , fa t ma s s index ; V A T , vis c e r a l ad ip os e tis s ue ; SA T , s u b c uta n eo us ad ipo se tis s u e. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 29, 2024. ; https://doi.org/10.1101/2024.12.25.24319619doi: medRxiv preprint 25 Table 2. Dif fer enc e s i n SHBG, s e x hormones a nd fa sting and OGTT-derive d par amete rs betwe e n menopausal and HI V groups. Pr e men opausa l P o s tm e n o pa us al P V alue a d just ed f o r ag e WLWOH ( n= 52) WL WH (n = 2 1) P va l u e WL WOH (n=2 25) WLWH (n= 41 ) P v a lue HI V M e n o H I V x Me n o S e x hormo n e s LH (IU/L ) 6. 7 ( 4. 0-1 4.5) 4 . 5 ( 3 . 1- 1 2 .0 ) 0 . 2 75 2 1.8 ( 16. 0-29. 3) 22. 7 (1 7 . 3 -3 1. 1 ) 0. 218 0 . 4 9 5 <0 .001 0. 4 4 8 F SH (IU/L ) 1 1. 9 ( 6. 0 - 23. 6 ) 8 . 3 ( 4 . 7- 1 8 .9 ) 0 . 2 1 2 5 5.3 ( 40. 8-69. 3) 63. 8 (4 3. 8 -7 5. 3 ) 0. 092 0 . 3 8 0 <0 .001 0. 1 2 8 E 2 ( p mo l/L ) 1 98.0 ( 77 .5 - 5 6 0 . 5) 3 3 3. 0 ( 96.0- 48 3. 0 ) 0 . 9 9 1 2 4 . 0 ( 15. 0-39. 0 ) 8. 0 ( 4. 3- 19. 5) <0. 00 1 < 0. 0 01 <0 . 0 0 1 <0 . 0 0 1 SH BG ( n m o l / L ) 55. 4 ( 4 3.4-8 1.7) 6 4.2 ( 5 1 . 3-1 11. 3 ) 0 . 1 7 6 5 0 . 6 ( 37. 8-65. 2 ) 7 2 . 1 ( 54. 2-10 3. 0) <0. 00 1 0 .32 6 0 .370 0 .18 0 T o ta l te s tos ter one ( nm ol/L) 0. 4 5 ( 0. 25-0 .56 ) 0. 24 ( 0 . 1 8-0.3 4 ) 0 . 0 0 8 0 .3 5 (0 .21 - 0 . 57) 0 .25 (0 .09 - 0. 48 ) 0 .0 23 0 .02 6 0 .038 0. 2 9 4 F r e e tes to ster o n e (p mol/L) 5. 4 ( 3. 5 - 6. 9) 2. 5 ( 1 . 3- 4.3) 0 . 0 12 4. 8 ( 2 . 5- 8.0) 2. 0 ( 1. 0- 3 . 6) <0. 00 1 0 . 0 11 0. 9 0 7 0. 9 2 1 Gl y c aem ic stat us , n ( % ) NGT 4 6 ( 8 0. 7 ) 19 ( 8 6 . 4) 0 . 7 0 3 1 54 ( 65. 8) 34 ( 77. 3) 0. 0 6 5 Ref Ref Ref IGM 5 ( 8. 8) 2 ( 9.1) 44 ( 1 8 . 8) 2 ( 4 . 5) 0 . 8 70 0 . 6 39 0 . 1 88 T 2D 6 ( 10. 5) 1 ( 4.5) 36 ( 1 5 . 4) 8 ( 1 8 . 2) 0 . 4 8 6 0 . 764 0 . 4 49 D ys g l ycaemia 1 1 (1 9. 3 ) 3 ( 1 3.6) 0 . 5 5 5 80 ( 3 4 . 2) 10 ( 22. 7) 0. 136 0 . 7 1 1 0 . 6 03 0 . 8 61 Fasti n g p ar a m e te rs inc l ud i n g part i c ipa nts w ith T 2 D n 57 2 2 2 3 2 4 4 Hb A 1 c (% A 1c) 5. 7 ( 5. 3 - 6. 3) 5 . 7 ( 5 . 3- 5.9) 0 . 5 6 9 6 . 0 ( 5 . 6- 6.6) 6. 1 ( 5. 7- 6 . 6) 0. 543 0 . 6 9 8 0 . 5 29 0 . 3 26 F as ti ng gl u c os e ( m mo l/L ) 5. 0 ( 4. 5 - 5. 5) 5 . 0 ( 4 . 7- 5.5) 0 . 5 7 7 5 . 0 ( 4 . 5- 5.7) 5. 2 ( 4. 5- 5 . 8) 0. 592 0 . 7 5 3 0 . 2 62 0 . 6 04 F as ti ng i ns ul i n ( mmol/L) 8. 5 ( 5. 2-1 4.0) 1 0 . 3 ( 6 . 3- 14 . 0) 0 . 6 9 7 9 . 0 ( 4 . 9- 1 4 .4 ) 9 . 9 ( 5 . 6- 16.2 ) 0. 497 0 . 3 9 2 0 . 6 27 0 . 756 F as ti ng C- pe pti de ( nm ol / L) 1. 6 ( 1. 3 - 2. 3) 1 . 8 ( 1 . 4- 2.2) 0 . 71 0 1 . 8 ( 1 . 3- 2.5) 1. 9 ( 1. 3- 2 . 7 ) 0. 457 0 . 5 1 2 0 . 5 22 0 . 8 28 HOM A-I R 1. 7 ( 1. 2 - 3. 6) 2 . 3 ( 1 . 3- 3.4) 0 . 70 6 2 . 0 ( 1 . 1- 3.6) 2. 4 ( 1. 3- 4 . 2) 0. 430 0 . 2 5 7 0 . 8 43 0 . 706 Ba s a l ins u li n c l e ara n c e 0 . 2 0 ( 0. 15-0. 26) 0. 18 ( 0.1 6-0 . 2 7) 0 . 9 6 9 0. 20 ( 0.1 6-0 . 2 7) 0.2 0 (0. 1 5 -0. 2 7 ) 0. 602 0 . 2 9 7 0 . 5 99 0 . 4 46 Fasti n g a nd OG TT - d eri ve d p ar a mete rs exc lu di ng pa rtici p a nts w ith T 2D n 51 2 2 1 9 9 3 7 Hb A 1 c (% A 1c) 5. 6 ( 5. 2 - 6. 1) 5 . 6 ( 5 . 3- 5.8) 0 . 77 5 6 . 0 ( 5 . 6- 6.3) 6. 1 ( 5. 6- 6 . 4) 0. 635 0 . 9 9 0 0 . 1 7 4 0 . 5 39 Fa s t i n g g l uc o s e ( m m ol / L ) 4. 8 ( 4. 4 - 5. 3) 5 . 0 ( 4 . 7- 5.4) 0 . 5 7 7 4 . 9 ( 4 . 4- 5.4) 5. 1 ( 4. 5- 5 . 6) 0. 492 0 . 4 8 1 0 . 4 85 0 . 6 38 F as ti ng i ns ul i n ( mmol/L) 8. 0 ( 5. 1-1 3.8) 1 0 . 2 ( 6 . 3- 13 . 4) 0 . 6 2 3 8 . 1 ( 4 . 7 - 1 3 .2 ) 9 . 6 ( 5 . 4- 15.7) 0. 357 0 . 3 0 1 0 . 5 18 0 . 8 14 F as ti ng C- pe pti de ( nm ol / L) 1. 6 ( 1. 3 - 2. 3) 1 . 8 ( 1 . 4- 2.1) 0 . 6 9 6 1 . 7 ( 1 . 3- 2.3) 1. 7 ( 1. 3- 2 . 7) 0. 667 0 . 3 1 8 0 . 5 23 0 . 4 85 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 29, 2024. ; https://doi.org/10.1101/2024.12.25.24319619doi: medRxiv preprint 26 HOM A-I R 1. 7 ( 1. 1 - 3. 0) 2 . 3 ( 1 . 3- 2.8) 0 . 5 3 7 1 . 7 ( 1 . 0- 3.2) 2. 2 ( 1. 2- 3 . 4) 0. 354 0 . 2 0 9 0 . 6 60 0 . 8 94 Ba s a l ins u li n c l e ara n c e 0 . 2 1 ( 0. 15-0. 27) 0. 18 ( 0.1 6-0 . 2 7) 0 . 8 4 0 0. 21 ( 0.1 6-0 . 2 8) 0.2 1 (0. 1 6 -0. 2 6 ) 0. 390 0 . 4 7 7 0 . 5 33 0 . 6 96 Ma t s u da in dex 5. 7 ( 3. 1 - 8. 1) 4 . 4 ( 3 . 0- 8.4) 0 . 7 6 3 5 . 3 ( 3 . 2- 8.1) 4. 9 ( 2. 8- 8 . 0) 0. 869 0 . 3 9 0 0 . 4 62 0 . 4 35 Ins ul i n o g e ni c in dex 32. 7 ( 1 7.5-5 6.2) 3 9.8 ( 19. 6-56. 5) 0 . 8 2 4 2 3 . 4 ( 13. 0-42. 3 ) 38. 4 ( 1 9. 5 -6 5. 5) 0. 00 2 0 .58 2 0 .49 0 0 .35 7 C - p e pt id e in d e x 3 .2 (1 .8 -5 .6 ) 3 .6 (2 .5 -4 . 3 ) 1 .00 0 2. 8 ( 1 . 6- 4.4) 4. 0 ( 2. 4- 6 . 6) 0. 01 9 0. 4 4 4 0 .77 7 0 .36 0 Dis p osi tio n i ndex 1 69.0 ( 73.5 - 2 5 3 . 2) 1 3 9. 7 ( 63.3- 23 2. 5 ) 0 . 76 0 126 .2 (5 7. 1 -2 15 .4 ) 21 4 .2 (9 0. 6 - 461 . 9) 0 .0 07 0 .41 8 0 .62 2 0 .019 Value s a re medi an (interq uar til e rang e ). W L WO H, women liv ing without H IV; WL WH , women living with HIV; NGT, norma l gluc os e tol e r ance ( fa sting pla sma gl ucose <6. 1 mmol/L & 120 minute post gl ucos e loa d <7.8 mmol/ L ) ; IFG, impaired fa s ting g lucos e (f as ting pl a s ma gluc ose: 6.1–6.9 mmol/L); IGT, impai red glucose t ole ra nce (120 minute pos t-g luco s e l oad: 7 .8– 11.0 mmo l / L ); T2D, type 2 dia bet es (fa s ting pla sma gl ucos e > 7. 0 mmol/L a nd/ or 12 0 minut e po s t -g luc os e l oad ≥ 11 .1 mmol / L and/ or on dia bete s medi cation. IGM, i mpair ed g lucos e met a bolism ( IFG and/or IGT ); Dy sglyc aemi a: I GM and/or type 2 diabe te s. S I , insulin sensit ivity; AIR g , acute insulin respons e t o glucos e ; DI, di sposi tion index; Sg, gluc ose ef fec tiv enes s . . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 29, 2024. ; https://doi.org/10.1101/2024.12.25.24319619doi: medRxiv preprint 27 Table 3. Ass ocia tion s be tween SH BG, an drogen hormon e s and glycaemic and in sulin parame ter s Parameter SHBG (Scaled robust Z-score) Total testosterone (Scaled robust Z-score) Free testosterone (Scaled robust Z-score) Odds ratio (95% CI) P value Odds ratio (95% CI) P value Odds ratio (95% CI) P value D ys glycaemia 0.60 (0.45 – 0.79)# <0.001 0.95 ( 0.80 – 1.12 ) 0.540 1.20 ( 0.12 – 11.97 ) 0.876 Standardised values (scaled robust Z-scores) Beta (95% CI) P value Beta (95% CI) P value Beta (95% CI) P value HbA1c -0 . 0 8 ( - 0 .18- 0 .02) 0.106 -0 . 0 6 ( - 0 .12 to 0.01 ) 0.106 -0 . 0 5 ( - 0 .12 to 0.02 ) 0.195 Fasting Gluco s e -0.13 (-0.22 to -0.04) 0.003 -0 . 0 2 ( - 0 .09 to 0.05 ) 0.541 -0 . 0 1 ( - 0 .09 to 0.06 ) 0.673 Fasting insulin -0.12 (-0.21 to -0.03) 0.011 0.07 ( - 0 .003 to 0.14 ) 0.060 0.06 ( - 0 .01 to 0.12 ) 0.079 HOMA - IR -0.14 (-0.23 to -0.05) 0.003 0.06 ( - 0 .02 to 0.13 ) 0.132 0.05 (-0.02 to 0.11) * 0.176 Basal in s ulin clear ance 0.14 (0.04 to 0.24) 0.005 -0 . 0 6 ( - 0 .14 to 0.02 ) 0.147 -0.09 (-0.16 to -0.01) 0.021 Ins u linogenic index -0 . 0 3 ( - 0 .15 to 0.09) 0.592 0.0002 ( - 0 .08 to 0.08 ) 0.995 0.03 ( - 0 .05 to 0.11 ) 0.535 C - p eptid e index 0.10 ( - 0 .03 to 0.22 ) 0.139 -0 . 0 5 ( - 0 .14 to 0.03 ) 0.204 0.001 ( - 0 .08 to 0.08 ) 0.977 Mat s uda ind ex 0.31 (0.21 to 0.41) * <0.001 -0 . 0 4 ( - 0 .12 to 0.04 ) 0.354 -0 . 0 4 ( - 0 .13 to 0.04 ) 0.289 D i s po s itio n index 0.29 (0.16 to 0.42) <0.001 -0 . 0 2 ( - 0 .11 to 0.08 ) 0.690 -0 . 0 1 ( - 0 .11 to 0.78 ) 0.770

s f r om logis tic and quan tile r egre ss ion model s adju s t ed f o r age, f a t mass i ndex and HIV stat us. Robu s t Z-s co re s , ca lculat ed from the m edia n and median abs o lute devia tion ( MA D ), wer e de rive d f or SHB G, tot al and f r ee te sto s t eron e and the glycaemic and in s ulin mea sure s t o facilita te c ompari s on s of th e . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 29, 2024. ; https://doi.org/10.1101/2024.12.25.24319619doi: medRxiv preprint 28 magnitude o f the a s sociati on s . The rob us t Z-s cor e s rep re s e nt the n umber of M ADs t hat a da ta p oint li e s from t he med ian of th e v ariabl e in the d ata set . Th e MADs wer e s ca led by a con s tan t factor o f 1.4826 so tha t th e robu s t Z - s c ore i s mo re direc tly comparable to a t radit ional Z-s cor e . # s ex hormon e x HIV int erac tion; *sex hormone x menopau s al s t atu s int erac tion. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 29, 2024. ; https://doi.org/10.1101/2024.12.25.24319619doi: medRxiv preprint Figure 1 . The a s sociati on betwe en SHB G and prevalen t dys gla emia by HIV stat us (P =0.055 f or SHB G x HIV in ter action ) . Hig her SHB G was a ss ocia ted with odds o f dys glycaemi a in WLWOH ( OR ( 95% conf idence in terval ( C I ) : 0 .50 ( 0.35 – 0 . 71) , P<0.001) bu t not in WLWH ( O R ( 95% C I ) : 0.90 ( 0 .62 – 1.31, P =0 .59 2 Model adju s t ed f o r age and FM I. SHB G p re s en ted a s s caled robu s t Z-s co re s , calcul ated f rom th e m edian a nd median ab s ol ute devia tion (MA D ) , with M A by a cons ta nt facto r of 1.4826 s o that th e robu s t Z-s co re i s compara ble to a tra dit ional Z-s core . 29 lo w e r 2 ). D s cal ed . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 29, 2024. ; https://doi.org/10.1101/2024.12.25.24319619doi: medRxiv preprint 30 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 29, 2024. ; https://doi.org/10.1101/2024.12.25.24319619doi: medRxiv preprint Figure 2 . The a s sociati on betwe en SHB G and f r ee t e s to ster one and in s ulin p aram ete rs by menop au s al s ta tu s , adju s ti ng f o r age, HI V sta tu s and fat ma ss i A) The a ss ocia tion be tween SH BG and in s ulin s en s i tivity ( Ma t s uda ind ex ) diff e rs b y menopausal sta tu s ( p=0 .043 f o r SHBG x menopa u s e ) , with th e a ss oci a being stronge r in preme nopau s al (s tan da rdised ß ( 95%CI): 0.46 ( 0 .22- 0 .70) p<0 .00 1) compared to po s tm enopau sal women ( stand ardi s ed ß ( 95%C I ) : 0 .2 9 0.43) , p<0 .001). B) The a ss oci ation b etw een free te sto s t eron e and HOM A-IR di f fe rs by menopa usal sta tu s (P =0.015 f or f re e te s t o s te rone x menopa use), ass ocia tion only s ignf icia nt in th e prem e nopau s al women (standa rdi s ed ß ( 95%C I ) : 0.40 ( 0 .13- 0 .67), p=0.005 ) and no t the post- m e nopau sal women ( stand ardi s ed ß ( 95%C I ) :0 .05 (- 0.02 – 0.1 2) , p=0.200 ) . 31 i ndex . a ti on 9 ( 0.16 - with the . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted December 29, 2024. ; https://doi.org/10.1101/2024.12.25.24319619doi: medRxiv preprint