Hypogonadism, adipose tissue inflammation, and adipose tissue circadian clock disruption promote metabolic dysfunction in Klinefelter syndrome

Hypogonadism, adipose tissue inflammation, and adipose tissue circadian clock disruption promote metabolic dysfunction in Klinefelter syndrome | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Hypogonadism, adipose tissue inflammation, and adipose tissue circadian clock disruption promote metabolic dysfunction in Klinefelter syndrome Jesper Just, Emma Hasselholm, Anne Skakkebæk, Claus Gravholt, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8850956/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted You are reading this latest preprint version Abstract Sex chromosome aneuploidies like Klinefelter syndrome (KS) represent a naturally occurring genetic model to investigate metabolic function at the intersection of sex chromosomes, hypogonadism and sex hormone replacement therapy. Men with KS have severe metabolic dysfunction and a more than 2.5-fold increased risk of type 2 diabetes. We assessed body composition, insulin sensitivity, and adipokine profile before and after testosterone replacement therapy (TRT) in KS. TRT does not completely normalize body composition or abolish insulin resistance in KS, indicating insufficient rescue of metabolic risk by correction of hypogonadism. We then assessed gene expression in KS adipose tissue related to the metabolic profile, demonstrating that hypogonadism independently increased adipose tissue expansion and inflammation in KS, and profound disruption of adipose tissue circadian rhythm tightly associated with insulin resistance and signs of biological aging in adipose tissue. We further provide a prediction model for body fat in KS. Our data reveals intricate interactions between sex and metabolism, evolving the pathophysiological understanding of KS. Health sciences/Endocrinology/Endocrine system and metabolic diseases/Gonadal disorders/Hypogonadism Health sciences/Endocrinology/Endocrine system and metabolic diseases/Obesity Health sciences/Endocrinology/Endocrine system and metabolic diseases/Metabolic syndrome Health sciences/Molecular medicine Biological sciences/Genetics/Genomics/Transcriptomics Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 INTRODUCTION Genetic sex and sex hormones have profound impact on adipose tissue deposition and metabolic function 1 . Sex chromosome aneuploidies, including Klinefelter syndrome (KS; 47,XXY), provide a unique integrated human model to investigate metabolic function in the context of altered sex chromosome dosage, hypogonadism, and hormone replacement therapy. Metabolic dysfunction in KS is prominent already at an early age 2 , with accumulation of body fat, dyslipidemia, insulin resistance and the metabolic syndrome present even in pre-pubertal boys with KS 3,4 . KS is characterized by hypergonadotropic hypogonadism, and metabolic dysfunction is aggravated after puberty, when hypogonadism contributes to a vicious metabolic cycle and further promote hyperinsulinemia, insulin resistance and accumulation of visceral fat 2,5 . As a result, rates of metabolic syndrome and type 2 diabetes are increased in KS 2 . In our most recent national cohort study, we demonstrated a more than 2.5-fold increased risk of type 2 diabetes in men with KS corresponding to a cumulative risk at around 20% 2,6,7 . Testosterone replacement therapy (TRT) in KS aims to abolish the negative consequences of hypogonadism, but the effects of TRT on metabolic function and the cardiometabolic profile in KS have not been clarified 2 . TRT has been shown to reduce body fat in KS 2,5,8 , but TRT is also associated with reduced high-density lipoprotein (HDL) leading to higher predicted coronary risk 8 . In addition, available studies find little or no effect of TRT on insulin resistance in KS 2 . We recently demonstrated that despite TRT leading to an almost halving of mortality risk 9 , this was not reflected by a comparable reduction in metabolic risk, and TRT did not alleviate the risk of type 2 diabetes in KS 10 . As such, it has become increasingly clear that metabolic dysfunction in KS is not solely a result of hypogonadism, but that individuals with KS are likely genetically predisposed to metabolic dysfunction irrespective of hypogonadism. It is also becoming clear that metabolic dysfunction in KS has a pivotal impact on the excess comorbidity risk seen with the syndrome; cardiovascular disease, thrombosis, steatosis of the liver, inflammatory and autoimmune disease among others 2 . To provide better care for men with KS, we need to improve our understanding of independent and specific contributions of hypogonadism and genetics to metabolic dysfunction in KS. Further, investigating KS as a human genetic model of metabolic dysfunction and hypogonadism could lead to new insights with general applicability. Here we present a large, comprehensive, and longitudinal assessment of metabolic function and body composition in men with KS stratified by TRT as either untreated (U-KS) or treated (T-KS). We describe the effect of TRT on a broad selection of metabolic risk markers and highlight areas of metabolic dysfunction in KS insensitive to TRT, implicating the extra X chromosome independent of hypogonadism. To elucidate the specific genetic imprint of KS, we investigate associations between metabolic function and the transcriptomics of adipose tissue, for the first time exposing a disrupted diurnal rhythm. We further evaluate the appliance of several metrics for prediction of body fat and present KS specific prediction models of body fat to suggest a qualitative tool for clinical monitoring of cardiometabolic risk in KS patients. RESULTS Participants A total of 149 unique men with KS and 178 controls were included at Visit 1 (Table 1, Fig. 1). Fifty-four men with KS had more than one visit, with five men still not receiving TRT at Visit 2. A total of 26 men with KS, who were untreated at Visit 1, initiated TRT at a later visit; 24 between their first and second visit, and the remaining two between the second and third visit (Figure 1). Mean age at Visit 1 was approximately 37 years in all groups, with around 2.5 years between visits (Fig. 1). Sex hormones As expected, total testosterone was lowest in U-KS, and TRT was associated with a 60% increase in testosterone levels among KS males (Table 1). Hypogonadism in U-KS was further substantiated by increased levels of LH, that were reduced to about a quarter with TRT. However, LH levels in T-KS were still higher than in controls. Also, T-KS had higher levels of estradiol and lower SHBG compared with U-KS and controls, corresponding to an almost 30% mean increase in 17β-estradiol and 20% mean reduction in SHBG following TRT (Table 1). Body composition Total body weight was lower in controls compared with the KS groups, but with no difference in BMI across the groups. Neither weight nor BMI was affected by TRT (Table 1). At Visit 1, body fat percentage was different between all groups, with the highest body fat observed in U-KS and the lowest in controls. Across the complete data set, TRT was associated with an approximate 15% reduction of body fat, corresponding to (Table 1, Figure 2). Conversely, lean mass at Visit 1 was lowest in U-KS, and across the entire data set we saw a mean increase of 2.6 kg of lean mass after TRT in KS (Table 1, Figure 2). This was reflected by an increase in nFFMI with TRT, but still, nFFMI in both U-KS and T-KS was reduced compared with controls. At Visit 1, an average (>19 kg/m 2 ) or above average (>21 kg/m 2 ) nFFMI was found in 56% of controls, but only in 32% of U-KS and 36% of T-KS, demonstrating the relative lower lean mass even with TRT in KS. Also, 48% of U-KS and 39% of T-KS had a low (<17.6 kg/m 2 ) or severely low (<16 kg/m 2 ) nFFMI compared to only 17% of controls indicating a high risk of sarcopenia. BMI was associated with body fat when adjusting for the interaction between grouping and BMI (mR 2 =0.65, p<0.001). From this model, at any given BMI in the clinically most relevant range (25-35 kg/m 2 ), body fat was higher in U-KS compared with the other groups, and higher in T-KS compared with controls (Fig. 2). We then classified our observations according to commonly agreed thresholds for obesity based on either BMI (>30 kg/m 2 ) or body fat (>25%). Of those with a BMI below the threshold for obesity, 44.0% of men with KS, but only 19.4% of controls, demonstrated body fat above the threshold for obesity, demonstrating a clear incongruence between interpreting BMI as a marker of body composition in KS versus controls (Figure 2). We further considered other commonly applied surrogate metrics for body fat. Numerically, mean WHRadjBMI was highest in U-KS and lowest in controls, but with very large variability and WHRadjBMI was not different between groups. Both VAI and LAP were elevated in the KS groups compared with controls, but with no effect of TRT. As such, neither of these metrics could detect changes in measured body fat with TRT in KS. Correspondingly, WHRadjBMI was not associated with body fat in KS (p=0.85), and neither was LAP (p=0.17) while VAI was only very weakly associated with body fat (β(95% CI); 2.1 (1.6 ; 2.6), p<0.001, mR 2 =0.05). Lipids LDL cholesterol levels were not different between groups at Visit 1 and there was no effect of TRT on LDL levels (Table 1). However, levels of HDL were different between all groups at Visit 1 with the highest values in controls and lowest in T-KS, and with a corresponding 0.2 mmol reduction in HDL after TRT (Table 1). Similarly, triglyceride levels were highest in T-KS and lowest in controls, but with no apparent effect of TRT on triglycerides levels. Glucose metabolism Fasting glucose was higher in U-KS compared with both T-KS and controls. Insulin was higher in both U-KS and T-KS compared with controls, which was converted into higher HOMA2 %B and HOMA2-IR in both KS groups compared with controls (Table 1). In mixed model regression, TRT was not associated with changes in either glucose, insulin, or any of the HOMA2 derivates. We further evaluated TyG as a surrogate marker of insulin resistance. TyG was increased in both KS groups, with no effect of TRT and a weak association with HOMA2-IR (β (95% CI) 0.74 (0.38; 1.10), p<0.001, mR 2 =0.19). We assessed levels of leptin and adiponectin; metabolically active hormones produced by fat cells. The highest levels of leptin were seen in U-KS and the lowest levels in controls, with TRT being associated with a 43% reduction in leptin levels in KS (Table 1). Adiponectin levels were lowest in T-KS and highest in U-KS with a 21% reduction in adiponectin levels following TRT. Also, leptin-to-adiponectin ratio was increased in both KS groups compared with controls at Visit 1 with no effect of TRT. Applying a mixed linear regression model as described in relation to TRT, log-transformed HOMA2 %S was tightly and inversely associated to body fat adjusting for group, group-group interaction, visit, age and change in age between visits (β(95% CI), p; -0,04 (-0.5 ; -0,3), p<0.001), Fig. 2) with the complete model explaining 69% of the variability in HOMA2 %S (cR 2 =0.69). In this model, body fat alone explained 25% of the variability in HOMA2 %S (mR 2 Partial for body fat = mR 2 full model - mR 2 without body fat = 0.40-0.15=0.25). Body fat was also inversely associated with HOMA2 %S in each of the inclusion groups separately and without adjustment for age, but with a better fit in T-KS compared with U-KS (p≤0.002 for all, Figure 2). Assessment of genetic factors inducing insulin resistance in KS Since TRT did not completely normalise body fat, nor alleviate insulin resistance in KS, we investigated the genomic underpinnings of adipose tissue in KS compared with healthy controls to identify KS-specific gene expression patterns that could explain differences in body fat and insulin resistance. We performed unsupervised multi-factor analysis (MOFA) and constructed 10 latent gene expression factors that captured most of the variance in the dataset (Extended Data Figure 1). We associated subject-specific factor weights for each factor (i.e. the score indicating how strongly each subject shows that gene expression pattern) with core clinical traits (testosterone, LH, body fat, leptin, adiponectin, HOMA2%B, HOMA2%S, HOMA2%IR), (Figure 3A, Extended Data Figure 2). Factor 1 was strongly associated with insulin resistance but was not KS-specific (KS and controls had similar subject-specific factor 1 weights) and was associated with body fat in both KS and controls (Figure 3 D-F). In contrast, factor 6 was both KS-specific and associated with insulin resistance (HOMA2S, HOMA2IR), independent of body fat, pointing towards an inherent genomic feature of KS (Figure 3 G-I). Genes with the largest absolute weights for factor 1 were predominantly inflammation-related, and their expression increased with higher factor 1 scores. These included several canonical mediators of innate immunity and inflammatory activation; complement and chemokine signalling components ( C5AR1 , CCR1 , CXCL16 ), activating Fc receptors ( FCGR3A , FCGR2A ), pro-inflammatory integrins ( ITGAX / ITGB2 ), and key mediators of tissue remodelling and lipid-driven inflammation ( SPP1 /osteopontin, MMP9 , PLA2G7 , CHI3L1 ) (Figure 4A). Enrichment analysis supported this, with the top enriched pathways being related to immune and inflammatory responses, lymphocyte activation, and cytokine production (Figure 4B). Conversely, downregulated genes were involved in processes such as amino acid metabolism, lipid homeostasis, and responses to glucose and insulin (Extended Data Figure 3). To further validate these findings, we used DNA methylation (DNAm) data obtained from the same adipose tissue biopsies to infer immune cell infiltration. In line with factor 1 weights, we observed a non-significant increase in immune cell proportions in men with KS, mainly driven by higher inferred levels of NK cells and macrophages/monocytes (Figure 4C, Extended Data Figure 4). Immune cell fraction was strongly associated with both body fat and insulin resistance in both KS and controls (Figure 4D-F), indicating that increased body fat percentage is universally associated with immune cell infiltration and activation in adipose tissue. For the KS-specific factor 6, KS had markedly lower factor weights than controls, and this factor appeared to be driven mainly by insulin-related pathways (Figure 5A). In KS, the factor 6 gene signature associated with higher insulin resistance in adipose tissue was characterized by reduced expression of key insulin/circadian regulators (including IRS2 , PDK4 , PFKFB3 , CIART , PER3 , ZBTB16 , and HLF ), increased expression of genes involved in lipid storage and steroid/xenobiotic metabolism ( CES1 , GPAM , DGAT2 , VLDLR , SULT1A1 / 2 , AKR1C3 , LEP ), together with markers of extracellular matrix remodelling and stress signalling ( TNMD , SFRP4 , STC1 , ADAMTS4 , SGK2 , PMEPA1 ). This profile was consistent with hypertrophic, fibrotic, and metabolically inflexible adipose tissue in the context of insulin resistance. Enrichment analysis supported this, showing that factor 6 genes downregulated in KS were enriched in pathways related to circadian regulation and rhythm, response to insulin stimulus and glucose metabolism (Figure 5B), whereas upregulated genes were enriched in pathways related to steroid, hormone, and lipid metabolism (Extended Data Figure 5). Due to this downregulation of genes related to circadian rhythm, we estimated the degree of circadian clock disruption in the adipose tissue using the TimeTeller algorithm. We observed a highly significant disruption of circadian rhythmicity in KS adipose tissue compared with controls (Figure 5C). Furthermore, using the deltaCCD algorithm on a core circadian clock gene set consisting of both circadian activators and repressors (Extended Data Table 1), KS adipose tissue showed a significantly larger deviation from the control clock co-expression pattern compared with controls (deltaCCD = 2.96, p = 0.001). Especially the core circadian activators were collectively downregulated (Figure 5D,E). To further investigate the potential impact of the circadian clock dysregulation within adipose tissue, we assessed age acceleration using an epigenetic clock based on DNAm patterns. Epigenetic age acceleration was significantly increased in adipose tissue from KS compared with controls (Figure 5F). Together, these findings provide strong evidence that the internal circadian clock in KS adipose tissue is disrupted, and that KS adipose tissue exhibits increased age acceleration, which appears to be linked to insulin resistance. Prediction of body fat in KS Body fat in KS appeared to be a good candidate marker for evaluating effects of TRT and assessing overall cardiometabolic risk in KS. However, continuous monitoring of treatment by direct assessment of body fat requires access to regular DEXA or CT scans, restricting the practical utility in the outpatient clinic. Building on this clinical need, we developed two XGBoost prediction models for estimating body fat in men with KS using routinely available inputs: (i) an anthropometric model (age, height, weight, waist circumference, hip circumference and treatment status) and (ii) an extended model including total testosterone. In the training dataset (visit 1, n = 134), the anthropometric model achieved a root mean squared error (RMSE) of 3.60, and mean absolute error (MAE) of 2.77 with R² 0.83 (Figure 6A). Feature importance analysis showed that predictions were driven primarily by weight, hip circumference and waist circumference, with smaller contributions from height, age and treatment status (Figure 6B). In the validation dataset (subsequent visits, n = 75), the prediction error was RMSE 4.17 (95% CI 3.48–4.94) and MAE 3.40 (95% CI 2.88–4.00) with R² 0.63 (Figure 6C). However, calibration indicated systematic miscalibration (intercept 9.37 [6.19–12.3], slope 0.681 [0.572–0.795]). Model updating by linear recalibration in the validation cohort reduced error to RMSE 3.52 and MAE 2.65 (R² unchanged), (Figure 6D). Inclusion of total testosterone as input, improved model fit in the training dataset (RMSE 3.18, MAE 2.37, R² 0.87) and modestly improved external validation performance (RMSE 3.93 [3.24–4.61], MAE 3.10 [2.55–3.67], R² 0.67) with calibration closer to ideal (intercept 7.81 [4.73–10.8], slope 0.762 [0.648–0.878]). After recalibration, error further decreased (RMSE 3.34, MAE 2.46). As an additional analysis aimed at personalized longitudinal monitoring, we anchored predictions to baseline body fat by applying a subject-specific offset derived from visit 1 (observed baseline body fat minus model-predicted baseline body fat) and adding this offset to the subsequent visit predictions, followed by linear recalibration. Using the anthropometric model, this personalized, calibrated approach achieved an RMSE of 3.00 and an MAE of 2.43 in visit 2 (R² 0.70)(Figure 6E). When total testosterone was included as an additional predictor, performance improved further, yielding an RMSE of 2.81 and an MAE of 2.27 (R² 0.74). Our models show that body fat in both U-KS and T-KS can be modelled with reasonably good precision using only anthropometric variables as input, and additional improvement of model fit if further including baseline visit offset. DISCUSSION The present study shows how pivotal adipose tissue is for metabolic health of males with KS. Metabolic dysfunction and increased autoimmunity emerged early as important clinical traits in KS 11 , and already in the late sixties, Johannes Nielsen from Aarhus University described increased incidence of diabetes and insulin resistance in men with KS 12,13 . Here we, for the first time, demonstrate that both hypogonadism and tissue specific multigenetic alterations are working in conjunction to promote metabolic dysfunction in KS. We clearly demonstrate that remission of hypogonadism through TRT does not fully resolve insulin resistance in men with KS. The finding of pervasive metabolic dysfunction in KS, irrespective of TRT, is in line with our recent national cohort-study, where TRT did not diminish the more than twofold increased risk of type 2 diabetes seen among men with KS compared with controls 2 . As such, the present study offers a link between KS specific gene expression in adipose tissue, adipose inflammation, increased insulin resistance, and disrupted circadian rhythm independently of hypogonadism, linking to real-world comorbidity risk and mortality. The findings from this unique integrated model, combining the clinical phenotype, biomarkers and genetic background, highlight the complexity of pathological mechanisms behind the distinct comorbidity patterns seen across various sex chromosome aneuploidies 14 , and the need for therapeutic approaches beyond simple replenishment of sex hormones to fully minimize cardiometabolic risk. Combined, our findings elucidate a complex pattern of TRT effects on metabolic function in KS, highlighting the intricate interaction of hypogonadism and metabolism in general, while also emphasizing the specific genetic backdrop of KS. Our data evolves the pathophysiological understanding of KS; from a condition solely related to androgen deficiency or restricted to X chromosome gene dosage effects, into a pervasive multigenetic syndrome. Intertwining of sex hormones, sex specific genetics, and metabolic function is further substantiated by the broadly observed sex specific metabolic phenotype among cis-males and cis-females 1 , and by diverging effects of hormone therapy on body composition and cardiometabolic risk markers among those receiving transgender hormone therapy 10 . Body composition was not normalized by TRT and men with KS had persistently higher body fat and lower nFFMI compared with controls, which are both associated with increased mortality 15 . As such, the loss of body fat and increase in nFFMI after TRT in this study is consistent with our previous finding of reduced mortality in T-KS compared with U-KS 9 , underscoring the importance of TRT in KS. Low HDL is commonly associated with increased cardiovascular risk, and TRT in KS was associated with a reduction in HDL. However, it is not known how specific dyslipidemic traits affect cardiovascular risk in KS. In our recent national study, TRT in KS was not associated with increased risk of major adverse cardiovascular events 9 , and as such commonly used cardiovascular risk scores might not apply directly to KS. TRT significantly impacted the adipokine profile in KS. Reflecting the changes in body fat, leptin was lowered in T-KS, but not to the level of controls. Adiponectin was also lowered, but to levels below that of the control population. The suppression of adiponectin by TRT has also been described in non-KS populations 16 . Importantly, this adipokine profile with high leptin, indicative of leptin resistance and concomitant low adiponectin is specifically associated with the highest risk of obesity, insulin resistance and type 2 diabetes, especially among individuals with higher body fat 17 . As such, the lowering of adiponectin by TRT could contribute to sustained insulin resistance in KS. Also, men with KS had higher leptin-to-adiponectin ratio than controls irrespective of TRT, corresponding with the sustained increased risk of type 2 diabetes in KS 10 , and further supporting the presence of non-hypogonadism dependent genetic risk factors. Our findings indicate that increased body fat, as in other populations, is associated with adipose tissue immune cell infiltration and inflammation, which in turn may impair insulin sensitivity 18 . In both KS and controls, higher body fat was associated with upregulated expression of innate immune and inflammatory genes in adipose tissue, gene signatures that are hallmarks of macrophage/monocyte and lymphocyte recruitment and activation, including M1-macophage polarization, and are consistent with obesity-driven immune cell infiltration 19 . The enrichment of pathways related to immune and cytokine activation further supports adiposity induced immune cell accumulation as a general consequence of adipose tissue expansion. Multiple immune cell types are known to infiltrate obese adipose tissue and propagate an inflammatory cascade that impairs insulin signalling 20 . The DNA methylation-based cell type deconvolution followed this trend, showing greater inferred NK cell and monocyte/macrophage proportions in adipose tissue from men with higher body fat. The strong correlation of immune cell fraction with insulin resistance across our cohort, highlights the fact that elevated body fat promotes adipose immune infiltration and insulin resistance. This mechanism may provide an explanation for why men with KS become insulin resistant and prone to develop metabolic syndrome. We uncovered a KS-specific genomic signature of insulin resistance strongly linked to circadian rhythm disruption (captured by Factor 6). This was driven by downregulated expression of core clock and insulin pathway genes (including PER3 , CIART , ZBTB16 , HLF , IRS2 , PDK4 , PFKFB3 ) alongside upregulation of genes involved in lipid storage, extracellular matrix remodelling and stress responses 21 . As such, intrinsic impairment of the biological circadian clock and metabolic flexibility may exacerbate insulin resistance in KS adipose tissue. Notably, circadian regulators in most peripheral tissues, including adipose, display diurnal rhythmic expression of metabolic genes, adipokines and adipocytokines. This plays a crucial role in insulin signalling and maintaining metabolic homeostasis as disruption of circadian rhythm and lowered expression of clock genes has been observed with obesity, metabolic stress, or sleep disturbances 22,23 . As demonstrated here, obesity and metabolic dysfunction are common in KS, and sleep disturbances have also been observed in KS from an early age 24,25 . Animal models have demonstrated that mutation of core clock genes leads to metabolic derangements 26 . Thus, the downregulation of circadian clock components in KS adipose tissue is mechanistically plausible as a contributor to insulin resistance, given the established link between circadian disruption and impaired insulin signalling 22 . Our analyses provided support for this, revealing significantly blunted or desynchronized circadian oscillations in KS adipose tissue. In other words, the adipose tissue of men with KS appeared to have lost the normal rhythmic gene expression patterns that help synchronize metabolic processes. High rates of insulin resistance are seen already among prepubertal boys with KS 27 . We speculate that early disruption of the circadian rhythm of peripheral tissues during childhood, driven by KS specific genetic dispositions, may be linked to both increased body fat and insulin resistance. The circadian dysregulation in KS was accompanied by signs of accelerated biological aging in the tissue (increased DNA methylation age relative to chronological age), which we recently showed was also present, albeit to a lesser degree, in blood 28 . Chronic metabolic stress and obesity have been associated with epigenetic age acceleration in metabolic tissues 29 , so the elevated “epigenetic age” of KS adipose tissue may reflect cumulative metabolic strain due to lifelong circadian disruption or metabolic imbalance. The cause of the disrupted circadian rhythm is presently not known. Men with KS also have an inherently different genetic and endocrine profile that could impact circadian regulation. Testosterone itself exhibits diurnal variation 30 , and chronic hypogonadism might blunt normal circadian metabolic cues. Also, the pervasive genome-wide changes to both the transcriptome and epigenome could impact circadian rhythm 31 . This suggests that traditional interventions (like testosterone replacement or weight loss) might need to be complemented by approaches that restore circadian alignment - for instance, pharmacological agents that target clock pathways, such as melanocortin receptor agonists, antidepressants and other new drugs currently being developed 32,33 . We provide a precise prediction of body fat among men with KS, as a simple tool for assessing cardiometabolic risk and effect of TRT. As demonstrated, the low muscle-to-fat ratio and slightly increased height in KS 34 will cause BMI to underestimate body fat, masking the underlying metabolic dysfunction. Also, BMI, WHRadjBMI, LAP, and VAI were incapable of detecting changes in body composition following TRT. We believe that direct assessment of body fat by DXA/CT or by KS-specific prediction of body fat, has potential to identify men with KS that are at an increased risk of cardiometabolic complications. We developed two prediction models that estimate body fat in KS from readily available anthropometrics. The anthropometric model achieved reasonable accuracy in external validation, supporting that clinically relevant variation in body fat can be captured without the need for repeated DXA/CT scans. Importantly, feature-importance patterns were biologically plausible, with weight, hip circumference and waist circumference contributing most to prediction, while age, height, and treatment status added smaller incremental information. This aligns with the expectation that overall and central adiposity dominate inter-individual differences in body fat, whereas treatment status likely influences body composition through pathways only partly reflected in simple anthropometrics. As an additional step to support longitudinal, patient-specific monitoring, we anchored follow-up predictions to baseline DXA, which improved prediction error. While anthropometrics explains much of the between-person variation in body fat, individuals with similar waist, hip, and weight can still differ systematically in true body composition. The baseline offset quantifies the person-specific deviation (DXA-measured body fat minus model-predicted body fat at Visit 1) and carries it forward to subsequent visits when DXA is unavailable. This converts a population-based estimate into an individualized estimate and is therefore well-suited for tracking changes over time. Despite maintained accuracy in the validation cohort, calibration analyses revealed systematic miscalibration, indicating that predicted values were on a different scale in the testing dataset (positive intercept and slope <1). This is a common finding when models are used across cohorts, reflecting differences in measurement procedures, or underlying distributions of predictors and outcomes. We therefore performed model updating by linear recalibration in the validation cohort, which improved absolute error and brought predictions closer to the observed scale. In practical terms, this suggests that the proposed tool is best implemented with local calibration when applied in new clinical settings or when measurement protocols differ. Ideally, input data should be collected prospectively from men with KS around the globe to allow development of a strong clinically applicable tool. The main strength of this study is the comprehensive, longitudinal, collective assessment of metabolic function and integration with gene expression in a large sample of men with KS. The breadth of the data set is unique and allows a systems biology approach to metabolic function in KS. Still, several limitations apply. Most prominently, all participants in the study have lived with prolonged hypogonadism prior to TRT, and thus the timing of TRT could be important. Although metabolic dysfunction is present already at a young age in KS 2 , it can be speculated that longer time spent in hypogonadism could lead to a metabolic memory legacy in metabolic tissues 35 , with permanent damage to pivotal metabolic processes that hinder subsequent reversal of metabolic dysfunction through TRT and correction of hypogonadism. Other studies are currently planned to assess the effect of TRT initiated during puberty on changes in body composition and metabolism 36 . Also, our study did not assess visceral fat directly but relies on total body fat percentage. Total fat and visceral fat are strongly associated 37 , but we might have achieved more precise estimates if assessing visceral fat directly with adjustment for total body fat - especially when wanting to use fat mass as a predictor for future cardiometabolic risk 38 . Along the same lines, the proposed models for assessing body fat in KS need to be verified in other cohorts, and ultimately to be tested against clinical cardiometabolic events. Also, most participants only had data from one visit, potentially allowing for selection bias regarding the longitudinal analyses. Our study represents a breakthrough in the understanding of metabolic dysfunction in KS demonstrating insufficient rescue of obesity and insulin sensitivity with TRT, while also identifying syndrome specific genetic markers, that are driving metabolic dysfunction independent of sex hormones. Our data calls for a paradigm shift in the clinical approach to cardiometabolic prophylactic care in KS, highlighting the need for continuous monitoring of body fat and insulin sensitivity and advancement of treatment approaches to alleviate metabolic risk. METHODS Participants and ethics We include data from three separate clinical research projects, collectively including participants from March 2002 until March 2019 8,39,40 . All studies included men with KS verified by karyotyping, and age-matched control males. All participants provided signed informed consent. Chronologically, study A was approved by the Danish Data Protection Agency and the Aarhus County Ethical Scientific Committee (# 20010155), study B by the Danish Data Protection Agency and the Region Midtjylland Ethical Committee (M-20080238) and registered with ClinicalTrials.gov (NCT00999310), and study C by the Danish Data Protection Agency (1-16-02-472-15), the Central Denmark Regional Committees on Health Research Ethics (1-10-72-131-15), and registered with ClinicalTrials.gov (NCT02526628). Study design Studies A and B were cross-sectional with a single visit, while study C had both a baseline and follow-up visit. However, as some individuals participated in more than one of the assessed studies, any individual could have data from as many as four visits combined. We thus define the number and order of visits based on the individual across all studies, with the first visit in any study assigned as Visit 1 and subsequent visits as Visit 2, 3, and 4, respectively. At each visit, men with KS were stratified by TRT status (U-KS/T-KS). Sex hormones Assays for sex hormones have varied over time. For study A, sex hormones were measured by specific radioimmunoassyas (RIA) 41 . For studies B and C, both testosterone and estradiol were measured by liquid chromatography tandem mass spectrometry, while luteinizing hormone (LH), and sex hormone binding globulin (SHBG) were quantitatively determined by immunoassays 34,42 . Glucose metabolism Data on glucose metabolism were available from studies A and C, except for fasting glucose which was also available from study B. For study A, serum insulin was determined by a commercial immunological kit and serum adiponectin was determined by a time-resolved immunofluorometric assay and leptin was determined by a commercial radioimmunoassay 39 . For study C, insulin was measured by ELISA (Mercodia, Sweden), adiponectin and leptin were measured by in house time-resolved immunofluorometric assays. The Homeostasis Model Assessment 2 (HOMA2) was calculated based on fasting glucose and insulin levels using the HOMA2 calculator macro (Version 2.2.3, Diabetes Trial Unit, University of Oxford, Oxford, United Kingdom). The calculator outputs HOMA2-%B as a marker of beta-cell function, HOMA2-%S a marker of insulin sensitivity and the reciprocal HOMA2-IR indicating level of insulin resistance. We further assessed the Triglyceride Glucose Index (TyG) as a marker of insulin resistance calculated as TyG = ln[(Triglycerides (mg/dL) * Fasting Glucose (mg/dL)) / 2] 43 . Lipids were assessed enzymatically using laboratory standard assays at Aarhus University Hospital. Body composition Total body fat percentage and lean mass were examined by DEXA scans using Hologic scanners 34,39,42 . Body mass index (BMI) was calculated as weight (kg) over height (m) squared. Waist-to-hip ratio adjusted for BMI (WHRadjBMI) was computed as the residuals from a linear regression with waist-to-hip ratio as the dependent variable and BMI, age, and age squared as covariates. The Visceral Adiposity Index (VAI) was calculated using the male sex specific formula as VAI = (Waist circumference / (39.68 + (1.88 * BMI))) * (Triglycerides/ 1.03) * (HDL / 1.31) 44 . The Lipid Accumulation Product (LAP) was calculated as LAP = (Waist Circumference - 65) * Triglycerides) 45 . The male normalized Fat-Free Mass Index (nFFMI) was calculated as nFFMI = (total lean mass(kg)/height(m 2 )+6.1*(1.8-height(m)) 46 . The genomic dataset Adipose tissue biopsies were taken in study C from the abdominal subcutaneous periumbilical region by liposuction and snap-frozen in liquid nitrogen, before DNA and RNA isolation and sequencing. The genomic dataset consisted of a subset of the full cohort, KS (n = 22) and Controls (n = 15). All DNA methylation (DNAm) and RNA sequencing (RNAseq) data from adipose tissue of the participants have been published previously 47 . RNA analysis For gene expression, fastq files underwent initial quality control using FastQC (Babraham Bioinformatics). Adaptor removal and trimming of low-quality ends were performed using Trim Galore with default settings (Babraham Bioinformatics). Transcript expression levels were quantified using Salmon 48 , with a decoy-aware transcriptome index based on the hg38 transcriptome. Transcript abundances were summarized to the gene level using the R package Tximeta 49 . Multi-Omics Factor Analysis (MOFA) of adipose tissue gene expression Unsupervised latent factor analysis of the adipose gene expression was performed using the MOFA2 framework 50 . Gene expression counts were transformed using the variance-stabilizing transformation (VST) implemented in DESeq2. The resulting VST expression matrix was used as an input for factor generation. To focus on major biological sources of variation, we restricted the analysis to the 2,000 most variable genes. A MOFA object was created with pre-processing options set using default parameters. We specified 10 latent factors to capture multiple independent axes of variation. To visualize factor-gene relationships, we generated heatmaps of expression for the most strongly weighted genes per factor. Associations between MOFA factors and clinical/biochemical traits were assessed with computed Pearson correlations between factor scores and sample-level covariates. Correlations were visualized as correlation matrices and interpreted using a nominal significance threshold of α = 0.05. To functionally characterize factors, we performed gene set enrichment analysis on MOFA feature weights using the run enrichment function (MOFA2). Enrichment was computed separately for positively and negatively weighted genes per factor. TimeTeller Circadian clock function was quantified at the sample level using TimeTeller 51 . TimeTeller was trained on a dataset from healthy volunteers, in which punch biopsies were collected every 4 h across 24 h and profiled 51 . The TimeTeller training model was fitted on the training data with inter-gene normalisation and 3 principal components, retaining the first 3 principal components of the panel expression at each circadian phase. All other parameters were kept at their recommended defaults. For the KS and control adipose gene expression data, raw gene-level counts were converted to log₂ counts per million (CPM). The expression matrix was supplied to TimeTeller and for each sample TimeTeller returned an estimated clock dysfunction score (Θ, 0–1), where higher Θ indicates poorer agreement with the healthy reference clock state - a more disrupted clock. deltaCCD Group-level disruption of the circadian clock gene network was quantified using the deltaCCD method (Hughey J, Outland E (2022). deltaccd: Quantify Rhythmic Gene Co-Expression Relative to a Reference. https://deltaccd.hugheylab.org, https://github.com/hugheylab/deltaccd). Briefly, deltaCCD compared the clock correlation distance (CCD) between groups relative to a reference clock-gene co-expression pattern. RNAseq counts from adipose tissue from KS and controls were normalised to log₂-CPM, yielding an expression matrix that included all measured genes. A circadian clock gene set was defined a priori, comprising ARNTL , CLOCK , PER1 , PER2 , PER3 , CRY1 , CRY2 , NR1D1 , NR1D2 , NPAS2 , DBP , TEF , HLF . For deltaCCD, samples were classified by genotype with controls specified as the reference group. CCDs were then calculated for each group. The ΔCCD was defined as CCD(KS) − CCD(Controls), so positive values indicate that the KS clock-gene co-expression pattern is more dissimilar from the reference than the 46,XY pattern (i.e. more disrupted). Statistical significance was assessed using permutation testing within the function calcDeltaCCD. DNA Methylation Raw intensity values for all CpG sites (Infinium MethylationEPIC) were imported and processed using the R package Minfi 52 . Cross-reactive probes and poorly performing probes, as indicated by a detection p-value < 0.01, were excluded from the analysis. The preprocess Funnorm normalization method 53 was applied to remove between-array variation inferred by control probes, followed by the conversion of methylation values to beta-values. Estimation of cell type proportions To investigate for differences in cell composition, the HEpiDISH algorithm was employed using a reference for solid tissues “ centEpiFibIC.m ” 54 and a reference that yield the proportions of 7 immune cell subtypes in addition to the solid fractions estimated “ centEpiFibIC.m ”. Normalized beta-values were used as input. Epigenetic age acceleration Using the R-package MethylClock (v1.16.0) 55 , we applied Horvath’s pan-tissue epigenetic clock, suitable for the analysis of multiple tissues, including adipose tissue 56 . Using this, we estimated the intrinsic age acceleration, defined as the residuals of regressing chronological age on epigenetic age. Statistical analysis Distribution of continuous outcomes was evaluated by quantile-quantile plots. Data are presented as mean ± SD or median (p25-p75). Baseline comparison across groups was performed by one-way ANOVA with Tukey post-hoc test or Kruskal Wallis test with Dunn post-hoc test dependent on distribution. We assessed the effects of TRT on repeated outcome measures using mixed-effects linear regression. Each participant was modeled with a random intercept to account for within-subject correlation, and robust (clustered) standard errors were used. Fixed effects included treatment status (treated vs untreated) and study visit (1–4). To account for age, we decomposed age into two components: baseline age (centered around the sample mean) to capture between-subject differences, and within-subject change in age across visits (years since baseline) to capture longitudinal aging effects. Conditional (cR 2 ) and marginal (mR 2 ) r-squared summarizing statistics from the mixed-effects model were derived as described by Nakagawa 57 . Analysis was done using Stata Now (Statacorp). Body fat was modelled in men with KS using extreme gradient boosting regression (XGBoost). The XGBoost algorithm was chosen because it performs well on structured clinical data and can flexibly capture non-linear relationships and interactions among anthropometric variables. It also includes built-in regularization and early stopping to reduce overfitting. Participants with missing body fat measurements were excluded. Treatment status was included as a predictor. Non-informative variables were removed, and the final predictor set comprised key anthropometric variables (hip, weight, waist, height, age). Total testosterone was added for a second model. Model hyperparameters were selected by random search over 40 candidate parameter sets using repeated 10-fold cross-validation (5 repeats) within the training dataset. Early stopping was applied (maximum 5000 boosting rounds; stopping after 50 rounds without improvement) to reduce overfitting. Predictive performance was evaluated on a validation cohort, using RMSE, MAE, and R². Calibration was assessed in the validation data by regressing observed body fat on predicted body fat to obtain a calibration intercept and slope. Uncertainty in RMSE/MAE and calibration parameters was quantified using nonparametric bootstrapping (2000 resamples). Feature importance was summarized using XGBoost gain, cover and frequency. Declarations DATA AVAILABILITY The clinical data reported here can be obtained by reasonable request to the authors in accordance with the General Data Protection Regulation. All sequence and methylation data from this study are deposited at the European Genome-phenome Archive (EGA) repository (ega-archive.org, EGAS00001006996, EGAS00001007020). Acknowledgements We would like to acknowledge bio analysts Lone Kvist, Maria Flink Schwartz, and Line Mentz for their assistance in obtaining clinical samples and Pamela Celis for isolation of DNA and RNA from samples. We also acknowledge the MOMA NGS Core Center, and GenomeDK. Author contributions CRediT statement Conceptualization (JJ, CHG, SC), data curation (JJ,SC), formal analysis (JJ, EH, SC), funding acquisition (JJ, CHG, SC), investigation (JJ, EH, AS, CHG, SC), methodology (JJ, EH, SC), project administration (AS, CHG, SC), resources (AS, CHG, SC), software (JJ, EH, SC), supervision (SC, CHG, JJ), visualization (JJ, EH, SC), writing – original draft (JJ, CHG, SC), writing – review & editing (JJ, EH, AS, CHG, SC). Approval of the final manuscript was granted by all authors. Funding This work was supported by Aarhus University; Novo Nordisk Foundation (NNF20OC0060610); the Independent Research Fund Denmark (2096-00165A, 0134-00130B); Sygesikringen danmark (2022-0189), the Danish Diabetes and Endocrine Academy which is funded by the Novo Nordisk Foundation, grant number NNF22SA0079901. References Mauvais-Jarvis, F. Sex differences in energy metabolism: natural selection, mechanisms and consequences. Nat Rev Nephrol 20 , 56-69 (2024). https://doi.org/10.1038/s41581-023-00781-2 Gravholt, C. H. et al. Klinefelter syndrome - integrating genetics, neuropsychology and endocrinology. Endocrine reviews (2018). https://doi.org/10.1210/er.2017-00212 Bardsley, M. Z., Falkner, B., Kowal, K. & Ross, J. L. Insulin resistance and metabolic syndrome in prepubertal boys with Klinefelter syndrome. Acta paediatrica (Oslo, Norway : 1992) 100 , 866-870 (2011). https://doi.org/10.1111/j.1651-2227.2011.02161.x Aksglaede, L., Molgaard, C., Skakkebaek, N. E. & Juul, A. Normal bone mineral content but unfavourable muscle/fat ratio in Klinefelter syndrome. Archives of disease in childhood 93 , 30-34 (2008). https://doi.org/10.1136/adc.2007.120675 Granato, S. et al. Epicardial fat: the role of testosterone and lipid metabolism in a cohort of patients with Klinefelter syndrome. Metabolism: clinical and experimental 95 , 21-26 (2019). https://doi.org/10.1016/j.metabol.2019.03.002 Demirci, I. et al. Triglyceride-glucose index levels in patients with Klinefelter syndrome and its relationship with endothelial dysfunction and insulin resistance: a cross-sectional observational study. Arch Endocrinol Metab 67 , 378-384 (2023). https://doi.org/10.20945/2359-3997000000594 Haymana, C. et al. Increased Endothelial Dysfunction and Insulin Resistance in Patients with Klinefelter Syndrome. Endocrine, metabolic & immune disorders drug targets 18 , 401-406 (2018). https://doi.org/10.2174/1871530318666180220114322 Chang, S. et al. Testosterone Replacement Therapy in Klinefelter Syndrome-Follow-up Study Associating Hemostasis and RNA Expression. The Journal of clinical endocrinology and metabolism 109 , 978-991 (2024). https://doi.org/10.1210/clinem/dgad658 Chang, S., Pedersen, L., Skakkebæk, A., Berglund, A. & Gravholt, C. H. Cardiovascular risk and mortality in men receiving testosterone replacement therapy for Klinefelter syndrome in Denmark: a retrospective cohort study. The Lancet Regional Health - Europe 51 , 101230 (2025). https://doi.org/10.1016/j.lanepe.2025.101230 Chang, S., Pedersen, L., Skakkebæk, A., Berglund, A. & Gravholt, C. H. Metabolism and type 2 diabetes across life stages in Klinefelter syndrome: a population-based cohort study. The Journal of clinical endocrinology and metabolism (2025). https://doi.org/10.1210/clinem/dgaf619 Burch, P. R. Klinefelter's syndrome, dizygotic twinning and diabetes mellitus. Nature 221 , 175-177 (1969). https://doi.org/10.1038/221175a0 Menzinger, G. et al. KLINEFELTER'S SYNDROME AND DIABETES MELLITUS. The Lancet 288 , 747-748 (1966). https://doi.org/https://doi.org/10.1016/S0140-6736(66)93008-X Nielsen, J., Johansen, K. & Yde, H. Frequency of diabetes mellitus in patients with Klinefelter's syndrome of different chromosome constitutions and the XYY syndrome. Plasma insulin and growth hormone level after a glucose load. The Journal of clinical endocrinology and metabolism 29 , 1062-1073 (1969). https://doi.org/10.1210/jcem-29-8-1062 Berglund, A., Chang, S., Lind-Holst, M., Stochholm, K. & Gravholt, C. H. The epidemiology of disorders of sex development. Best Pract Res Clin Endocrinol Metab , 102002 (2025). https://doi.org/10.1016/j.beem.2025.102002 Sedlmeier, A. M. et al. Relation of body fat mass and fat-free mass to total mortality: results from 7 prospective cohort studies. Am J Clin Nutr 113 , 639-646 (2021). https://doi.org/10.1093/ajcn/nqaa339 Page, S. T. et al. Testosterone administration suppresses adiponectin levels in men. Journal of andrology 26 , 85-92 (2005). Wannamethee, S. G. et al. Adipokines and risk of type 2 diabetes in older men. Diabetes care 30 , 1200-1205 (2007). https://doi.org/10.2337/dc06-2416 Rohm, T. V., Meier, D. T., Olefsky, J. M. & Donath, M. Y. Inflammation in obesity, diabetes, and related disorders. Immunity 55 , 31-55 (2022). https://doi.org/10.1016/j.immuni.2021.12.013 Turner, L., Wanasinghe, A. I., Brunori, P. & Santosa, S. Is Adipose Tissue Inflammation the Culprit of Obesity-Associated Comorbidities? Obes Rev 26 , e13956 (2025). https://doi.org/10.1111/obr.13956 Guzik, T. J., Skiba, D. S., Touyz, R. M. & Harrison, D. G. The role of infiltrating immune cells in dysfunctional adipose tissue. Cardiovasc Res 113 , 1009-1023 (2017). https://doi.org/10.1093/cvr/cvx108 Takahashi, J. S. Transcriptional architecture of the mammalian circadian clock. Nat Rev Genet 18 , 164-179 (2017). https://doi.org/10.1038/nrg.2016.150 Hepler, C. & Bass, J. Circadian mechanisms in adipose tissue bioenergetics and plasticity. Genes Dev 37 , 454-473 (2023). https://doi.org/10.1101/gad.350759.123 Stenvers, D. J., Scheer, F., Schrauwen, P., la Fleur, S. E. & Kalsbeek, A. Circadian clocks and insulin resistance. Nat Rev Endocrinol 15 , 75-89 (2019). https://doi.org/10.1038/s41574-018-0122-1 Fjermestad, K. W., Finnbakk, R. R., Solbakk, A. K., Gravholt, C. H. & Huster, R. J. Subjective versus objective sleep in men with Klinefelter syndrome. Orphanet J Rare Dis 18 , 260 (2023). https://doi.org/10.1186/s13023-023-02822-2 Paparella, R. et al. Evaluation of sleep disorders in children and adolescents affected by Klinefelter syndrome. Eur J Pediatr 184 , 129 (2025). https://doi.org/10.1007/s00431-024-05944-2 Marcheva, B. et al. Disruption of the clock components CLOCK and BMAL1 leads to hypoinsulinaemia and diabetes. Nature 466 , 627-631 (2010). https://doi.org/10.1038/nature09253 Bardsley, M. Z., Falkner, B., Kowal, K. & Ross, J. L. Insulin resistance and metabolic syndrome in prepubertal boys with Klinefelter syndrome. Acta Paediatr. 100 , 866-870 (2011). Hasselholm, E. B., Just, J., Chang, S., Skakkebæk, A. & Gravholt, C. H. Epigenetic age acceleration in Turner and Klinefelter syndrome: Correlations with clinical aging markers. Clin Epigenetics 17 , 147 (2025). https://doi.org/10.1186/s13148-025-01963-4 de Toro-Martín, J. et al. Body mass index is associated with epigenetic age acceleration in the visceral adipose tissue of subjects with severe obesity. Clin Epigenetics 11 , 172 (2019). https://doi.org/10.1186/s13148-019-0754-6 Bremner, W. J., Vitiello, M. V. & Prinz, P. N. Loss of circadian rhythmicity in blood testosterone levels with aging in normal men. J Clin.Endocrinol Metab. 56 , 1278-1281 (1983). Skakkebaek, A. et al. DNA hypermethylation and differential gene expression associated with Klinefelter syndrome. Sci. Rep. 8 , 13740-31780 (2018). Bhatnagar, A., Murray, G. & Ray, S. Circadian biology to advance therapeutics for mood disorders. Trends Pharmacol Sci 44 , 689-704 (2023). https://doi.org/10.1016/j.tips.2023.07.008 Sulli, G., Manoogian, E. N. C., Taub, P. R. & Panda, S. Training the Circadian Clock, Clocking the Drugs, and Drugging the Clock to Prevent, Manage, and Treat Chronic Diseases. Trends Pharmacol Sci 39 , 812-827 (2018). https://doi.org/10.1016/j.tips.2018.07.003 Chang, S. et al. Anthropometry in Klinefelter Syndrome - Multifactorial Influences Due to CAG Length, Testosterone Treatment and Possibly Intrauterine Hypogonadism. Journal of Clinical Endocrinology & Metabolism 100 , E508-E517 (2015). https://doi.org/10.1210/jc.2014-2834 Dong, H. et al. Metabolic memory: mechanisms and diseases. Signal Transduct Target Ther 9 , 38 (2024). https://doi.org/10.1038/s41392-024-01755-x Caspersen, I. D. et al. Effect of testosterone treatment during puberty in boys with Klinefelter syndrome (The TIPY Study): protocol for a nationwide randomised, double-blinded, placebo-controlled study. BMJ open 15 , e095628 (2025). https://doi.org/10.1136/bmjopen-2024-095628 Martin, A. D., Janssens, V., Caboor, D., Clarys, J. P. & Marfell-Jones, M. J. Relationships between visceral, trunk and whole-body adipose tissue weights by cadaver dissection. Annals of human biology 30 , 668-677 (2003). https://doi.org/10.1080/03014460310001599590 Gast, K. B. et al. Individual contributions of visceral fat and total body fat to subclinical atherosclerosis: The NEO study. Atherosclerosis 241 , 547-554 (2015). https://doi.org/10.1016/j.atherosclerosis.2015.05.026 Bojesen, A. et al. The metabolic syndrome is frequent in Klinefelter's syndrome and is associated with abdominal obesity and hypogonadism. Diabetes care 29 , 1591-1598 (2006). https://doi.org/10.2337/dc06-0145 Skakkebaek, A. et al. Neuropsychology and brain morphology in Klinefelter syndrome - the impact of genetics. Andrology (2014). https://doi.org/10.1111/j.2047-2927.2014.00229.x Lykkesfeldt, G. et al. Abnormal androgen and oestrogen metabolism in men with steroid sulphatase deficiency and recessive X-linked ichthyosis. Clinical endocrinology 23 , 385-393 (1985). https://doi.org/10.1111/j.1365-2265.1985.tb01096.x Chang, S. et al. Testosterone treatment and association with thrombin generation and coagulation inhibition in Klinefelter syndrome: A cross-sectional study. Thrombosis research 182 , 175-181 (2019). https://doi.org/10.1016/j.thromres.2019.08.011 Guerrero-Romero, F. et al. The product of triglycerides and glucose, a simple measure of insulin sensitivity. Comparison with the euglycemic-hyperinsulinemic clamp. The Journal of clinical endocrinology and metabolism 95 , 3347-3351 (2010). https://doi.org/10.1210/jc.2010-0288 Amato, M. C. et al. Visceral Adiposity Index: a reliable indicator of visceral fat function associated with cardiometabolic risk. Diabetes care 33 , 920-922 (2010). https://doi.org/10.2337/dc09-1825 Kahn, H. S. The lipid accumulation product is better than BMI for identifying diabetes: a population-based comparison. Diabetes care 29 , 151-153 (2006). https://doi.org/10.2337/diacare.29.1.151 Kouri, E. M., Pope, H. G., Jr., Katz, D. L. & Oliva, P. Fat-free mass index in users and nonusers of anabolic-androgenic steroids. Clinical journal of sport medicine : official journal of the Canadian Academy of Sport Medicine 5 , 223-228 (1995). https://doi.org/10.1097/00042752-199510000-00003 Viuff, M. et al. X chromosome dosage and the genetic impact across human tissues. Genome Med 15 , 21 (2023). https://doi.org/10.1186/s13073-023-01169-4 Patro, R., Duggal, G., Love, M. I., Irizarry, R. A. & Kingsford, C. Salmon provides fast and bias-aware quantification of transcript expression. Nat Methods 14 , 417-419 (2017). https://doi.org/10.1038/nmeth.4197 Love, M. I. et al. Tximeta: Reference sequence checksums for provenance identification in RNA-seq. PLoS Comput Biol 16 , e1007664 (2020). https://doi.org/10.1371/journal.pcbi.1007664 Argelaguet, R. et al. Multi-Omics Factor Analysis-a framework for unsupervised integration of multi-omics data sets. Mol Syst Biol 14 , e8124 (2018). https://doi.org/10.15252/msb.20178124 Vlachou, D. et al. TimeTeller: A tool to probe the circadian clock as a multigene dynamical system. PLoS Comput Biol 20 , e1011779 (2024). https://doi.org/10.1371/journal.pcbi.1011779 Fortin, J. P., Triche, T. J., Jr. & Hansen, K. D. Preprocessing, normalization and integration of the Illumina HumanMethylationEPIC array with minfi. Bioinformatics 33 , 558-560 (2017). https://doi.org/10.1093/bioinformatics/btw691 Fortin, J. P. et al. Functional normalization of 450k methylation array data improves replication in large cancer studies. Genome Biol 15 , 503 (2014). https://doi.org/10.1186/s13059-014-0503-2 Salas, L. A. et al. Enhanced cell deconvolution of peripheral blood using DNA methylation for high-resolution immune profiling. Nat Commun 13 , 761 (2022). https://doi.org/10.1038/s41467-021-27864-7 Pelegí-Sisó, D., de Prado, P., Ronkainen, J., Bustamante, M. & González, J. R. methylclock: a Bioconductor package to estimate DNA methylation age. Bioinformatics 37 , 1759-1760 (2021). https://doi.org/10.1093/bioinformatics/btaa825 Horvath, S. DNA methylation age of human tissues and cell types. Genome Biol. 14 , R115 (2013). Nakagawa, S. & Schielzeth, H. A general and simple method for obtaining R2 from generalized linear mixed-effects models. Methods in Ecology and Evolution 4 , 133-142 (2013). https://doi.org/https://doi.org/10.1111/j.2041-210x.2012.00261.x Additional Declarations There is NO Competing Interest. Supplementary Files ExtendedDataPDF.pdf Extended Data Cite Share Download PDF Status: Under Review Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8850956","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":589562139,"identity":"c870db2d-e7a5-4056-9d56-795b66bb460e","order_by":0,"name":"Jesper Just","email":"","orcid":"https://orcid.org/0000-0002-3825-0000","institution":"Aarhus University","correspondingAuthor":false,"prefix":"","firstName":"Jesper","middleName":"","lastName":"Just","suffix":""},{"id":589562140,"identity":"6c1015af-33b5-4c60-a28c-fb9d4a95be94","order_by":1,"name":"Emma Hasselholm","email":"","orcid":"","institution":"Aarhus University","correspondingAuthor":false,"prefix":"","firstName":"Emma","middleName":"","lastName":"Hasselholm","suffix":""},{"id":589562141,"identity":"3e8cacf0-2041-4fc3-a12b-f0cf368b7f17","order_by":2,"name":"Anne Skakkebæk","email":"","orcid":"","institution":"Aarhus University Hospital","correspondingAuthor":false,"prefix":"","firstName":"Anne","middleName":"","lastName":"Skakkebæk","suffix":""},{"id":589562142,"identity":"879adb56-b057-4dbf-a500-567f12129644","order_by":3,"name":"Claus Gravholt","email":"","orcid":"","institution":"Aarhus University Hospital","correspondingAuthor":false,"prefix":"","firstName":"Claus","middleName":"","lastName":"Gravholt","suffix":""},{"id":589562138,"identity":"c25ebf48-0195-4296-b036-ea08a06616b4","order_by":4,"name":"Simon Chang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA8klEQVRIiWNgGAWjYDACZgiVAMSMDxgKJCBcngPEaWE2YDAgRgsDQgubBIMBA2Et8u3ciY8LGOzy+Ge3X6v4YWCRx8B+xoDhzRncWgwO8242nsGQXCxx50zZzR4DiWIGnhwDxjk38Ghh5t0mzcNwILHhRk7abaBfEhsYcgyYeT7gcVgzVMt8oJZisBb+N/i1MByGatlwI/0YM1iLBMgWfA4D+YXHIDlx440cZkmgXxLbJJ4VHJyDx/vy/Wc3PuapsEucdyP94YcfFXWJ/fzJGx+8OYbHYRC7QAQPJFLYgPgAIQ1QwP6ASIWjYBSMglEw0gAAcQZNf17chjoAAAAASUVORK5CYII=","orcid":"https://orcid.org/0000-0003-1130-3659","institution":"Aarhus University Hospital","correspondingAuthor":true,"prefix":"","firstName":"Simon","middleName":"","lastName":"Chang","suffix":""}],"badges":[],"createdAt":"2026-02-11 11:19:13","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8850956/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8850956/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":102559169,"identity":"e21461c3-26e0-4055-bf0c-c448bbfc0fc0","added_by":"auto","created_at":"2026-02-13 03:27:20","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":34941,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eParticipant flow across visits.\u003c/strong\u003e Number of participants (n) are given at each visit and for those men with KS changing treatment status between visits. Mean ± SD of age in years is also given at each visit. U-KS: Untreated Klinefelter syndrome, T-KS: Testosterone treated Klinefelter syndrome.\u003c/p\u003e","description":"","filename":"Figure1Patientflow.png","url":"https://assets-eu.researchsquare.com/files/rs-8850956/v1/f92dd8e42e92da3ee39698fa.png"},{"id":102747176,"identity":"c6f3dae6-2aac-4369-bf43-e71b938e9c15","added_by":"auto","created_at":"2026-02-16 09:04:07","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":79335,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eBody composition and association with BMI and insulin sensitivity.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eUpper panel: Body composition in KS. Change (%) in body fat percentage and lean mass (kg) in paired data available for 21 men with Klinefelter syndrome following initiation of testosterone replacement therapy (TRT). Middle panel: Association between body mass index (BMI) and body fat percentage in untreated KS (U-KS), treated KS (T-KS), and controls. Lines represent linear fit of predicted values based on mixed-effects modelling allowing for assessment of repeated measures. Dashed lines represent the threshold values for obesity based on either BMI or body fat. Lower panel: Association between body fat percentage and HOMA2 %S in U-KS, T-KS and controls. Scatter plot of observations with linear fit of predicted values derived from mixed model regression accounting for repeated measures.mR\u003csup\u003e2\u003c/sup\u003e; marginal R\u003csup\u003e2\u003c/sup\u003e.\u003c/p\u003e","description":"","filename":"Fig2combined.png","url":"https://assets-eu.researchsquare.com/files/rs-8850956/v1/1f62ffad84462654a849a155.png"},{"id":102746876,"identity":"2d0db7c9-69b2-4bbf-b4ef-990cc1ce1464","added_by":"auto","created_at":"2026-02-16 09:02:28","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":88881,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eAdipose tissue gene expression factors associated with body fat and insulin resistance in Klinefelter syndrome\u003c/strong\u003e. (A) Heatmap of Pearson correlation coefficients (R) between MOFA latent factors 1 and 6 and clinical/metabolic traits (KS, testosterone, LH, body fat, leptin, adiponectin, HOMA2 %B, HOMA2 %S, HOMA2 %IR) in men with KS on TRT (T-KS, n = 22) and controls (n = 15). Non-significant correlations (p ≥ 0.05) are labelled NS. (B–C) Boxplots showing factor 1 (B) and factor 6 (C) loadings in controls and T-KS (p values from Wilcoxon rank-sum tests). (D–F) Scatter plots of factor 1 loadings versus body fat (D), insulin sensitivity (HOMA2 %S) (E), and β-cell function (HOMA2 %B) (F). (G–I) Scatter plots of factor 6 loadings versus body fat (G), HOMA2-%S (H), and HOMA2-%B (I). Each point represents an individual (controls, grey triangles; T-KS, red circles); lines indicate linear regression fits with corresponding Pearson R and p values shown in each panel.\u003c/p\u003e","description":"","filename":"F1.png","url":"https://assets-eu.researchsquare.com/files/rs-8850956/v1/0427e4f2532c138748433908.png"},{"id":102559170,"identity":"9775814a-9a17-4772-b5bf-70bf0768ec5b","added_by":"auto","created_at":"2026-02-13 03:27:20","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":130222,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eInflammatory/immune gene signature (MOFA factor 1) and inferred immune cell infiltration in adipose tissue from men with KS\u003c/strong\u003e. (A) Heatmap of the top weighted genes contributing to MOFA factor 1 in adipose tissue from men with KS (T-KS, n = 22) and 46,XY controls (n = 15). (B) Gene ontology enrichment analysis of the positively weighted factor 1 genes, showing the most significantly enriched immune- and inflammation-related biological process terms. Bars indicate −log₁₀(p value). (C) Estimated total immune cell proportion (summed immune fractions) in adipose tissue inferred from DNA methylation deconvolution (Wilcoxon rank-sum p value). (D–F) Associations between estimated immune cell proportion and body fat percentage (D), insulin sensitivity (HOMA2-%S) (E), and β-cell function (HOMA2-%B) (F). Each point represents an individual (controls, grey; T-KS, red); lines indicate linear regression fits with corresponding Pearson correlation coefficients (R) and p values shown in each panel.\u003c/p\u003e","description":"","filename":"F2.png","url":"https://assets-eu.researchsquare.com/files/rs-8850956/v1/532a7628fb405047f153e7e4.png"},{"id":102559175,"identity":"71814cea-9bb7-4880-a928-42ec8abe6a88","added_by":"auto","created_at":"2026-02-13 03:27:20","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":139089,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eKS-specific circadian/insulin-resistance gene signature (MOFA factor 6) in adipose tissue with disrupted circardian clock function and increased epigenetic age acceleration\u003c/strong\u003e. (A) Heatmap of the top positively (red) and negatively (blue) weighted genes contributing to MOFA factor 6 in adipose tissue from men with KS (T-KS, n = 22) and 46,XY controls (n = 15). (B) Gene ontology enrichment analysis of factor 6, showing the most significantly enriched biological process terms for genes downregulated in T-KS relative to controls, highlighting circadian regulation and insulin/glucose metabolic processes. Points indicate −log₁₀(p value). (C) Boxplot of the TimeTeller circadian clock disruption score (Θ) in adipose tissue from controls and T-KS; higher Θ indicates greater deviation from a healthy reference clock; p value from Wilcoxon rank-sum test. (D,E) Mean expression of circadian core activators and repressors. (F) Boxplot of intrinsic epigenetic age acceleration (Horvath IEAA) estimated from adipose DNA methylation in controls and T-KS; p value from Wilcoxon rank-sum test.\u003c/p\u003e","description":"","filename":"F3.png","url":"https://assets-eu.researchsquare.com/files/rs-8850956/v1/aa95a1e5980623ed4275668a.png"},{"id":102559172,"identity":"ed99a489-19f7-4be8-ba6b-2a76e84ff51f","added_by":"auto","created_at":"2026-02-13 03:27:20","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":56754,"visible":true,"origin":"","legend":"\u003cp\u003ePrediction model of body fat in Klinefelter syndrome. (A) Agreement between predicted and DXA-measured body fat in the development dataset (visit 1) for the anthropometric model; dashed red line indicates perfect agreement (line of equality), and the blue line shows the linear calibration fit (observed vs predicted). (B) Relative feature importance (gain) for the anthropometric model. (C) External validation in subsequent visits before recalibration, showing systematic miscalibration. (D) Validation results after model updating by linear recalibration, improving agreement on the observed scale. (E) Personalized longitudinal monitoring: predictions anchored to individual baseline DXA (visit 1) using a subject-specific offset and subsequently recalibrated, yielding improved follow-up accuracy.\u003c/p\u003e","description":"","filename":"F4.png","url":"https://assets-eu.researchsquare.com/files/rs-8850956/v1/8ea4bfdec639a49845ea0702.png"},{"id":107479640,"identity":"69cd8a1c-0b65-4b71-90ca-2dea914f7bb1","added_by":"auto","created_at":"2026-04-22 01:38:12","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1076060,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8850956/v1/07f4d607-866d-4912-9ecd-994cd7f765fd.pdf"},{"id":102747376,"identity":"545a9a38-6ea5-4222-95ec-b0ce6816e7bc","added_by":"auto","created_at":"2026-02-16 09:04:40","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":274197,"visible":true,"origin":"","legend":"Extended Data","description":"","filename":"ExtendedDataPDF.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8850956/v1/cab810075e293d53415d6048.pdf"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"Hypogonadism, adipose tissue inflammation, and adipose tissue circadian clock disruption promote metabolic dysfunction in Klinefelter syndrome","fulltext":[{"header":"INTRODUCTION","content":"\u003cp\u003eGenetic sex and sex hormones have profound impact on adipose tissue deposition and metabolic function \u003csup\u003e1\u003c/sup\u003e. Sex chromosome aneuploidies, including Klinefelter syndrome (KS; 47,XXY), provide a unique integrated human model to investigate metabolic function in the context of altered sex chromosome dosage, hypogonadism, and hormone replacement therapy. Metabolic dysfunction in KS \u0026nbsp;is prominent already at an early age\u003csup\u003e2\u003c/sup\u003e, with accumulation of body fat, dyslipidemia, insulin resistance and the metabolic syndrome present even in pre-pubertal boys with KS\u003csup\u003e3,4\u003c/sup\u003e. KS is characterized by hypergonadotropic hypogonadism, and metabolic dysfunction is aggravated after puberty, when hypogonadism contributes to a vicious metabolic cycle and further promote hyperinsulinemia, insulin resistance and accumulation of visceral fat\u003csup\u003e2,5\u003c/sup\u003e. As a result, rates of metabolic syndrome and type 2 diabetes are increased in KS\u003csup\u003e2\u003c/sup\u003e. In our most recent national cohort study, we demonstrated a more than 2.5-fold increased risk of type 2 diabetes in men with KS corresponding to a cumulative risk at around 20%\u003csup\u003e2,6,7\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eTestosterone replacement therapy (TRT) in KS aims to abolish the negative consequences of hypogonadism, but the effects of TRT on metabolic function and the cardiometabolic profile in KS have not been clarified\u003csup\u003e2\u003c/sup\u003e. TRT has been shown to reduce body fat in KS\u003csup\u003e2,5,8\u003c/sup\u003e, but TRT is also associated with reduced high-density lipoprotein (HDL) leading to higher predicted coronary risk\u003csup\u003e8\u003c/sup\u003e. In addition, available studies find little or no effect of TRT on insulin resistance in KS\u003csup\u003e2\u003c/sup\u003e. We recently demonstrated that despite TRT leading to an almost halving of mortality risk\u003csup\u003e9\u003c/sup\u003e, this was not reflected by a comparable reduction in metabolic risk, and TRT did not alleviate the risk of type 2 diabetes in KS\u003csup\u003e10\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eAs such, it has become increasingly clear that metabolic dysfunction in KS is not solely a result of hypogonadism, but that individuals with KS are likely genetically predisposed to metabolic dysfunction irrespective of hypogonadism. It is also becoming clear that metabolic dysfunction in KS has a pivotal impact on the excess comorbidity risk seen with the syndrome; cardiovascular disease, thrombosis, steatosis of the liver, inflammatory and autoimmune disease among others\u003csup\u003e2\u003c/sup\u003e. To provide better care for men with KS, we need to improve our understanding of independent and specific contributions of hypogonadism and genetics to metabolic dysfunction in KS. Further, investigating KS as a human genetic model of metabolic dysfunction and hypogonadism could lead to new insights with general applicability.\u003c/p\u003e\n\u003cp\u003eHere we present a large, comprehensive, and longitudinal assessment of metabolic function and body composition in men with KS stratified by TRT as either untreated (U-KS) or treated (T-KS). We describe the effect of TRT on a broad selection of metabolic risk markers and highlight areas of metabolic dysfunction in KS insensitive to TRT, implicating the extra X chromosome independent of hypogonadism. To elucidate the specific genetic imprint of KS, we investigate associations between metabolic function and the transcriptomics of adipose tissue, for the first time exposing a disrupted diurnal rhythm. We further evaluate the appliance of several metrics for prediction of body fat and present KS specific prediction models of body fat to suggest a qualitative tool for clinical monitoring of cardiometabolic risk in KS patients.\u0026nbsp;\u003c/p\u003e"},{"header":"RESULTS","content":"\u003ch2\u003eParticipants\u0026nbsp;\u003c/h2\u003e\n\u003cp\u003eA total of 149 unique men with KS and 178 controls were included at Visit 1 (Table 1, Fig. 1). Fifty-four men with KS had more than one visit, with five men still not receiving TRT at Visit 2. A total of 26 men with KS, who were untreated at Visit 1, initiated TRT at a later visit; 24 between their first and second visit, and the remaining two between the second and third visit (Figure 1). Mean age at Visit 1 was approximately 37 years in all groups, with around 2.5 years between visits (Fig. 1).\u0026nbsp;\u003c/p\u003e\n\u003ch2\u003eSex hormones\u0026nbsp;\u003c/h2\u003e\n\u003cp\u003eAs expected, total testosterone was lowest in U-KS, and TRT was associated with a 60% increase in testosterone levels among KS males (Table 1). Hypogonadism in U-KS was further substantiated by increased levels of LH, that were reduced to about a quarter with TRT. However, LH levels in T-KS were still higher than in controls. Also, T-KS had higher levels of estradiol and lower SHBG compared with U-KS and controls, corresponding to an almost 30% mean increase in 17\u0026beta;-estradiol and 20% mean reduction in SHBG following TRT (Table 1).\u003c/p\u003e\n\u003ch2\u003eBody composition\u0026nbsp;\u003c/h2\u003e\n\u003cp\u003eTotal body weight was lower in controls compared with the KS groups, but with no difference in BMI across the groups. Neither weight nor BMI was affected by TRT (Table 1). \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAt Visit 1, body fat percentage was different between all groups, with the highest body fat observed in U-KS and the lowest in controls. Across the complete data set, TRT was associated with an approximate 15% reduction of body fat, corresponding to (Table 1, Figure 2). Conversely, lean mass at Visit 1 was lowest in U-KS, and across the entire data set we saw a mean increase of 2.6 kg of lean mass after TRT in KS (Table 1, Figure 2). This was reflected by an increase in nFFMI with TRT, but still, nFFMI in both U-KS and T-KS was reduced compared with controls. At Visit 1, an average (\u0026gt;19 kg/m\u003csup\u003e2\u003c/sup\u003e) or above average (\u0026gt;21 kg/m\u003csup\u003e2\u003c/sup\u003e) nFFMI was found in 56% of controls, but only in 32% of U-KS and 36% of T-KS, demonstrating the relative lower lean mass even with TRT in KS. Also, 48% of U-KS and 39% of T-KS had a low (\u0026lt;17.6 kg/m\u003csup\u003e2\u003c/sup\u003e) or severely low (\u0026lt;16 kg/m\u003csup\u003e2\u003c/sup\u003e) nFFMI compared to only 17% of controls indicating a high risk of sarcopenia.\u003c/p\u003e\n\u003cp\u003eBMI was associated with body fat when adjusting for the interaction between grouping and BMI (mR\u003csup\u003e2\u003c/sup\u003e=0.65, p\u0026lt;0.001). From this model, at any given BMI in the clinically most relevant range (25-35 kg/m\u003csup\u003e2\u003c/sup\u003e), body fat was higher in U-KS compared with the other groups, and higher in T-KS compared with controls (Fig. 2). We then classified our observations according to commonly agreed thresholds for obesity based on either BMI (\u0026gt;30 kg/m\u003csup\u003e2\u003c/sup\u003e) or body fat (\u0026gt;25%). Of those with a BMI below the threshold for obesity, 44.0% of men with KS, but only 19.4% of controls, demonstrated body fat above the threshold for obesity, demonstrating a clear incongruence between interpreting BMI as a marker of body composition in KS versus controls (Figure 2).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWe further considered other commonly applied surrogate metrics for body fat. Numerically, mean WHRadjBMI was highest in U-KS and lowest in controls, but with very large variability and WHRadjBMI was not different between groups. Both VAI and LAP were elevated in the KS groups compared with controls, but with no effect of TRT. As such, neither of these metrics could detect changes in measured body fat with TRT in KS. Correspondingly, WHRadjBMI was not associated with body fat in KS (p=0.85), and neither was LAP (p=0.17) while VAI was only very weakly associated with body fat (\u0026beta;(95% CI); 2.1 (1.6 ; 2.6), p\u0026lt;0.001, mR\u003csup\u003e2\u003c/sup\u003e=0.05).\u003c/p\u003e\n\u003ch2\u003eLipids\u003c/h2\u003e\n\u003cp\u003eLDL cholesterol levels were not different between groups at Visit 1 and there was no effect of TRT on LDL levels (Table 1). However, levels of HDL were different between all groups at Visit 1 with the highest values in controls and lowest in T-KS, and with a corresponding 0.2 mmol reduction in HDL after TRT (Table 1). Similarly, triglyceride levels were highest in T-KS and lowest in controls, but with no apparent effect of TRT on triglycerides levels.\u003c/p\u003e\n\u003ch2\u003eGlucose metabolism\u003c/h2\u003e\n\u003cp\u003eFasting glucose was higher in U-KS compared with both T-KS and controls. Insulin was higher in both U-KS and T-KS compared with controls, which was converted into higher HOMA2 %B and HOMA2-IR in both KS groups compared with controls (Table 1). In mixed model regression, TRT was not associated with changes in either glucose, insulin, or any of the HOMA2 derivates. We further evaluated TyG as a surrogate marker of insulin resistance. TyG was increased in both KS groups, with no effect of TRT and a weak association with HOMA2-IR (\u0026beta; (95% CI) 0.74 (0.38; 1.10), p\u0026lt;0.001, mR\u003csup\u003e2\u003c/sup\u003e=0.19).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWe assessed levels of leptin and adiponectin; metabolically active hormones produced by fat cells. The highest levels of leptin were seen in U-KS and the lowest levels in controls, with TRT being associated with a 43% reduction in leptin levels in KS (Table 1). Adiponectin levels were lowest in T-KS and highest in U-KS with a 21% reduction in adiponectin levels following TRT. Also, leptin-to-adiponectin ratio was increased in both KS groups compared with controls at Visit 1 with no effect of TRT.\u003c/p\u003e\n\u003cp\u003eApplying a mixed linear regression model as described in relation to TRT, log-transformed HOMA2 %S was tightly and inversely associated to body fat adjusting for group, group-group interaction, visit, age and change in age between visits (\u0026beta;(95% CI), p; -0,04 (-0.5 ; -0,3), p\u0026lt;0.001), Fig. 2) with the complete model explaining 69% of the variability in HOMA2 %S (cR\u003csup\u003e2\u003c/sup\u003e=0.69). In this model, body fat alone explained 25% of the variability in HOMA2 %S (mR\u003csup\u003e2\u003c/sup\u003e\u003csub\u003ePartial for body fat\u003c/sub\u003e= mR\u003csup\u003e2\u003c/sup\u003e\u003csub\u003efull model\u003c/sub\u003e- mR\u003csup\u003e2\u003c/sup\u003e\u003csub\u003ewithout body fat\u003c/sub\u003e= 0.40-0.15=0.25). Body fat was also inversely associated with HOMA2 %S in each of the inclusion groups separately and without adjustment for age, but with a better fit in T-KS compared with U-KS (p\u0026le;0.002 for all, Figure 2).\u003c/p\u003e\n\u003ch2\u003eAssessment of genetic factors inducing insulin resistance in KS\u003c/h2\u003e\n\u003cp\u003eSince TRT did not completely normalise body fat, nor alleviate insulin resistance in KS, we investigated the genomic underpinnings of adipose tissue in KS compared with healthy controls to identify KS-specific gene expression patterns that could explain differences in body fat and insulin resistance.\u003c/p\u003e\n\u003cp\u003eWe performed unsupervised multi-factor analysis (MOFA) and constructed 10 latent gene expression factors that captured most of the variance in the dataset (Extended Data Figure 1). We associated subject-specific factor weights for each factor (i.e. the score indicating how strongly each subject shows that gene expression pattern) with core clinical traits (testosterone, LH, body fat, leptin, adiponectin, HOMA2%B, HOMA2%S, HOMA2%IR), (Figure 3A, Extended Data Figure 2). Factor 1 was strongly associated with insulin resistance but was not KS-specific (KS and controls had similar subject-specific factor 1 weights) and was associated with body fat in both KS and controls (Figure 3 D-F). In contrast, factor 6 was both KS-specific and associated with insulin resistance (HOMA2S, HOMA2IR), independent of body fat, pointing towards an inherent genomic feature of KS (Figure 3 G-I).\u003c/p\u003e\n\u003cp\u003eGenes with the largest absolute weights for factor 1 were predominantly inflammation-related, and their expression increased with higher factor 1 scores. These included several canonical mediators of innate immunity and inflammatory activation; complement and chemokine signalling components (\u003cem\u003eC5AR1\u003c/em\u003e, \u003cem\u003eCCR1\u003c/em\u003e, \u003cem\u003eCXCL16\u003c/em\u003e), activating Fc receptors (\u003cem\u003eFCGR3A\u003c/em\u003e, \u003cem\u003eFCGR2A\u003c/em\u003e), pro-inflammatory integrins (\u003cem\u003eITGAX\u003c/em\u003e/\u003cem\u003eITGB2\u003c/em\u003e), and key mediators of tissue remodelling and lipid-driven inflammation (\u003cem\u003eSPP1\u003c/em\u003e/osteopontin, \u003cem\u003eMMP9\u003c/em\u003e, \u003cem\u003ePLA2G7\u003c/em\u003e, \u003cem\u003eCHI3L1\u003c/em\u003e) (Figure 4A). Enrichment analysis supported this, with the top enriched pathways being related to immune and inflammatory responses, lymphocyte activation, and cytokine production (Figure 4B). Conversely, downregulated genes were involved in processes such as amino acid metabolism, lipid homeostasis, and responses to glucose and insulin (Extended Data Figure 3).\u003c/p\u003e\n\u003cp\u003eTo further validate these findings, we used DNA methylation (DNAm) data obtained from the same adipose tissue biopsies to infer immune cell infiltration. In line with factor 1 weights, we observed a non-significant increase in immune cell proportions in men with KS, mainly driven by higher inferred levels of NK cells and macrophages/monocytes (Figure 4C, Extended Data Figure \u0026nbsp;4). Immune cell fraction was strongly associated with both body fat and insulin resistance in both KS and controls (Figure 4D-F), indicating that increased body fat percentage is universally associated with immune cell infiltration and activation in adipose tissue.\u003c/p\u003e\n\u003cp\u003eFor the KS-specific factor 6, KS had markedly lower factor weights than controls, and this factor appeared to be driven mainly by insulin-related pathways (Figure 5A). In KS, the factor 6 gene signature associated with higher insulin resistance in adipose tissue was characterized by reduced expression of key insulin/circadian regulators (including \u003cem\u003eIRS2\u003c/em\u003e, \u003cem\u003ePDK4\u003c/em\u003e, \u003cem\u003ePFKFB3\u003c/em\u003e, \u003cem\u003eCIART\u003c/em\u003e, \u003cem\u003ePER3\u003c/em\u003e, \u003cem\u003eZBTB16\u003c/em\u003e, and \u003cem\u003eHLF\u003c/em\u003e), increased expression of genes involved in lipid storage and steroid/xenobiotic metabolism (\u003cem\u003eCES1\u003c/em\u003e, \u003cem\u003eGPAM\u003c/em\u003e, \u003cem\u003eDGAT2\u003c/em\u003e, \u003cem\u003eVLDLR\u003c/em\u003e, \u003cem\u003eSULT1A1\u003c/em\u003e/\u003cem\u003e2\u003c/em\u003e, \u003cem\u003eAKR1C3\u003c/em\u003e, \u003cem\u003eLEP\u003c/em\u003e), together with markers of extracellular matrix remodelling and stress signalling (\u003cem\u003eTNMD\u003c/em\u003e, \u003cem\u003eSFRP4\u003c/em\u003e, \u003cem\u003eSTC1\u003c/em\u003e, \u003cem\u003eADAMTS4\u003c/em\u003e, \u003cem\u003eSGK2\u003c/em\u003e, \u003cem\u003ePMEPA1\u003c/em\u003e). This profile was consistent with hypertrophic, fibrotic, and metabolically inflexible adipose tissue in the context of insulin resistance. Enrichment analysis supported this, showing that factor 6 genes downregulated in KS were enriched in pathways related to circadian regulation and rhythm, response to insulin stimulus and glucose metabolism (Figure 5B), whereas upregulated genes were enriched in pathways related to steroid, hormone, and lipid metabolism (Extended Data Figure 5).\u003c/p\u003e\n\u003cp\u003eDue to this downregulation of genes related to circadian rhythm, we estimated the degree of circadian clock disruption in the adipose tissue using the TimeTeller algorithm. We observed a highly significant disruption of circadian rhythmicity in KS adipose tissue compared with controls (Figure 5C). Furthermore, using the deltaCCD algorithm on a core circadian clock gene set consisting of both circadian activators and repressors (Extended Data Table 1), KS adipose tissue showed a significantly larger deviation from the control clock co-expression pattern compared with controls (deltaCCD = 2.96, p = 0.001). Especially the core circadian activators were collectively downregulated (Figure 5D,E).\u003c/p\u003e\n\u003cp\u003eTo further investigate the potential impact of the circadian clock dysregulation within adipose tissue, we assessed age acceleration using an epigenetic clock based on DNAm patterns. Epigenetic age acceleration was significantly increased in adipose tissue from KS compared with controls (Figure 5F). Together, these findings provide strong evidence that the internal circadian clock in KS adipose tissue is disrupted, and that KS adipose tissue exhibits increased age acceleration, which appears to be linked to insulin resistance.\u003c/p\u003e\n\u003ch2\u003ePrediction of body fat in KS\u003c/h2\u003e\n\u003cp\u003eBody fat in KS appeared to be a good candidate marker for evaluating effects of TRT and assessing overall cardiometabolic risk in KS. However, continuous monitoring of treatment by direct assessment of body fat requires access to regular DEXA or CT scans, restricting the practical utility in the outpatient clinic.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eBuilding on this clinical need, we developed two XGBoost prediction models for estimating body fat in men with KS using routinely available inputs: (i) an anthropometric model (age, height, weight, waist circumference, hip circumference and treatment status) and (ii) an extended model including total testosterone. In the training dataset (visit 1, n = 134), the anthropometric model achieved a root mean squared error (RMSE) of 3.60, and mean absolute error (MAE) of 2.77 with R\u0026sup2; 0.83 (Figure 6A). Feature importance analysis showed that predictions were driven primarily by weight, hip circumference and waist circumference, with smaller contributions from height, age and treatment status (Figure 6B). In the validation dataset (subsequent visits, n = 75), the prediction error was RMSE 4.17 (95% CI 3.48\u0026ndash;4.94) and MAE 3.40 (95% CI 2.88\u0026ndash;4.00) with R\u0026sup2; 0.63 (Figure 6C). However, calibration indicated systematic miscalibration (intercept 9.37 [6.19\u0026ndash;12.3], slope 0.681 [0.572\u0026ndash;0.795]). Model updating by linear recalibration in the validation cohort reduced error to RMSE 3.52 and MAE 2.65 (R\u0026sup2; unchanged), (Figure 6D). \u0026nbsp;Inclusion of total testosterone as input, improved model fit in the training dataset (RMSE 3.18, MAE 2.37, R\u0026sup2; 0.87) and modestly improved external validation performance (RMSE 3.93 [3.24\u0026ndash;4.61], MAE 3.10 [2.55\u0026ndash;3.67], R\u0026sup2; 0.67) with calibration closer to ideal (intercept 7.81 [4.73\u0026ndash;10.8], slope 0.762 [0.648\u0026ndash;0.878]). After recalibration, error further decreased (RMSE 3.34, MAE 2.46). As an additional analysis aimed at personalized longitudinal monitoring, we anchored predictions to baseline body fat by applying a subject-specific offset derived from visit 1 (observed baseline body fat minus model-predicted baseline body fat) and adding this offset to the subsequent visit predictions, followed by linear recalibration. Using the anthropometric model, this personalized, calibrated approach achieved an RMSE of 3.00 and an MAE of 2.43 in visit 2 (R\u0026sup2; 0.70)(Figure 6E). When total testosterone was included as an additional predictor, performance improved further, yielding an RMSE of 2.81 and an MAE of 2.27 (R\u0026sup2; 0.74). Our models show that body fat in both U-KS and T-KS can be modelled with reasonably good precision using only anthropometric variables as input, and additional improvement of model fit if further including baseline visit offset.\u003c/p\u003e"},{"header":"DISCUSSION","content":"\u003cp\u003eThe present study shows how pivotal adipose tissue is for metabolic health of males with KS. Metabolic dysfunction and increased autoimmunity emerged early as important clinical traits in KS\u003csup\u003e11\u003c/sup\u003e, and already in the late sixties, Johannes Nielsen from Aarhus University described increased incidence of diabetes and insulin resistance in men with KS \u003csup\u003e12,13\u003c/sup\u003e. Here we, for the first time, demonstrate that both hypogonadism and tissue specific multigenetic alterations are working in conjunction to promote metabolic dysfunction in KS. We clearly demonstrate that remission of hypogonadism through TRT does not fully resolve insulin resistance in men with KS. The finding of pervasive metabolic dysfunction in KS, irrespective of TRT, is in line with our recent national cohort-study, where TRT did not diminish the more than twofold increased risk of type 2 diabetes seen among men with KS compared with controls \u003csup\u003e2\u003c/sup\u003e. As such, the present study offers a link between KS specific gene expression in adipose tissue, adipose inflammation, increased insulin resistance, and disrupted circadian rhythm independently of hypogonadism, linking to real-world comorbidity risk and mortality. The findings from this unique integrated model, combining the clinical phenotype, biomarkers and genetic background, highlight the complexity of pathological mechanisms behind the distinct comorbidity patterns seen across various sex chromosome aneuploidies \u003csup\u003e14\u003c/sup\u003e, and the need for therapeutic approaches beyond simple replenishment of sex hormones to fully minimize cardiometabolic risk. Combined, our findings elucidate a complex pattern of TRT effects on metabolic function in KS, highlighting the intricate interaction of hypogonadism and metabolism in general, while also emphasizing the specific genetic backdrop of KS. Our data evolves the pathophysiological understanding of KS; from a condition solely related to androgen deficiency or restricted to X chromosome gene dosage effects, into a pervasive multigenetic syndrome. \u0026nbsp;Intertwining of sex hormones, sex specific genetics, and metabolic function is further substantiated by the broadly observed sex specific metabolic phenotype among cis-males and cis-females\u003csup\u003e1\u003c/sup\u003e, and by diverging effects of hormone therapy on body composition and cardiometabolic risk markers among those receiving transgender hormone therapy \u003csup\u003e10\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eBody composition was not normalized by TRT and men with KS had persistently higher body fat and lower nFFMI compared with controls, which are both associated with increased mortality \u003csup\u003e15\u003c/sup\u003e. As such, the loss of body fat and increase in nFFMI after TRT in this study is consistent with our previous finding of reduced mortality in T-KS compared with U-KS\u003csup\u003e9\u003c/sup\u003e, underscoring the importance of TRT in KS. Low HDL is commonly associated with increased cardiovascular risk, and TRT in KS was associated with a reduction in HDL. However, it is not known how specific dyslipidemic traits affect cardiovascular risk in KS. In our recent national study, TRT in KS was not associated with increased risk of major adverse cardiovascular events \u003csup\u003e9\u003c/sup\u003e, and as such commonly used cardiovascular risk scores might not apply directly to KS.\u003c/p\u003e\n\u003cp\u003eTRT significantly impacted the adipokine profile in KS. Reflecting the changes in body fat, leptin was lowered in T-KS, but not to the level of controls. Adiponectin was also lowered, but to levels below that of the control population. The suppression of adiponectin by TRT has also been described in non-KS populations \u003csup\u003e16\u003c/sup\u003e. Importantly, this adipokine profile with high leptin, indicative of leptin resistance and concomitant low adiponectin is specifically associated with the highest risk of obesity, insulin resistance and type 2 diabetes, especially among individuals with higher body fat \u003csup\u003e17\u003c/sup\u003e. As such, the lowering of adiponectin by TRT could contribute to sustained insulin resistance in KS. Also, men with KS had higher leptin-to-adiponectin ratio than controls irrespective of TRT, corresponding with the sustained increased risk of type 2 diabetes in KS \u003csup\u003e10\u003c/sup\u003e, and further supporting the presence of non-hypogonadism dependent genetic risk factors. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eOur findings indicate that increased body fat, as in other populations, is associated with adipose tissue immune cell infiltration and inflammation, which in turn may impair insulin sensitivity\u003csup\u003e18\u003c/sup\u003e. In both KS and controls, higher body fat was associated with upregulated expression of innate immune and inflammatory genes in adipose tissue, gene signatures that are hallmarks of macrophage/monocyte and lymphocyte recruitment and activation, including M1-macophage polarization, and are consistent with obesity-driven immune cell infiltration \u003csup\u003e19\u003c/sup\u003e. The enrichment of pathways related to immune and cytokine activation further supports adiposity induced immune cell accumulation as a general consequence of adipose tissue expansion. Multiple immune cell types are known to infiltrate obese adipose tissue and propagate an inflammatory cascade that impairs insulin signalling \u003csup\u003e20\u003c/sup\u003e. The DNA methylation-based cell type deconvolution followed this trend, showing greater inferred NK cell and monocyte/macrophage proportions in adipose tissue from men with higher body fat. The strong correlation of immune cell fraction with insulin resistance across our cohort, highlights the fact that elevated body fat promotes adipose immune infiltration and insulin resistance. This mechanism may provide an explanation for why men with KS become insulin resistant and prone to develop metabolic syndrome.\u003c/p\u003e\n\u003cp\u003eWe uncovered a KS-specific genomic signature of insulin resistance strongly linked to circadian rhythm disruption (captured by Factor 6). This was driven by downregulated expression of core clock and insulin pathway genes (including \u003cem\u003ePER3\u003c/em\u003e, \u003cem\u003eCIART\u003c/em\u003e, \u003cem\u003eZBTB16\u003c/em\u003e, \u003cem\u003eHLF\u003c/em\u003e, \u003cem\u003eIRS2\u003c/em\u003e, \u003cem\u003ePDK4\u003c/em\u003e, \u003cem\u003ePFKFB3\u003c/em\u003e) alongside upregulation of genes involved in lipid storage, extracellular matrix remodelling and stress responses \u003csup\u003e21\u003c/sup\u003e. As such, intrinsic impairment of the biological circadian clock and metabolic flexibility may exacerbate insulin resistance in KS adipose tissue. Notably, circadian regulators in most peripheral tissues, including adipose, display diurnal rhythmic expression of metabolic genes, adipokines and adipocytokines. This plays a crucial role in insulin signalling and maintaining metabolic homeostasis as disruption of circadian rhythm and lowered expression of clock genes has been observed with obesity, metabolic stress, or sleep disturbances \u003csup\u003e22,23\u003c/sup\u003e. As demonstrated here, obesity and metabolic dysfunction are common in KS, and sleep disturbances have also been observed in KS from an early age \u003csup\u003e24,25\u003c/sup\u003e. Animal models have demonstrated that mutation of core clock genes leads to metabolic derangements\u003csup\u003e26\u003c/sup\u003e. Thus, the downregulation of circadian clock components in KS adipose tissue is mechanistically plausible as a contributor to insulin resistance, given the established link between circadian disruption and impaired insulin signalling\u003csup\u003e22\u003c/sup\u003e. Our analyses provided support for this, revealing significantly blunted or desynchronized circadian oscillations in KS adipose tissue. In other words, the adipose tissue of men with KS appeared to have lost the normal rhythmic gene expression patterns that help synchronize metabolic processes.\u003c/p\u003e\n\u003cp\u003eHigh rates of insulin resistance are seen already among prepubertal boys with KS \u003csup\u003e27\u003c/sup\u003e. We speculate that early disruption of the circadian rhythm of peripheral tissues during childhood, driven by KS specific genetic dispositions, may be linked to both increased body fat and insulin resistance.\u003c/p\u003e\n\u003cp\u003eThe circadian dysregulation in KS was accompanied by signs of accelerated biological aging in the tissue (increased DNA methylation age relative to chronological age), which we recently showed was also present, albeit to a lesser degree, in blood \u003csup\u003e28\u003c/sup\u003e. Chronic metabolic stress and obesity have been associated with epigenetic age acceleration in metabolic tissues \u003csup\u003e29\u003c/sup\u003e, so the elevated “epigenetic age” of KS adipose tissue may reflect cumulative metabolic strain due to lifelong circadian disruption or metabolic imbalance.\u003c/p\u003e\n\u003cp\u003eThe cause of the disrupted circadian rhythm is presently not known.\u0026nbsp;Men with KS also have an inherently different genetic and endocrine profile that could impact circadian regulation. Testosterone itself exhibits diurnal variation \u003csup\u003e30\u003c/sup\u003e, and chronic hypogonadism might blunt normal circadian metabolic cues. Also, the pervasive genome-wide changes to both the transcriptome and epigenome could impact circadian rhythm \u003csup\u003e31\u003c/sup\u003e. This suggests that traditional interventions (like testosterone replacement or weight loss) might need to be complemented by approaches that restore circadian alignment - for instance, pharmacological agents that target clock pathways, such as melanocortin receptor agonists, antidepressants and other new drugs currently being developed \u003csup\u003e32,33\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWe provide a precise prediction of body fat among men with KS, as a simple tool for assessing cardiometabolic risk and effect of TRT. \u0026nbsp;As demonstrated, the low muscle-to-fat ratio and slightly increased height in KS \u003csup\u003e34\u003c/sup\u003e will cause BMI to underestimate body fat, masking the underlying metabolic dysfunction. Also, BMI, WHRadjBMI, LAP, and VAI were incapable of detecting changes in body composition following TRT. We believe that direct assessment of body fat by DXA/CT or by KS-specific prediction of body fat, has potential to identify men with KS that are at an increased risk of cardiometabolic complications.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWe developed two prediction models that estimate body fat in KS from readily available anthropometrics. The anthropometric model achieved reasonable accuracy in external validation, supporting that clinically relevant variation in body fat can be captured without the need for repeated DXA/CT scans. Importantly, feature-importance patterns were biologically plausible, with weight, hip circumference and waist circumference contributing most to prediction, while age, height, and treatment status added smaller incremental information. This aligns with the expectation that overall and central adiposity dominate inter-individual differences in body fat, whereas treatment status likely influences body composition through pathways only partly reflected in simple anthropometrics.\u003c/p\u003e\n\u003cp\u003eAs an additional step to support longitudinal, patient-specific monitoring, we anchored follow-up predictions to baseline DXA, which improved prediction error. While anthropometrics explains much of the between-person variation in body fat, individuals with similar waist, hip, and weight can still differ systematically in true body composition. The baseline offset quantifies the person-specific deviation (DXA-measured body fat minus model-predicted body fat at Visit 1) and carries it forward to subsequent visits when DXA is unavailable. This converts a population-based estimate into an individualized estimate and is therefore well-suited for tracking changes over time.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eDespite maintained accuracy in the validation cohort, calibration analyses revealed systematic miscalibration, indicating that predicted values were on a different scale in the testing dataset (positive intercept and slope \u0026lt;1). This is a common finding when models are used across cohorts, reflecting differences in measurement procedures, or underlying distributions of predictors and outcomes. We therefore performed model updating by linear recalibration in the validation cohort, which improved absolute error and brought predictions closer to the observed scale. In practical terms, this suggests that the proposed tool is best implemented with local calibration when applied in new clinical settings or when measurement protocols differ. Ideally, input data should be collected prospectively from men with KS around the globe to allow development of a strong clinically applicable tool.\u003c/p\u003e\n\u003cp\u003eThe main strength of this study is the comprehensive, longitudinal, collective assessment of metabolic function and integration with gene expression in a large sample of men with KS. The breadth of the data set is unique and allows a systems biology approach to metabolic function in KS. Still, several limitations apply. Most prominently, all participants in the study have lived with prolonged hypogonadism prior to TRT, and thus the timing of TRT could be important. Although metabolic dysfunction is present already at a young age in KS \u003csup\u003e2\u003c/sup\u003e, it can be speculated that longer time spent in hypogonadism could lead to a metabolic memory legacy in metabolic tissues \u003csup\u003e35\u003c/sup\u003e, with permanent damage to pivotal metabolic processes that hinder subsequent reversal of metabolic dysfunction through TRT and correction of hypogonadism. Other studies are currently planned to assess the effect of TRT initiated during puberty on changes in body composition and metabolism\u003csup\u003e36\u003c/sup\u003e. Also, our study did not assess visceral fat directly but relies on total body fat percentage. Total fat and visceral fat are strongly associated \u003csup\u003e37\u003c/sup\u003e, but we might have achieved more precise estimates if assessing visceral fat directly with adjustment for total body fat - especially when wanting to use fat mass as a predictor for future cardiometabolic risk\u003csup\u003e38\u003c/sup\u003e. Along the same lines, the proposed models for assessing body fat in KS need to be verified in other cohorts, and ultimately to be tested against clinical cardiometabolic events. Also, most participants only had data from one visit, potentially allowing for selection bias regarding the longitudinal analyses.\u003c/p\u003e\n\u003cp\u003eOur study represents a breakthrough in the understanding of metabolic dysfunction in KS demonstrating insufficient rescue of obesity and insulin sensitivity with TRT, while also identifying syndrome specific genetic markers, that are driving metabolic dysfunction independent of sex hormones. Our data calls for a paradigm shift in the clinical approach to cardiometabolic prophylactic care in KS, highlighting the need for continuous monitoring of body fat and insulin sensitivity and advancement of treatment approaches to alleviate metabolic risk.\u003c/p\u003e"},{"header":"METHODS","content":"\u003ch2\u003eParticipants and ethics\u003c/h2\u003e\n\u003cp\u003eWe include data from three separate clinical research projects, collectively including participants from March 2002 until March 2019 \u003csup\u003e8,39,40\u003c/sup\u003e. All studies included men with KS verified by karyotyping, and age-matched control males. All participants provided signed informed consent. Chronologically, study A was approved by the Danish Data Protection Agency and the Aarhus County Ethical Scientific Committee (# 20010155), study B by the Danish Data Protection Agency and the Region Midtjylland Ethical Committee (M-20080238) and registered with ClinicalTrials.gov (NCT00999310), and study C by the Danish Data Protection Agency (1-16-02-472-15), the Central Denmark Regional Committees on Health Research Ethics (1-10-72-131-15), and registered with ClinicalTrials.gov (NCT02526628).\u0026nbsp;\u003c/p\u003e\n\u003ch2\u003eStudy design\u003c/h2\u003e\n\u003cp\u003eStudies A and B were cross-sectional with a single visit, while study C had both a baseline and follow-up visit. However, as some individuals participated in more than one of the assessed studies, any individual could have data from as many as four visits combined. We thus define the number and order of visits based on the individual across all studies, with the first visit in any study assigned as Visit 1 and subsequent visits as Visit 2, 3, and 4, respectively. At each visit, men with KS were stratified by TRT status (U-KS/T-KS).\u0026nbsp;\u003c/p\u003e\n\u003ch2\u003eSex hormones\u0026nbsp;\u003c/h2\u003e\n\u003cp\u003eAssays for sex hormones have varied over time. For study A, sex hormones were measured by specific radioimmunoassyas (RIA)\u003csup\u003e41\u003c/sup\u003e. For studies B and C, both testosterone and estradiol were measured by liquid chromatography tandem mass spectrometry, while luteinizing hormone (LH), and sex hormone binding globulin (SHBG) were quantitatively determined by immunoassays \u003csup\u003e34,42\u003c/sup\u003e.\u003c/p\u003e\n\u003ch2\u003eGlucose metabolism\u003c/h2\u003e\n\u003cp\u003eData on glucose metabolism were available from studies A and C, except for fasting glucose which was also available from study B. For study A, serum insulin was determined by a commercial immunological kit and serum adiponectin was determined by a time-resolved immunofluorometric assay and leptin was determined by a commercial radioimmunoassay \u003csup\u003e39\u003c/sup\u003e. For study C, insulin was measured by ELISA (Mercodia, Sweden), adiponectin and leptin were measured by in house time-resolved immunofluorometric assays. \u0026nbsp;The Homeostasis Model Assessment 2 (HOMA2) was calculated based on fasting glucose and insulin levels using the HOMA2 calculator macro (Version 2.2.3, Diabetes Trial Unit, University of Oxford, Oxford, United Kingdom). The calculator outputs HOMA2-%B as a marker of beta-cell function, HOMA2-%S a marker of insulin sensitivity and the reciprocal HOMA2-IR indicating level of insulin resistance. We further assessed the Triglyceride Glucose Index (TyG) as a marker of insulin resistance calculated as TyG = ln[(Triglycerides (mg/dL) * Fasting Glucose (mg/dL)) / 2]\u003csup\u003e43\u003c/sup\u003e. Lipids were assessed enzymatically using laboratory standard assays at Aarhus University Hospital.\u003c/p\u003e\n\u003ch2\u003eBody composition\u0026nbsp;\u003c/h2\u003e\n\u003cp\u003eTotal body fat percentage and lean mass were examined by DEXA scans using Hologic scanners \u003csup\u003e34,39,42\u003c/sup\u003e. Body mass index (BMI) was calculated as weight (kg) over height (m) squared. Waist-to-hip ratio adjusted for BMI (WHRadjBMI) was computed as the residuals from a linear regression with waist-to-hip ratio as the dependent variable and BMI, age, and age squared as covariates. The Visceral Adiposity Index (VAI) was calculated using the male sex specific formula as VAI = (Waist circumference / (39.68 + (1.88 * BMI))) * (Triglycerides/ 1.03) * (HDL / 1.31)\u003csup\u003e44\u003c/sup\u003e. The Lipid Accumulation Product (LAP) was calculated as LAP = (Waist Circumference - 65) * Triglycerides)\u003csup\u003e45\u003c/sup\u003e. \u0026nbsp; The male normalized Fat-Free Mass Index (nFFMI) was calculated as nFFMI = (total lean mass(kg)/height(m\u003csup\u003e2\u003c/sup\u003e)+6.1*(1.8-height(m))\u003csup\u003e46\u003c/sup\u003e.\u003c/p\u003e\n\u003ch2\u003eThe genomic dataset\u0026nbsp;\u003c/h2\u003e\n\u003cp\u003eAdipose tissue biopsies were taken in study C from the abdominal subcutaneous periumbilical region by liposuction and snap-frozen in liquid nitrogen, before DNA and RNA isolation and sequencing. The genomic dataset consisted of a subset of the full cohort, KS (n = 22) and Controls (n = 15). All DNA methylation (DNAm) and RNA sequencing (RNAseq) data from adipose tissue of the participants have been published previously \u003csup\u003e47\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003ch2\u003eRNA analysis\u003c/h2\u003e\n\u003cp\u003eFor gene expression, fastq files underwent initial quality control using FastQC (Babraham Bioinformatics). Adaptor removal and trimming of low-quality ends were performed using Trim Galore with default settings (Babraham Bioinformatics). Transcript expression levels were quantified using Salmon \u003csup\u003e48\u003c/sup\u003e, with a decoy-aware transcriptome index based on the hg38 transcriptome. Transcript abundances were summarized to the gene level using the R package Tximeta \u003csup\u003e49\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003ch3\u003eMulti-Omics Factor Analysis (MOFA) of adipose tissue gene expression\u003c/h3\u003e\n\u003cp\u003eUnsupervised latent factor analysis of the adipose gene expression was performed using the MOFA2 framework \u003csup\u003e50\u003c/sup\u003e. Gene expression counts were transformed using the variance-stabilizing transformation (VST) implemented in DESeq2. The resulting VST expression matrix was used as an input for factor generation. To focus on major biological sources of variation, we restricted the analysis to the 2,000 most variable genes. A MOFA object was created with pre-processing options set using default parameters. We specified 10 latent factors to capture multiple independent axes of variation. To visualize factor-gene relationships, we generated heatmaps of expression for the most strongly weighted genes per factor. Associations between MOFA factors and clinical/biochemical traits were assessed with computed Pearson correlations between factor scores and sample-level covariates. Correlations were visualized as correlation matrices and interpreted using a nominal significance threshold of α = 0.05. To functionally characterize factors, we performed gene set enrichment analysis on MOFA feature weights using the run enrichment function (MOFA2). Enrichment was computed separately for positively and negatively weighted genes per factor.\u003c/p\u003e\n\u003ch3\u003eTimeTeller\u003c/h3\u003e\n\u003cp\u003eCircadian clock function was quantified at the sample level using TimeTeller \u003csup\u003e51\u003c/sup\u003e. TimeTeller was trained on a dataset from healthy volunteers, in which punch biopsies were collected every 4 h across 24 h and profiled \u003csup\u003e51\u003c/sup\u003e. The TimeTeller training model was fitted on the training data with inter-gene normalisation and 3 principal components, retaining the first 3 principal components of the panel expression at each circadian phase. All other parameters were kept at their recommended defaults. For the KS and control adipose gene expression data, raw gene-level counts were converted to log₂ counts per million (CPM). The expression matrix was supplied to TimeTeller and for each sample TimeTeller returned an estimated clock dysfunction score (Θ, 0–1), where higher Θ indicates poorer agreement with the healthy reference clock state - a more disrupted clock.\u003c/p\u003e\n\u003ch3\u003edeltaCCD\u0026nbsp;\u003c/h3\u003e\n\u003cp\u003eGroup-level disruption of the circadian clock gene network was quantified using the deltaCCD method (Hughey J, Outland E (2022). deltaccd: Quantify Rhythmic Gene Co-Expression Relative to a Reference. https://deltaccd.hugheylab.org, https://github.com/hugheylab/deltaccd). Briefly, deltaCCD compared the clock correlation distance (CCD) between groups relative to a reference clock-gene co-expression pattern. RNAseq counts from adipose tissue from KS and controls were normalised to log₂-CPM, yielding an expression matrix that included all measured genes. A circadian clock gene set was defined a priori, comprising \u003cem\u003eARNTL\u003c/em\u003e, \u003cem\u003eCLOCK\u003c/em\u003e, \u003cem\u003ePER1\u003c/em\u003e, \u003cem\u003ePER2\u003c/em\u003e, \u003cem\u003ePER3\u003c/em\u003e, \u003cem\u003eCRY1\u003c/em\u003e, \u003cem\u003eCRY2\u003c/em\u003e, \u003cem\u003eNR1D1\u003c/em\u003e, \u003cem\u003eNR1D2\u003c/em\u003e, \u003cem\u003eNPAS2\u003c/em\u003e, \u003cem\u003eDBP\u003c/em\u003e, \u003cem\u003eTEF\u003c/em\u003e, \u003cem\u003eHLF\u003c/em\u003e. For deltaCCD, samples were classified by genotype with controls specified as the reference group. CCDs were then calculated for each group. The ΔCCD was defined as CCD(KS) − CCD(Controls), so positive values indicate that the KS clock-gene co-expression pattern is more dissimilar from the reference than the 46,XY pattern (i.e. more disrupted). Statistical significance was assessed using permutation testing within the function calcDeltaCCD.\u003c/p\u003e\n\u003ch2\u003eDNA Methylation\u003c/h2\u003e\n\u003cp\u003eRaw intensity values for all CpG sites (Infinium MethylationEPIC) were imported and processed using the R package Minfi \u003csup\u003e52\u003c/sup\u003e. Cross-reactive probes and poorly performing probes, as indicated by a detection p-value \u0026lt; 0.01, were excluded from the analysis. The preprocess Funnorm normalization method \u003csup\u003e53\u003c/sup\u003e was applied to remove between-array variation inferred by control probes, followed by the conversion of methylation values to beta-values.\u0026nbsp;\u003c/p\u003e\n\u003ch3\u003eEstimation of cell type proportions\u003c/h3\u003e\n\u003cp\u003eTo investigate for differences in cell composition, the HEpiDISH algorithm was employed using a reference for solid tissues “\u003ccode\u003ecentEpiFibIC.m\u003c/code\u003e”\u0026nbsp;\u003csup\u003e54\u003c/sup\u003e and a reference that yield the proportions of 7 immune cell subtypes in addition to the solid fractions estimated “\u003ccode\u003ecentEpiFibIC.m\u003c/code\u003e”. Normalized beta-values were used as input. \u0026nbsp; \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n\u003ch3\u003eEpigenetic age acceleration\u003c/h3\u003e\n\u003cp\u003eUsing the R-package MethylClock (v1.16.0) \u003csup\u003e55\u003c/sup\u003e, we applied Horvath’s pan-tissue epigenetic clock, suitable for the analysis of multiple tissues, including adipose tissue \u003csup\u003e56\u003c/sup\u003e. Using this, we estimated the intrinsic age acceleration, defined as the residuals of regressing chronological age on epigenetic age. \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n\u003ch2\u003eStatistical analysis\u003c/h2\u003e\n\u003cp\u003eDistribution of continuous outcomes was evaluated by quantile-quantile plots. Data are presented as mean ± SD or median (p25-p75). Baseline comparison across groups was performed by one-way ANOVA with Tukey post-hoc test or Kruskal Wallis test with Dunn post-hoc test dependent on distribution.\u003c/p\u003e\n\u003cp\u003eWe assessed the effects of TRT on repeated outcome measures using mixed-effects linear regression. Each participant was modeled with a random intercept to account for within-subject correlation, and robust (clustered) standard errors were used. Fixed effects included treatment status (treated vs untreated) and study visit (1–4). To account for age, we decomposed age into two components: baseline age (centered around the sample mean) to capture between-subject differences, and within-subject change in age across visits (years since baseline) to capture longitudinal aging effects. Conditional (cR\u003csup\u003e2\u003c/sup\u003e) and marginal (mR\u003csup\u003e2\u003c/sup\u003e) r-squared summarizing statistics from the mixed-effects model were derived as described by Nakagawa \u003csup\u003e57\u003c/sup\u003e. Analysis was done using Stata Now (Statacorp).\u003c/p\u003e\n\u003cp\u003eBody fat was modelled in men with KS using extreme gradient boosting regression (XGBoost). The XGBoost algorithm was chosen because it performs well on structured clinical data and can flexibly capture non-linear relationships and interactions among anthropometric variables. It also includes built-in regularization and early stopping to reduce overfitting. Participants with missing body fat measurements were excluded. Treatment status was included as a predictor. Non-informative variables were removed, and the final predictor set comprised key anthropometric variables (hip, weight, waist, height, age). Total testosterone was added for a second model. Model hyperparameters were selected by random search over 40 candidate parameter sets using repeated 10-fold cross-validation (5 repeats) within the training dataset. Early stopping was applied (maximum 5000 boosting rounds; stopping after 50 rounds without improvement) to reduce overfitting. Predictive performance was evaluated on a validation cohort, using RMSE, MAE, and R². Calibration was assessed in the validation data by regressing observed body fat on predicted body fat to obtain a calibration intercept and slope. Uncertainty in RMSE/MAE and calibration parameters was quantified using nonparametric bootstrapping (2000 resamples). Feature importance was summarized using XGBoost gain, cover and frequency.\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eDATA AVAILABILITY\u003c/h2\u003e\n\u003cp\u003eThe clinical data reported here can be obtained by reasonable request to the authors in accordance with the General Data Protection Regulation. All sequence and methylation data from this study are deposited at the European Genome-phenome Archive (EGA) repository (ega-archive.org, EGAS00001006996, EGAS00001007020).\u003c/p\u003e\n\u003ch2\u003eAcknowledgements\u003c/h2\u003e\n\u003cp\u003eWe would like to acknowledge bio analysts Lone Kvist, Maria Flink Schwartz, and Line Mentz for their assistance in obtaining clinical samples and Pamela Celis for isolation of DNA and RNA from samples. We also acknowledge the MOMA NGS Core Center, and GenomeDK. \u0026nbsp;\u003c/p\u003e\n\u003ch2\u003eAuthor contributions\u003c/h2\u003e\n\u003cp\u003eCRediT statement\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eConceptualization (JJ, CHG, SC), data curation (JJ,SC), formal analysis (JJ, EH, SC), funding acquisition (JJ, CHG, SC), investigation (JJ, EH, AS, CHG, SC), methodology (JJ, EH, SC), project administration (AS, CHG, SC), resources (AS, CHG, SC), software (JJ, EH, SC), supervision (SC, CHG, JJ), visualization (JJ, EH, SC), writing \u0026ndash; original draft (JJ, CHG, SC), writing \u0026ndash; review \u0026amp; editing (JJ, EH, AS, CHG, SC).\u003c/p\u003e\n\u003cp\u003eApproval of the final manuscript was granted by all authors.\u0026nbsp;\u003c/p\u003e\n\u003ch2\u003eFunding\u003c/h2\u003e\n\u003cp\u003eThis work was supported by Aarhus University; Novo Nordisk Foundation (NNF20OC0060610); the Independent Research Fund Denmark (2096-00165A, 0134-00130B); Sygesikringen danmark (2022-0189), the Danish Diabetes and Endocrine Academy which is funded by the Novo Nordisk Foundation, grant number NNF22SA0079901.\u0026nbsp;\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eMauvais-Jarvis, F. Sex differences in energy metabolism: natural selection, mechanisms and consequences. \u003cem\u003eNat Rev Nephrol\u003c/em\u003e\u003cstrong\u003e20\u003c/strong\u003e, 56-69 (2024). https://doi.org/10.1038/s41581-023-00781-2\u003c/li\u003e\n\u003cli\u003eGravholt, C. H.\u003cem\u003e et al.\u003c/em\u003e Klinefelter syndrome - integrating genetics, neuropsychology and endocrinology. \u003cem\u003eEndocrine reviews\u003c/em\u003e (2018). https://doi.org/10.1210/er.2017-00212\u003c/li\u003e\n\u003cli\u003eBardsley, M. Z., Falkner, B., Kowal, K. \u0026amp; Ross, J. L. Insulin resistance and metabolic syndrome in prepubertal boys with Klinefelter syndrome. \u003cem\u003eActa paediatrica (Oslo, Norway : 1992)\u003c/em\u003e\u003cstrong\u003e100\u003c/strong\u003e, 866-870 (2011). https://doi.org/10.1111/j.1651-2227.2011.02161.x\u003c/li\u003e\n\u003cli\u003eAksglaede, L., Molgaard, C., Skakkebaek, N. E. \u0026amp; Juul, A. Normal bone mineral content but unfavourable muscle/fat ratio in Klinefelter syndrome. \u003cem\u003eArchives of disease in childhood\u003c/em\u003e\u003cstrong\u003e93\u003c/strong\u003e, 30-34 (2008). https://doi.org/10.1136/adc.2007.120675\u003c/li\u003e\n\u003cli\u003eGranato, S.\u003cem\u003e et al.\u003c/em\u003e Epicardial fat: the role of testosterone and lipid metabolism in a cohort of patients with Klinefelter syndrome. \u003cem\u003eMetabolism: clinical and experimental\u003c/em\u003e\u003cstrong\u003e95\u003c/strong\u003e, 21-26 (2019). https://doi.org/10.1016/j.metabol.2019.03.002\u003c/li\u003e\n\u003cli\u003eDemirci, I.\u003cem\u003e et al.\u003c/em\u003e Triglyceride-glucose index levels in patients with Klinefelter syndrome and its relationship with endothelial dysfunction and insulin resistance: a cross-sectional observational study. \u003cem\u003eArch Endocrinol Metab\u003c/em\u003e\u003cstrong\u003e67\u003c/strong\u003e, 378-384 (2023). https://doi.org/10.20945/2359-3997000000594\u003c/li\u003e\n\u003cli\u003eHaymana, C.\u003cem\u003e et al.\u003c/em\u003e Increased Endothelial Dysfunction and Insulin Resistance in Patients with Klinefelter Syndrome. \u003cem\u003eEndocrine, metabolic \u0026amp; immune disorders drug targets\u003c/em\u003e\u003cstrong\u003e18\u003c/strong\u003e, 401-406 (2018). https://doi.org/10.2174/1871530318666180220114322\u003c/li\u003e\n\u003cli\u003eChang, S.\u003cem\u003e et al.\u003c/em\u003e Testosterone Replacement Therapy in Klinefelter Syndrome-Follow-up Study Associating Hemostasis and RNA Expression. \u003cem\u003eThe Journal of clinical endocrinology and metabolism\u003c/em\u003e\u003cstrong\u003e109\u003c/strong\u003e, 978-991 (2024). https://doi.org/10.1210/clinem/dgad658\u003c/li\u003e\n\u003cli\u003eChang, S., Pedersen, L., Skakkeb\u0026aelig;k, A., Berglund, A. \u0026amp; Gravholt, C. H. Cardiovascular risk and mortality in men receiving testosterone replacement therapy for Klinefelter syndrome in Denmark: a retrospective cohort study. \u003cem\u003eThe Lancet Regional Health - Europe\u003c/em\u003e\u003cstrong\u003e51\u003c/strong\u003e, 101230 (2025). https://doi.org/10.1016/j.lanepe.2025.101230\u003c/li\u003e\n\u003cli\u003e Chang, S., Pedersen, L., Skakkeb\u0026aelig;k, A., Berglund, A. \u0026amp; Gravholt, C. H. Metabolism and type 2 diabetes across life stages in Klinefelter syndrome: a population-based cohort study. \u003cem\u003eThe Journal of clinical endocrinology and metabolism\u003c/em\u003e (2025). https://doi.org/10.1210/clinem/dgaf619\u003c/li\u003e\n\u003cli\u003e Burch, P. R. Klinefelter's syndrome, dizygotic twinning and diabetes mellitus. \u003cem\u003eNature\u003c/em\u003e\u003cstrong\u003e221\u003c/strong\u003e, 175-177 (1969). https://doi.org/10.1038/221175a0\u003c/li\u003e\n\u003cli\u003e Menzinger, G.\u003cem\u003e et al.\u003c/em\u003e KLINEFELTER'S SYNDROME AND DIABETES MELLITUS. \u003cem\u003eThe Lancet\u003c/em\u003e\u003cstrong\u003e288\u003c/strong\u003e, 747-748 (1966). https://doi.org/https://doi.org/10.1016/S0140-6736(66)93008-X\u003c/li\u003e\n\u003cli\u003e Nielsen, J., Johansen, K. \u0026amp; Yde, H. Frequency of diabetes mellitus in patients with Klinefelter's syndrome of different chromosome constitutions and the XYY syndrome. Plasma insulin and growth hormone level after a glucose load. \u003cem\u003eThe Journal of clinical endocrinology and metabolism\u003c/em\u003e\u003cstrong\u003e29\u003c/strong\u003e, 1062-1073 (1969). https://doi.org/10.1210/jcem-29-8-1062\u003c/li\u003e\n\u003cli\u003e Berglund, A., Chang, S., Lind-Holst, M., Stochholm, K. \u0026amp; Gravholt, C. H. The epidemiology of disorders of sex development. \u003cem\u003eBest Pract Res Clin Endocrinol Metab\u003c/em\u003e, 102002 (2025). https://doi.org/10.1016/j.beem.2025.102002\u003c/li\u003e\n\u003cli\u003e Sedlmeier, A. M.\u003cem\u003e et al.\u003c/em\u003e Relation of body fat mass and fat-free mass to total mortality: results from 7 prospective cohort studies. \u003cem\u003eAm J Clin Nutr\u003c/em\u003e\u003cstrong\u003e113\u003c/strong\u003e, 639-646 (2021). https://doi.org/10.1093/ajcn/nqaa339\u003c/li\u003e\n\u003cli\u003e Page, S. T.\u003cem\u003e et al.\u003c/em\u003e Testosterone administration suppresses adiponectin levels in men. \u003cem\u003eJournal of andrology\u003c/em\u003e\u003cstrong\u003e26\u003c/strong\u003e, 85-92 (2005).\u003c/li\u003e\n\u003cli\u003e Wannamethee, S. G.\u003cem\u003e et al.\u003c/em\u003e Adipokines and risk of type 2 diabetes in older men. \u003cem\u003eDiabetes care\u003c/em\u003e\u003cstrong\u003e30\u003c/strong\u003e, 1200-1205 (2007). https://doi.org/10.2337/dc06-2416\u003c/li\u003e\n\u003cli\u003e Rohm, T. V., Meier, D. T., Olefsky, J. M. \u0026amp; Donath, M. Y. Inflammation in obesity, diabetes, and related disorders. \u003cem\u003eImmunity\u003c/em\u003e\u003cstrong\u003e55\u003c/strong\u003e, 31-55 (2022). https://doi.org/10.1016/j.immuni.2021.12.013\u003c/li\u003e\n\u003cli\u003e Turner, L., Wanasinghe, A. I., Brunori, P. \u0026amp; Santosa, S. Is Adipose Tissue Inflammation the Culprit of Obesity-Associated Comorbidities? \u003cem\u003eObes Rev\u003c/em\u003e\u003cstrong\u003e26\u003c/strong\u003e, e13956 (2025). https://doi.org/10.1111/obr.13956\u003c/li\u003e\n\u003cli\u003e Guzik, T. J., Skiba, D. S., Touyz, R. M. \u0026amp; Harrison, D. G. The role of infiltrating immune cells in dysfunctional adipose tissue. \u003cem\u003eCardiovasc Res\u003c/em\u003e\u003cstrong\u003e113\u003c/strong\u003e, 1009-1023 (2017). https://doi.org/10.1093/cvr/cvx108\u003c/li\u003e\n\u003cli\u003e Takahashi, J. S. Transcriptional architecture of the mammalian circadian clock. \u003cem\u003eNat Rev Genet\u003c/em\u003e\u003cstrong\u003e18\u003c/strong\u003e, 164-179 (2017). https://doi.org/10.1038/nrg.2016.150\u003c/li\u003e\n\u003cli\u003e Hepler, C. \u0026amp; Bass, J. Circadian mechanisms in adipose tissue bioenergetics and plasticity. \u003cem\u003eGenes Dev\u003c/em\u003e\u003cstrong\u003e37\u003c/strong\u003e, 454-473 (2023). https://doi.org/10.1101/gad.350759.123\u003c/li\u003e\n\u003cli\u003e Stenvers, D. J., Scheer, F., Schrauwen, P., la Fleur, S. E. \u0026amp; Kalsbeek, A. Circadian clocks and insulin resistance. \u003cem\u003eNat Rev Endocrinol\u003c/em\u003e\u003cstrong\u003e15\u003c/strong\u003e, 75-89 (2019). https://doi.org/10.1038/s41574-018-0122-1\u003c/li\u003e\n\u003cli\u003e Fjermestad, K. W., Finnbakk, R. R., Solbakk, A. K., Gravholt, C. H. \u0026amp; Huster, R. J. Subjective versus objective sleep in men with Klinefelter syndrome. \u003cem\u003eOrphanet J Rare Dis\u003c/em\u003e\u003cstrong\u003e18\u003c/strong\u003e, 260 (2023). https://doi.org/10.1186/s13023-023-02822-2\u003c/li\u003e\n\u003cli\u003e Paparella, R.\u003cem\u003e et al.\u003c/em\u003e Evaluation of sleep disorders in children and adolescents affected by Klinefelter syndrome. \u003cem\u003eEur J Pediatr\u003c/em\u003e\u003cstrong\u003e184\u003c/strong\u003e, 129 (2025). https://doi.org/10.1007/s00431-024-05944-2\u003c/li\u003e\n\u003cli\u003e Marcheva, B.\u003cem\u003e et al.\u003c/em\u003e Disruption of the clock components CLOCK and BMAL1 leads to hypoinsulinaemia and diabetes. \u003cem\u003eNature\u003c/em\u003e\u003cstrong\u003e466\u003c/strong\u003e, 627-631 (2010). https://doi.org/10.1038/nature09253\u003c/li\u003e\n\u003cli\u003e Bardsley, M. Z., Falkner, B., Kowal, K. \u0026amp; Ross, J. L. Insulin resistance and metabolic syndrome in prepubertal boys with Klinefelter syndrome. \u003cem\u003eActa Paediatr.\u003c/em\u003e\u003cstrong\u003e100\u003c/strong\u003e, 866-870 (2011).\u003c/li\u003e\n\u003cli\u003e Hasselholm, E. B., Just, J., Chang, S., Skakkeb\u0026aelig;k, A. \u0026amp; Gravholt, C. H. Epigenetic age acceleration in Turner and Klinefelter syndrome: Correlations with clinical aging markers. \u003cem\u003eClin Epigenetics\u003c/em\u003e\u003cstrong\u003e17\u003c/strong\u003e, 147 (2025). https://doi.org/10.1186/s13148-025-01963-4\u003c/li\u003e\n\u003cli\u003e de Toro-Mart\u0026iacute;n, J.\u003cem\u003e et al.\u003c/em\u003e Body mass index is associated with epigenetic age acceleration in the visceral adipose tissue of subjects with severe obesity. \u003cem\u003eClin Epigenetics\u003c/em\u003e\u003cstrong\u003e11\u003c/strong\u003e, 172 (2019). https://doi.org/10.1186/s13148-019-0754-6\u003c/li\u003e\n\u003cli\u003e Bremner, W. J., Vitiello, M. V. \u0026amp; Prinz, P. N. Loss of circadian rhythmicity in blood testosterone levels with aging in normal men. \u003cem\u003eJ Clin.Endocrinol Metab.\u003c/em\u003e\u003cstrong\u003e56\u003c/strong\u003e, 1278-1281 (1983).\u003c/li\u003e\n\u003cli\u003e Skakkebaek, A.\u003cem\u003e et al.\u003c/em\u003e DNA hypermethylation and differential gene expression associated with Klinefelter syndrome. \u003cem\u003eSci. Rep.\u003c/em\u003e\u003cstrong\u003e8\u003c/strong\u003e, 13740-31780 (2018).\u003c/li\u003e\n\u003cli\u003e Bhatnagar, A., Murray, G. \u0026amp; Ray, S. Circadian biology to advance therapeutics for mood disorders. \u003cem\u003eTrends Pharmacol Sci\u003c/em\u003e\u003cstrong\u003e44\u003c/strong\u003e, 689-704 (2023). https://doi.org/10.1016/j.tips.2023.07.008\u003c/li\u003e\n\u003cli\u003e Sulli, G., Manoogian, E. N. C., Taub, P. R. \u0026amp; Panda, S. Training the Circadian Clock, Clocking the Drugs, and Drugging the Clock to Prevent, Manage, and Treat Chronic Diseases. \u003cem\u003eTrends Pharmacol Sci\u003c/em\u003e\u003cstrong\u003e39\u003c/strong\u003e, 812-827 (2018). https://doi.org/10.1016/j.tips.2018.07.003\u003c/li\u003e\n\u003cli\u003e Chang, S.\u003cem\u003e et al.\u003c/em\u003e Anthropometry in Klinefelter Syndrome - Multifactorial Influences Due to CAG Length, Testosterone Treatment and Possibly Intrauterine Hypogonadism. \u003cem\u003eJournal of Clinical Endocrinology \u0026amp; Metabolism\u003c/em\u003e\u003cstrong\u003e100\u003c/strong\u003e, E508-E517 (2015). https://doi.org/10.1210/jc.2014-2834\u003c/li\u003e\n\u003cli\u003e Dong, H.\u003cem\u003e et al.\u003c/em\u003e Metabolic memory: mechanisms and diseases. \u003cem\u003eSignal Transduct Target Ther\u003c/em\u003e\u003cstrong\u003e9\u003c/strong\u003e, 38 (2024). https://doi.org/10.1038/s41392-024-01755-x\u003c/li\u003e\n\u003cli\u003e Caspersen, I. D.\u003cem\u003e et al.\u003c/em\u003e Effect of testosterone treatment during puberty in boys with Klinefelter syndrome (The TIPY Study): protocol for a nationwide randomised, double-blinded, placebo-controlled study. \u003cem\u003eBMJ open\u003c/em\u003e\u003cstrong\u003e15\u003c/strong\u003e, e095628 (2025). https://doi.org/10.1136/bmjopen-2024-095628\u003c/li\u003e\n\u003cli\u003e Martin, A. D., Janssens, V., Caboor, D., Clarys, J. P. \u0026amp; Marfell-Jones, M. J. Relationships between visceral, trunk and whole-body adipose tissue weights by cadaver dissection. \u003cem\u003eAnnals of human biology\u003c/em\u003e\u003cstrong\u003e30\u003c/strong\u003e, 668-677 (2003). https://doi.org/10.1080/03014460310001599590\u003c/li\u003e\n\u003cli\u003e Gast, K. B.\u003cem\u003e et al.\u003c/em\u003e Individual contributions of visceral fat and total body fat to subclinical atherosclerosis: The NEO study. \u003cem\u003eAtherosclerosis\u003c/em\u003e\u003cstrong\u003e241\u003c/strong\u003e, 547-554 (2015). https://doi.org/10.1016/j.atherosclerosis.2015.05.026\u003c/li\u003e\n\u003cli\u003e Bojesen, A.\u003cem\u003e et al.\u003c/em\u003e The metabolic syndrome is frequent in Klinefelter's syndrome and is associated with abdominal obesity and hypogonadism. \u003cem\u003eDiabetes care\u003c/em\u003e\u003cstrong\u003e29\u003c/strong\u003e, 1591-1598 (2006). https://doi.org/10.2337/dc06-0145\u003c/li\u003e\n\u003cli\u003e Skakkebaek, A.\u003cem\u003e et al.\u003c/em\u003e Neuropsychology and brain morphology in Klinefelter syndrome - the impact of genetics. \u003cem\u003eAndrology\u003c/em\u003e (2014). https://doi.org/10.1111/j.2047-2927.2014.00229.x\u003c/li\u003e\n\u003cli\u003e Lykkesfeldt, G.\u003cem\u003e et al.\u003c/em\u003e Abnormal androgen and oestrogen metabolism in men with steroid sulphatase deficiency and recessive X-linked ichthyosis. \u003cem\u003eClinical endocrinology\u003c/em\u003e\u003cstrong\u003e23\u003c/strong\u003e, 385-393 (1985). https://doi.org/10.1111/j.1365-2265.1985.tb01096.x\u003c/li\u003e\n\u003cli\u003e Chang, S.\u003cem\u003e et al.\u003c/em\u003e Testosterone treatment and association with thrombin generation and coagulation inhibition in Klinefelter syndrome: A cross-sectional study. \u003cem\u003eThrombosis research\u003c/em\u003e\u003cstrong\u003e182\u003c/strong\u003e, 175-181 (2019). https://doi.org/10.1016/j.thromres.2019.08.011\u003c/li\u003e\n\u003cli\u003e Guerrero-Romero, F.\u003cem\u003e et al.\u003c/em\u003e The product of triglycerides and glucose, a simple measure of insulin sensitivity. Comparison with the euglycemic-hyperinsulinemic clamp. \u003cem\u003eThe Journal of clinical endocrinology and metabolism\u003c/em\u003e\u003cstrong\u003e95\u003c/strong\u003e, 3347-3351 (2010). https://doi.org/10.1210/jc.2010-0288\u003c/li\u003e\n\u003cli\u003e Amato, M. C.\u003cem\u003e et al.\u003c/em\u003e Visceral Adiposity Index: a reliable indicator of visceral fat function associated with cardiometabolic risk. \u003cem\u003eDiabetes care\u003c/em\u003e\u003cstrong\u003e33\u003c/strong\u003e, 920-922 (2010). https://doi.org/10.2337/dc09-1825\u003c/li\u003e\n\u003cli\u003e Kahn, H. S. The lipid accumulation product is better than BMI for identifying diabetes: a population-based comparison. \u003cem\u003eDiabetes care\u003c/em\u003e\u003cstrong\u003e29\u003c/strong\u003e, 151-153 (2006). https://doi.org/10.2337/diacare.29.1.151\u003c/li\u003e\n\u003cli\u003e Kouri, E. M., Pope, H. G., Jr., Katz, D. L. \u0026amp; Oliva, P. Fat-free mass index in users and nonusers of anabolic-androgenic steroids. \u003cem\u003eClinical journal of sport medicine : official journal of the Canadian Academy of Sport Medicine\u003c/em\u003e\u003cstrong\u003e5\u003c/strong\u003e, 223-228 (1995). https://doi.org/10.1097/00042752-199510000-00003\u003c/li\u003e\n\u003cli\u003e Viuff, M.\u003cem\u003e et al.\u003c/em\u003e X chromosome dosage and the genetic impact across human tissues. \u003cem\u003eGenome Med\u003c/em\u003e\u003cstrong\u003e15\u003c/strong\u003e, 21 (2023). https://doi.org/10.1186/s13073-023-01169-4\u003c/li\u003e\n\u003cli\u003e Patro, R., Duggal, G., Love, M. I., Irizarry, R. A. \u0026amp; Kingsford, C. Salmon provides fast and bias-aware quantification of transcript expression. \u003cem\u003eNat Methods\u003c/em\u003e\u003cstrong\u003e14\u003c/strong\u003e, 417-419 (2017). https://doi.org/10.1038/nmeth.4197\u003c/li\u003e\n\u003cli\u003e Love, M. I.\u003cem\u003e et al.\u003c/em\u003e Tximeta: Reference sequence checksums for provenance identification in RNA-seq. \u003cem\u003ePLoS Comput Biol\u003c/em\u003e\u003cstrong\u003e16\u003c/strong\u003e, e1007664 (2020). https://doi.org/10.1371/journal.pcbi.1007664\u003c/li\u003e\n\u003cli\u003e Argelaguet, R.\u003cem\u003e et al.\u003c/em\u003e Multi-Omics Factor Analysis-a framework for unsupervised integration of multi-omics data sets. \u003cem\u003eMol Syst Biol\u003c/em\u003e\u003cstrong\u003e14\u003c/strong\u003e, e8124 (2018). https://doi.org/10.15252/msb.20178124\u003c/li\u003e\n\u003cli\u003e Vlachou, D.\u003cem\u003e et al.\u003c/em\u003e TimeTeller: A tool to probe the circadian clock as a multigene dynamical system. \u003cem\u003ePLoS Comput Biol\u003c/em\u003e\u003cstrong\u003e20\u003c/strong\u003e, e1011779 (2024). https://doi.org/10.1371/journal.pcbi.1011779\u003c/li\u003e\n\u003cli\u003e Fortin, J. P., Triche, T. J., Jr. \u0026amp; Hansen, K. D. Preprocessing, normalization and integration of the Illumina HumanMethylationEPIC array with minfi. \u003cem\u003eBioinformatics\u003c/em\u003e\u003cstrong\u003e33\u003c/strong\u003e, 558-560 (2017). https://doi.org/10.1093/bioinformatics/btw691\u003c/li\u003e\n\u003cli\u003e Fortin, J. P.\u003cem\u003e et al.\u003c/em\u003e Functional normalization of 450k methylation array data improves replication in large cancer studies. \u003cem\u003eGenome Biol\u003c/em\u003e\u003cstrong\u003e15\u003c/strong\u003e, 503 (2014). https://doi.org/10.1186/s13059-014-0503-2\u003c/li\u003e\n\u003cli\u003e Salas, L. A.\u003cem\u003e et al.\u003c/em\u003e Enhanced cell deconvolution of peripheral blood using DNA methylation for high-resolution immune profiling. \u003cem\u003eNat Commun\u003c/em\u003e\u003cstrong\u003e13\u003c/strong\u003e, 761 (2022). https://doi.org/10.1038/s41467-021-27864-7\u003c/li\u003e\n\u003cli\u003e Peleg\u0026iacute;-Sis\u0026oacute;, D., de Prado, P., Ronkainen, J., Bustamante, M. \u0026amp; Gonz\u0026aacute;lez, J. R. methylclock: a Bioconductor package to estimate DNA methylation age. \u003cem\u003eBioinformatics\u003c/em\u003e\u003cstrong\u003e37\u003c/strong\u003e, 1759-1760 (2021). https://doi.org/10.1093/bioinformatics/btaa825\u003c/li\u003e\n\u003cli\u003e Horvath, S. DNA methylation age of human tissues and cell types. \u003cem\u003eGenome Biol.\u003c/em\u003e\u003cstrong\u003e14\u003c/strong\u003e, R115 (2013).\u003c/li\u003e\n\u003cli\u003e Nakagawa, S. \u0026amp; Schielzeth, H. A general and simple method for obtaining R2 from generalized linear mixed-effects models. \u003cem\u003eMethods in Ecology and Evolution\u003c/em\u003e\u003cstrong\u003e4\u003c/strong\u003e, 133-142 (2013). https://doi.org/https://doi.org/10.1111/j.2041-210x.2012.00261.x\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"nature-portfolio","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"Nature Portfolio","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"ejp","reportingPortfolio":"","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-8850956/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8850956/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Sex chromosome aneuploidies like Klinefelter syndrome (KS) represent a naturally occurring genetic model to investigate metabolic function at the intersection of sex chromosomes, hypogonadism and sex hormone replacement therapy. Men with KS have severe metabolic dysfunction and a more than 2.5-fold increased risk of type 2 diabetes. We assessed body composition, insulin sensitivity, and adipokine profile before and after testosterone replacement therapy (TRT) in KS. TRT does not completely normalize body composition or abolish insulin resistance in KS, indicating insufficient rescue of metabolic risk by correction of hypogonadism. We then assessed gene expression in KS adipose tissue related to the metabolic profile, demonstrating that hypogonadism independently increased adipose tissue expansion and inflammation in KS, and profound disruption of adipose tissue circadian rhythm tightly associated with insulin resistance and signs of biological aging in adipose tissue. We further provide a prediction model for body fat in KS. Our data reveals intricate interactions between sex and metabolism, evolving the pathophysiological understanding of KS.","manuscriptTitle":"Hypogonadism, adipose tissue inflammation, and adipose tissue circadian clock disruption promote metabolic dysfunction in Klinefelter syndrome","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-02-13 03:27:15","doi":"10.21203/rs.3.rs-8850956/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"communications-medicine","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"commsmed","sideBox":"Learn more about [Communications Medicine](http://www.nature.com/commsmed)","snPcode":"43856","submissionUrl":"https://mts-commsmed.nature.com/cgi-bin/main.plex","title":"Communications Medicine","twitterHandle":"@commsmedicine","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Communications Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"bca97c7f-aabb-4af1-ba6b-3f4c3845d283","owner":[],"postedDate":"February 13th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":62728860,"name":"Health sciences/Endocrinology/Endocrine system and metabolic diseases/Gonadal disorders/Hypogonadism"},{"id":62728861,"name":"Health sciences/Endocrinology/Endocrine system and metabolic diseases/Obesity"},{"id":62728862,"name":"Health sciences/Endocrinology/Endocrine system and metabolic diseases/Metabolic syndrome"},{"id":62728863,"name":"Health sciences/Molecular medicine"},{"id":62728864,"name":"Biological sciences/Genetics/Genomics/Transcriptomics"}],"tags":[],"updatedAt":"2026-03-20T13:16:12+00:00","versionOfRecord":[],"versionCreatedAt":"2026-02-13 03:27:15","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8850956","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8850956","identity":"rs-8850956","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}