Abstract
High cortisol production in Cushing’s syndrome leads to fat centralisation. The influence of modest cortisol variations on body shape, however, is less clear. We examined potentially causal associations between morning plasma cortisol and body shape and obesity with inverse-variance weighted random-effects models in a two-sample Mendelian randomisation analysis. We used publicly available summary statistics from the CORtisol NETwork (CORNET) consortium, UK Biobank, and the Genetic Investigation of Anthropometric Traits (GIANT) consortium. Only in women, morning plasma cortisol (proxied by ten genetic polymorphisms) was associated positively with waist size reflected in waist-to-hip index (WHI, 0.035 standard deviation (SD) units change per one SD cortisol increase; 95% confidence interval (0.002–0.067); p = 0.036) and “a body shape index” (ABSI; 0.039 (0.006–0.071); p = 0.021). There was no evidence for associations with hip index (HI) or body mass index (BMI). Among individual polymorphisms, rs7450600 stood out (chromosome 6; Long Intergenic Non-Protein-Coding RNA 473 gene, LINC00473). Morning plasma cortisol proxied by rs7450600 was associated positively with WHI and inversely with HI and BMI in women and men. Our findings support a causal association of higher morning plasma cortisol with larger waist size in women and highlight LINC00473 as a genetic link between morning plasma cortisol and body shape.
Similar content being viewed by others
Introduction
Cortisol is an adrenal glucocorticoid hormone and plays a key role in the regulation of energy expenditure, fat distribution, and lipid metabolism1. It is well known that a prolonged exposure to very high glucocorticoid levels, as in Cushing’s syndrome, leads to fat centralisation and development of abdominal obesity2, which are reversed after adrenalectomy3. More modest cortisol increases, as in functional adrenal incidentalomas with autonomous cortisol production, similarly lead to larger visceral fat depots4. It is less clear, however, whether more modest variations of cortisol levels can alter body shape and contribute to the development of abdominal obesity5. Investigating the determinants of body shape is important because fat depots are differentially associated with cardiometabolic conditions, positively for visceral fat accumulation but inversely for gluteofemoral fat accumulation6. Although cross-sectional observational studies have shown that individuals with abdominal obesity have higher free cortisol excretion and higher responsiveness to stimulation with corticotropin releasing hormone (CRH) compared to individuals with peripheral fat distribution7, confirming a causal relationship would require prospective interventional studies, which are harder to perform. Mendelian randomisation (MR) analysis is a more practical approach, which uses single nucleotide polymorphisms (SNPs) as instrumental variables (IVs) and permits the interpretation of associations between the genetically predicted exposure and the outcome as potentially causal, conditional on a set of assumptions8.
The main genetic region associated with the inter-individual variability of morning plasma cortisol was identified in a genome-wide association meta-analysis (GWAMA) by the CORtisol NETwork (CORNET) consortium and includes the SERPINA6 gene (coding corticosteroid binding globulin (CBG), the main cortisol carrier protein) and SERPINA1 gene (coding α1-antitrypsin, which inhibits the cleavage and inactivation of CBG by neutrophil elastase)9. A more recently updated GWAMA has confirmed the importance of the SERPINA6/SERPINA1 locus but could not identify further loci with genome-wide significance, despite doubling the sample size and tripling the number of genetic polymorphisms10. Although mutations in SERPINA6 have been associated with attenuated clinical features of cortisol deficit11 and common genetic polymorphisms in the SERPINA6 locus have been associated with altered cortisol release10, it cannot be assumed that all polymorphisms related to CBG levels in blood would additionally affect cortisol release. Hence, there is a need to diversify the genetic instrument for cortisol and to include genetic variants outside the SERPINA6 locus.
Based on IVs derived from the first GWAMA9, an MR study has shown inverse associations between genetically predicted morning plasma cortisol and general obesity class I12. These findings, however, could not be confirmed using body mass index (BMI) on a continuous scale and IVs for cortisol identified in the updated GWAMA10. No study, to our knowledge, has so far used MR to examine potentially causal associations of cortisol with body shape. Moreover, waist and hip circumferences and the waist-to-hip ratio (WHR), which have traditionally been used as indices of body shape, are correlated strongly positively with BMI and thus reflect general obesity in addition to abdominal obesity and body shape13. The allometric “a body shape index” (ABSI), on the other hand, reflects waist circumference among individuals with the same weight and height and is thus uncorrelated with BMI14. In analogy to ABSI, hip index (HI) and waist-to-hip index (WHI) have been defined as the allometric equivalents of hip circumference and WHR, correspondingly, and are uncorrelated with BMI15,16.
In this study, we have examined, separately in women and men, whether a MR analysis provides evidence for a potentially causal association of genetically predicted morning plasma cortisol with body shape (as reflected in the allometric indices WHI, ABSI, and HI) and with general obesity (as reflected in BMI). To highlight the similarity of traditional body shape indices with BMI and their difference from allometric body shape indices, we have compared WHR with BMI and WHI.
Methods
Overview
MR analysis involves the use of genetic variants in an IV analysis framework and is based on the principle that genotypes are not associated with environmental confounders and are not affected by reverse causation8. The following three assumptions must be satisfied for the genetic instrument and the corresponding causal inference to be deemed valid: (i) the genetic variants are strongly associated with the exposure (morning plasma cortisol); (ii) the genetic variants have no direct effect on the outcome (body shape and obesity) and influence the outcome only via the exposure (morning plasma cortisol); (iii) the genetic variants are not associated with confounders (measured or unmeasured) of the exposure-outcome association.
We conducted a two-sample MR study, using publicly available summary statistics for gene-exposure and gene-outcome association estimates based on distinct and non-overlapping populations consisting of participants with European ancestry. We report our findings according to the MR-STROBE guidance17.
Data sources
The genetic instrument for the exposure, morning plasma cortisol, was based on publicly available summary statistics from the latest GWAMA update of the CORNET consortium, which included 8,452,427 SNPs for up to 25,314 individuals from 17 population-based cohorts of European ancestry10. Linear regression had been performed jointly in women and men with z-scores of log-transformed morning plasma cortisol (standard deviation (SD) scale) and adjustment for sex, age, and ten cohort-specific principal components of genetic ancestry10. Sex-specific analyses were not available, as the authors had explained in their previous GWAMA, that they could not identify sex differences when examining separately women and men9.
Sex-specific genetic variants associated with the outcome body shape were extracted from the publicly available summary statistics of a genome-wide association study (GWAS) of allometric body shape indices based on 219,872 women and 186,825 men of European ancestry from the UK Biobank cohort15. WHI, ABSI, and HI had been calculated according to the general formula: Z × Weightβ × Heightγ, where Z represented either WHR, waist circumference, or hip circumference, correspondingly, and the power coefficients β and γ had been calibrated for UK Biobank, i.e. they were derived based on UK Biobank data. Bayesian linear mixed-models (BOLT-LMM) had been performed, following inverse normal transformation of WHI, ABSI, and HI to SD scale (normalised SD-unit) using Blom’s method, and adjusting the models for age, age2, and an indicator of genotyping array15. Analyses combining women and men were not available.
Sex-specific genetic variants associated with the outcome general obesity reflected in BMI and with the traditional body shape index WHR were extracted from the publicly available summary statistics of a meta-analysis based on individuals of European ancestry (434,794 women and 374,756 men for BMI; 381,152 women and 316,772 men for WHR)18, which combined a GWAS with a linear mixed model in UK Biobank data and publicly available GWAS summary statistics obtained from the Genetic Investigation of Anthropometric Traits (GIANT) consortium19. BMI and WHR were expressed in SD units. No GWAS summary statistics of equivalent quality were available for waist and hip circumferences.
Selection of genetic instruments
To improve the reliability of our findings, we included in the selection of genetic instruments SNPs with gene-exposure associations estimated in at least 20,000 individuals (7,300,058 SNPs, 86.4% of all available SNPs). We used the SNP2GENE function of the web-based platform for Functional Mapping and Annotation (FUMA) v1.6.0 to perform positional mapping, clumping, and annotation of genetic variants20. Linkage disequilibrium (LD) mapping was based on the 1000G Phase3 EUR reference panel. All relevant genetic variants were mapped to genes within a maximum distance of 1 kb based on their genomic position on GRCh37 (hg19) and Ensemble genes v110 with ANNOVAR employed in FUMA21. We used the deleteriousness score (Combined Annotation Dependent Depletion (CADD) score) provided by FUMA as a measure of pathogenicity of a given genetic variant and evaluated this with respect to the recommended cut-off 12.3722.
In our main analysis, following the algorithm and terminology used in FUMA20, independent significant SNPs were defined as genetic variants associated with the exposure with significance p < 5 × 10–6 and only in a weak LD with each other at r2 < 0.6 (first clumping step). Genetic variants associated with the exposure at p < 0.05 and in LD with an independent significant SNP at r2 ≥ 0.6 constituted the corresponding high-LD block with candidate SNPs. In a second clumping step, independent significant SNPs were clumped to identify lead SNPs associated with the exposure at p < 5 × 10–6 and independent from each other at r2 < 0.05 (the lowest r2 accepted by FUMA). The lead SNPs identified by FUMA constituted the main genetic instrument (IVA). Lead SNPs with LD block boundaries closer than 250 kb were joined in a genetic locus. We used a more lenient significance criterion in our main analysis (p < 5 × 10–6) to allow for the inclusion of a wider range of genetic loci, since the conventional conservative genome-wide significance level (p < 5 × 10–8) had previously identified only the CBG-related locus on chromosome 1410. We applied, however, a conservative LD threshold (r2 < 0.05 rather than r2 < 0.3 used in Ref.10), to ensure independence of the selected SNPs. We used FUMA for IV selection because the lead SNPs are equivalent to clumping with PLINK at the specified p-value and r2 thresholds, but FUMA additionally permits a visualisation of the LD structure of the relevant genetic region20.
In a secondary analysis, we used FUMA to obtain an alternative genetic instrument (IVB), defining lead SNPs with a conservative genome-wide significance threshold at p < 5 × 10–8 but with a lenient threshold for independence at r2 < 0.3, similarly to Crawford et al.10.
We orientated the genetic variants of the outcome such that the effect allele corresponded to the minor allele (mean allele frequency ≤ 0.5). We aligned the genetic variants of the exposure to match the minor allele of the outcome, irrespective of the allele frequency of the exposure. Thus, the signs of the gene-outcome and gene-exposure association estimates for each genetic variant correspond to the allele representing the minor allele in the morning plasma cortisol dataset. For lead SNPs with unavailable gene-exposure association estimates, we used replacement candidate SNPs with p < 5 × 10–6 and r2 ≥ 0.6 with the corresponding lead SNP, when such were available. In the text below, SNPs are labelled with chromosome_rsID.
To characterise further the genetic determinants of morning plasma cortisol, we performed gene-based association analysis with Mutimarker Analysis of GenoMic Annotation (MAGMA) v.1.0823, including only genetic variants with association estimates based on at least 20,000 individuals. Gene-based analysis derives a SNP-wide mean model for each individual protein-coding gene. Significant were considered genes with p < 0.05, after Bonferroni correction for the number of identified protein-coding genes. Due to the limited power of the available GWAMA of morning plasma cortisol, we used gene Q-Q plots to identify additional genes of potential interest which showed substantial nominal significance but could not reach the Bonferroni adjusted significance cut-off.
Statistical MR analysis
To calculate causal estimates for each individual SNP included in the genetic instrument, we used the Wald ratio (SNP-outcome divided by SNP-exposure regression coefficient)24. To provide some information for potential pathogenicity, we performed sensitivity analyses with the corresponding high-CADD analogue, defined as the candidate SNP from the corresponding LD block with p < 5 × 10–5 and the highest CADD > 12.37, when SNPs fulfilling these criteria were available.
To calculate causal estimates for morning plasma cortisol overall, we combined the SNP-specific Wald ratio estimates in an inverse variance weighted (IVW) analysis24. In our main analysis (IVA), which included independent SNPs across the genome, we used random-effects IVW because this allows the mean effects of the individual SNPs to differ due to horizontal pleiotropy and provides an unbiased estimate when horizontal pleiotropy is balanced25. In our secondary analysis (IVB), which included partly correlated SNPs from the same genetic locus, we used fixed-effect IVW because this assumes that all SNPs have the same effect. In sensitivity analysis, we performed fixed-effect IVW for IVA and random-effects IVW for IVB, to check the robustness of our findings. For validation, we examined heterogeneity in the IVW estimates with the Cochran Q statistic and the corresponding test for heterogeneity (pheterogeneity) and the I2 metric of inconsistency25. To detect outliers, we additionally performed a sensitivity analysis with MR-PRESSO (Mendelian Randomization Pleiotropy RESidual Sum and Outlier), which includes a global test for heterogeneity (considering all SNPs jointly), a local test for heterogeneity (identifying outlier SNPs), and a distortion test (comparing the causal estimate before and after removal of the detected outliers)26. To assess horizontal pleiotropy, we used as sensitivity analyses two methods which make different IV assumptions: (i) weighted median method, which allows for some SNPs to be invalid instruments, as long as they account for less than 50% of the information27; (ii) MR-Egger regression, which provides consistent estimates when all SNPs are invalid, conditional on the InSIDE (Instrument Strength Independent of Direct Effect) assumption, stating that the direct (pleiotropic) effects of the genetic variants on the outcome are independent of the associations of the genetic variants with the exposure28. We used the MR-Egger intercept to evaluate potential IV violations, as this provides an estimate of the average pleiotropic effect and a test for directional pleiotropy (an intercept with p-value < 0.05 indicates the presence of horizontal pleiotropy)28.
To assess the strength of each SNP as a genetic instrument, we calculated an F statistic using as approximation the squared ratio of the gene-exposure association and the corresponding standard error (βX2/σX2). To minimise weak instrument bias, we considered as acceptable strength of the gene-exposure association F > 1029. To evaluate the reliability of MR Egger estimates, we used the I2GX statistic, which when low (< 90%) indicates a violation of the NOME (NO Measurement Error) assumption that the exposure is measured without measurement error30.
The estimates obtained from the MR analysis quantify the change in each outcome on an SD scale (SDchange) per one SD increase in genetically predicted morning plasma cortisol. All p-values were two-sided. MR associations with p < 0.05 were considered statistically significant.
Data analyses and visualisation were performed with R version 4.1.3 (using the “MendelianRandomisation” v.0.7.0 and “ggplot2” v.3.3.5 packages)31. MR-PRESSO was run in R version 4.3.1.
Ethics approval and consent to participate
This research did not involve individual level data and used only publicly available summary statistics generated from previously published studies, referenced in the manuscript, which had obtained ethics approval and informed consent from study participants in accordance with the Declaration of Helsinki.
Results
Genetic instruments
Our main selection strategy (p < 5 × 10–6, r2 < 0.05) identified a genetic instrument including ten lead SNPs (IVA): three in the SERPINA6 locus (14_rs11620763, 14_rs9989237, 14_rs7161231) and seven on other chromosomes: 3_rs1868602 (mapped to TMEM108 gene on chromosome 3), 4_rs13151695 (EEF1A1P9), 5_rs115656533 (SERINC5), 5_rs6873320 (SERINC5), 6_rs7450600 (LINC00473), 9_rs140738399 (RPS6P12), 10_rs142967045 (KIAA1598) (Table 1). Morning plasma cortisol was lower for the minor allele of 4_rs13151695, 5_rs115656533, 5_rs6873320, 6_rs7450600, 10_rs142967045, and 14_rs11620763 and higher for the minor allele of the other four lead SNPs. Summary statistics for all ten lead SNPs were available for allometric body shape indices. Three SNPs, however, were unavailable for BMI and WHR, so we replaced 4_rs13151695 with 4_rs9996658 (r2 = 1.000 between the two SNPs) and 14_rs11620763 with 14_rs7141205 (r2 = 1.000), but there was no appropriate replacement for 9_rs140738399. Thus, the adapted IVA# included nine lead SNPs in the analyses for BMI and WHR. To facilitate the comparisons with WHR, we performed an additional matching analysis for WHI, using the adapted IVA#. None of the SNPs included in IVA or their substitutes had substantial deleteriousness score (largest CADD = 7.24). There were, however, high-LD candidate SNPs with CADD > 12.37 for three of the lead SNPs. Thus, 3_rs6776118 was the high-CADD analogue for 3_rs1868602 (r2 = 0.748 between the two SNPs), 4_ rs13104830 for 4_rs13151695 (r2 = 0.946), and 6_rs480621 for 6_rs7450600 (r2 = 0.901) (Table 1, Fig. 1).
The secondary selection strategy (p < 5 × 10–8, r2 < 0.3) reproduced as lead SNPs for IVB the four SNPs identified by Crawford et al.10 in the SERPINA6 locus (14_rs11620763, 14_rs7146221, 14_rs9989237, 14_rs2736898) (Table 1, Fig. 1).
The F statistics for SNPs outside chromosome 14 were around 20 (17.2 to 25.2) and were higher for SNPs within the SERPINA6 locus (27.5 to 81.4), but only SNPs on chromosome 14 had shown significant heterogeneity between the studies contributing to the GWAMA (Table 1).
Few lead SNPs showed gene-outcome associations with nominal statistical significance, except 6_rs7450600 and its high-CADD analogue 6_rs480621 (LINC00473), which were associated positively with HI and BMI in women and men (Supplementary Table S1 for WHI, ABSI, and HI; Supplementary Table S2 for BMI).
In the gene-based analysis, only SERPINA1 and SERPINA6 were associated with morning plasma cortisol at significance level p < 0.05, after Bonferroni adjustment for 18,814 identified protein-coding genes. The gene Q–Q plot, however, indicated that five more genes had higher than expected nominal significance including (in descending order of the unadjusted p-value): SERPINA10 (chromosome 14), ETNK1 (chromosome 12), TMEM108 (chromosome 3), SERINC5 (chromosome 5), and PPP4R4 (chromosome 14), all with p ≤ 0.0001 (Supplementary Fig. S1).
MR associations
Only in women, morning plasma cortisol proxied genetically by IVA was associated positively with WHI (SDchange = 0.035; 95% confidence interval (CI) 0.002–0.067; p = 0.036) and ABSI (SDchange = 0.039; 95% confidence interval (CI) 0.006–0.071; p = 0.021), with no evidence for heterogeneity between individual SNPs. There was no evidence for associations with WHI or ABSI in men, or with HI and BMI in women or men. There was, however, evidence for heterogeneity for HI in women and men and for WHI and BMI in men.
Among individual SNPs, morning plasma cortisol proxied genetically by 6_rs7450600 (LINC00473) was associated positively with WHI in both women (SDchange = 0.167; 95% CI 0.062–0.272; p = 0.002) and men (SDchange = 0.233; 95% CI 0.117–0.348; p = 8 × 10–5) (Fig. 1). There were associations with ABSI in the positive direction without nominal significance, but the positive associations with WHI were mainly accounted for by inverse associations with HI in both women (SDchange = − 0.168; 95% CI − 0.277 to − 0.058; p = 0.003) and men (SDchange = − 0.203; 95% CI − 0.322 to − 0.084; p = 0.0008). In addition, morning plasma cortisol proxied genetically by 6_rs7450600 was associated inversely with BMI in both women (SDchange = − 0.133; 95% CI − 0.233 to − 0.032; p = 0.010) and men (SDchange = − 0.170; 95% CI − 0.280 to − 0.060; p = 0.003). Only in women, morning plasma cortisol proxied genetically by 4_rs13151695 (EEF1A1P9) was associated positively with ABSI (SDchange = 0.133; 95% CI 0.022–0.244; p = 0.019) as well as with HI (SDchange = 0.149; 95% CI 0.035–0.263; p = 0.010) but not with WHI or BMI (Fig. 1). Further, in women and men, morning plasma cortisol proxied genetically by 10_rs142967045 (KIAA1598) was associated inversely with BMI, less prominently in women (SDchange = − 0.141; 95% CI − 0.284 to 0.003; p = 0.055) than in men (SDchange = − 0.189; 95% CI − 0.345 to − 0.032; p = 0.018). Last, only in women, morning plasma cortisol proxied genetically by 14_rs9989237 (SERPINA6) was associated positively with HI (SDchange = 0.095; 95% CI 0.013–0.177; p = 0.024) (Fig. 2).
Morning plasma cortisol proxied genetically by IVB (p < 5 × 10–8, r2 < 0.3) was associated positively with ABSI in women only (SDchange = 0.065; 95% CI 0.020–0.110; p = 0.005) and with BMI in men only (SDchange = 0.048; 95% CI 0.0004–0.096; p = 0.048). A positive association with HI was observed in both women (SDchange = 0.088; 95% CI 0.042–0.134; p = 0.0002) and men (SDchange = 0.055; 95% CI 0.004–0.105; p = 0.034). There was no evidence for heterogeneity between individual SNPs, although proxying morning plasma cortisol by 14_rs7146221 (SERPINA6), showed the strongest associations (Fig. 3).
Sensitivity analyses
Proxying morning plasma cortisol by the high-CADD variants 3_rs6776118, 4_ rs13104830, and 6_rs480621 showed similar association patterns to the corresponding lead SNPs in IVA (3_rs1868602, 4_rs13151695, and 6_rs7450600) (Fig. 4).
Morning plasma cortisol proxied genetically by IVA# showed intermediate associations with WHR compared to WHI and BMI, with the most prominent difference noted for 6_rs7450600, for which a positive association with WHI and an inverse with BMI corresponded to a nil association with WHR in both women and men (Fig. 5).
Association estimates obtained with fixed-effect IVW and random-effects IVW were identical when there was no evidence for heterogeneity (low I2) and showed no material difference even when there was evidence for heterogeneity, except for a wider confidence interval of the positive association of morning plasma cortisol proxied by IVB with ABSI in women when using random-effects IVW (SDchange = 0.065; 95% CI − 0.0006 to 0.131; p = 0.052) (Supplementary Table S3). The global test of MR-PRESSO, similarly to Cochran’s Q, indicated heterogeneity for HI in women and for WHI and BMI in men when using IVA, and the local tests identified 6_rs7450600 as an outlier, but there was no evidence for distortion and the conclusion of no association with these indices remained. Association estimates obtained with the weighted median method confirmed the findings of the main IVW random-effects analysis. There was no evidence for pleiotropy based on the MR Egger intercept, but MR Egger estimates had very wide confidence intervals for all analyses and low I2GX < 50 (Supplementary Table S3).
Discussion
Morning plasma cortisol was associated positively with WHI and ABSI in women, when proxied genetically by SNPs across the genome, and inversely with HI and BMI in women and men, when proxied individually by 6_rs7450600 in LINC00473 locus. Morning plasma cortisol, however, was associated positively with HI in women and men when proxied genetically by SNPs confined to the CBG-related locus SERPINA6. Our findings are compatible with a causal association of higher morning plasma cortisol with larger waist size in women but show conflicting causal conclusions for hip size.
The interpretation of the associations of total morning plasma cortisol levels with any outcome should account for considerable biological and analytical caveats. Importantly, more than 90% of total blood cortisol is bound to CBG, while only free cortisol is traditionally considered biologically active32. On the one hand, CBG levels could be relevant to cortisol action since genetic variations in CBG can affect its affinity for cortisol and can modulate cortisol release and action at tissue level33. Crawford et al. have, indeed, shown that eQTL and GWAMA signals colocalise within a region of SERPINA6 represented by 14_rs2736898 and have argued that this polymorphism can influence cortisol release at tissue level10. On the other hand, however, a similar functionality cannot be assumed for other polymorphisms in the SERPINA6 locus and, when total plasma cortisol levels simply reflect CBG levels, the cortisol-driven negative feed-back to the hypothalamus-pituitary-adrenal (HPA) axis would adjust adrenal cortisol production and maintain constant free cortisol levels34. Further limiting the informativeness of circulating cortisol is the local interconversion between the active cortisol and the inactive cortisone by 11β-hydroxysteroid dehydrogenases, which regulate cortisol availability and action at tissue level35. Furthermore, a single cortisol measurement does not reflect the circadian and ultradian patterns of cortisol secretion and their alterations in disease states36. Similarly, blood levels are a short-term measure of cortisol status, while the glucocorticoid receptor (GR) shows a differential metabolic response to acute and chronic cortisol exposure, with lipolysis in the former and fat accumulation and obesity in the latter case37. There are also variations in the glucocorticoid sensitivity of the GR, which affect cortisol action at tissue level38. The above considerations determine the need to expand IVs for morning plasma cortisol outside the CBG-related SERPINA6 region and to interpret with caution causal inference based solely on IVs in this region39. In addition to the biological considerations, there are also analytical limitations in cortisol measurements by immunoassays, which have been used by most studies contributing to the CORNET consortium10, as immunoassays have low specificity and can be subject to interference by cortisol precursors40. Such interferences would be particularly problematic in obesity, when there is an underlying adrenal disfunction but cortisol levels are not particularly high5. Bearing in mind the highlighted limitations of morning plasma cortisol measurements, we have compared below our findings with the results of previous studies and have discussed the MR assumptions.
The available knowledge on associations of cortisol with obesity is largely derived from population-based observational studies with a relatively small sample size and inconsistent findings, almost always with a cross-sectional design, often including only women or only men, and considering exclusively waist circumference and WHR as indices of body shape, with little attention paid to hip size5. Focusing on morning blood cortisol measurements, in agreement with our results, a meta-analysis of 26 observational studies found little evidence for associations with BMI, although there was a tendency towards lower levels in obese compared to non-obese individuals41. There was similarly little evidence for association with BMI on a continuous scale when using IVs located in SERPINA6 in a two-sample MR10. A previously described inverse association with obesity class I (BMI ≥ 30 to < 35 kg/m2) and not above, although based on cortisol-related SNPs across the genome12, had derived IVs using the earlier smaller-size GWAMA and these did not overlap with our IVs. In contrast and in compliance with the expectations of the Cushing’s syndrome paradigm, a large meta-analysis of cortisol levels in hair, which are not dependent on CBG levels, provided robust observational evidence in support of a positive association with BMI (122 studies; 26,527 participants), including in the analyses restricted to studies using mass-spectrometry-based measurements as opposed to immunoassays42. Considering body shape indices, the associations with ABSI as a measure of waist size in our study were consistent between IVA, which included a wider range of genetic instruments, and IVB, which was confined to SEPRINA6 locus. They also agree with the findings of the large meta-analysis of cortisol levels in hair, which reported positive associations with waist circumference (24 studies; 11,004 participants) and WHR (16 studies; 6,786 participants)42. For HI, however, which to date has only been examined in our study, there was a marked difference between the null association with IVA and the positive association with IVB in both women and men. Considering the relatedness of total plasma cortisol and CBG levels32, a positive association with HI likely reflects an influence of oestrogens on CBG levels, because CBG is higher in women compared to men43 and even higher in women using oestrogens44. At the same time, peripheral oestrogen production is highest in gluteofemoral adipose tissue45 and serum oestradiol is higher for larger HI46.
Notwithstanding the described associations, our study has shown that there is very little overlap between the genetic determinants of morning plasma cortisol, on the one hand, and body shape indices and BMI, on the other hand, especially given that the latter were based on biobank size datasets with thousands of SNPs reaching genome-wide significance15,18. This is not completely unexpected, since the genetic determinants of body shape indices reflected mainly links with morphogenesis and embryogenesis and not a dynamic influence of steroid hormones15. The main notable exception was 6_rs7450600 variant and its high-CADD analogue 6_rs480621 in LINC00473 (Long Intergenic Non-Protein-Coding RNA 473) locus. These SNPs showed similar association patterns in women and men, which are thus more likely to be reproducible because sex-specific datasets could be considered a discovery and a validation cohort for each other. The LINC00473 gene is especially interesting because the corresponding long non-coding RNA stimulates thermogenesis in brown adipose tissue and shows higher expression in supraclavicular than in abdominal subcutaneous adipose tissue but lower expression in obesity, which has impaired thermogenesis47. Given that in our study, BMI and HI were higher for the minor allele C of 6_rs7450600, this would likely represent a loss of function variant. The corresponding lower morning plasma cortisol levels are compatible with the role of glucocorticoids for stimulation of thermogenesis48. LINC00473 is also considered oncogenic, as it is upregulated in various cancers49, promoting proliferation, migration, and invasion, and is associated with worse cancer survival50. Thus, the role of LINC00473 in cortisol-related pathology merits further investigation. Of some further interest for cortisol is also TMEM108 (transmembrane protein 108) gene, which included multiple SNPs in LD and was ranked high in the gene-based MAGMA analysis. There was also a high-CADD analogue to the lead SNP but neither showed associations with body shape indices or BMI when used as genetic proxy of morning plasma cortisol in our study. TMEM108 has not shown associations with body weight in animal studies either but has shown metabolic effects for glucose and lipid metabolism51.
Considering the MR assumptions, although our genetic variants were reasonably strongly associated with the exposure, a caveat remains that those outside the SERPINA6 locus did not reach genome-wide significance and would require further validation. A major impediment is the low SNP-based heritability of morning plasma cortisol52, considerably lower than the estimates based on twin studies53. This drives a need for a substantial further increase in sample size of a future GWAMA and a need to examine genetic instruments for longer-term measures of cortisol exposure independent of CBG levels, such as cortisol levels in hair52. Regarding pleiotropy, it could be reasonably assumed that SERPINA6 variants influence body shape only via modulating cortisol levels, but further knowledge of the mechanisms underlying the remaining variants in IVA would be required to clarify their relationships with body shape, especially since some variants stood out with clearer and stronger associations than others. The MR Egger test for pleiotropy would not be reliable, given that a low I2GX indicated that the NOME assumption is unlikely to hold. A violation of NOME is not unexpected, since cortisol exposure is determined by cortisol availability and action at cellular level, which may differ from total cortisol measured in blood due to the biological and analytical considerations discussed above. Regarding unmeasured confounders, these could not be disregarded either, as cortisol interacts biologically with sex steroids. Thus, in addition to increasing CBG levels44, oestrogens counter abdominal obesity via modification of the sensitivity of the glucocorticoid receptor54, while testosterone can supress adrenal cortisol production55 and is associated with body shape in a sex-specific pattern46. Multivariable analyses, however, were not feasible in our study because sex-steroid-related IVs suffer from the same limitation as cortisol-related IVs. Sex steroids are primarily bound to sex hormone binding globulin56 and their free fractions are unknown, their measurement is hampered by the low specificity and sensitivity of immunoassays and the diurnal and menstrual variations57,58, and they are subject to local tissue interconversions between the active and inactive forms59.
A strength of our study is the broader range of SNPs included in our genetic instrument, which was based on the latest and largest GWAMA of morning plasma cortisol. We have also used allometric indices to evaluate body shape independent of BMI and have re-iterated that traditional body shape indices resemble BMI in their associations and are thus unable to evaluate body shape independent of obesity. Nevertheless, a major limitation of our study is the relatively low power of the GWAMA of morning plasma cortisol compared to the biobank size GWAS of body shape indices and BMI, determining a need for validation of genetic variants outside the SERPINA6 locus. Further, cortisol measurements in the studies contributing data to CORNET had been performed mainly with immunoassays10, which lack specificity and allow for a larger measurement error. Furthermore, while samples collected during the morning hours had been used, their timing and their relationship with the time of awakening had not been standardised, thus contributing to a larger variability. In addition, a single measurement would not reflect diurnal variations and longer-term cortisol status. Importantly, no sex-specific genetic association estimates for morning plasma cortisol were available, when HPA axis responsiveness shows sex differences60, and we were thus unable to derive sex-specific IVs. Similarly, age-specific estimates and separate estimates for pre- and post-menopausal women were also lacking, so we were unable to examine the influence of age and menopausal status on the observed associations. The publicly available summary statistics were derived from studies including only participants with European ancestry, so we could not examine ethnic differences. Not least, plasma cortisol as well as body shape are complex traits and are unlikely to fully reflect the underlying biological traits of interest, cortisol exposure at cellular level and body composition, correspondingly.
In conclusion, our findings support a causal association of higher morning plasma cortisol with larger waist size in women and highlight LINC00473 as a potential link between morning plasma cortisol levels and body shape, which merits further investigation.
Data availability
The datasets analysed in the current study are publicly available and can be accessed from the following locations. The summary statistics for morning plasma cortisol generated by Crawford et al.10, for women and men combined, can be downloaded from https://datashare.ed.ac.uk/handle/10283/3836. The summary statistics for allometric body shape indices generated by Christakoudi et al.15, separately for women and men, can be downloaded from the NHGRI-EBI GWAS Catalog at https://www.ebi.ac.uk/gwas/publications/34021172 (select option FTP Download). The summary statistics for BMI and WHR generated by Pulit et al.18, separately for women and men, can be downloaded from https://zenodo.org/record/1251813#.XCLJ7vZKhE4. The associated FUMA results will be made publicly available upon acceptance at https://fuma.ctglab.nl.
Abbreviations
- ABSI:
-
A body shape index
- BMI:
-
Body mass index
- CADD:
-
Combined annotation dependent depletion score
- CI:
-
Confidence interval
- CORNET:
-
CORtisol NETwork consortium
- FUMA:
-
Functional Mapping and Annotation
- GIANT:
-
Genetic Investigation of Anthropometric Traits consortium
- GWAMA:
-
Genome-wide association meta-analysis
- HI:
-
Hip index
- MAGMA:
-
Multimarker Analysis of GenoMic Annotation
- SD:
-
Standard deviation
- SNP:
-
Single nucleotide polymorphism
- WHI:
-
Waist-to-hip index
- WHR:
-
Waist-to-hip ratio
References
Mir, N., Chin, S. A., Riddell, M. C. & Beaudry, J. L. Genomic and non-genomic actions of glucocorticoids on adipose tissue lipid metabolism. Int. J. Mol. Sci. 22, 503. https://doi.org/10.3390/ijms22168503 (2021).
Delivanis, D. A. et al. Impact of hypercortisolism on skeletal muscle mass and adipose tissue mass in patients with adrenal adenomas. Clin. Endocrinol. (Oxf.) 88, 209–216. https://doi.org/10.1111/cen.13512 (2018).
Hong, N. et al. Changes of computed tomography-based body composition after adrenalectomy in patients with endogenous hypercortisolism. Clin. Endocrinol. (Oxf.) 90, 267–276. https://doi.org/10.1111/cen.13902 (2019).
Yener, S. et al. Autonomous cortisol secretion in adrenal incidentalomas and increased visceral fat accumulation during follow-up. Clin. Endocrinol. (Oxf.) 87, 425–432. https://doi.org/10.1111/cen.13408 (2017).
Incollingo Rodriguez, A. C. et al. Hypothalamic–pituitary–adrenal axis dysregulation and cortisol activity in obesity: A systematic review. Psychoneuroendocrinology 62, 301–318. https://doi.org/10.1016/j.psyneuen.2015.08.014 (2015).
Piché, M. E., Tchernof, A. & Després, J. P. Obesity phenotypes, diabetes, and cardiovascular diseases. Circ. Res. 126, 1477–1500. https://doi.org/10.1161/circresaha.120.316101 (2020).
Pasquali, R. et al. The hypothalamic–pituitary–adrenal axis in obese women with different patterns of body fat distribution. J. Clin. Endocrinol. Metab. 77, 341–346. https://doi.org/10.1210/jcem.77.2.8393881 (1993).
Haycock, P. C. et al. Best (but oft-forgotten) practices: The design, analysis, and interpretation of Mendelian randomization studies. Am. J. Clin. Nutr. 103, 965–978. https://doi.org/10.3945/ajcn.115.118216 (2016).
Bolton, J. L. et al. Genome wide association identifies common variants at the SERPINA6/SERPINA1 locus influencing plasma cortisol and corticosteroid binding globulin. PLoS Genet. 10, e1004474. https://doi.org/10.1371/journal.pgen.1004474 (2014).
Crawford, A. A. et al. Variation in the SERPINA6/SERPINA1 locus alters morning plasma cortisol, hepatic corticosteroid binding globulin expression, gene expression in peripheral tissues, and risk of cardiovascular disease. J. Hum. Genet. 66, 625–636. https://doi.org/10.1038/s10038-020-00895-6 (2021).
Meyer, E. J. et al. CBG Montevideo: A clinically novel SERPINA6 mutation leading to haploinsufficiency of corticosteroid-binding globulin. J. Endocr. Soc. 5, 115. https://doi.org/10.1210/jendso/bvab115 (2021).
Qi, X., Cui, B. & Cao, M. The role of morning plasma cortisol in obesity: A bidirectional Mendelian randomization study. J. Clin. Endocrinol. Metab. 107, e1954–e1960. https://doi.org/10.1210/clinem/dgac008 (2022).
Christakoudi, S., Tsilidis, K. K., Evangelou, E. & Riboli, E. Association of body-shape phenotypes with imaging measures of body composition in the UK Biobank cohort: Relevance to colon cancer risk. BMC Cancer 21, 1106. https://doi.org/10.1186/s12885-021-08820-6 (2021).
Krakauer, N. Y. & Krakauer, J. C. A new body shape index predicts mortality hazard independently of body mass index. PLoS ONE 7, e39504. https://doi.org/10.1371/journal.pone.0039504 (2012).
Christakoudi, S., Evangelou, E., Riboli, E. & Tsilidis, K. K. GWAS of allometric body-shape indices in UK Biobank identifies loci suggesting associations with morphogenesis, organogenesis, adrenal cell renewal and cancer. Sci. Rep. 11, 10688. https://doi.org/10.1038/s41598-021-89176-6 (2021).
Krakauer, N. Y. & Krakauer, J. C. An anthropometric risk index based on combining height, weight, waist, and hip measurements. J. Obes. 2016, 8094275. https://doi.org/10.1155/2016/8094275 (2016).
Skrivankova, V. W. et al. Strengthening the reporting of observational studies in epidemiology using Mendelian randomization: The STROBE-MR statement. JAMA 326, 1614–1621. https://doi.org/10.1001/jama.2021.18236 (2021).
Pulit, S. L. et al. Meta-analysis of genome-wide association studies for body fat distribution in 694,649 individuals of European ancestry. Hum. Mol. Genet. 28, 166–174. https://doi.org/10.1093/hmg/ddy327 (2019).
Shungin, D. et al. New genetic loci link adipose and insulin biology to body fat distribution. Nature 518, 187–196. https://doi.org/10.1038/nature14132 (2015).
Watanabe, K., Taskesen, E., Van Bochoven, A. & Posthuma, D. Functional mapping and annotation of genetic associations with FUMA. Nat. Commun. 8, 1826. https://doi.org/10.1038/s41467-017-01261-5 (2017).
Wang, K., Li, M. & Hakonarson, H. ANNOVAR: Functional annotation of genetic variants from high-throughput sequencing data. Nucleic Acids Res. 38, e164. https://doi.org/10.1093/nar/gkq603 (2010).
Kircher, M. et al. A general framework for estimating the relative pathogenicity of human genetic variants. Nat. Genet. 46, 310–315. https://doi.org/10.1038/ng.2892 (2014).
De Leeuw, C. A., Mooij, J. M., Heskes, T. & Posthuma, D. MAGMA: Generalized gene-set analysis of GWAS data. PLoS Comput. Biol. 11, e1004219. https://doi.org/10.1371/journal.pcbi.1004219 (2015).
Hemani, G. et al. The MR-base platform supports systematic causal inference across the human phenome. Elife 7, 34408. https://doi.org/10.7554/eLife.34408 (2018).
Bowden, J. et al. A framework for the investigation of pleiotropy in two-sample summary data Mendelian randomization. Stat. Med. 36, 1783–1802. https://doi.org/10.1002/sim.7221 (2017).
Verbanck, M., Chen, C. Y., Neale, B. & Do, R. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nat. Genet. 50, 693–698. https://doi.org/10.1038/s41588-018-0099-7 (2018).
Bowden, J., Davey Smith, G., Haycock, P. C. & Burgess, S. Consistent estimation in Mendelian randomization with some invalid instruments using a weighted median estimator. Genet. Epidemiol. 40, 304–314. https://doi.org/10.1002/gepi.21965 (2016).
Bowden, J., Davey Smith, G. & Burgess, S. Mendelian randomization with invalid instruments: Effect estimation and bias detection through Egger regression. Int. J. Epidemiol. 44, 512–525. https://doi.org/10.1093/ije/dyv080 (2015).
Pierce, B. L., Ahsan, H. & Vanderweele, T. J. Power and instrument strength requirements for Mendelian randomization studies using multiple genetic variants. Int. J. Epidemiol. 40, 740–752. https://doi.org/10.1093/ije/dyq151 (2011).
Bowden, J. et al. Assessing the suitability of summary data for two-sample Mendelian randomization analyses using MR-Egger regression: The role of the I2 statistic. Int. J. Epidemiol. 45, 1961–1974. https://doi.org/10.1093/ije/dyw220 (2016).
R Core Team. R: A Language and Environment for Statistical Computing. https://www.r-project.org (Accessed 27 April 2023) (R Foundation for Statistical Computing, 2022).
Henley, D., Lightman, S. & Carrell, R. Cortisol and CBG—Getting cortisol to the right place at the right time. Pharmacol. Ther. 166, 128–135. https://doi.org/10.1016/j.pharmthera.2016.06.020 (2016).
Bae, Y. J. & Kratzsch, J. Corticosteroid-binding globulin: Modulating mechanisms of bioavailability of cortisol and its clinical implications. Best Pract. Res. Clin. Endocrinol. Metab. 29, 761–772. https://doi.org/10.1016/j.beem.2015.09.001 (2015).
Spencer, R. L. & Deak, T. A users guide to HPA axis research. Physiol. Behav. 178, 43–65. https://doi.org/10.1016/j.physbeh.2016.11.014 (2017).
Chapman, K., Holmes, M. & Seckl, J. 11β-hydroxysteroid dehydrogenases: Intracellular gate-keepers of tissue glucocorticoid action. Physiol. Rev. 93, 1139–1206. https://doi.org/10.1152/physrev.00020.2012 (2013).
Lightman, S. L., Birnie, M. T. & Conway-Campbell, B. L. Dynamics of ACTH and cortisol secretion and implications for disease. Endocr. Rev. 41, 2. https://doi.org/10.1210/endrev/bnaa002 (2020).
John, K., Marino, J. S., Sanchez, E. R. & Hinds, T. D. The glucocorticoid receptor: Cause of or cure for obesity? Am. J. Physiol. Endocrinol. Metab. 310, E249–E257. https://doi.org/10.1152/ajpendo.00478.2015 (2016).
Van Der Valk, E. S., Savas, M. & Van Rossum, E. F. C. Stress and obesity: Are there more susceptible individuals? Curr. Obes. Rep. 7, 193–203. https://doi.org/10.1007/s13679-018-0306-y (2018).
Larsson, S. C., Lee, W. H., Kar, S., Burgess, S. & Allara, E. Assessing the role of cortisol in cancer: A wide-ranged Mendelian randomisation study. Br. J. Cancer 125, 1025–1029. https://doi.org/10.1038/s41416-021-01505-8 (2021).
Casals, G. & Hanzu, F. A. Cortisol measurements in cushing’s syndrome: Immunoassay or mass spectrometry? Ann. Lab. Med. 40, 285–296. https://doi.org/10.3343/alm.2020.40.4.285 (2020).
Tenk, J. et al. In obesity, HPA axis activity does not increase with BMI, but declines with aging: A meta-analysis of clinical studies. PLoS ONE 11, e0166842. https://doi.org/10.1371/journal.pone.0166842 (2016).
Van Der Valk, E. et al. Cross-sectional relation of long-term glucocorticoids in hair with anthropometric measurements and their possible determinants: A systematic review and meta-analysis. Obes. Rev. 23, e13376. https://doi.org/10.1111/obr.13376 (2022).
Purnell, J. Q., Brandon, D. D., Isabelle, L. M., Loriaux, D. L. & Samuels, M. H. Association of 24-hour cortisol production rates, cortisol-binding globulin, and plasma-free cortisol levels with body composition, leptin levels, and aging in adult men and women. J. Clin. Endocrinol. Metab. 89, 281–287. https://doi.org/10.1210/jc.2003-030440 (2004).
Panton, K. K. et al. New reference intervals for cortisol, cortisol binding globulin and free cortisol index in women using ethinyl estradiol. Scand. J. Clin. Lab. Investig. 79, 314–319. https://doi.org/10.1080/00365513.2019.1622031 (2019).
Bulun, S. E. & Simpson, E. R. Competitive reverse transcription-polymerase chain reaction analysis indicates that levels of aromatase cytochrome P450 transcripts in adipose tissue of buttocks, thighs, and abdomen of women increase with advancing age. J. Clin. Endocrinol. Metab. 78, 428–432. https://doi.org/10.1210/jcem.78.2.8106632 (1994).
Christakoudi, S., Riboli, E., Evangelou, E. & Tsilidis, K. K. Associations of body shape phenotypes with sex steroids and their binding proteins in the UK Biobank cohort. Sci. Rep. 12, 10774. https://doi.org/10.1038/s41598-022-14439-9 (2022).
Tran, K. V. et al. Human thermogenic adipocyte regulation by the long noncoding RNA LINC00473. Nat. Metab. 2, 397–412. https://doi.org/10.1038/s42255-020-0205-x (2020).
Scotney, H. et al. Glucocorticoids modulate human brown adipose tissue thermogenesis in vivo. Metabolism 70, 125–132. https://doi.org/10.1016/j.metabol.2017.01.024 (2017).
Li, L., Zhang, X., Liu, N., Chen, X. & Peng, C. LINC00473: A novel oncogenic long noncoding RNA in human cancers. J. Cell. Physiol. 236, 4174–4183. https://doi.org/10.1002/jcp.30176 (2021).
He, Y., Chen, Z. & He, J. The clinical prognostic value of lncRNA LINC00473 in cancer patients: A meta-analysis. Medicine 101, e32465. https://doi.org/10.1097/md.0000000000032465 (2022).
Yu, J. et al. The candidate schizophrenia risk gene Tmem108 regulates glucose metabolism homeostasis. Front. Endocrinol. 12, 770145. https://doi.org/10.3389/fendo.2021.770145 (2021).
Neumann, A. et al. The low single nucleotide polymorphism heritability of plasma and saliva cortisol levels. Psychoneuroendocrinology 85, 88–95. https://doi.org/10.1016/j.psyneuen.2017.08.011 (2017).
Bartels, M., Van Den Berg, M., Sluyter, F., Boomsma, D. I. & De Geus, E. J. Heritability of cortisol levels: Review and simultaneous analysis of twin studies. Psychoneuroendocrinology 28, 121–137. https://doi.org/10.1016/s0306-4530(02)00003-3 (2003).
Quinn, M. A., Xu, X., Ronfani, M. & Cidlowski, J. A. Estrogen deficiency promotes hepatic steatosis via a glucocorticoid receptor-dependent mechanism in mice. Cell Rep. 22, 2690–2701. https://doi.org/10.1016/j.celrep.2018.02.041 (2018).
Rubinow, D. R. et al. Testosterone suppression of CRH-stimulated cortisol in men. Neuropsychopharmacology 30, 1906–1912. https://doi.org/10.1038/sj.npp.1300742 (2005).
Hammond, G. L. Plasma steroid-binding proteins: Primary gatekeepers of steroid hormone action. J. Endocrinol. 230, R13–R25. https://doi.org/10.1530/joe-16-0070 (2016).
Kanakis, G. A., Tsametis, C. P. & Goulis, D. G. Measuring testosterone in women and men. Maturitas 125, 41–44. https://doi.org/10.1016/j.maturitas.2019.04.203 (2019).
Ketha, H., Girtman, A. & Singh, R. J. Estradiol assays—The path ahead. Steroids 99, 39–44. https://doi.org/10.1016/j.steroids.2014.08.009 (2015).
Kiani, A. K. et al. Steroid-converting enzymes in human adipose tissues and fat deposition with a focus on AKR1C enzymes. Eur. Rev. Med. Pharmacol. Sci. 25, 23–32. https://doi.org/10.26355/eurrev_202112_27330 (2021).
Oyola, M. G. & Handa, R. J. Hypothalamic–pituitary–adrenal and hypothalamic–pituitary–gonadal axes: Sex differences in regulation of stress responsivity. Stress 20, 476–494. https://doi.org/10.1080/10253890.2017.1369523 (2017).
Funding
This work was supported by the National Institute for Health Research (NIHR) Imperial Biomedical Research Centre (BRC), which provided infrastructure support for the Department of Epidemiology and Biostatistics at Imperial College London (UK). The open access fee was paid from the Imperial College London Open Access Fund. The funders had no role in the design and conduct of the study, the collection, analysis, and interpretation of the data, or the preparation, review, and approval of the manuscript, or in the decision to submit the manuscript for publication.
Author information
Authors and Affiliations
Contributions
S.C., E.R., and K.K.T. conceived and designed the study. K.K.T. provided statistical advice. A.-G.A. identified and obtained publicly available summary statistics, performed the initial statistical analysis and sensitivity analyses. S.C. led the research and finalised the statistical analysis. SC had full access to all data used in this study and takes responsibility for the integrity of the data and the accuracy of the data analysis. S.C. drafted the paper with contributions from A.-G.A., E.R., and K.K.T. S.C., A.-G.A., and K.K.T. designed the figures and tables. All authors: S.C., A.-G.A., E.R., and K.K.T. were involved in the interpretation of the results and the critical revisions of the paper. All authors have read and approved the manuscript.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Publisher's note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary Information
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Christakoudi, S., Asimakopoulos, AG., Riboli, E. et al. Links between the genetic determinants of morning plasma cortisol and body shape: a two-sample Mendelian randomisation study. Sci Rep 14, 3230 (2024). https://doi.org/10.1038/s41598-024-53727-4
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41598-024-53727-4
Keywords
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.