Genetically Predicted Thyroid-Stimulating Hormone and Coronary Artery Disease: Lipid-Mediated Causal Pathways, PCSK9 Protein Validation, and a Suppression Effect Revealed by Multi-Layered Mendelian Randomization

Genetically Predicted Thyroid-Stimulating Hormone and Coronary Artery Disease: Lipid-Mediated Causal Pathways, PCSK9 Protein Validation, and a Suppression Effect Revealed by Multi-Layered Mendelian Randomization | 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 Research Article Genetically Predicted Thyroid-Stimulating Hormone and Coronary Artery Disease: Lipid-Mediated Causal Pathways, PCSK9 Protein Validation, and a Suppression Effect Revealed by Multi-Layered Mendelian Randomization Feiyu Gu, Ling Xue This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9273655/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background Subclinical hypothyroidism (SCH) affects 4–10% of the general population and has been inconsistently linked to coronary artery disease (CAD). Whether thyroid-stimulating hormone (TSH) causally influences CAD risk, and through which biological pathways, remains unresolved. We conducted a multi-layered Mendelian randomization (MR) study integrating mediation analysis, proteomics-based drug-target MR, and genetic colocalization to dissect the causal architecture of the TSH–CAD relationship. Methods Two-sample MR was performed using 63 genome-wide significant instruments for TSH (Zhou et al., HUNT/ThyroidOmics, N = 119,715) against CAD (CARDIoGRAMplusC4D, 60,801 cases/123,504 controls). Two-step mediation MR with Sobel test evaluated five candidate pathways: metabolic (LDL-C, triglycerides, HDL-C), inflammatory (CRP), and coagulation (fibrinogen). Drug-target MR used five independent PCSK9 cis-pQTL instruments from deCODE proteomics (N = 35,363). Bayesian colocalization was performed at the PCSK9 locus. Sensitivity analyses included IVW (fixed and random effects), weighted median, MR-Egger, MR-PRESSO, Steiger directionality test, and reverse MR (CAD→TSH). Results Genetically predicted TSH showed no overall association with CAD (IVW: OR = 0.987, 95% CI: 0.943–1.032, P = 0.557; 62 SNPs). However, TSH was robustly associated with increased LDL-C (β = 0.077, P = 1.4×10⁻⁵; confirmed by weighted median P = 0.043 and MR-PRESSO corrected P = 7.4×10⁻⁴), with no evidence of pleiotropy (Egger intercept P = 0.983). The triglyceride association was borderline (IVW P = 0.053; MR-PRESSO corrected P = 0.010). No robust associations were found with HDL-C (P = 0.291) or CRP (IVW P = 0.039; MR-PRESSO corrected P = 0.838 after removing 6 outliers). Two-step mediation MR identified LDL-C as the sole significant mediator (indirect effect β = 0.029, Sobel P < 0.001); the triglyceride pathway was borderline (Sobel P = 0.059). A suppression effect was identified: the harmful LDL-mediated indirect effect (+ 0.029) was offset by a protective direct effect (− 0.043), yielding the observed null total effect (suppression ratio = − 2.1). Drug-target MR confirmed that genetically proxied PCSK9 protein levels causally increased CAD risk (IVW: OR = 1.210, P = 5.3×10⁻⁷; weighted median P = 5.2×10⁻⁴; Egger intercept P = 0.923). Reverse MR confirmed no evidence of reverse causality (CAD→TSH: β=−0.003, P = 0.825). Strong colocalization was observed between LDL-C and CAD at the PCSK9 locus (PPH4 = 0.915), but not for triglycerides (PPH4 = 0.251) or CRP (PPH4 = 0.002). Conclusions TSH influences CAD risk predominantly through an LDL-C–mediated pathway involving the TSHR–SREBP2–PCSK9–LDLR axis, as validated by convergent evidence from mediation MR, proteomics-based MR, and colocalization. The null total effect reflects a suppression phenomenon where harmful lipid-mediated and protective lipid-independent effects of TSH counterbalance each other. These findings support prioritizing lipid-lowering therapy—particularly PCSK9-targeted interventions—over TSH normalization alone for cardiovascular risk management in SCH. Cardiac & Cardiovascular Systems Mendelian randomization thyroid-stimulating hormone coronary artery disease LDL cholesterol PCSK9 mediation analysis suppression effect subclinical hypothyroidism proteomics colocalization Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Introduction Reverse MR was performed to assess potential reverse causality from CAD to TSH. Genome-wide significant SNPs for CAD (P < 5×10⁻⁸) were clumped and matched in the complete TSH GWAS (N = 22,397,080 variants) using rsID-based coordinate bridging through the GLGC lipid GWAS, which provided hg19 positions for the hg38-based CAD instruments. IVW analysis was then performed to estimate the causal effect of CAD liability on TSH levels. Subclinical hypothyroidism (SCH), defined as elevated thyroid-stimulating hormone (TSH) with normal free thyroxine (FT4) levels, affects 4–10% of the general population and is particularly prevalent among women and the elderly [ 1 , 2 ]. A large individual-participant data meta-analysis of 55,287 individuals demonstrated that SCH with TSH ≥ 10 mIU/L is associated with increased coronary heart disease events (hazard ratio 1.89) and mortality (hazard ratio 1.58) [ 3 ]. However, for mild SCH (TSH 4.5–10 mIU/L), which constitutes the majority of cases, the cardiovascular risk remains controversial [ 4 , 5 ]. Randomized trials of levothyroxine replacement in elderly SCH patients have not demonstrated cardiovascular benefit [ 6 ], raising the question of whether TSH elevation is causally related to coronary artery disease (CAD) or merely a marker of other processes. At the molecular level, TSH exerts direct hepatic effects independent of thyroid hormones. TSH binds to TSH receptors (TSHR) on hepatocytes and activates the cAMP/PKA/CREB signaling cascade, which upregulates sterol regulatory element-binding protein 2 (SREBP2) [ 7 ]. SREBP2 transcriptionally activates proprotein convertase subtilisin/kexin type 9 (PCSK9), a serine protease that promotes lysosomal degradation of LDL receptors (LDLR), thereby reducing hepatic LDL-cholesterol (LDL-C) clearance [ 8 , 9 ]. Clinical studies have confirmed that serum TSH positively correlates with circulating PCSK9 levels in both euthyroid subjects and patients with SCH [ 10 , 11 ]. This molecular pathway suggests that elevated TSH may increase cardiovascular risk specifically through lipid-mediated mechanisms, but this hypothesis has not been tested using causal inference methods. Mendelian randomization (MR) leverages genetic variants as instrumental variables to estimate causal effects free from confounding and reverse causation [ 12 ]. Several previous MR studies have examined the total effect of thyroid function on CAD, consistently finding null or borderline results [ 13 – 16 ]. However, these studies assessed only the total effect without dissecting mediating pathways. Importantly, a null total effect does not exclude the possibility that TSH influences CAD through specific pathways whose effects are masked by opposing mechanisms—a phenomenon known as a suppression effect [ 17 ]. Furthermore, no previous study has integrated proteomics-based MR (using protein quantitative trait loci, pQTL) or genetic colocalization to validate the molecular mediators of the TSH–CAD relationship. Here, we conducted a comprehensive multi-layered MR analysis to: (1) estimate the total causal effect of TSH on CAD; (2) dissect five candidate mediating pathways (LDL-C, triglycerides, HDL-C, CRP, fibrinogen) using two-step mediation MR with Sobel testing; (3) assess the direct effect of TSH independent of lipids using multivariable MR (MVMR); (4) validate the PCSK9-mediated mechanism using drug-target MR with cis-pQTL from the deCODE proteomics study; and (5) confirm shared causal variants through Bayesian colocalization at the PCSK9 locus. Methods Study design and reporting This study employed a multi-layered two-sample MR framework (Fig. 1 ). All analyses used publicly available summary-level genome-wide association study (GWAS) data; no individual-level data were accessed. The study adhered to the Strengthening the Reporting of Observational Studies in Epidemiology Using Mendelian Randomization (STROBE-MR) guidelines [ 18 ]. Data sources Genetic associations with TSH were obtained from a meta-analysis of the HUNT study, Michigan Genomics Initiative, and ThyroidOmics Consortium (N = 119,715 individuals of European ancestry) [ 19 ]. Genetic associations with CAD were obtained from the CARDIoGRAMplusC4D Consortium, comprising 60,801 CAD cases and 123,504 controls of predominantly European ancestry [ 20 ]. Lipid trait associations (LDL-C, triglycerides, HDL-C) were obtained from the Global Lipids Genetics Consortium (GLGC; N ≈ 188,000) [ 21 ]. C-reactive protein (CRP) associations were from UK Biobank (N ≈ 500,000) [ 22 ]. PCSK9 protein quantitative trait loci (pQTL) were from the deCODE Genetics proteomics study (N = 35,363 Icelandic individuals) [ 23 ]. Genetic instrument selection For TSH, we selected single nucleotide polymorphisms (SNPs) reaching genome-wide significance (P < 5×10⁻⁸) and performed distance-based clumping using a 10-megabase (Mb) window to ensure independence, retaining 63 instruments. SNPs with F-statistic < 10 were excluded to minimize weak instrument bias; the median F-statistic was 88.5 (range: 10.1–709.8). Of these 63 instruments, 62 were available in the CARDIoGRAMplusC4D data and 30 in the GLGC lipid data, reflecting differences in genotyping platform coverage. For the PCSK9 pQTL analysis, five independent cis-pQTL instruments were selected from pre-clumped deCODE data (P < 5×10⁻⁸, r²<0.001, 10 Mb window), with F-statistics ranging from 34.1 to 529.3. All instruments satisfied the three core MR assumptions of relevance, independence, and exclusion restriction, as evaluated by F-statistics, heterogeneity tests, and pleiotropy diagnostics. Statistical analyses Primary MR analyses employed the inverse-variance weighted (IVW) method under both fixed-effect and multiplicative random-effects models. Sensitivity analyses included: (i) the weighted median method, which provides consistent estimates when up to 50% of instruments are invalid [ 24 ]; (ii) MR-Egger regression, which tests for directional pleiotropy through its intercept term [ 25 ]; and (iii) MR-PRESSO, which identifies and removes outlier instruments contributing to horizontal pleiotropy [ 26 ]. Heterogeneity was assessed using Cochran’s Q statistic and the I² index. The Steiger directionality test was applied to confirm the correct causal direction [ 27 ]. Two-step mediation MR was performed for five candidate mediators. In Step 1, we estimated the causal effect of TSH on each mediator using IVW with the TSH instruments. In Step 2, we estimated the causal effect of each mediator on CAD using independently selected instruments from the respective mediator GWAS. The indirect (mediated) effect was calculated as the product of the Step 1 and Step 2 coefficients (β₁×β₂), with significance assessed by the Sobel test [ 28 ]. The direct effect was estimated as the difference between the total effect and the sum of significant indirect effects. Multivariable MR (MVMR) was performed to estimate the direct effect of TSH on CAD after simultaneously adjusting for LDL-C and triglycerides, using the subset of TSH instruments with available effect estimates for all three exposures in the CAD GWAS [ 29 ]. Drug-target MR used PCSK9 cis-pQTL instruments to estimate the causal effects of circulating PCSK9 protein levels on LDL-C and CAD. An analogous analysis was performed using a cis-SNP for HMG-CoA reductase (HMGCR; rs12916), the statin target. Bayesian colocalization was conducted at the PCSK9 locus (chr1:54.5–56.5 Mb, GRCh37) using approximate Bayes factors [ 30 ] to test whether LDL-C and CAD share the same causal genetic variant (hypothesis H4, with PPH4 > 0.80 considered strong evidence of colocalization). All analyses were performed in Python 3.10 using scipy, numpy, and pandas. Analysis code is available at [GitHub URL to be added upon acceptance]. Results Total causal effect of TSH on CAD Genetically predicted TSH was not significantly associated with CAD in the primary IVW analysis (β=−0.014, OR = 0.987, 95% CI: 0.943–1.032, P = 0.557; 62 SNPs; Table 1 ). Consistent null results were obtained from the random-effects IVW (P = 0.699), weighted median (β=−0.046, P = 0.211), and MR-PRESSO corrected estimate (β=−0.015, P = 0.525 after removal of 3 outliers). MR-Egger yielded a nominally significant protective estimate (β=−0.158, P = 0.020) with evidence of directional pleiotropy (intercept = 0.010, P = 0.014). Importantly, as demonstrated in the mediation analysis below, this apparent pleiotropy is consistent with a suppression effect wherein opposing causal pathways create the appearance of directional pleiotropy in the aggregate analysis. The Steiger test confirmed the correct causal direction (R² exposure > > R² outcome, P < 0.001). Heterogeneity was moderate (Cochran’s Q = 140.4, I²=56.6%). Table 1 Mendelian randomization estimates for the causal effects of TSH on CAD, candidate mediators, and PCSK9 protein on CAD. Exposure→Outcome N SNP IVW β IVW P WM P Egger P Egger int P PRESSO P I² (%) TSH→CAD 62 -0.014 0.557 0.211 0.020 0.014 0.525 56.6 TSH→LDL-C 30 0.077 1.4×10⁻⁵ 0.043 0.418 0.983 7.4×10⁻⁴ 84.1 TSH→TG 30 0.031 0.053 0.465 0.653 0.777 0.010 43.3 TSH→HDL-C 30 -0.017 0.291 1.000 0.649 0.405 0.418 71.3 TSH→CRP 61 -0.010 0.039 0.319 0.034 0.048 0.838 85.3 CAD→TSH 23 -0.003 0.825 — — — — — PCSK9→CAD 5 0.191 5.3×10⁻⁷ 5.2×10⁻⁴ 0.245 0.923 — 75.9 IVW, inverse-variance weighted; WM, weighted median; PRESSO, MR-PRESSO outlier-corrected; int, intercept. Bold indicates P < 0.05. Causal effects of TSH on candidate mediators Genetically predicted TSH was robustly associated with increased LDL-C (β = 0.077, P = 1.41×10⁻⁵; 30 SNPs; Table 1 ), with consistent results from the weighted median (β = 0.058, P = 0.043) and MR-PRESSO corrected estimate (β = 0.062, P = 7.4×10⁻⁴ after removal of 2 outliers). There was no evidence of directional pleiotropy (MR-Egger intercept P = 0.983). The Steiger test confirmed correct directionality (P < 0.001). TSH showed a borderline association with triglycerides in the primary IVW analysis (β = 0.031, P = 0.053; 30 SNPs), which became significant after MR-PRESSO outlier correction (β = 0.042, P = 0.010 after removal of 1 outlier). There was no pleiotropy (Egger intercept P = 0.777). The weighted median estimate was not significant (β = 0.018, P = 0.465). TSH was not significantly associated with HDL-C (IVW β=−0.017, P = 0.291). For CRP, a nominal IVW association was observed (β=−0.010, P = 0.039; 61 SNPs), but this was abolished after MR-PRESSO correction (β = 0.001, P = 0.838 after removal of 6 outliers), with borderline pleiotropy (Egger intercept P = 0.048), indicating that the IVW signal was driven by pleiotropic outlier instruments. Fibrinogen could not be evaluated due to insufficient clumped instruments in the outcome GWAS. Two-step mediation MR LDL-C was identified as the sole significant mediator of the TSH–CAD relationship (Table 2 ). The indirect effect through LDL-C was β = 0.029 (Sobel Z = 4.15, P < 0.001), representing the product of the TSH→LDL-C effect (β = 0.077) and the established LDL-C→CAD effect (β = 0.368, P = 2.0×10⁻⁶⁷, from 71 GLGC instruments [ 21 ]). The triglyceride pathway showed a borderline indirect effect (β = 0.008, Sobel P = 0.059). Neither HDL-C (Sobel P = 0.294) nor CRP (Sobel P = 0.055) demonstrated significant mediation. Table 2 Two-step mediation MR results: indirect effects of TSH on CAD through candidate mediators. Mediator Step 1 β Step 1 P Step 2 β Step 2 P Indirect Sobel P Coloc PPH4 LDL-C 0.077 1.4×10⁻⁵ 0.368 2.0×10⁻⁶⁷ 0.029 < 0.001 0.915 TG 0.031 0.053 0.259 6.5×10⁻¹⁹ 0.008 0.059 0.251 HDL-C -0.017 0.291 -0.268 6.8×10⁻²¹ 0.005 0.294 — CRP -0.010 0.039* 0.127 2.2×10⁻⁷ -0.001 0.055 0.002 *CRP Step 1 IVW P = 0.039 was attenuated to P = 0.838 after MR-PRESSO correction (6 outliers removed). Coloc PPH4 values are from Bayesian colocalization at the PCSK9 locus (chr1:54.5–56.5 Mb). Suppression effect A classical suppression effect was identified in the TSH–CAD relationship. The total effect was null (β=−0.014), which decomposed into a significant harmful indirect effect through LDL-C (β=+0.029, P < 0.001) and an estimated protective direct effect (β=−0.043). The suppression ratio was − 2.1, indicating that the absolute magnitude of the indirect effect exceeded that of the total effect and operated in the opposite direction to the direct effect. This pattern indicates that TSH simultaneously exerts two opposing influences on CAD: a deleterious effect mediated through lipid metabolism and a potentially protective effect through lipid-independent mechanisms. Multivariable MR MVMR analysis using 30 SNPs with data available across TSH, LDL-C, triglycerides, and CAD showed that the direct effect of TSH was null (β=−0.002, P = 0.978) after adjusting for lipids. However, neither LDL-C (P = 0.914) nor triglycerides (P = 0.840) reached significance in the joint model, reflecting limited statistical power due to the small instrument overlap and potential multicollinearity. These results are presented as exploratory and should be interpreted with caution. Reverse Mendelian randomization To exclude reverse causality, we performed reverse MR using CAD as the exposure and TSH as the outcome. From the CARDIoGRAMplusC4D GWAS, 2,046 genome-wide significant SNPs were identified, yielding 39 independent instruments after distance-based clumping (10 Mb window). Using rsID-based coordinate bridging through the GLGC dataset, 23 instruments were successfully matched in the TSH GWAS. IVW analysis showed no evidence that genetically predicted CAD liability influences TSH levels (β=−0.003, SE = 0.012, P = 0.825), confirming the absence of reverse causality and supporting the directionality of effects from TSH to downstream outcomes. PCSK9 pQTL drug-target MR and colocalization Drug-target MR using five independent PCSK9 cis-pQTL instruments from deCODE confirmed that genetically proxied PCSK9 protein levels causally increased CAD risk (IVW: β = 0.191, OR = 1.210, 95% CI: 1.123–1.303, P = 5.3×10⁻⁷; Table 1 ). Directional consistency was observed with the weighted median (β = 0.211, P = 5.2×10⁻⁴) and MR-Egger (β = 0.205, P = 0.245), with no evidence of directional pleiotropy (Egger intercept P = 0.923). There was evidence of heterogeneity among the five instruments (Q = 16.6, I²=75.9%, P = 0.002), which is expected in cis-MR analyses where multiple regulatory mechanisms within the same gene region may operate [ 31 ]. PCSK9 protein levels were strongly associated with LDL-C (β = 0.537, P = 6.0×10⁻¹⁷⁰). A parallel analysis using the HMGCR cis-SNP rs12916 confirmed that genetically proxied LDL-C reduction via the statin pathway also reduced CAD risk (OR per 1-SD LDL-C reduction = 0.613, P = 1.6×10⁻⁴). Bayesian colocalization at the PCSK9 locus (chr1:54.5–56.5 Mb) revealed strong evidence that LDL-C and CAD share the same causal variant (PPH4 = 0.915; 1,913 SNPs), with the top shared variant being rs11591147 (chr1:55505647, a well-characterized PCSK9 loss-of-function variant). By contrast, triglycerides (PPH4 = 0.251) and CRP (PPH4 = 0.002) did not colocalize with CAD at this locus, consistent with the specificity of the LDL-C–mediated pathway. As a negative control, LDL-C and CRP showed no colocalization (PPH4 = 0.040), confirming that the LDL-C signal at PCSK9 is independent of inflammatory pathways. Discussion This multi-layered MR study provides, to our knowledge, the first evidence that the TSH–CAD relationship is predominantly mediated through the LDL-C pathway, with the molecular mechanism traceable to the TSHR–SREBP2–PCSK9–LDLR axis. By integrating two-step mediation MR, proteomics-based drug-target MR, and genetic colocalization, we demonstrate convergent evidence across multiple independent analytical frameworks that genetically elevated TSH increases LDL-C through PCSK9 upregulation, which in turn increases CAD risk. The identification of a suppression effect—wherein this harmful lipid-mediated pathway is offset by a protective direct effect—provides a mechanistic explanation for the consistently null total effect of TSH on CAD reported in all prior MR studies [ 13 – 16 ]. Our central finding—that TSH causally increases LDL-C (β = 0.077, P = 1.4×10⁻⁵)—is consistent with decades of clinical evidence linking SCH to elevated LDL-C [ 32 ] and with the demonstration that levothyroxine treatment reduces LDL-C by approximately 9–14% in SCH patients [ 33 ]. The molecular mechanism is well characterized: TSH activates hepatocyte TSHR signaling, leading to SREBP2-mediated transcriptional upregulation of PCSK9 [ 8 , 9 ]. Elevated PCSK9 promotes LDLR degradation, reducing LDL-C clearance. Our PCSK9 pQTL-MR results—using five independent cis-pQTL instruments from a population (deCODE, Icelandic) entirely independent of the TSH and CAD GWAS populations—directly confirm this pathway: each standard deviation increase in genetically proxied PCSK9 protein increases CAD risk by 21% (OR = 1.210, P = 5.3×10⁻⁷). The strong colocalization between LDL-C and CAD at the PCSK9 locus (PPH4 = 0.915) further confirms that the causal signal passes through this specific molecular node. The suppression effect we identified has important methodological and clinical implications. Methodologically, it explains why all prior MR studies—including the well-powered analyses by Ellervik et al. [ 14 ] (N = 195,055), Marouli et al. [ 15 ] (N > 700,000), and Sterenborg et al. [ 16 ] (16 cardiovascular outcomes)—found null total effects of TSH on CAD. A suppression effect occurs when indirect effects through different pathways operate in opposite directions, canceling each other in the total effect estimate [ 17 ]. In our data, the harmful indirect effect through LDL-C (+ 0.029) was counterbalanced by a protective direct effect (− 0.043). This also explains the significant MR-Egger intercept (P = 0.014) observed in the total effect analysis: when opposing causal pathways coexist, the IVW estimate (which assumes homogeneous effects) conflates them, creating the appearance of directional pleiotropy. Clinically, this suppression effect suggests that interventions targeting only one arm of the TSH effect—for example, lowering TSH with levothyroxine—may not optimally reduce cardiovascular risk, as this simultaneously attenuates both the harmful lipid-mediated effect and the potentially protective direct effect. The nature of the protective direct effect of TSH warrants discussion. Several biological mechanisms are plausible. First, mild reductions in thyroid hormone activity decrease heart rate, cardiac contractility, and myocardial oxygen consumption, which may be protective in the setting of existing coronary atherosclerosis [ 34 ]. Second, TSHR is expressed on vascular endothelial cells, where TSH signaling may promote nitric oxide production via the PI3K/Akt/eNOS pathway [ 35 ]. Third, TSH may stabilize atherosclerotic plaques by inhibiting matrix metalloproteinase activity [ 36 ]. These mechanisms are speculative and require direct experimental validation; we emphasize that the direct effect is derived as a residual (total minus indirect) and cannot be definitively attributed to a specific pathway from these data alone. Our findings have direct therapeutic implications for cardiovascular risk management in SCH. First, they support lipid-lowering therapy as the primary cardiovascular intervention in SCH, independent of levothyroxine treatment decisions. The PCSK9-mediated pathway identified in our study suggests that PCSK9 inhibitors may be particularly effective in this population, as they directly target the molecular mechanism through which TSH elevates LDL-C. This is consistent with the clinical observation that SCH patients show elevated PCSK9 levels that correlate with TSH [ 10 , 11 ]. Second, the suppression effect implies that TSH normalization with levothyroxine may have more complex cardiovascular effects than previously assumed, potentially explaining the neutral results of randomized trials [ 6 ]. A combined strategy of levothyroxine plus lipid-lowering therapy may be more effective than either alone, although this hypothesis requires prospective testing. Strengths Key strengths include: (i) the multi-layered analytical framework providing convergent evidence from mediation MR, pQTL-MR, and colocalization; (ii) the use of PCSK9 pQTL from an independent Icelandic population (deCODE), minimizing sample overlap bias; (iii) comprehensive sensitivity analyses across six MR methods; (iv) the novel identification of a suppression effect that explains previously puzzling null results; and (v) the integration of molecular biology evidence (TSH–PCSK9–LDLR pathway) with genetic epidemiology, bridging the bench-to-bedside gap; and (vi) confirmation of causal directionality through both Steiger testing and formal reverse MR (CAD→TSH: P = 0.825). Limitations Several limitations should be acknowledged. First, instrument selection employed distance-based clumping with a conservative 10 Mb window rather than LD-based clumping; however, at this window size, residual LD is negligible for common variants in European populations, and the resulting instrument set showed strong F-statistics (median 88.5) with no evidence of weak instrument bias. Second, MVMR had limited statistical power due to the small number of instruments with complete multi-trait data (n = 30) and should be interpreted as exploratory. Third, partial sample overlap between the TSH, lipid, and CAD GWAS consortia may exist, although the principal exposure GWAS (HUNT/ThyroidOmics) and the pQTL data (deCODE, Iceland) are from independent populations. Fourth, all data were from European-ancestry cohorts, and generalizability to other populations requires further investigation. Fifth, the triglyceride-mediated pathway showed sensitivity to outlier instruments (IVW P = 0.053 vs. MR-PRESSO corrected P = 0.010), warranting cautious interpretation of this specific pathway. Finally, the suppression effect decomposition relies on the assumption that the identified mediators capture the dominant indirect pathways; unmeasured mediators operating in the same direction could alter the estimated direct effect. Conclusions This multi-layered MR study demonstrates that TSH influences CAD risk predominantly through the LDL-C pathway, mediated by the TSHR–SREBP2–PCSK9–LDLR molecular axis. The null total effect of TSH on CAD observed in this and prior studies reflects a suppression phenomenon in which harmful lipid-mediated effects are counterbalanced by a protective direct effect of uncertain mechanism. These findings support prioritizing lipid-lowering therapy—particularly PCSK9-targeted interventions—as the cornerstone of cardiovascular risk management in patients with subclinical hypothyroidism, and suggest that the cardiovascular consequences of levothyroxine treatment may be more nuanced than current guidelines assume. Declarations Ethics approval and consent to participate: Not applicable. This study exclusively used publicly available, de-identified summary-level GWAS data. No individual-level data were accessed. Consent for publication: Not applicable. Availability of data and materials: All GWAS summary statistics are publicly available from the sources cited. TSH GWAS: GCST010653 (GWAS Catalog). CAD GWAS: CARDIoGRAMplusC4D (GWAS Catalog GCST003116). Lipid GWAS: GLGC (GWAS Catalog GCST002216/2222/2223). PCSK9 pQTL: deCODE (https://www.decode.com/summarydata/). Analysis code will be deposited at [GitHub URL] upon acceptance. Competing interests: The authors declare no competing interests. Funding: [none] Authors’ contributions: FG conceived the study, designed the analytical framework, performed all statistical analyses, generated all figures and tables, and drafted the manuscript. LX provided clinical interpretation, supervised the study, and critically revised the manuscript. Both authors read and approved the final manuscript. Acknowledgments: We thank the investigators of the HUNT study, ThyroidOmics Consortium, CARDIoGRAMplusC4D Consortium, Global Lipids Genetics Consortium, UK Biobank, and deCODE Genetics for making their GWAS summary statistics publicly available. References Hollowell JG et al (2002) Serum TSH, T(4), and thyroid antibodies in the United States population (1988 to 1994): NHANES III. J Clin Endocrinol Metab 87:489–499 Canaris GJ et al (2000) The Colorado thyroid disease prevalence study. Arch Intern Med 160:526–534 Rodondi N et al (2010) Subclinical hypothyroidism and the risk of coronary heart disease and mortality. JAMA 304:1365–1374 Razvi S et al (2007) The beneficial effect of L-thyroxine on cardiovascular risk factors, endothelial function, and quality of life in subclinical hypothyroidism. J Clin Endocrinol Metab 92:1715–1723 Biondi B, Cooper DS (2019) Subclinical hypothyroidism. Lancet 393:1320–1330 Stott DJ et al (2017) Thyroid hormone therapy for older adults with subclinical hypothyroidism. N Engl J Med 376:2534–2544 Ma S et al (2018) TSH promotes metabolic syndrome by acting on the hypothalamus–pituitary–thyroid axis. Trends Endocrinol Metab 29:386–393 Song Y et al (2015) TSH exhibits the impact on LDLR/LDL-c via up-regulating hepatic PCSK9 expression. Metabolism 64:1395–1401 Liu J et al (2022) Update on dyslipidemia in hypothyroidism: the mechanism of dyslipidemia in hypothyroidism. Endocr Connect 11:e210002 Sadik NA et al (2022) The relationship of circulating PCSK9 with TSH and lipid profile in newly diagnosed hypothyroidism. Endocr Metab Immune Disord Drug Targets 22:741–749 Kwakernaak AJ et al (2014) Thyroid function is associated with PCSK9 and non-HDL cholesterol. J Clin Endocrinol Metab 99:E1346–E1353 Davey Smith G, Hemani G (2014) Mendelian randomization: genetic anchors for causal inference in epidemiological studies. Hum Mol Genet 23:R89–R98 Ellervik C et al (2017) Thyroid function and ischemic heart disease: a Mendelian randomization study. Sci Rep 7:7592 Sterenborg RBTM et al (2019) Thyroid function and dysfunction in relation to 16 cardiovascular diseases. Circ Genom Precis Med 12:e002468 Marouli E et al (2021) Higher thyrotropin leads to unfavorable lipid profile and somewhat higher cardiovascular disease risk. BMC Med 19:266 Ruan W et al (2024) Thyroid function effect on cardiac structure, cardiac function, and disease risk. Heart Rhythm 21:2272–2281 MacKinnon DP et al (2002) A comparison of methods to test mediation and other intervening variable effects. Psychol Methods 7:83–104 Skrivankova VW et al (2021) Strengthening the reporting of observational studies in epidemiology using Mendelian randomization: the STROBE-MR statement. JAMA 326:1614–1621 Zhou W et al (2018) Efficiently controlling for case–control imbalance and sample relatedness in large-scale genetic association studies. Nat Genet 50:1335–1341 Nikpay M et al (2015) A comprehensive 1,000 Genomes-based GWAS meta-analysis of coronary artery disease. Nat Genet 47:1121–1130 Willer CJ et al (2013) Discovery and refinement of loci associated with lipid levels. Nat Genet 45:1274–1283 Bycroft C et al (2018) The UK Biobank resource with deep phenotyping and genomic data. Nature 562:203–209 Ferkingstad E et al (2021) Large-scale integration of the plasma proteome with genetics and disease. Nat Genet 53:1712–1721 Bowden J et al (2016) Consistent estimation in Mendelian randomization with some invalid instruments using a weighted median estimator. Genet Epidemiol 40:304–314 Bowden J et al (2015) Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. Int J Epidemiol 44:512–525 Verbanck M et al (2018) Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization. Nat Genet 50:693–698 Hemani G et al (2017) Orienting the causal relationship between imprecisely measured traits using GWAS summary data. PLoS Genet 13:e1007081 Sobel ME (1982) Asymptotic confidence intervals for indirect effects in structural equation models. Sociol Methodol 13:290–312 Burgess S, Thompson SG (2015) Multivariable Mendelian randomization: the use of pleiotropic genetic variants to estimate causal effects. Am J Epidemiol 181:251–260 Giambartolomei C et al (2014) Bayesian test for colocalisation between pairs of genetic association studies. PLoS Genet 10:e1004383 Schmidt AF et al (2020) Genetic drug target validation using Mendelian randomisation. Nat Commun 11:3255 Duntas LH (2002) Thyroid disease and lipids. Thyroid 12:287–293 Razvi S et al (2017) Subclinical hypothyroidism: to treat or not to treat, that is the question! A systematic review with meta-analysis on lipid profile. Endocr Connect 6:R188–R199 Klein I, Danzi S (2007) Thyroid disease and the heart. Circulation 116:1725–1735 Tian L et al (2014) TSH stimulates the proliferation of vascular smooth muscle cells. Endocrine 46:651–658 Diekman MJ et al (2001) Determinants of changes in plasma homocysteine in hypothyroidism and hyperthyroidism. Clin Endocrinol 54:197–204 Additional Declarations The authors declare no competing interests. Supplementary Files Table1MRresults.csv Table1_MR_results.csv：Supplementary Table 1. Complete Mendelian randomization results for all exposure-outcome pairs, including IVW, random-effects IVW, weighted median, MR-Egger, MR-PRESSO, heterogeneity statistics, and Steiger directionality test. Table2mediation.csv Table2_mediation.csv：Supplementary Table 2. Two-step mediation MR results showing Step 1 (TSH to mediator) and Step 2 (mediator to CAD) effects, indirect effects, Sobel test statistics, and colocalization posterior probabilities at the PCSK9 locus. SuppFig1forestplots.png Figure 6 (SuppFig1forestplots.png)：Forest plots showing individual SNP Wald ratio estimates for (A) TSH to CAD (62 SNPs), (B) TSH to LDL-C (30 SNPs), and (C) TSH to triglycerides (30 SNPs). The vertical red line indicates the overall IVW estimate. Diamond markers represent individual instrument estimates with 95% confidence intervals. SuppFig2leaveoneout.png Figure 7 (SuppFig2leaveoneout.png)：Leave-one-out sensitivity analysis for (A) TSH to CAD, (B) TSH to LDL-C, and (C) TSH to triglycerides. Each point represents the IVW estimate after sequentially removing one SNP. The vertical red line indicates the estimate using all SNPs. No single SNP substantially altered the overall results. SuppFig3mediationsummary.png Figure 8 (SuppFig3mediationsummary.png)：Mediation analysis summary. (A) Step 1 causal effects of TSH on candidate mediators (IVW estimates with 95% CI). Red diamonds indicate significant associations (P<0.05); gray diamonds indicate non-significant. (B) Indirect effects with Sobel test P-values. Only LDL-C reached significance (Sobel P<0.001); TG was borderline (P=0.059). Cite Share Download PDF Status: Posted 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-9273655","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":614942264,"identity":"37063bac-1d78-47d8-813e-17b2bad4564e","order_by":0,"name":"Feiyu Gu","email":"","orcid":"","institution":"Jinzhou Medical University","correspondingAuthor":false,"prefix":"","firstName":"Feiyu","middleName":"","lastName":"Gu","suffix":""},{"id":614942301,"identity":"260d9e4e-897c-42d1-a63c-4b6604de1648","order_by":1,"name":"Ling Xue","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA5UlEQVRIiWNgGAWjYDAC5gMMEmCGBGPjAyDFw0dQC1sCSIsBSEuzAUgLGwlaGNjA1hHUIu/GY3jjQ80fOfPZzW2VX3PsZNgYmB8+uoFHi+ExHmPLGccMjGXuHGy7LbstGegwNmPjHHxa5veYSfM2GCTOkEhsuy25jRmohYdNGq+WNh6ElmLJbfWEtcizIWlh/LjtMGEtBmxsxUC/GBtLSCQ2SzNuO87DxkzAL/JtzBuBISYnJyGR/vDjz23V9vzszQ8f47XlAIcBnMPMAybxKAfb0sD+AM5h/EFA9SgYBaNgFIxMAACjiUCyXiThWAAAAABJRU5ErkJggg==","orcid":"","institution":"Third Affiliated Hospital of Jinzhou Medical University","correspondingAuthor":true,"prefix":"","firstName":"Ling","middleName":"","lastName":"Xue","suffix":""}],"badges":[],"createdAt":"2026-03-31 03:10:50","currentVersionCode":1,"declarations":{"humanSubjects":false,"vertebrateSubjects":false,"conflictsOfInterestStatement":false,"humanSubjectEthicalGuidelines":false,"humanSubjectConsent":false,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":false},"doi":"10.21203/rs.3.rs-9273655/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9273655/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":105910743,"identity":"ecb4f4f3-f5ac-4216-84b0-6e011cdd074b","added_by":"auto","created_at":"2026-04-01 10:50:39","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":333793,"visible":true,"origin":"","legend":"\u003cp\u003eStudy design and multi-layered analytical framework. Eight analytical layers are shown: Layer 1, total effect of TSH on CAD; Layer 2, two-step mediation MR through five candidate pathways (LDL-C, triglycerides, HDL-C, CRP, fibrinogen); Layer 3, multivariable MR adjusting for LDL-C and triglycerides; Layer 4, suppression effect decomposition (total = indirect + direct); Layer 5, molecular mechanism network (TSHR–SREBP2–PCSK9–LDLR); Layer 6, PCSK9 pQTL drug-target MR using deCODE cis-pQTL; Layer 7, Bayesian colocalization at the PCSK9 locus; Layer 8, comprehensive sensitivity analyses (IVW, weighted median, MR-Egger, MR-PRESSO, Steiger, leave-one-out).\u003c/p\u003e","description":"","filename":"Figure1studydesign.png","url":"https://assets-eu.researchsquare.com/files/rs-9273655/v1/7a3a43f7a9ff8e7f470ed3d7.png"},{"id":105910705,"identity":"860ce52a-07d5-4386-a511-315bb5a39923","added_by":"auto","created_at":"2026-04-01 10:50:23","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":316156,"visible":true,"origin":"","legend":"\u003cp\u003eScatter plots showing SNP-level associations between TSH genetic instruments and outcomes. (A) TSH vs. CAD. (B) TSH vs. LDL-C. (C) TSH vs. triglycerides. Lines represent IVW (red), weighted median (blue), and MR-Egger (green) regression fits.\u003c/p\u003e","description":"","filename":"Figure2scatterplots.png","url":"https://assets-eu.researchsquare.com/files/rs-9273655/v1/c57be5c58015d92f4bbd473b.png"},{"id":105910933,"identity":"79ad8d11-1e38-410f-8b0a-e64140649ebc","added_by":"auto","created_at":"2026-04-01 10:51:43","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":200576,"visible":true,"origin":"","legend":"\u003cp\u003eComplete causal network diagram integrating all MR findings. Solid red arrows indicate MR-confirmed significant causal effects; dashed gray arrows indicate non-significant associations. The molecular layer (TSH→TSHR→SREBP2→PCSK9→LDLR→LDL-C) is supported by both MR evidence and published experimental data. Drug targets (PCSK9 inhibitors, statins) are highlighted in orange. The suppression effect (harmful indirect +0.029 vs. protective direct −0.043) is shown in the center.\u003c/p\u003e","description":"","filename":"Figure3causalnetwork.png","url":"https://assets-eu.researchsquare.com/files/rs-9273655/v1/060ea4506e6aef3184e37551.png"},{"id":105909930,"identity":"423c63de-4099-48f3-ab93-f5e52f29e3d6","added_by":"auto","created_at":"2026-04-01 10:45:52","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":151359,"visible":true,"origin":"","legend":"\u003cp\u003ePCSK9 drug-target MR and colocalization results. (A) Forest plot of PCSK9 cis-pQTL MR estimates for CAD across three methods (IVW, weighted median, MR-Egger) using five independent instruments from deCODE proteomics. (B) Bar chart of colocalization posterior probabilities (PPH4) at the PCSK9 locus for LDL-C vs. CAD (0.915), TG vs. CAD (0.251), CRP vs. CAD (0.002), and LDL-C vs. CRP (0.040, negative control).\u003c/p\u003e","description":"","filename":"Figure4pcsk9coloc.png","url":"https://assets-eu.researchsquare.com/files/rs-9273655/v1/188cbc883412879a2598b95f.png"},{"id":105910793,"identity":"e120e115-a8a3-4f3a-9988-c57e90424eff","added_by":"auto","created_at":"2026-04-01 10:50:44","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":163994,"visible":true,"origin":"","legend":"\u003cp\u003eFunnel plots for (A) TSH→CAD, (B) TSH→LDL-C, and (C) TSH→TG analyses, with individual SNP Wald ratio estimates plotted against precision (1/SE). The vertical red line indicates the IVW point estimate.\u003c/p\u003e","description":"","filename":"Figure5funnelplots.png","url":"https://assets-eu.researchsquare.com/files/rs-9273655/v1/8aa4df9d691063f070c45d76.png"},{"id":105913781,"identity":"6f33576e-9535-4c57-bda3-6027d904f610","added_by":"auto","created_at":"2026-04-01 11:06:50","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1713500,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9273655/v1/2ba37283-c3eb-4e10-ab3e-83038f406c09.pdf"},{"id":105911719,"identity":"d96d468f-181b-4230-94ae-fd017c25bafd","added_by":"auto","created_at":"2026-04-01 10:54:48","extension":"csv","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":1052,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTable1_MR_results.csv：\u003c/strong\u003eSupplementary Table 1. Complete Mendelian randomization results for all exposure-outcome pairs, including IVW, random-effects IVW, weighted median, MR-Egger, MR-PRESSO, heterogeneity statistics, and Steiger directionality test.\u003c/p\u003e","description":"","filename":"Table1MRresults.csv","url":"https://assets-eu.researchsquare.com/files/rs-9273655/v1/201bd96785f1cb7c90b3e187.csv"},{"id":105909927,"identity":"db4f0f84-d18c-4017-b517-a1d6d0ca61d1","added_by":"auto","created_at":"2026-04-01 10:45:52","extension":"csv","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":410,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTable2_mediation.csv：\u003c/strong\u003eSupplementary Table 2. Two-step mediation MR results showing Step 1 (TSH to mediator) and Step 2 (mediator to CAD) effects, indirect effects, Sobel test statistics, and colocalization posterior probabilities at the PCSK9 locus.\u003c/p\u003e","description":"","filename":"Table2mediation.csv","url":"https://assets-eu.researchsquare.com/files/rs-9273655/v1/06d632d6488a3610e490797b.csv"},{"id":105911777,"identity":"d04b1224-c8f9-49df-9a22-c5c058048cbb","added_by":"auto","created_at":"2026-04-01 10:55:03","extension":"png","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":131985,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFigure 6 (SuppFig1forestplots.png)：\u003c/strong\u003eForest plots showing individual SNP Wald ratio estimates for (A) TSH to CAD (62 SNPs), (B) TSH to LDL-C (30 SNPs), and (C) TSH to triglycerides (30 SNPs). The vertical red line indicates the overall IVW estimate. Diamond markers represent individual instrument estimates with 95% confidence intervals.\u003c/p\u003e","description":"","filename":"SuppFig1forestplots.png","url":"https://assets-eu.researchsquare.com/files/rs-9273655/v1/210d06f407ee2c92a333b62c.png"},{"id":105910845,"identity":"15019d8f-9efa-40dc-a80f-9a7d27651315","added_by":"auto","created_at":"2026-04-01 10:51:29","extension":"png","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":133133,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFigure 7 (SuppFig2leaveoneout.png)：\u003c/strong\u003eLeave-one-out sensitivity analysis for (A) TSH to CAD, (B) TSH to LDL-C, and (C) TSH to triglycerides. Each point represents the IVW estimate after sequentially removing one SNP. The vertical red line indicates the estimate using all SNPs. No single SNP substantially altered the overall results.\u003c/p\u003e","description":"","filename":"SuppFig2leaveoneout.png","url":"https://assets-eu.researchsquare.com/files/rs-9273655/v1/141831d8baa8cce317573611.png"},{"id":105909924,"identity":"63208a4f-d675-49be-9528-6a06a84a837d","added_by":"auto","created_at":"2026-04-01 10:45:49","extension":"png","order_by":5,"title":"","display":"","copyAsset":false,"role":"supplement","size":124330,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFigure 8 (SuppFig3mediationsummary.png)：\u003c/strong\u003eMediation analysis summary. (A) Step 1 causal effects of TSH on candidate mediators (IVW estimates with 95% CI). Red diamonds indicate significant associations (P\u0026lt;0.05); gray diamonds indicate non-significant. (B) Indirect effects with Sobel test P-values. Only LDL-C reached significance (Sobel P\u0026lt;0.001); TG was borderline (P=0.059).\u003c/p\u003e","description":"","filename":"SuppFig3mediationsummary.png","url":"https://assets-eu.researchsquare.com/files/rs-9273655/v1/0047e288be506a7588e3b05a.png"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"\u003cp\u003e\u003cstrong\u003eGenetically Predicted Thyroid-Stimulating Hormone and Coronary Artery Disease: Lipid-Mediated Causal Pathways, PCSK9 Protein Validation, and a Suppression Effect Revealed by Multi-Layered Mendelian Randomization\u003c/strong\u003e\u003c/p\u003e","fulltext":[{"header":"Introduction","content":"\u003cp\u003eReverse MR was performed to assess potential reverse causality from CAD to TSH. Genome-wide significant SNPs for CAD (P\u0026thinsp;\u0026lt;\u0026thinsp;5\u0026times;10⁻⁸) were clumped and matched in the complete TSH GWAS (N\u0026thinsp;=\u0026thinsp;22,397,080 variants) using rsID-based coordinate bridging through the GLGC lipid GWAS, which provided hg19 positions for the hg38-based CAD instruments. IVW analysis was then performed to estimate the causal effect of CAD liability on TSH levels.\u003c/p\u003e \u003cp\u003eSubclinical hypothyroidism (SCH), defined as elevated thyroid-stimulating hormone (TSH) with normal free thyroxine (FT4) levels, affects 4\u0026ndash;10% of the general population and is particularly prevalent among women and the elderly [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. A large individual-participant data meta-analysis of 55,287 individuals demonstrated that SCH with TSH\u0026thinsp;\u0026ge;\u0026thinsp;10 mIU/L is associated with increased coronary heart disease events (hazard ratio 1.89) and mortality (hazard ratio 1.58) [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. However, for mild SCH (TSH 4.5\u0026ndash;10 mIU/L), which constitutes the majority of cases, the cardiovascular risk remains controversial [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. Randomized trials of levothyroxine replacement in elderly SCH patients have not demonstrated cardiovascular benefit [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e], raising the question of whether TSH elevation is causally related to coronary artery disease (CAD) or merely a marker of other processes.\u003c/p\u003e \u003cp\u003eAt the molecular level, TSH exerts direct hepatic effects independent of thyroid hormones. TSH binds to TSH receptors (TSHR) on hepatocytes and activates the cAMP/PKA/CREB signaling cascade, which upregulates sterol regulatory element-binding protein 2 (SREBP2) [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. SREBP2 transcriptionally activates proprotein convertase subtilisin/kexin type 9 (PCSK9), a serine protease that promotes lysosomal degradation of LDL receptors (LDLR), thereby reducing hepatic LDL-cholesterol (LDL-C) clearance [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. Clinical studies have confirmed that serum TSH positively correlates with circulating PCSK9 levels in both euthyroid subjects and patients with SCH [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. This molecular pathway suggests that elevated TSH may increase cardiovascular risk specifically through lipid-mediated mechanisms, but this hypothesis has not been tested using causal inference methods.\u003c/p\u003e \u003cp\u003eMendelian randomization (MR) leverages genetic variants as instrumental variables to estimate causal effects free from confounding and reverse causation [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]. Several previous MR studies have examined the total effect of thyroid function on CAD, consistently finding null or borderline results [\u003cspan additionalcitationids=\"CR14 CR15\" citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. However, these studies assessed only the total effect without dissecting mediating pathways. Importantly, a null total effect does not exclude the possibility that TSH influences CAD through specific pathways whose effects are masked by opposing mechanisms\u0026mdash;a phenomenon known as a suppression effect [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. Furthermore, no previous study has integrated proteomics-based MR (using protein quantitative trait loci, pQTL) or genetic colocalization to validate the molecular mediators of the TSH\u0026ndash;CAD relationship.\u003c/p\u003e \u003cp\u003eHere, we conducted a comprehensive multi-layered MR analysis to: (1) estimate the total causal effect of TSH on CAD; (2) dissect five candidate mediating pathways (LDL-C, triglycerides, HDL-C, CRP, fibrinogen) using two-step mediation MR with Sobel testing; (3) assess the direct effect of TSH independent of lipids using multivariable MR (MVMR); (4) validate the PCSK9-mediated mechanism using drug-target MR with cis-pQTL from the deCODE proteomics study; and (5) confirm shared causal variants through Bayesian colocalization at the PCSK9 locus.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eStudy design and reporting\u003c/h2\u003e \u003cp\u003eThis study employed a multi-layered two-sample MR framework (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). All analyses used publicly available summary-level genome-wide association study (GWAS) data; no individual-level data were accessed. The study adhered to the Strengthening the Reporting of Observational Studies in Epidemiology Using Mendelian Randomization (STROBE-MR) guidelines [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e].\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eData sources\u003c/h3\u003e\n\u003cp\u003eGenetic associations with TSH were obtained from a meta-analysis of the HUNT study, Michigan Genomics Initiative, and ThyroidOmics Consortium (N\u0026thinsp;=\u0026thinsp;119,715 individuals of European ancestry) [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]. Genetic associations with CAD were obtained from the CARDIoGRAMplusC4D Consortium, comprising 60,801 CAD cases and 123,504 controls of predominantly European ancestry [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]. Lipid trait associations (LDL-C, triglycerides, HDL-C) were obtained from the Global Lipids Genetics Consortium (GLGC; N\u0026thinsp;\u0026asymp;\u0026thinsp;188,000) [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]. C-reactive protein (CRP) associations were from UK Biobank (N\u0026thinsp;\u0026asymp;\u0026thinsp;500,000) [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]. PCSK9 protein quantitative trait loci (pQTL) were from the deCODE Genetics proteomics study (N\u0026thinsp;=\u0026thinsp;35,363 Icelandic individuals) [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e].\u003c/p\u003e\n\u003ch3\u003eGenetic instrument selection\u003c/h3\u003e\n\u003cp\u003eFor TSH, we selected single nucleotide polymorphisms (SNPs) reaching genome-wide significance (P\u0026thinsp;\u0026lt;\u0026thinsp;5\u0026times;10⁻⁸) and performed distance-based clumping using a 10-megabase (Mb) window to ensure independence, retaining 63 instruments. SNPs with F-statistic\u0026thinsp;\u0026lt;\u0026thinsp;10 were excluded to minimize weak instrument bias; the median F-statistic was 88.5 (range: 10.1\u0026ndash;709.8). Of these 63 instruments, 62 were available in the CARDIoGRAMplusC4D data and 30 in the GLGC lipid data, reflecting differences in genotyping platform coverage. For the PCSK9 pQTL analysis, five independent cis-pQTL instruments were selected from pre-clumped deCODE data (P\u0026thinsp;\u0026lt;\u0026thinsp;5\u0026times;10⁻⁸, r\u0026sup2;\u0026lt;0.001, 10 Mb window), with F-statistics ranging from 34.1 to 529.3. All instruments satisfied the three core MR assumptions of relevance, independence, and exclusion restriction, as evaluated by F-statistics, heterogeneity tests, and pleiotropy diagnostics.\u003c/p\u003e\n\u003ch3\u003eStatistical analyses\u003c/h3\u003e\n\u003cp\u003ePrimary MR analyses employed the inverse-variance weighted (IVW) method under both fixed-effect and multiplicative random-effects models. Sensitivity analyses included: (i) the weighted median method, which provides consistent estimates when up to 50% of instruments are invalid [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]; (ii) MR-Egger regression, which tests for directional pleiotropy through its intercept term [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e]; and (iii) MR-PRESSO, which identifies and removes outlier instruments contributing to horizontal pleiotropy [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e]. Heterogeneity was assessed using Cochran\u0026rsquo;s Q statistic and the I\u0026sup2; index. The Steiger directionality test was applied to confirm the correct causal direction [\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eTwo-step mediation MR was performed for five candidate mediators. In Step 1, we estimated the causal effect of TSH on each mediator using IVW with the TSH instruments. In Step 2, we estimated the causal effect of each mediator on CAD using independently selected instruments from the respective mediator GWAS. The indirect (mediated) effect was calculated as the product of the Step 1 and Step 2 coefficients (β₁\u0026times;β₂), with significance assessed by the Sobel test [\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]. The direct effect was estimated as the difference between the total effect and the sum of significant indirect effects.\u003c/p\u003e \u003cp\u003eMultivariable MR (MVMR) was performed to estimate the direct effect of TSH on CAD after simultaneously adjusting for LDL-C and triglycerides, using the subset of TSH instruments with available effect estimates for all three exposures in the CAD GWAS [\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eDrug-target MR used PCSK9 cis-pQTL instruments to estimate the causal effects of circulating PCSK9 protein levels on LDL-C and CAD. An analogous analysis was performed using a cis-SNP for HMG-CoA reductase (HMGCR; rs12916), the statin target. Bayesian colocalization was conducted at the PCSK9 locus (chr1:54.5\u0026ndash;56.5 Mb, GRCh37) using approximate Bayes factors [\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e] to test whether LDL-C and CAD share the same causal genetic variant (hypothesis H4, with PPH4\u0026thinsp;\u0026gt;\u0026thinsp;0.80 considered strong evidence of colocalization).\u003c/p\u003e \u003cp\u003eAll analyses were performed in Python 3.10 using scipy, numpy, and pandas. Analysis code is available at [GitHub URL to be added upon acceptance].\u003c/p\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eTotal causal effect of TSH on CAD\u003c/h2\u003e \u003cp\u003eGenetically predicted TSH was not significantly associated with CAD in the primary IVW analysis (β=\u0026minus;0.014, OR\u0026thinsp;=\u0026thinsp;0.987, 95% CI: 0.943\u0026ndash;1.032, P\u0026thinsp;=\u0026thinsp;0.557; 62 SNPs; Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). Consistent null results were obtained from the random-effects IVW (P\u0026thinsp;=\u0026thinsp;0.699), weighted median (β=\u0026minus;0.046, P\u0026thinsp;=\u0026thinsp;0.211), and MR-PRESSO corrected estimate (β=\u0026minus;0.015, P\u0026thinsp;=\u0026thinsp;0.525 after removal of 3 outliers). MR-Egger yielded a nominally significant protective estimate (β=\u0026minus;0.158, P\u0026thinsp;=\u0026thinsp;0.020) with evidence of directional pleiotropy (intercept\u0026thinsp;=\u0026thinsp;0.010, P\u0026thinsp;=\u0026thinsp;0.014). Importantly, as demonstrated in the mediation analysis below, this apparent pleiotropy is consistent with a suppression effect wherein opposing causal pathways create the appearance of directional pleiotropy in the aggregate analysis. The Steiger test confirmed the correct causal direction (R\u0026sup2; exposure\u0026thinsp;\u0026gt;\u0026thinsp;\u0026gt;\u0026thinsp;R\u0026sup2; outcome, P\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Heterogeneity was moderate (Cochran\u0026rsquo;s Q\u0026thinsp;=\u0026thinsp;140.4, I\u0026sup2;=56.6%).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eMendelian randomization estimates for the causal effects of TSH on CAD, candidate mediators, and PCSK9 protein on CAD.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"9\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eExposure\u0026rarr;Outcome\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eN SNP\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eIVW β\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eIVW P\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eWM P\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eEgger P\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eEgger int P\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003ePRESSO P\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c9\"\u003e \u003cp\u003eI\u0026sup2; (%)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTSH\u0026rarr;CAD\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e62\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e-0.014\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.557\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.211\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.020\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.014\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e0.525\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e56.6\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTSH\u0026rarr;LDL-C\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e30\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.077\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.4\u0026times;10⁻⁵\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.043\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.418\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.983\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e7.4\u0026times;10⁻⁴\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e84.1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTSH\u0026rarr;TG\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e30\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.031\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.053\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.465\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.653\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.777\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e0.010\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e43.3\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTSH\u0026rarr;HDL-C\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e30\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e-0.017\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.291\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.649\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.405\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e0.418\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e71.3\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTSH\u0026rarr;CRP\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e61\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e-0.010\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.039\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.319\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.034\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.048\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e0.838\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e85.3\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCAD\u0026rarr;TSH\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e23\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e-0.003\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.825\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePCSK9\u0026rarr;CAD\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.191\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e5.3\u0026times;10⁻⁷\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e5.2\u0026times;10⁻⁴\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.245\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.923\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e75.9\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"9\"\u003eIVW, inverse-variance weighted; WM, weighted median; PRESSO, MR-PRESSO outlier-corrected; int, intercept. Bold indicates P\u0026thinsp;\u0026lt;\u0026thinsp;0.05.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eCausal effects of TSH on candidate mediators\u003c/h3\u003e\n\u003cp\u003eGenetically predicted TSH was robustly associated with increased LDL-C (β\u0026thinsp;=\u0026thinsp;0.077, P\u0026thinsp;=\u0026thinsp;1.41\u0026times;10⁻⁵; 30 SNPs; Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e), with consistent results from the weighted median (β\u0026thinsp;=\u0026thinsp;0.058, P\u0026thinsp;=\u0026thinsp;0.043) and MR-PRESSO corrected estimate (β\u0026thinsp;=\u0026thinsp;0.062, P\u0026thinsp;=\u0026thinsp;7.4\u0026times;10⁻⁴ after removal of 2 outliers). There was no evidence of directional pleiotropy (MR-Egger intercept P\u0026thinsp;=\u0026thinsp;0.983). The Steiger test confirmed correct directionality (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001).\u003c/p\u003e \u003cp\u003eTSH showed a borderline association with triglycerides in the primary IVW analysis (β\u0026thinsp;=\u0026thinsp;0.031, P\u0026thinsp;=\u0026thinsp;0.053; 30 SNPs), which became significant after MR-PRESSO outlier correction (β\u0026thinsp;=\u0026thinsp;0.042, P\u0026thinsp;=\u0026thinsp;0.010 after removal of 1 outlier). There was no pleiotropy (Egger intercept P\u0026thinsp;=\u0026thinsp;0.777). The weighted median estimate was not significant (β\u0026thinsp;=\u0026thinsp;0.018, P\u0026thinsp;=\u0026thinsp;0.465).\u003c/p\u003e \u003cp\u003eTSH was not significantly associated with HDL-C (IVW β=\u0026minus;0.017, P\u0026thinsp;=\u0026thinsp;0.291). For CRP, a nominal IVW association was observed (β=\u0026minus;0.010, P\u0026thinsp;=\u0026thinsp;0.039; 61 SNPs), but this was abolished after MR-PRESSO correction (β\u0026thinsp;=\u0026thinsp;0.001, P\u0026thinsp;=\u0026thinsp;0.838 after removal of 6 outliers), with borderline pleiotropy (Egger intercept P\u0026thinsp;=\u0026thinsp;0.048), indicating that the IVW signal was driven by pleiotropic outlier instruments. Fibrinogen could not be evaluated due to insufficient clumped instruments in the outcome GWAS.\u003c/p\u003e\n\u003ch3\u003eTwo-step mediation MR\u003c/h3\u003e\n\u003cp\u003eLDL-C was identified as the sole significant mediator of the TSH\u0026ndash;CAD relationship (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). The indirect effect through LDL-C was β\u0026thinsp;=\u0026thinsp;0.029 (Sobel Z\u0026thinsp;=\u0026thinsp;4.15, P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), representing the product of the TSH\u0026rarr;LDL-C effect (β\u0026thinsp;=\u0026thinsp;0.077) and the established LDL-C\u0026rarr;CAD effect (β\u0026thinsp;=\u0026thinsp;0.368, P\u0026thinsp;=\u0026thinsp;2.0\u0026times;10⁻⁶⁷, from 71 GLGC instruments [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]). The triglyceride pathway showed a borderline indirect effect (β\u0026thinsp;=\u0026thinsp;0.008, Sobel P\u0026thinsp;=\u0026thinsp;0.059). Neither HDL-C (Sobel P\u0026thinsp;=\u0026thinsp;0.294) nor CRP (Sobel P\u0026thinsp;=\u0026thinsp;0.055) demonstrated significant mediation.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eTwo-step mediation MR results: indirect effects of TSH on CAD through candidate mediators.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"8\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\"\u0026times;\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMediator\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eStep 1 β\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eStep 1 P\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eStep 2 β\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eStep 2 P\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eIndirect\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eSobel P\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003eColoc PPH4\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLDL-C\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.077\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.4\u0026times;10⁻⁵\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.368\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026times;\" colname=\"c5\"\u003e \u003cp\u003e2.0\u0026times;10⁻⁶⁷\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.029\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e0.915\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTG\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.031\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.053\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.259\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026times;\" colname=\"c5\"\u003e \u003cp\u003e6.5\u0026times;10⁻\u0026sup1;⁹\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.008\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.059\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e0.251\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHDL-C\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e-0.017\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.291\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e-0.268\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026times;\" colname=\"c5\"\u003e \u003cp\u003e6.8\u0026times;10⁻\u0026sup2;\u0026sup1;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.005\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.294\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCRP\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e-0.010\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.039*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.127\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026times;\" colname=\"c5\"\u003e \u003cp\u003e2.2\u0026times;10⁻⁷\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e-0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.055\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e0.002\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"8\"\u003e*CRP Step 1 IVW P\u0026thinsp;=\u0026thinsp;0.039 was attenuated to P\u0026thinsp;=\u0026thinsp;0.838 after MR-PRESSO correction (6 outliers removed). Coloc PPH4 values are from Bayesian colocalization at the PCSK9 locus (chr1:54.5\u0026ndash;56.5 Mb).\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eSuppression effect\u003c/h2\u003e \u003cp\u003eA classical suppression effect was identified in the TSH\u0026ndash;CAD relationship. The total effect was null (β=\u0026minus;0.014), which decomposed into a significant harmful indirect effect through LDL-C (β=+0.029, P\u0026thinsp;\u0026lt;\u0026thinsp;0.001) and an estimated protective direct effect (β=\u0026minus;0.043). The suppression ratio was \u0026minus;\u0026thinsp;2.1, indicating that the absolute magnitude of the indirect effect exceeded that of the total effect and operated in the opposite direction to the direct effect. This pattern indicates that TSH simultaneously exerts two opposing influences on CAD: a deleterious effect mediated through lipid metabolism and a potentially protective effect through lipid-independent mechanisms.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eMultivariable MR\u003c/h2\u003e \u003cp\u003eMVMR analysis using 30 SNPs with data available across TSH, LDL-C, triglycerides, and CAD showed that the direct effect of TSH was null (β=\u0026minus;0.002, P\u0026thinsp;=\u0026thinsp;0.978) after adjusting for lipids. However, neither LDL-C (P\u0026thinsp;=\u0026thinsp;0.914) nor triglycerides (P\u0026thinsp;=\u0026thinsp;0.840) reached significance in the joint model, reflecting limited statistical power due to the small instrument overlap and potential multicollinearity. These results are presented as exploratory and should be interpreted with caution.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eReverse Mendelian randomization\u003c/h2\u003e \u003cp\u003eTo exclude reverse causality, we performed reverse MR using CAD as the exposure and TSH as the outcome. From the CARDIoGRAMplusC4D GWAS, 2,046 genome-wide significant SNPs were identified, yielding 39 independent instruments after distance-based clumping (10 Mb window). Using rsID-based coordinate bridging through the GLGC dataset, 23 instruments were successfully matched in the TSH GWAS. IVW analysis showed no evidence that genetically predicted CAD liability influences TSH levels (β=\u0026minus;0.003, SE\u0026thinsp;=\u0026thinsp;0.012, P\u0026thinsp;=\u0026thinsp;0.825), confirming the absence of reverse causality and supporting the directionality of effects from TSH to downstream outcomes.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003ePCSK9 pQTL drug-target MR and colocalization\u003c/h2\u003e \u003cp\u003eDrug-target MR using five independent PCSK9 cis-pQTL instruments from deCODE confirmed that genetically proxied PCSK9 protein levels causally increased CAD risk (IVW: β\u0026thinsp;=\u0026thinsp;0.191, OR\u0026thinsp;=\u0026thinsp;1.210, 95% CI: 1.123\u0026ndash;1.303, P\u0026thinsp;=\u0026thinsp;5.3\u0026times;10⁻⁷; Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). Directional consistency was observed with the weighted median (β\u0026thinsp;=\u0026thinsp;0.211, P\u0026thinsp;=\u0026thinsp;5.2\u0026times;10⁻⁴) and MR-Egger (β\u0026thinsp;=\u0026thinsp;0.205, P\u0026thinsp;=\u0026thinsp;0.245), with no evidence of directional pleiotropy (Egger intercept P\u0026thinsp;=\u0026thinsp;0.923). There was evidence of heterogeneity among the five instruments (Q\u0026thinsp;=\u0026thinsp;16.6, I\u0026sup2;=75.9%, P\u0026thinsp;=\u0026thinsp;0.002), which is expected in cis-MR analyses where multiple regulatory mechanisms within the same gene region may operate [\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e]. PCSK9 protein levels were strongly associated with LDL-C (β\u0026thinsp;=\u0026thinsp;0.537, P\u0026thinsp;=\u0026thinsp;6.0\u0026times;10⁻\u0026sup1;⁷⁰). A parallel analysis using the HMGCR cis-SNP rs12916 confirmed that genetically proxied LDL-C reduction via the statin pathway also reduced CAD risk (OR per 1-SD LDL-C reduction\u0026thinsp;=\u0026thinsp;0.613, P\u0026thinsp;=\u0026thinsp;1.6\u0026times;10⁻⁴).\u003c/p\u003e \u003cp\u003eBayesian colocalization at the PCSK9 locus (chr1:54.5\u0026ndash;56.5 Mb) revealed strong evidence that LDL-C and CAD share the same causal variant (PPH4\u0026thinsp;=\u0026thinsp;0.915; 1,913 SNPs), with the top shared variant being rs11591147 (chr1:55505647, a well-characterized PCSK9 loss-of-function variant). By contrast, triglycerides (PPH4\u0026thinsp;=\u0026thinsp;0.251) and CRP (PPH4\u0026thinsp;=\u0026thinsp;0.002) did not colocalize with CAD at this locus, consistent with the specificity of the LDL-C\u0026ndash;mediated pathway. As a negative control, LDL-C and CRP showed no colocalization (PPH4\u0026thinsp;=\u0026thinsp;0.040), confirming that the LDL-C signal at PCSK9 is independent of inflammatory pathways.\u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis multi-layered MR study provides, to our knowledge, the first evidence that the TSH\u0026ndash;CAD relationship is predominantly mediated through the LDL-C pathway, with the molecular mechanism traceable to the TSHR\u0026ndash;SREBP2\u0026ndash;PCSK9\u0026ndash;LDLR axis. By integrating two-step mediation MR, proteomics-based drug-target MR, and genetic colocalization, we demonstrate convergent evidence across multiple independent analytical frameworks that genetically elevated TSH increases LDL-C through PCSK9 upregulation, which in turn increases CAD risk. The identification of a suppression effect\u0026mdash;wherein this harmful lipid-mediated pathway is offset by a protective direct effect\u0026mdash;provides a mechanistic explanation for the consistently null total effect of TSH on CAD reported in all prior MR studies [\u003cspan additionalcitationids=\"CR14 CR15\" citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eOur central finding\u0026mdash;that TSH causally increases LDL-C (β\u0026thinsp;=\u0026thinsp;0.077, P\u0026thinsp;=\u0026thinsp;1.4\u0026times;10⁻⁵)\u0026mdash;is consistent with decades of clinical evidence linking SCH to elevated LDL-C [\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e] and with the demonstration that levothyroxine treatment reduces LDL-C by approximately 9\u0026ndash;14% in SCH patients [\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e]. The molecular mechanism is well characterized: TSH activates hepatocyte TSHR signaling, leading to SREBP2-mediated transcriptional upregulation of PCSK9 [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. Elevated PCSK9 promotes LDLR degradation, reducing LDL-C clearance. Our PCSK9 pQTL-MR results\u0026mdash;using five independent cis-pQTL instruments from a population (deCODE, Icelandic) entirely independent of the TSH and CAD GWAS populations\u0026mdash;directly confirm this pathway: each standard deviation increase in genetically proxied PCSK9 protein increases CAD risk by 21% (OR\u0026thinsp;=\u0026thinsp;1.210, P\u0026thinsp;=\u0026thinsp;5.3\u0026times;10⁻⁷). The strong colocalization between LDL-C and CAD at the PCSK9 locus (PPH4\u0026thinsp;=\u0026thinsp;0.915) further confirms that the causal signal passes through this specific molecular node.\u003c/p\u003e \u003cp\u003eThe suppression effect we identified has important methodological and clinical implications. Methodologically, it explains why all prior MR studies\u0026mdash;including the well-powered analyses by Ellervik et al. [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e] (N\u0026thinsp;=\u0026thinsp;195,055), Marouli et al. [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e] (N\u0026thinsp;\u0026gt;\u0026thinsp;700,000), and Sterenborg et al. [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e] (16 cardiovascular outcomes)\u0026mdash;found null total effects of TSH on CAD. A suppression effect occurs when indirect effects through different pathways operate in opposite directions, canceling each other in the total effect estimate [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. In our data, the harmful indirect effect through LDL-C (+\u0026thinsp;0.029) was counterbalanced by a protective direct effect (\u0026minus;\u0026thinsp;0.043). This also explains the significant MR-Egger intercept (P\u0026thinsp;=\u0026thinsp;0.014) observed in the total effect analysis: when opposing causal pathways coexist, the IVW estimate (which assumes homogeneous effects) conflates them, creating the appearance of directional pleiotropy. Clinically, this suppression effect suggests that interventions targeting only one arm of the TSH effect\u0026mdash;for example, lowering TSH with levothyroxine\u0026mdash;may not optimally reduce cardiovascular risk, as this simultaneously attenuates both the harmful lipid-mediated effect and the potentially protective direct effect.\u003c/p\u003e \u003cp\u003eThe nature of the protective direct effect of TSH warrants discussion. Several biological mechanisms are plausible. First, mild reductions in thyroid hormone activity decrease heart rate, cardiac contractility, and myocardial oxygen consumption, which may be protective in the setting of existing coronary atherosclerosis [\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e]. Second, TSHR is expressed on vascular endothelial cells, where TSH signaling may promote nitric oxide production via the PI3K/Akt/eNOS pathway [\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e]. Third, TSH may stabilize atherosclerotic plaques by inhibiting matrix metalloproteinase activity [\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e]. These mechanisms are speculative and require direct experimental validation; we emphasize that the direct effect is derived as a residual (total minus indirect) and cannot be definitively attributed to a specific pathway from these data alone.\u003c/p\u003e \u003cp\u003eOur findings have direct therapeutic implications for cardiovascular risk management in SCH. First, they support lipid-lowering therapy as the primary cardiovascular intervention in SCH, independent of levothyroxine treatment decisions. The PCSK9-mediated pathway identified in our study suggests that PCSK9 inhibitors may be particularly effective in this population, as they directly target the molecular mechanism through which TSH elevates LDL-C. This is consistent with the clinical observation that SCH patients show elevated PCSK9 levels that correlate with TSH [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. Second, the suppression effect implies that TSH normalization with levothyroxine may have more complex cardiovascular effects than previously assumed, potentially explaining the neutral results of randomized trials [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. A combined strategy of levothyroxine plus lipid-lowering therapy may be more effective than either alone, although this hypothesis requires prospective testing.\u003c/p\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eStrengths\u003c/h2\u003e \u003cp\u003eKey strengths include: (i) the multi-layered analytical framework providing convergent evidence from mediation MR, pQTL-MR, and colocalization; (ii) the use of PCSK9 pQTL from an independent Icelandic population (deCODE), minimizing sample overlap bias; (iii) comprehensive sensitivity analyses across six MR methods; (iv) the novel identification of a suppression effect that explains previously puzzling null results; and (v) the integration of molecular biology evidence (TSH\u0026ndash;PCSK9\u0026ndash;LDLR pathway) with genetic epidemiology, bridging the bench-to-bedside gap; and (vi) confirmation of causal directionality through both Steiger testing and formal reverse MR (CAD\u0026rarr;TSH: P\u0026thinsp;=\u0026thinsp;0.825).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003eLimitations\u003c/h2\u003e \u003cp\u003eSeveral limitations should be acknowledged. First, instrument selection employed distance-based clumping with a conservative 10 Mb window rather than LD-based clumping; however, at this window size, residual LD is negligible for common variants in European populations, and the resulting instrument set showed strong F-statistics (median 88.5) with no evidence of weak instrument bias. Second, MVMR had limited statistical power due to the small number of instruments with complete multi-trait data (n\u0026thinsp;=\u0026thinsp;30) and should be interpreted as exploratory. Third, partial sample overlap between the TSH, lipid, and CAD GWAS consortia may exist, although the principal exposure GWAS (HUNT/ThyroidOmics) and the pQTL data (deCODE, Iceland) are from independent populations. Fourth, all data were from European-ancestry cohorts, and generalizability to other populations requires further investigation. Fifth, the triglyceride-mediated pathway showed sensitivity to outlier instruments (IVW P\u0026thinsp;=\u0026thinsp;0.053 vs. MR-PRESSO corrected P\u0026thinsp;=\u0026thinsp;0.010), warranting cautious interpretation of this specific pathway. Finally, the suppression effect decomposition relies on the assumption that the identified mediators capture the dominant indirect pathways; unmeasured mediators operating in the same direction could alter the estimated direct effect.\u003c/p\u003e \u003c/div\u003e"},{"header":"Conclusions","content":"\u003cp\u003eThis multi-layered MR study demonstrates that TSH influences CAD risk predominantly through the LDL-C pathway, mediated by the TSHR\u0026ndash;SREBP2\u0026ndash;PCSK9\u0026ndash;LDLR molecular axis. The null total effect of TSH on CAD observed in this and prior studies reflects a suppression phenomenon in which harmful lipid-mediated effects are counterbalanced by a protective direct effect of uncertain mechanism. These findings support prioritizing lipid-lowering therapy\u0026mdash;particularly PCSK9-targeted interventions\u0026mdash;as the cornerstone of cardiovascular risk management in patients with subclinical hypothyroidism, and suggest that the cardiovascular consequences of levothyroxine treatment may be more nuanced than current guidelines assume.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate:\u0026nbsp;\u003c/strong\u003eNot applicable. This study exclusively used publicly available, de-identified summary-level GWAS data. No individual-level data were accessed.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication:\u0026nbsp;\u003c/strong\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials:\u0026nbsp;\u003c/strong\u003eAll GWAS summary statistics are publicly available from the sources cited. TSH GWAS: GCST010653 (GWAS Catalog). CAD GWAS: CARDIoGRAMplusC4D (GWAS Catalog GCST003116). Lipid GWAS: GLGC (GWAS Catalog GCST002216/2222/2223). PCSK9 pQTL: deCODE (https://www.decode.com/summarydata/). Analysis code will be deposited at [GitHub URL] upon acceptance.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests:\u0026nbsp;\u003c/strong\u003eThe authors declare no competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding:\u0026nbsp;\u003c/strong\u003e[none]\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026rsquo; contributions:\u0026nbsp;\u003c/strong\u003eFG conceived the study, designed the analytical framework, performed all statistical analyses, generated all figures and tables, and drafted the manuscript. LX provided clinical interpretation, supervised the study, and critically revised the manuscript. Both authors read and approved the final manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgments:\u0026nbsp;\u003c/strong\u003eWe thank the investigators of the HUNT study, ThyroidOmics Consortium, CARDIoGRAMplusC4D Consortium, Global Lipids Genetics Consortium, UK Biobank, and deCODE Genetics for making their GWAS summary statistics publicly available.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eHollowell JG et al (2002) Serum TSH, T(4), and thyroid antibodies in the United States population (1988 to 1994): NHANES III. J Clin Endocrinol Metab 87:489\u0026ndash;499\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCanaris GJ et al (2000) The Colorado thyroid disease prevalence study. Arch Intern Med 160:526\u0026ndash;534\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRodondi N et al (2010) Subclinical hypothyroidism and the risk of coronary heart disease and mortality. JAMA 304:1365\u0026ndash;1374\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRazvi S et al (2007) The beneficial effect of L-thyroxine on cardiovascular risk factors, endothelial function, and quality of life in subclinical hypothyroidism. J Clin Endocrinol Metab 92:1715\u0026ndash;1723\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBiondi B, Cooper DS (2019) Subclinical hypothyroidism. Lancet 393:1320\u0026ndash;1330\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStott DJ et al (2017) Thyroid hormone therapy for older adults with subclinical hypothyroidism. N Engl J Med 376:2534\u0026ndash;2544\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMa S et al (2018) TSH promotes metabolic syndrome by acting on the hypothalamus\u0026ndash;pituitary\u0026ndash;thyroid axis. Trends Endocrinol Metab 29:386\u0026ndash;393\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSong Y et al (2015) TSH exhibits the impact on LDLR/LDL-c via up-regulating hepatic PCSK9 expression. Metabolism 64:1395\u0026ndash;1401\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiu J et al (2022) Update on dyslipidemia in hypothyroidism: the mechanism of dyslipidemia in hypothyroidism. Endocr Connect 11:e210002\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSadik NA et al (2022) The relationship of circulating PCSK9 with TSH and lipid profile in newly diagnosed hypothyroidism. Endocr Metab Immune Disord Drug Targets 22:741\u0026ndash;749\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKwakernaak AJ et al (2014) Thyroid function is associated with PCSK9 and non-HDL cholesterol. J Clin Endocrinol Metab 99:E1346\u0026ndash;E1353\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDavey Smith G, Hemani G (2014) Mendelian randomization: genetic anchors for causal inference in epidemiological studies. Hum Mol Genet 23:R89\u0026ndash;R98\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEllervik C et al (2017) Thyroid function and ischemic heart disease: a Mendelian randomization study. Sci Rep 7:7592\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSterenborg RBTM et al (2019) Thyroid function and dysfunction in relation to 16 cardiovascular diseases. Circ Genom Precis Med 12:e002468\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMarouli E et al (2021) Higher thyrotropin leads to unfavorable lipid profile and somewhat higher cardiovascular disease risk. BMC Med 19:266\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRuan W et al (2024) Thyroid function effect on cardiac structure, cardiac function, and disease risk. Heart Rhythm 21:2272\u0026ndash;2281\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMacKinnon DP et al (2002) A comparison of methods to test mediation and other intervening variable effects. Psychol Methods 7:83\u0026ndash;104\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSkrivankova VW et al (2021) Strengthening the reporting of observational studies in epidemiology using Mendelian randomization: the STROBE-MR statement. JAMA 326:1614\u0026ndash;1621\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhou W et al (2018) Efficiently controlling for case\u0026ndash;control imbalance and sample relatedness in large-scale genetic association studies. Nat Genet 50:1335\u0026ndash;1341\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNikpay M et al (2015) A comprehensive 1,000 Genomes-based GWAS meta-analysis of coronary artery disease. Nat Genet 47:1121\u0026ndash;1130\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWiller CJ et al (2013) Discovery and refinement of loci associated with lipid levels. Nat Genet 45:1274\u0026ndash;1283\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBycroft C et al (2018) The UK Biobank resource with deep phenotyping and genomic data. Nature 562:203\u0026ndash;209\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFerkingstad E et al (2021) Large-scale integration of the plasma proteome with genetics and disease. Nat Genet 53:1712\u0026ndash;1721\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBowden J et al (2016) Consistent estimation in Mendelian randomization with some invalid instruments using a weighted median estimator. Genet Epidemiol 40:304\u0026ndash;314\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBowden J et al (2015) Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. Int J Epidemiol 44:512\u0026ndash;525\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVerbanck M et al (2018) Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization. Nat Genet 50:693\u0026ndash;698\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHemani G et al (2017) Orienting the causal relationship between imprecisely measured traits using GWAS summary data. PLoS Genet 13:e1007081\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSobel ME (1982) Asymptotic confidence intervals for indirect effects in structural equation models. Sociol Methodol 13:290\u0026ndash;312\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBurgess S, Thompson SG (2015) Multivariable Mendelian randomization: the use of pleiotropic genetic variants to estimate causal effects. Am J Epidemiol 181:251\u0026ndash;260\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGiambartolomei C et al (2014) Bayesian test for colocalisation between pairs of genetic association studies. PLoS Genet 10:e1004383\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSchmidt AF et al (2020) Genetic drug target validation using Mendelian randomisation. Nat Commun 11:3255\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDuntas LH (2002) Thyroid disease and lipids. Thyroid 12:287\u0026ndash;293\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRazvi S et al (2017) Subclinical hypothyroidism: to treat or not to treat, that is the question! A systematic review with meta-analysis on lipid profile. Endocr Connect 6:R188\u0026ndash;R199\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKlein I, Danzi S (2007) Thyroid disease and the heart. Circulation 116:1725\u0026ndash;1735\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTian L et al (2014) TSH stimulates the proliferation of vascular smooth muscle cells. Endocrine 46:651\u0026ndash;658\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDiekman MJ et al (2001) Determinants of changes in plasma homocysteine in hypothyroidism and hyperthyroidism. Clin Endocrinol 54:197\u0026ndash;204\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"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":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Mendelian randomization, thyroid-stimulating hormone, coronary artery disease, LDL cholesterol, PCSK9, mediation analysis, suppression effect, subclinical hypothyroidism, proteomics, colocalization","lastPublishedDoi":"10.21203/rs.3.rs-9273655/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9273655/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eSubclinical hypothyroidism (SCH) affects 4\u0026ndash;10% of the general population and has been inconsistently linked to coronary artery disease (CAD). Whether thyroid-stimulating hormone (TSH) causally influences CAD risk, and through which biological pathways, remains unresolved. We conducted a multi-layered Mendelian randomization (MR) study integrating mediation analysis, proteomics-based drug-target MR, and genetic colocalization to dissect the causal architecture of the TSH\u0026ndash;CAD relationship.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eTwo-sample MR was performed using 63 genome-wide significant instruments for TSH (Zhou et al., HUNT/ThyroidOmics, N\u0026thinsp;=\u0026thinsp;119,715) against CAD (CARDIoGRAMplusC4D, 60,801 cases/123,504 controls). Two-step mediation MR with Sobel test evaluated five candidate pathways: metabolic (LDL-C, triglycerides, HDL-C), inflammatory (CRP), and coagulation (fibrinogen). Drug-target MR used five independent PCSK9 cis-pQTL instruments from deCODE proteomics (N\u0026thinsp;=\u0026thinsp;35,363). Bayesian colocalization was performed at the PCSK9 locus. Sensitivity analyses included IVW (fixed and random effects), weighted median, MR-Egger, MR-PRESSO, Steiger directionality test, and reverse MR (CAD\u0026rarr;TSH).\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eGenetically predicted TSH showed no overall association with CAD (IVW: OR\u0026thinsp;=\u0026thinsp;0.987, 95% CI: 0.943\u0026ndash;1.032, P\u0026thinsp;=\u0026thinsp;0.557; 62 SNPs). However, TSH was robustly associated with increased LDL-C (β\u0026thinsp;=\u0026thinsp;0.077, P\u0026thinsp;=\u0026thinsp;1.4\u0026times;10⁻⁵; confirmed by weighted median P\u0026thinsp;=\u0026thinsp;0.043 and MR-PRESSO corrected P\u0026thinsp;=\u0026thinsp;7.4\u0026times;10⁻⁴), with no evidence of pleiotropy (Egger intercept P\u0026thinsp;=\u0026thinsp;0.983). The triglyceride association was borderline (IVW P\u0026thinsp;=\u0026thinsp;0.053; MR-PRESSO corrected P\u0026thinsp;=\u0026thinsp;0.010). No robust associations were found with HDL-C (P\u0026thinsp;=\u0026thinsp;0.291) or CRP (IVW P\u0026thinsp;=\u0026thinsp;0.039; MR-PRESSO corrected P\u0026thinsp;=\u0026thinsp;0.838 after removing 6 outliers). Two-step mediation MR identified LDL-C as the sole significant mediator (indirect effect β\u0026thinsp;=\u0026thinsp;0.029, Sobel P\u0026thinsp;\u0026lt;\u0026thinsp;0.001); the triglyceride pathway was borderline (Sobel P\u0026thinsp;=\u0026thinsp;0.059). A suppression effect was identified: the harmful LDL-mediated indirect effect (+\u0026thinsp;0.029) was offset by a protective direct effect (\u0026minus;\u0026thinsp;0.043), yielding the observed null total effect (suppression ratio\u0026thinsp;=\u0026thinsp;\u0026minus;\u0026thinsp;2.1). Drug-target MR confirmed that genetically proxied PCSK9 protein levels causally increased CAD risk (IVW: OR\u0026thinsp;=\u0026thinsp;1.210, P\u0026thinsp;=\u0026thinsp;5.3\u0026times;10⁻⁷; weighted median P\u0026thinsp;=\u0026thinsp;5.2\u0026times;10⁻⁴; Egger intercept P\u0026thinsp;=\u0026thinsp;0.923). Reverse MR confirmed no evidence of reverse causality (CAD\u0026rarr;TSH: β=\u0026minus;0.003, P\u0026thinsp;=\u0026thinsp;0.825). Strong colocalization was observed between LDL-C and CAD at the PCSK9 locus (PPH4\u0026thinsp;=\u0026thinsp;0.915), but not for triglycerides (PPH4\u0026thinsp;=\u0026thinsp;0.251) or CRP (PPH4\u0026thinsp;=\u0026thinsp;0.002).\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e \u003cp\u003eTSH influences CAD risk predominantly through an LDL-C\u0026ndash;mediated pathway involving the TSHR\u0026ndash;SREBP2\u0026ndash;PCSK9\u0026ndash;LDLR axis, as validated by convergent evidence from mediation MR, proteomics-based MR, and colocalization. The null total effect reflects a suppression phenomenon where harmful lipid-mediated and protective lipid-independent effects of TSH counterbalance each other. These findings support prioritizing lipid-lowering therapy\u0026mdash;particularly PCSK9-targeted interventions\u0026mdash;over TSH normalization alone for cardiovascular risk management in SCH.\u003c/p\u003e","manuscriptTitle":"Genetically Predicted Thyroid-Stimulating Hormone and Coronary Artery Disease: Lipid-Mediated Causal Pathways, PCSK9 Protein Validation, and a Suppression Effect Revealed by Multi-Layered Mendelian Randomization","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-04-01 10:17:37","doi":"10.21203/rs.3.rs-9273655/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"867d1bce-0271-4e22-87df-3d64c386cc2c","owner":[],"postedDate":"April 1st, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":65431695,"name":"Cardiac \u0026 Cardiovascular Systems"}],"tags":[],"updatedAt":"2026-04-01T10:17:37+00:00","versionOfRecord":[],"versionCreatedAt":"2026-04-01 10:17:37","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9273655","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9273655","identity":"rs-9273655","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}