Cardiac Biomarkers to Refine Pretest Probability for Coronary Obstruction and Predict Survival After Revascularization in Chronic Coronary Syndrome Authors: Andrej Teren, Jeffrey Netto, Joachim Thiery, Holger Thiele, Christoph Stellbrink, Dennis Lawin, Thorsten Lawrenz, Anselm Artur Derda, Sylvia Henger, Holger Kirsten, Markus Scholz, Thorsten Kaiser Published: 2026 ABSTRACT AbstractBackgroundThe diagnostic and prognostic roles of cardiovascular biomarkers in chronic coronary syndrome remain unclear.MethodsPatients undergoing coronary angiography for suspected chronic coronary syndrome were prospectively enrolled (n=2251; median follow‐up 12.6 years). Obstructive coronary artery disease was defined as ≥50% stenosis in ≥1 major epicardial vessel. HsTnT (high‐sensitivity cardiac troponin T), NT‐proBNP (N‐terminal pro‐B‐type natriuretic peptide), high‐sensitivity C‐reactive protein, interleukin‐6, and copeptin were measured. Receiver operating characteristic analysis and Cox regression with biomarker × treatment interaction testing were performed.ResultsOverall, 888 patients (39.4%) had obstructive coronary artery disease. Only hsTnT showed meaningful diagnostic capacity (area under the curve 0.669; risk factor‐weighted clinical likelihood area under the curve 0.663), with incremental benefit inversely proportional to risk factor‐weighted clinical likelihood category (Δ area under the curve: very low 10.4%, low 8.0%, intermediate/high 5.0%). NT‐proBNP was the strongest mortality predictor across optimal medical therapy (hazard ratio [HR], 1.488 [95% CI, 1.288–1.720], P<0.001), percutaneous coronary intervention (HR, 1.220 [95% CI, 1.020–1.458], P=0.029), and coronary artery bypass grafting (HR, 1.220 [95% CI, 1.049–1.420], P=0.010). Interaction analysis validated a data‐derived 150 pg/mL threshold (P=0.032): below it, mortality was comparably low regardless of revascularization (HR, 0.98 [95% CI, 0.67–1.43], P=0.910); above it, baseline risk was markedly elevated (HR, 5.75 [95% CI, 4.10–8.00], P<0.001) and revascularization associated with 40% mortality reduction, with substantial residual risk (HR, 3.43 [95% CI, 2.70–4.40], P<0.001).ConclusionsHsTnT provides risk factor‐weighted clinical likelihood category‐specific incremental diagnostic value. NT‐proBNP is a universal prognostic marker identifying patients with distinct revascularization‐associated mortality reduction driven by differential baseline risk.RegistrationURL: https://clinicaltrials.gov/study/NCT00497887; Unique Identifier: NCT00497887.Keywords: B‐type natriuretic peptide, chronic coronary syndrome, pretest probability, revascularization, troponinSubject Categories: Cardiovascular Disease, Secondary Prevention Nonstandard Abbreviations and Acronyms chronic coronary syndrome European Society of Cardiology high‐sensitivity cardiac troponin T optimal medical therapy pretest probability risk factor‐weighted clinical likelihood standardized mean difference Clinical Perspective What Is New? hsTnT (High‐sensitivity troponin T) provides incremental diagnostic value for coronary obstruction that follows an inverse gradient, with greatest benefit in very low pretest likelihood patients (Δ area under the curve 10.4%). hsTnT (High‐sensitivity troponin T) provides incremental diagnostic value for coronary obstruction that follows an inverse gradient, with greatest benefit in very low pretest likelihood patients (Δ area under the curve 10.4%). NT‐proBNP (N‐terminal pro‐B‐type natriuretic peptide) ≥150 pg/mL identifies a subgroup with chronic coronary syndrome with markedly elevated baseline mortality risk in whom revascularization is associated with a 40% mortality reduction, yet substantial residual risk persists. NT‐proBNP (N‐terminal pro‐B‐type natriuretic peptide) ≥150 pg/mL identifies a subgroup with chronic coronary syndrome with markedly elevated baseline mortality risk in whom revascularization is associated with a 40% mortality reduction, yet substantial residual risk persists. What Are the Clinical Implications? hsTnT testing adds most diagnostic value when selectively applied in low pretest likelihood patients, and elevated NT‐proBNP signals the need for comprehensive medical management and long‐term surveillance beyond revascularization alone. hsTnT testing adds most diagnostic value when selectively applied in low pretest likelihood patients, and elevated NT‐proBNP signals the need for comprehensive medical management and long‐term surveillance beyond revascularization alone. Coronary artery disease (CAD) remains a leading cause of morbidity and mortality worldwide, with chronic coronary syndrome (CCS) representing a significant clinical burden. 1 , 2 Accurate detection of coronary obstruction and prediction of long‐term outcomes remain challenging. 3 , 4 Current diagnostic approaches rely on clinical assessment, noninvasive imaging, and invasive coronary angiography, with inherent limitations. 5 The 2024 European Society of Cardiology (ESC) guidelines introduced a revised pretest probability (PTP) model based on risk factor—weighted clinical likelihood (RF‐CL) incorporating additional risk factors (ie, diabetes, smoking, dyslipidemia, hypertension, and family history) beyond the traditional age‐sex‐symptom triad. 2 Direct comparison of novel biomarker‐based approaches to improve the predictive value of guideline‐recommended PTP models remains limited. Moreover, differences in revascularization benefit between patients with CCS and myocardial infarction (MI) highlight the need for tailored therapeutic and secondary‐preventive strategies. 6 Cardiac‐specific biomarkers hsTnT (high‐sensitivity cardiac troponin T) and NT‐proBNP (N‐terminal pro‐B‐type natriuretic peptide) have demonstrated utility in acute MI, and to a lesser extent in CCS settings. 6 , 7 , 8 NT‐proBNP, in particular, represents a cornerstone biomarker across the entire spectrum of cardiovascular diseases, demonstrating strong diagnostic and prognostic value in patients with heart failure from acute decompensation through chronic stable disease, with guideline‐recommended thresholds for diagnosis and risk stratification. 7 In CCS, these biomarkers may serve dual roles in both diagnostic and prognostic applications. Inflammatory biomarkers hsCRP (high‐sensitivity C‐reactive protein) and IL‐6 (interleukin‐6), and copeptin also warrant systematic investigation, as these biomarkers may offer incremental value in both diagnostic assessment and prognostic stratification of patients with CCS. 3 , 9 Beyond stenosis detection, identifying patients most likely to benefit from revascularization remains an unresolved challenge. 10 Recent meta‐analysis of 19 randomized trials demonstrated that percutaneous coronary intervention reduces cardiovascular mortality (odds ratio [OR], 0.77, P=0.02) in CCS, yet benefit varies across subgroups, underscoring the need for biomarker‐based patient selection beyond anatomic evaluation alone. 11 This study aimed to evaluate the diagnostic performance of 5 circulating biomarkers against 2024 PTP model, assesses incremental diagnostic value stratified by baseline risk categories, and determines prognostic utility across different treatment strategies to identify biomarker thresholds predicting long‐term survival after revascularization. METHODS Data Availability Statement The data that support the findings of this study are available from the Leipzig Research Centre for Civilization Diseases (LIFE), University of Leipzig. Restrictions apply to the availability of these data, which were used under license for this study. Data are available from the corresponding author upon reasonable request and with permission of the LIFE‐Heart study steering committee, subject to institutional data sharing agreements and ethical approval. Study Design and Population This analysis used data from the LIFE‐Heart study, a prospective longitudinal cohort examining patients with suspected CCS at the Heart Center Leipzig at Leipzig University. The study aimed to identify genetic, biochemical, and environmental markers associated with coronary and extracoronary atherosclerosis development. 12 The protocol adhered to Declaration of Helsinki principles and was approved by the Ethics Committee of the Medical Faculty, University of Leipzig (registration 276‐2005; ClinicalTrials.gov NCT00497887). Written informed consent was obtained from all participants. Blood samples were collected prospectively at enrollment before coronary angiography, with subsequent clinical follow‐up for vital status and cardiovascular outcomes. Primary end point was all‐cause mortality, ascertained through civil registry linkage, ensuring complete vital status follow‐up. We here consider a subset of LIFE‐Heart comprising 3305 patients (2122 male, 1183 female) with suspected CCS based on clinical symptoms or positive noninvasive testing, undergoing first‐time coronary angiography. Exclusion criteria for the present analysis included acute MI, acute decompensated heart failure, pregnancy, lactation, and severe systemic diseases (autoimmune diseases with immunomodulatory therapy, dialysis‐requiring conditions, infectious diseases, and malignancy or cancer treatment within 2 years). After excluding further 1054 patients (821 for incomplete clinical symptoms and cardiovascular risk factor assessment, 121 for indeterminate angiographic findings, 112 for hsTnT ≥52 pg/mL), 2251 patients were eligible for the present analysis. Median follow‐up was 12.6 years (range 0.1–16.2 years). Assessment of Cardiovascular Risk Factors and Coronary Status Traditional atherosclerotic risk factors were evaluated through standardized clinical interviews, biometric assessments, and laboratory analyses. Key comorbidities considered as confounding variables included diabetes (ie, history of diabetes or glycated hemoglobin ≥6.5%), obesity (ie, body mass index ≥30 kg/m2), dyslipidemia (ie, history of dyslipidemia or administration of lipid‐modifying agents—Anatomical Therapeutic Chemical group C10), hypertension (ie, history of arterial hypertension, administration of antihypertensives, or current blood pressure ≥140/90 mm Hg), and impaired renal function, that is estimated glomerular filtration rate applying Chronic Kidney Disease Epidemiology Collaboration formula <90 mL/min per 1.73 m2. Medications were recorded according to Anatomical Therapeutic Chemical classification codes. Coronary angiography was performed following institutional protocols. Participants were stratified based on angiographic findings considering ≥50% luminal reduction as significant coronary obstruction. In order to refine the adjustment for angiographic extent of CAD, number of vessels with coronary obstruction and Gensini score were calculated using predefined standards. 13 The selection of coronary therapeutic modality in case of relevant coronary obstruction—optimal medical therapy (OMT), percutaneous coronary intervention (PCI), or coronary artery bypass grafting (CABG)—was determined at the clinical discretion of the physician performing the index coronary angiography. Laboratory Procedures Venous blood samples were collected at the day of index coronary angiography and transported at 4 °C to the laboratory within 5 hours. Samples underwent immediate centrifugation, analysis, and storage in liquid nitrogen gas phase at −80 °C. All procedures followed ISO 15189 and ISO 17025 standards. In the majority of patients (87.3%), samples were obtained immediately before catheterization (median 1.2 hours before arterial access, interquartile range [IQR], 0.5–4.1 hours). In a subset of cases (12.7%), samples were collected shortly after angiography (median 2.3 hours post procedure). Standard lipid profile, that is low‐density lipoprotein cholesterol, high‐density lipoprotein cholesterol, and triglycerides plus serum creatinine were analyzed according to LIFE‐Heart protocol. 12 Five biomarkers were selected for comprehensive evaluation based on their established pathophysiological relevance and clinical utility in cardiovascular disease: cardiac‐specific markers (hsTnT, NT‐proBNP) representing myocardial injury and wall stress, inflammatory markers (hsCRP, IL‐6) reflecting atherosclerotic disease activity, and copeptin as a marker of chronic neurohumoral activation. HsTnT, NT‐proBNP, hsCRP, and IL‐6 were measured by electrochemiluminescence on a Roche Cobas 8000 analyzer (Roche Diagnostics, Mannheim, Germany). Serum copeptin was determined using immunological trace technology (ThermoFisher/Brahms, Waltham, MA, USA). To ensure that biomarker measurements reflected chronic cardiovascular disease states rather than acute clinical conditions, we implemented predefined upper threshold exclusions. Patients with NT‐proBNP ≥1000 pg/mL were excluded to eliminate those with acute decompensated heart failure, as this threshold represents the established diagnostic cutoff for acute heart failure in the ESC guidelines. 7 Copeptin values ≥13.8 pmol/L were excluded to eliminate potential cases of diabetes insipidus or acute severe illness, as this threshold approximates diagnostic cutoffs for posterior pituitary dysfunction. 14 For inflammatory markers, we excluded patients with hsCRP ≥10 mg/L and IL‐6 ≥43.5 pg/mL to eliminate acute inflammatory states, infections, or acute phase reactions that would confound assessment of chronic cardiovascular inflammation. Importantly, for biomarkers other than hsTnT, exclusions were applied only when examining the diagnostic or prognostic value of the corresponding biomarker. Statistical Analysis Statistical analyses were performed using SPSS version 29.0 and R statistical software. Continuous variables are presented as medians with IQRs and categorical variables as frequencies and percentages. Complete case analysis was performed. Among 2251 patients in the primary cohort, missing data were minimal: waist–hip ratio 93 patients (4.1%), ejection fraction 34 patients (1.5%), follow‐up time 6 patients (0.3%). Patients with missing outcome or covariate data were excluded from respective analyses (sensitivity analysis cohort: 718/766=93.8% retention). Complete missing data analysis is provided in Table S1. Missing data did not substantially differ by CAD status or mortality. Patients were dichotomized based on angiographic findings: obstructive CAD (ie, ≥50% stenosis in at least 1 major epicardial vessel) versus no or nonobstructive disease (ie, <50% stenosis). Pretest probability was assessed using RF‐CL (ie age, sex, symptoms, diabetes, current smoking, dyslipidemia, hypertension, family history of premature CAD defined as MI or coronary revascularization before age 60 years in first‐degree relative). In order to examine the incremental discrimination value of biomarkers, patients were subsequently categorized based on RF‐CL as Category 1—very low likelihood (PTP <5%), Category 2—low likelihood (5%–15%), Category 3—intermediate/high likelihood (>15%) of relevant coronary obstruction. The more comprehensive cardiovascular risk factor model included RF‐CL variables plus low‐density lipoprotein cholesterol, high‐density lipoprotein cholesterol, triglycerides, body mass index, waist–hip ratio, administration of antihypertensive medication, administration of statins, history of hyperlipoproteinemia, systolic and diastolic blood pressure (averaged from 3 standardized measurements), left ventricular ejection fraction assessed by standardized transthoracic echocardiography and estimated glomerular filtration rate applying the Chronic Kidney Disease Epidemiology Collaboration formula. For diagnostic performance assessment, receiver operating characteristic curves were constructed for all 5 biomarkers. Area under the curve (AUC) values with 95% CIs were tested for differences using DeLong's method. 15 The biomarker demonstrating meaningful diagnostic performance was then compared directly with RF‐CL model. Optimal cutoff was determined using Youden's index. 16 Incremental discriminatory value was quantified by the change in AUC (ΔAUC) obtained by adding each biomarker at its optimal cutoff to the baseline model, assessed both overall and within predefined RF‐CL categories. Bootstrap validation with 500 iterations established robust CIs. 8 Bootstrap internal validation (250 iterations) assessed optimistic bias for data‐derived thresholds following Harrell–Steyerberg methodology, rederiving optimal cutoffs in each bootstrap sample and calculating optimism as the mean difference between apparent and test performance. Model calibration was evaluated using Hosmer–Lemeshow goodness‐of‐fit test and calibration plots comparing observed versus predicted probabilities across deciles of predicted risk. Brier score assessed overall model performance, with lower scores indicating better calibration. Clinical utility of identified biomarker diagnostic value was evaluated through decision curve analysis, quantifying net benefit across threshold probabilities by comparing true positives gained against false positives incurred. Survival analysis employed Kaplan–Meier curves with log‐rank tests and Cox proportional hazards regression to perform safety analysis for false negatives to detect coronary obstruction (ie, CAD cases reclassified downward based on biomarker profile) and to determine which biomarker provided strongest prognostic discrimination. Patients were stratified by coronary anatomy and treatment strategy, that is, patients with obstructive CAD treated OMT alone versus with PCI, and CABG, respectively. Cox proportional hazards regression with backward stepwise elimination was performed using differential adjustment strategies: safety analyses for discriminative value to detect coronary obstruction adjusted for RF‐CL variables, and prognostic models adjusted for number of vessels with coronary obstruction, Gensini score, and comprehensive cardiovascular risk factors. Hazard ratios (HRs) were calculated per 1 SD increase to enable cross‐biomarker comparison. To assess the stability of biomarker effect estimates, sensitivity analyses were performed forcing all candidate covariates into logistic and Cox regression models without selection (ENTER method) (Table S2A and S2B). Proportional hazards assumptions were verified using Schoenfeld residuals and log–log plots for all Cox models. No substantial violations were detected for primary covariates (all P>0.05 for time‐dependence tests; Figure S1). Time‐dependent ROC analysis using the survivalROC package in R determined optimal NT‐proBNP cutoff for mortality prediction. Cumulative sensitivity and specificity were calculated across biomarker values at 1, 3, 5, and 10 years, with optimal cutoff identified using Youden's index. Bootstrap validation (500 iterations) established CIs. Bootstrap validation (250 iterations) was performed to assess threshold optimism, rederiving optimal cutoffs in each bootstrap sample using time‐dependent ROC analysis for 5‐year mortality. To rigorously address potential confounding by indication, we performed 1:1 propensity score matching without replacement using all known confounders for the subsequent analysis. Propensity scores for receiving revascularization were estimated using multivariable logistic regression. Treated patients were matched to controls using nearest‐neighbor matching with caliper width set at 0.2 times the SD of the logit propensity score. Matching quality was assessed using standardized mean differences (SMD), with |SMD| <0.1 indicating excellent balance and |SMD| <0.2 considered acceptable. Covariate balance is detailed in Table S3 and visualized in Figure S2. In the matched cohort, we performed Cox proportional hazards regression with minimal covariate adjustment (age, sex) given the achieved balance. The NT‐proBNP × revascularization interaction was analyzed using a 4‐group framework with “No Revascularization + NT‐proBNP below the post hoc data‐derived threshold” as the reference group and tested using a product term interaction. Additionally, a continuous dose–response modeling was performed using restricted cubic splines. All statistical tests were 2 sided, and P values <0.05 were considered statistically significant. Given the exploratory and hypothesis‐generating nature of this single‐center observational study, no formal correction for multiple comparisons was prespecified. All 5 biomarkers were evaluated systematically for both diagnostic and prognostic utility; no single analysis was designated confirmatory a priori. All P values are reported continuously, and all findings should be interpreted as hypothesis generating. RESULTS Baseline Characteristics Among 2251 patients, 888 (39.4%) had obstructive CAD. Table 1 presents baseline characteristics stratified by angiographic findings. Patients with obstructive CAD were older (median 67 versus 63 years, P<0.001), more frequently male (75.7% versus 60.9%, P<0.001), and exhibited higher prevalence of traditional cardiovascular risk factors. Table 1Baseline Characteristics of Study Population Stratified by Presence of Obstructive CCSCharacteristicsNo CAD%/25–75 IQRCAD%/25–75 IQR P valueNo. or medianNo. or medianNo.136360.6%88839.4%…Female sex, N60544.4%19321.7%<0.001Age, y61.653.8–69.865.757.3–71.1<0.001Body mass index, kg/m2 29.326.4–32.829.226.4–32.60.370Waist–hip ratio0.960.88–1.031.010.95–1.01<0.001Diabetes, N33524.6%30234.0%<0.001Family history with myocardial infarction, N15811.6%11312.7%0.439Current smoker, N21315.6%18821.2%0.003Systolic blood pressure, mm Hg136125–149138126–1530.095Diastolic blood pressure, mm Hg84.577.5–92.082.574.6–89.4<0.001Antihypertensive medication, N114483.9%77687.4%<0.001Hyperlipidemia, N116585.5%81792.0%0.002Lipid‐lowering medication, N45933.7%41747.0%<0.001Low‐density lipoprotein cholesterol, mmol/L3.262.61–3.913.432.66–4.23<0.001High‐density lipoprotein cholesterol, mmol/L1321.10–1.611.221.01–1.48<0.001Triglycerides, mmol/L1.631.13–2.411.741.27–2.48<0.001Estimated glomerular filtration rate, in mL/min per 1.73 m2 87.875.2–96.884.971.9–94.8<0.001Left‐ventricular ejection fraction, %6155–666054–650.022Typical angina46634.2%43749.2%<0.001Atypical angina21315.6%859.6%<0.001Nonanginal chest pain31623.2%10912.3%…Dyspnea New York Heart Association class ≥270952.0%47453.4%0.546No chest pain/no dyspnea26211.6%1938.6%0.407European Society of Cardiology pretest probability 20245%26019.1475.3<0.0015%–15%63746.731335.2>15%46634.252859.2Percutaneous coronary intervention……34438.7…Coronary artery bypass grafting……26730.1High‐sensitivity troponin T, pg/mL6.584.0–10.79.866.55–15.56<0.001N‐terminal pro‐B‐type natriuretic peptide, pg/mL111.550.6–272.2172.175.3–415.7<0.001C‐reactive protein, mg/L1.961.02–4.192.351.12–5.07<0.001Interleukin‐6, pg/mL2.381.50–4.022.991.80–5.50<0.001Copeptin, pmol/L4.723.05–7.865.913.86–9.85<0.001CAD indicates obstructive coronary artery disease; IQR, interquartile range. Baseline Characteristics of Study Population Stratified by Presence of Obstructive CCS CAD indicates obstructive coronary artery disease; IQR, interquartile range. Left ventricular function was preserved (ejection fraction ≥50%) in 1850 patients (82.2%), with mildly reduced (40%–49%) in 267 (11.9%) and reduced (<40%) in 100 (4.4%). Heart failure symptom burden by New York Heart Association functional class ≥2 was documented in 1183 patients (52.6%). Biomarker concentrations demonstrated systematic elevation in patients with obstructive CAD. Median hsTnT was 10.0 ng/L (IQR, 6.0–16.0) in obstructive CAD versus 7.0 ng/L (IQR, 5.0–11.0) in nonobstructive disease (P<0.001). NT‐proBNP showed similar gradient: 151 pg/mL (IQR, 74–340) versus 98 pg/mL (IQR, 49–198), P<0.001. Inflammatory markers exhibited more modest differences. Obstructive CAD prevalence showed progressive separation across RF‐CL categories (low likelihood 23.6%, intermediate likelihood 32.9%, high likelihood 53.1%) (chi‐square P<0.001). Diagnostic Performance of Biomarkers to Detect Anatomical Coronary Obstruction Comparative ROC analysis across all 5 biomarkers revealed marked heterogeneity in diagnostic performance for detecting anatomical coronary obstruction. Only hsTnT demonstrated clinically relevant diagnostic capacity with AUC 0.669 (95% CI, 0.644–0.692). NT‐proBNP showed modest discrimination with AUC 0.609 (95% CI, 0.585–0.633), whereas inflammatory markers demonstrated limited diagnostic value: hsCRP AUC 0.536 (95% CI, 0.511–0.561), IL‐6 AUC 0.547 (95% CI, 0.522–0.572), and copeptin AUC 0.524 (95% CI, 0.499–0.549). Direct comparison with ESC 2024 PTP model (Figure 1) demonstrated that hsTnT performance was comparable to RF‐CL (AUC, 0.663 [95% CI, 0.642–0.689], P=0.842 for comparison). The comprehensive cardiovascular risk factor model achieved strong baseline performance (AUC, 0.731 [95% CI, 0.709–0.753]), with hsTnT addition providing limited additional benefit (AUC, 0.747, improvement +0.016, P=0.013). Optimal hsTnT cutoff by Youden index was 7.0 ng/L (sensitivity 72.4%, specificity 54.0%). This threshold provided the basis for incremental value assessment beyond current RF‐CL model (Table S4). Bootstrap validation (250 iterations) confirmed minimal bias: optimal cutoff median 6.97 ng/L (95% CI, 6.62–8.47), chosen 7.0 ng/L at 70th percentile, AUC optimism 0.001 (optimism‐corrected AUC 0.668). Comparative diagnostic performance of hsTnT with ESC 2024 PTP (considering age, sex, symptoms plus diabetes, smoking, dyslipidemia, hypertension, family history of premature CAD). AUC indicates area under the curve; CAD, coronary artery disease; CRF, comprehensive cardiovascular risk factor model (ie, ESC 2024 PTP variables plus); ESC, European Society of Cardiology; hsTnT, high‐sensitivity troponin T; PTP, pretest probability; and RF‐CL, risk factor‐weighted clinical likelihood. Incremental value of the hsTnT 7.0 ng/L cutoff stratified by RF‐CL categories demonstrated an inverse relationship between baseline pretest clinical likelihood and biomarker benefit (Table 2, Table S4). In very low likelihood patients (RF‐CL <5%), hsTnT addition yielded substantial discrimination improvement (ΔAUC=10.4%, P<0.001). Low pretest likelihood patients (RF‐CL 5%–15%) showed moderate benefit: ΔAUC=8.0%, P<0.001. Intermediate/high likelihood patients (RF‐CL >15%) exhibited the smallest incremental value: ΔAUC=5.0%, P<0.001 (Table 2). Table 2AUC Improvement by ESC 2024 RF‐CL Risk Category to Detect Coronary ObstructionRF‐CL categoryNo.Prevalence obstructive CAD (%)AUC (RF‐CL)AUC (RF‐CL+ hsTnT 7.0 ng/L)Delta AUC 95% CI (%) P valueVery low (<5%)30715.3%0.6000.7040.104 (0.062 to 0.146)<0.001Low (5%–15%)95032.9%0.5310.6110.08 (0.052 to 0.108)<0.001Intermediate/high (>15%)99453.1%0.5430.5930.05 (0.024 to 0.076)<0.001Overall225139.4%0.6630.6940.031 (−0.001 to 0.064)<0.001Comparative discriminative ability of RF‐CL model alone vs combined with hsTnT, stratified by baseline risk categories. AUC indicates area under the curve; CAD, coronary artery disease; ESC, European Society of Cardiology; hsTnT, high‐sensitivity Troponin T; and RF‐CL, risk factor weighted clinical likelihood. AUC Improvement by ESC 2024 RF‐CL Risk Category to Detect Coronary Obstruction Comparative discriminative ability of RF‐CL model alone vs combined with hsTnT, stratified by baseline risk categories. AUC indicates area under the curve; CAD, coronary artery disease; ESC, European Society of Cardiology; hsTnT, high‐sensitivity Troponin T; and RF‐CL, risk factor weighted clinical likelihood. Model calibration analysis (Figure S3A) demonstrated that the RF‐CL model showed acceptable fit (Hosmer–Lemeshow χ2=12.4, P=0.134, Brier score=0.224), with addition of hsTnT ≥7.0 ng/L cutoff improving calibration (χ2=8.9, P=0.351, Brier score=0.216, ΔBrier=−0.008; P<0.001). Decision curve analysis (Figure S3B) revealed that hsTnT provided clinical net benefit across threshold probabilities of 20% to 60%, with greatest benefit in the 30% to 45% range. Reclassification patterns (Table S5) showed differential biomarker impact by disease status: among 888 patients with obstructive CAD, 218 (24.5%) were reclassified upward and 228 (25.7%) downward, whereas among 1363 patients without obstructive CAD, 521 (38.2%) were appropriately reclassified downward versus 312 (22.9%) upward. Safety analysis (Table S6) revealed that unadjusted analysis showed false negative patients (obstructive CAD with hsTnT <7.0 ng/L, n=249) had significantly higher mortality than true negatives (no obstructive CAD with hsTnT <7.0 ng/L) in the very low likelihood category (HR, 1.83 [95% CI, 1.16–2.89], P=0.009). However, adjustment for RF‐CL risk factor variables substantially attenuated this apparent excess mortality to nonsignificance: very low CL (adjusted HR, 1.29 [95% CI, 0.81–2.06], P=0.283), low CL (adjusted HR, 0.90 [95% CI, 0.53–1.54], P=0.707), and overall (adjusted HR, 1.10 [95% CI, 0.77–1.56], P=0.608). RF‐CL factors as such (ie, not used as scoring components in categorical RF‐CL model) explained 65% to 81% of the excess risk, demonstrating that the unadjusted mortality differences were primarily attributable to baseline cardiovascular risk factor burden rather than low hsTnT itself. Long‐Term Survival Outcomes During median 12.6‐year follow‐up, 684 deaths occurred (30.4%) in 2251 patients. Overall survival rates were no/nonobstructive CAD 74% (95% CI, 72%–77%, n=1363), obstructive CAD with PCI 69% (64%–74%, n=344), OMT 64% (58%–69%, n=277), and CABG 54% (48%–60%, n=267; log‐rank P<0.001). Patients with no/nonobstructive CAD had significantly better survival than all groups with obstructive CAD (all P≤0.016), PCI showed superior survival versus both OMT (P=0.038) and CABG (P<0.001), whereas OMT and CABG did not differ significantly (P=0.16). Figure 2 presents full‐adjusted Cox regression HRs across patient subgroups. NT‐proBNP demonstrated universal prognostic utility across all categories with consistent effect magnitudes. Per 1 SD increase, NT‐proBNP conferred increased mortality risk: OMT HR 1.488 (95% CI, 1.288–1.720, P<0.001), PCI HR 1.220 (95% CI, 1.020–1.458, P=0.029), and CABG HR 1.220 (95% CI, 1.049–1.420, P=0.010). Treatment‐stratified biomarker prognostic performance Cox regression hazard ratios (per 1 SD increase) for 5 biomarkers across patient subgroups. Significant associations (P<0.05) shown in red (increased risk); nonsignificant in gray. NT‐proBNP demonstrated universal prognostic utility across all categories. CABG indicates coronary artery bypass grafting; CAD, coronary artery disease; HDL, high‐density lipoprotein; HR, hazard ratio; hsCRP, high‐sensitivity; C‐reactive protein; hsTnT, high‐sensitivity troponin T; IL‐6, Interleukin 6; LDL, low‐density lipoprotein; NT‐proBNP, N‐terminal pro‐B‐type natriuretic peptide; OMT, optimal medical therapy; and PCI, percutaneous coronary intervention. Other biomarkers showed inconsistent prognostic associations. HsTnT and hsCRP maintained significance in OMT and nonobstructive CAD but not in revascularized patients. Notably, copeptin demonstrated the strongest prognostic association in patients with CABG (HR 1.389 [95% CI, 1.130–1.697], P=0.001), exhibiting the highest hazard ratio among all biomarkers in this treatment subgroup. Time‐dependent ROC analysis across multiple follow‐up time points (1, 3, 5, and 10 years) was performed separately for subgroups with PCI (n=344, 108 deaths) and CABG (n=267, 99 deaths) using Youden's index optimization. Optimal NT‐proBNP cutoffs varied across time points: PCI subgroup ranged from 119.4 pg/mL (5‐year) to 233.1 pg/mL (1 year); the subgroup with CABG ranged from 142.8 pg/mL (5 year) to 175.9 pg/mL (1 year). Based on this analysis, we propose 150 pg/mL as a threshold, approximating the median across all time points and both revascularization strategies (median 153.4 pg/mL; range 119.4–233.1 pg/mL) while aligning with long‐term mortality prediction cutoffs most relevant for sustained treatment effects. Among all biomarkers evaluated, NT‐proBNP demonstrated the most consistent discriminative performance across intervention types, with excellent prognostic capacity at 5 years post revascularization. Patients with NT‐proBNP <150 pg/mL (n=1219) achieved 10‐year survival of 90.6% versus 68.8% in those ≥150 pg/mL (n=1026; fully adjusted HR, 1.69 [95% CI, 1.41–2.03], P<0.001) in fully adjusted model including coronary disease severity and ejection fraction (Figure 3). Bootstrap validation (250 iterations) confirmed threshold robustness: optimal cutoff median 128.8 pg/mL (95% CI, 119.2–199.0), chosen 150 pg/mL at 74th percentile, negligible AUC optimism (−0.000). Cox proportional hazard survival curves stratified by NT‐proBNP cutoff 150 pg/mL Cox proportional hazards survival analysis demonstrating prognostic divergence based on NT‐proBNP stratification at 150 pg/mL cutoff. Green curve: NT‐proBNP <150 pg/mL (n=1219 patients); Red curve: NT‐proBNP ≥150 pg/mL (n=1026 patients). Survival probabilities: 5‐y (96.9% vs 87.4%), 10‐y (90.6% vs 68.8%). NT‐proBNP indicates N‐terminal pro‐B‐type natriuretic peptide. Kaplan–Meier 10‐year survival estimates by NT‐proBNP level and treatment strategy revealed that among patients with NT‐proBNP <150 pg/mL, the OMT group achieved 91.2% survival versus 90.8% in revascularization group (absolute risk difference, +0.4% [95% CI, −3.2% to +4.0%]). Among patients with NT‐proBNP ≥150 pg/mL, OMT group achieved 68.1% survival versus 76.4% in revascularization group (absolute risk difference, +8.3% [95% CI, +3.1% to +13.5%]). For the subsequent analysis, propensity score matching yielded 572 matched pairs (n=1144 total) comparing revascularization versus OMT. Excellent covariate balance was achieved (mean absolute SMD 0.069, range 0.001–0.142; Table S3, Figure S2), with all covariates showing |SMD| <0.15. Using a 4‐group framework with no revascularization + NT‐proBNP <150 pg/mL as reference with the lowest risk, we observed distinct mortality patterns (Figure 4). In patients with NT‐proBNP <150 pg/mL revascularization associated with mortality comparable to the low‐risk reference subgroup (HR, 0.98 [95% CI, 0.67–1.43], P=0.910). Patients with NT‐proBNP ≥150 pg/mL demonstrated markedly elevated baseline risk (HR, 5.75 [95% CI, 4.10–8.00], P<0.001 without revascularization) with 40% relative mortality reduction following revascularization (HR, 3.43 [95% CI, 2.70–4.40], P<0.001), though substantial residual risk persisted. The interaction term confirmed differential treatment effect (HR, 0.609 [95% CI, 0.387–0.959], P=0.032). Absolute mortality rates over 12.6 years were 18.1% (reference), 17.6% (low NT‐proBNP + revascularization), 52.1% (high NT‐proBNP alone), and 38.4% (high NT‐proBNP + revascularization). Restricted cubic spline modeling demonstrated a nonlinear trend between continuous NT‐proBNP and mortality risk (P for nonlinearity=0.057), with the dose–response curve suggesting 3 distinct prognostic phenotypes that broadly correspond to the dichotomous threshold identified at 150 pg/mL (Figure S4). Propensity score matched interaction analysis—revascularization × NT‐proBNP Forest plot demonstrating significant NT‐proBNP × revascularization interaction (interaction term HR, 0.609 [95% CI, 0.387–0.959], P=0.032) from propensity score matched analysis (19 confounders, 572 matched pairs, mean |SMD|=0.069). The reference group comprises patients without revascularization and NT‐proBNP <150 pg/mL. Hazard ratios for the 3 remaining groups are expressed relative to this reference. HDL indicates high‐density lipoprotein; HR, hazard ratio; LDL, low‐density lipoprotein; MI, myocardial infarction; NT‐proBNP, N‐terminal pro‐B‐type natriuretic peptide; PSM, propensity score matching; Revasc, revascularization (ie, percutaneous coronary intervention or coronary artery bypass grafting); and SMD, standardized mean difference. DISCUSSION This comprehensive biomarker evaluation in patients with CCS yields 3 principal findings: (1) among 5 biomarkers assessed, only hsTnT provides clinically meaningful diagnostic capacity for obstructive CAD (AUC, 0.669), with performance comparable to RF‐CL assessment (0.663) and demonstrating inverse relationship between baseline RF‐CL and incremental discriminative value; (2) NT‐proBNP emerges as superior prognostic biomarker with universal discrimination across all treatment categories; and (3) NT‐proBNP 150 pg/mL threshold identifies distinct prognostic phenotypes with differential survival outcomes by coronary revascularization status. These findings extend our prior work 17 through direct comparison with the currently recommended PTP model, quantification of the incremental diagnostic value of hsTnT across RF‐CL categories, and identification of a NT‐proBNP threshold associated with differential survival patterns by revascularization status—addressing critical needs for biomarker‐guided treatment selection in contemporary CCS management. HsTnT's diagnostic capacity reflects its direct relationship to myocardial injury from flow‐limiting stenoses through repetitive ischemia–reperfusion (18, 19), achieving performance comparable to ESC 2024 PTP. Clinical likelihood‐stratified analysis with the optimized 7.0 ng/L cutoff demonstrated marked heterogeneity: patients with very low pretest likelihood (ΔAUC, 10.4%) showed greatest benefit, moderate benefit in low likelihood patients, whereas intermediate/high likelihood patients (ΔAUC=5.0%) showed the smallest benefit. This inverse gradient pattern likely reflects information content redundancy: in patients with multiple traditional risk factors and typical symptoms (high RF‐CL), biomarker measurement adds limited new information, whereas in patients with few risk factors and atypical symptoms (low RF‐CL), biomarker elevation meaningfully shifts probability estimates after adjustment for RF‐CL variables. Crucially, safety validation demonstrated that false negative patients showed no excess mortality after adjustment for RF‐CL variables, supporting the clinical applicability of this biomarker‐guided approach. However, the substantial unadjusted mortality risk attributable to baseline cardiovascular risk factors underscores that safe reclassification does not obviate aggressive risk factor optimization or ongoing surveillance. A very low likelihood classification should, even after hsTnT implementation, guide deferred rather than omitted testing, with reassessment frequency individualized to baseline risk burden. 1 , 2 , 20 , 21 These findings align with SCOT‐HEART (Scottish Computed Tomography of the Heart) trial (n=1430) demonstrating hsTnI (high‐sensitivity troponin I) improves pretest probability estimation (OR, 1.35 [95% CI, 1.25–1.46] per 2‐fold increment) 22 and PROMISE (Prospective Multicenter Imaging Study for Evaluation of Chest Pain; n=1838) showing hsTnI independently predicts obstructive CAD presence and severity (OR, 1.25 per IQR increase), 23 though neither specifically assessed incremental discriminative value beyond current ESC guideline‐recommended diagnostic approaches or quantified category‐specific benefits. NT‐proBNP emerged as the strongest mortality predictor across all patient categories, substantially outperforming hsTnT. This differential utility—hsTnT for diagnosis, NT‐proBNP for prognosis—reflects distinct pathophysiology: hsTnT indicates injury relevant to obstruction detection, whereas NT‐proBNP reflects ventricular stress, remodeling, and neurohumoral activation directly predicting mortality. 3 , 4 , 24 NT‐proBNP was the only biomarker maintaining prognostic discrimination across all therapeutic strategies. These findings contrast with the MICA (Myocardial Injury in Patients Referred for Coronary Angiography) study demonstrating hsTnI maintains independent prognostic value in patients with stable angina. 25 However, our results align with previous reports demonstrating robust major adverse cardiac events prediction by NT‐proBNP in patients with CCS with normal ejection fraction, 26 while providing novel contributions: (1) treatment‐interaction analysis demonstrating differential survival associations by NT‐proBNP status and (2) 12.6‐year median follow‐up far exceeding typical 2‐ to 5‐year investigations. Notably, biomarker‐specific patterns emerged, with copeptin showing particularly strong performance in patients with CABG. As a stable surrogate of arginine vasopressin release reflecting chronic neuroendocrine activation, copeptin's superior discriminative capacity in this subgroup may indicate more advanced coronary disease burden with greater myocardial dysfunction. 27 The proposed 150 pg/mL NT‐proBNP cutoff is supported by continuous spline analysis showing good discrimination at this level (Figure S4). This post hoc derived threshold is clinically meaningful, situated between ESC guideline thresholds for ruling out heart failure (125 pg/mL) and diagnosing acute heart failure (>300 pg/mL), 7 capturing early ventricular dysfunction and neurohumoral activation that precede overt heart failure. 28 The significant interaction between NT‐proBNP levels and revascularization survival demonstrated differential prognostic patterns: patients with baseline NT‐proBNP <150 pg/mL showed comparably low mortality irrespective of revascularization. In contrast, NT‐proBNP ≥150 pg/mL identified patients with substantially elevated baseline mortality (HR, 5.75; P<0.001) and significant mortality reduction with revascularization (HR, 3.43; P<0.001), albeit with persistent residual risk. Importantly, revascularization provides benefits beyond mortality including symptom relief and improved quality of life, which may be particularly relevant for patients with elevated NT‐proBNP. Clinically, this threshold may inform risk stratification: patients with NT‐proBNP ≥150 pg/mL appear to represent a higher‐risk phenotype requiring intensified medical therapy, closer surveillance, and potentially novel therapeutic approaches. 9 , 20 , 21 This biomarker‐guided approach parallels NT‐proBNP‐guided heart failure therapy, 7 though prospective validation is needed before clinical implementation. Several limitations warrant consideration. The single‐center observational design with single time point measurements precludes definitive causal inference and lacks external validation in independent cohorts, limiting generalizability. Most importantly, treatment allocation was clinician directed. Despite rigorous propensity score matching achieving good balance (mean |SMD| 0.069) with 19 confounders, unmeasured confounding cannot be excluded (Table S7). Our analysis focused on mortality without assessing symptom relief, quality of life, or heart failure hospitalization—important revascularization benefits particularly relevant for high‐risk patients. We used invasive angiography with anatomical stenosis ≥50%—the established standard 29 —which demonstrates substantial discordance to functional relevance, 30 potentially explaining moderate biomarker diagnostic performance. Fractional flow reserve and nonhyperemic pressure indices were not systematically performed. Exercise testing was available in 79.9% but did not add incremental diagnostic value over hsTnT (P=0.98) and showed no association with anatomic CAD presence (P=0.53). However, comprehensive sensitivity analyses using stricter anatomic thresholds (≥75%/left main coronary artery≥50%, differing by only 5.4% from ≥50% cohort) demonstrated stable diagnostic and prognostic performance (Table S8), supporting robustness of findings across stenosis severity definitions. Guideline‐recommended noninvasive imaging now preferred after RF‐CL assessment 2 was unavailable, precluding comparison with currently recommended modalities. Outcome assessment was limited by unavailable cause‐specific mortality and missing implantable cardioverter‐defibrillator implantation data. Relatively small treatment subgroup samples (particularly CABG, n=267) may limit statistical power. The 150 pg/mL threshold was data derived using optimization methods, risking overfitting. Although internal bootstrap validation supports model stability, generalizability to other populations, assays, or time periods remains unproven. The threshold should be considered exploratory pending independent replication. Finally, although 12.6‐year follow‐up is substantial, contemporary therapeutic advances may alter the prognostic landscape. Conclusions In conclusion, this comprehensive biomarker analysis demonstrates differential diagnostic versus prognostic utility in CCS. HsTnT provides risk factor‐weighted, clinical likelihood category‐dependent diagnostic value following an inverse gradient pattern (greatest benefit in very low‐risk patients), and NT‐proBNP emerged as the universal prognostic biomarker. The 150 pg/mL NT‐proBNP threshold identified 2 distinct phenotypes: patients below showed comparably low mortality regardless of revascularization (HR, 0.98; P=0.910), whereas patients above demonstrated elevated baseline mortality (HR, 5.75; P<0.001) with 40% relative reduction following revascularisation (HR, 3.43; P<0.001) but persistent 3.43‐fold residual risk. Elevated NT‐proBNP thus identifies a persistently high‐risk phenotype requiring comprehensive management beyond anatomic intervention, including guideline‐directed medical therapy, intensive risk factor modification, and close surveillance. Given the observational design with inherent confounding, these hypothesis‐generating findings warrant prospective multicenter randomized controlled trials validation to determine whether biomarker‐guided strategies improve outcomes and cost‐effectiveness in contemporary CCS management. Sources of Funding This work was supported by the Leipzig Research Centre for Civilization Diseases, University of Leipzig, which is funded by means of the European Union, by the European Regional Development Fund, and by means of the Free State of Saxony within the framework of the excellence initiative. Disclosures Conflict of interest: none declared. Supporting information Tables S1–S8 Figures S1–S4 Acknowledgments The authors acknowledge all participants of the LIFE‐Heart study for spending their time and donating blood, as well as for their continued participation in our follow‐up endeavors. We thank Annegret Unger and Kay Olischer for running the study ambulance, and Ronny Dathe for database development. Author contributions: Andrej Teren and Jeffrey Netto conceived and designed the study. Andrej Teren, Joachim Thiery, and Holger Thiele contributed to data acquisition. Andrej Teren, Holger Kirsten, and MS performed statistical analysis. Andrej Teren drafted the article. All authors (Andrej Teren, Jeffrey Netto, Joachim Thiery, Holger Thiele, Christoph Stellbrink, Dennis Lawin, Thorsten Lawrenz, Anselm Artur Derda, Sylvia Henger, Holger Kirsten, Markus Scholz, Thorsten Kaiser) contributed to data interpretation, critically revised the article for important intellectual content and approved the final version for submission. Andrej Teren takes responsibility for the integrity of the data and the accuracy of the data analysis. This article was sent to Krishnaraj S Rathod, MBBS, BMedSci, MRCP, Associate Editor, for review by expert referees, editorial decision, and final disposition. Supplemental Material is available at https://www.ahajournals.org/doi/suppl/10.1161/JAHA.125.048737 For Sources of Funding and Disclosures, see page 12. REFERENCES Associated Data Supplementary Materials Tables S1–S8 Figures S1–S4 Data Availability Statement The data that support the findings of this study are available from the Leipzig Research Centre for Civilization Diseases (LIFE), University of Leipzig. Restrictions apply to the availability of these data, which were used under license for this study. Data are available from the corresponding author upon reasonable request and with permission of the LIFE‐Heart study steering committee, subject to institutional data sharing agreements and ethical approval.