A nomogram integrating CTPA parameters and serum biomarkers to predict 30-day mortality and refine risk stratification in non-high-risk acute pulmonary embolism

Abstract

Objective To explore the predictive value of CT pulmonary angiography (CTPA) imaging parameters combined with serum biomarkers for 30-d all-cause mortality in non-high-risk acute pulmonary embolism (APE), and to compare parameters between simplified Pulmonary Embolism Severity Index (sPESI)-based risk strata. Methods This retrospective study enrolled 250 first-episode non-high-risk APE patients (50 deaths, 200 survivors). CTPA-derived pulmonary artery obstruction index (PAOI), right/left ventricular diameter ratio (RV/LV), main pulmonary/ascending aorta ratio, and serum biomarkers high-sensitivity cardiac troponin I (hs-cTnI), brain natriuretic peptide (BNP), D-dimer, and blood lactate (BLa) were collected. Independent risk factors were identified using Firth penalized likelihood logistic regression to construct a nomogram, which was internally validated. The incremental predictive value over sPESI was quantified. Cox regression was used to assess prognostic values and interactions. Spearman correlation analysis was conducted for relationships between imaging and biomarkers. Patients were stratified by sPESI into intermediate-risk (≥ 1 point) and low-risk (0 points) groups for comparative and survival analyses. Results Death vs. survival groups differed significantly in syncope, age >80 years, SBP <100 mmHg, PAOI, RV/LV, hs-cTnI, BNP, and BLa (P < 0.05). Independent predictors were syncope (OR = 5.846, 95% CI: 1.552–22.030), SBP <100 mmHg (OR = 6.118, 95% CI: 1.899–19.704), RV/LV (OR = 1.285, 95% CI: 1.028–1.607), BNP (OR = 1.002, 95% CI: 1.000–1.004), and BLa (OR = 2.162, 95% CI: 1.372–3.406). The nomogram showed good discrimination (AUC = 0.878), calibration, and clinical net benefit, significantly outperforming sPESI (AUC = 0.713, P < 0.001). Cox regression confirmed several predictors but sPESI was not an independent factor (HR = 0.478, P = 0.178). PAOI and RV/LV correlated positively with hs-cTnI, BNP, and BLa (r = 0.345–0.427, P < 0.001). Intermediate-risk patients had higher mortality, PAOI, RV/LV, BNP, and BLa than low-risk patients (P < 0.05). Conclusion The nomogram incorporating syncope, SBP <100 mmHg, RV/LV, BNP, and BLa, is an effective tool for predicting 30-day mortality in non-high-risk APE patients, demonstrating superior performance over sPESI. Combining CTPA imaging and serum biomarkers offers independent and complementary prognostic information, enhancing risk stratification.