Prognostic Significance of CT-Derived Pectoralis Muscle Phenotypes in Patients Undergoing Venoarterial Extracorporeal Membrane Oxygenation Support Authors: Jung-Chi Hsu, Yu-Cheng Huang, Chen-Hsu Pai, Ling-Yi Wei, Chih-Hsien Wang, Nai-Hsin Chi, Shu-Chien Huang, Jeng-Wei Chen, Heng-Wen Chou, Ron-Bin Hsu, Nai-Kuan Chou, Hsi-Yu Yu, Yih-Sharng Chen Published: 2026 Venoarterial extracorporeal membrane oxygenation (VA-ECMO) is a critical rescue therapy for refractory cardiogenic shock and extracorporeal cardiopulmonary resuscitation (ECPR).1,2 Although our previous study demonstrated a U-shaped association between body mass index (BMI) and ECPR mortality,3 BMI fails to differentiate lean muscle mass from adiposity. Computed tomography (CT) enables objective assessment of skeletal muscle quantity and quality, which have been associated with adverse outcomes in critical illness,4 yet evidence in extracorporeal membrane oxygenation (ECMO) remains limited and has largely relied on abdominal psoas measurements.5 The pectoralis major is routinely visible on chest CT and may better reflect overall muscle status and respiratory function,6 but its prognostic value in VA-ECMO remains unclear. We therefore evaluated CT-defined pectoralis muscle phenotypes and their association with in-hospital mortality in the VA-ECMO cohort and ECPR subgroup. Methods We included consecutive adults (age ≥18 years) who received VA-ECMO at National Taiwan University Hospital from 2018 to 2023 and underwent chest CT within 7 days before or after cannulation. Patients with incomplete CT images or missing clinical data were excluded. VA-ECMO was initiated per institutional protocol for refractory cardiogenic shock or cardiac arrest. ECPR was defined as VA-ECMO initiation during ongoing cardiopulmonary resuscitation in patients with witnessed arrest and no return of spontaneous circulation. The primary outcome was in-hospital mortality. At the fourth thoracic vertebral level, bilateral pectoralis major muscles were assessed using a semi-automated segmentation approach on the picture archiving and communication system, with manual delineation followed by automatic calculation of cross-sectional area and mean attenuation within −29 to +150 HU; all measurements were reviewed by a radiologist. Interobserver reproducibility was good (intraclass correlation coefficient 0.825 for cross-sectional area and 0.851 for mean attenuation). Patients were classified by cohort-specific medians as preserved muscle, isolated sarcopenia, isolated myosteatosis, or combined sarcopenia-myosteatosis. Outcomes were assessed using Kaplan-Meier curves and multivariable Cox models; the overall multivariable model adjusted for age, sex, hypertension, chronic obstructive pulmonary disease, chronic kidney disease, congestive heart failure, and intra-aortic balloon pump use, and the ECPR model adjusted for age, sex, BMI, hypertension, diabetes mellitus, and chronic kidney disease. Proportional hazards were evaluated using Schoenfeld residuals, with no significant violations observed for the primary exposure variable. This study was approved by the Research Ethics Committee of National Taiwan University Hospital (NTUH-REC No. 202406133RINB), and informed consent was waived. Results Of 811 adults receiving VA-ECMO, 537 had qualifying CT-derived pectoralis measurements and were included; 274 were excluded because CT measurements were unavailable. Missing data were minimal (BMI: 31 of 537, 5.8%), and in-hospital mortality was similar between included and excluded patients (60.5% [325 of 537] vs 60.6% [166 of 274]). Median follow-up was 13 days (IQR: 4-32) overall and 8 days (IQR: 2-21) in the ECPR subgroup, in which in-hospital mortality was 68.9% (222 of 322). Among ECPR patients with available CT and cannulation dates (n = 321), CT was performed before ECMO in 25 (7.8%), on the day of cannulation in 217 (67.6%), and after ECMO initiation in 79 (24.6%), with corresponding mortality of 88.0% (22 of 25), 67.3% (146 of 217), and 68.4% (54 of 79). Patients were classified as preserved muscle (group 1, n = 193), isolated sarcopenia (group 2, n = 76), isolated myosteatosis (group 3, n = 76), or combined sarcopenia-myosteatosis (group 4, n = 192). Group 4 patients were older, more often female, and had more chronic kidney disease, whereas acute myocardial infarction was more common in group 1. At day 30, Kaplan-Meier cumulative incidence of in-hospital death was 63.9% (95% CI: 54.9%-71.1%), 51.1% (95% CI: 37.2%-62.0%), 58.6% (95% CI: 45.0%-68.9%), and 50.2% (95% CI: 41.9%-57.3%) in groups 1-4 in the overall cohort, and 65.7% (95% CI: 54.3%-74.3%), 89.0% (95% CI: 63.3%-96.7%), 83.8% (95% CI: 63.8%-92.8%), and 67.4% (95% CI: 56.4%-75.6%) in the ECPR subgroup (Figure 1). In the overall cohort, group 1 had higher mortality than group 4 in crude and age- and sex-adjusted analyses (HR: 1.45; 95% CI: 1.12-1.88; P = 0.005; and HR: 1.73; 95% CI: 1.29-2.31; P < 0.001, respectively); this association persisted in a sensitivity multivariable model excluding BMI and ECPR (HR: 1.56; 95% CI: 1.16-2.09; P = 0.003). In the ECPR subgroup, groups 2 and 3 had higher mortality than group 4 in crude analysis (HR: 1.66; 95% CI: 1.09-2.53; P = 0.018; and HR: 1.52; 95% CI: 1.01-2.30; P = 0.044) and after multivariable adjustment (HR: 1.89; 95% CI: 1.21-2.95; P = 0.005; and HR: 1.93; 95% CI: 1.22-3.05; P = 0.005). Differences were most evident in the Survival after Veno-Arterial ECMO (SAVE) Score Class III in the overall cohort and SAVE Class IV in the ECPR subgroup (HR: 2.62; 95% CI: 1.40-4.89; P = 0.003; and HR: 1.90; 95% CI: 1.08-3.36; P = 0.027).Figure 1Unadjusted and Age- and Sex-Adjusted Survival Curves by Pectoralis Muscle Phenotype(A) Unadjusted Kaplan-Meier cumulative incidence curves for in-hospital death in the overall VA-ECMO cohort. (B) Corresponding age- and sex-adjusted Cox-based survival curves for the overall cohort. (C) Unadjusted Kaplan-Meier cumulative incidence curves for the ECPR subgroup. (D) Corresponding age- and sex-adjusted Cox-based survival curves for the ECPR subgroup. Numbers at risk are displayed below each panel at 0, 5, 10, 15, 20, 25, and 30 days. ECPR = extracorporeal cardiopulmonary resuscitation; VA-ECMO = venoarterial extracorporeal membrane oxygenation. Unadjusted and Age- and Sex-Adjusted Survival Curves by Pectoralis Muscle Phenotype (A) Unadjusted Kaplan-Meier cumulative incidence curves for in-hospital death in the overall VA-ECMO cohort. (B) Corresponding age- and sex-adjusted Cox-based survival curves for the overall cohort. (C) Unadjusted Kaplan-Meier cumulative incidence curves for the ECPR subgroup. (D) Corresponding age- and sex-adjusted Cox-based survival curves for the ECPR subgroup. Numbers at risk are displayed below each panel at 0, 5, 10, 15, 20, 25, and 30 days. ECPR = extracorporeal cardiopulmonary resuscitation; VA-ECMO = venoarterial extracorporeal membrane oxygenation. Discussion CT-derived pectoralis muscle phenotypes provided context-specific prognostic information in VA-ECMO beyond conventional body composition indices. In the overall cohort, the preserved muscle phenotype was associated with higher mortality, whereas in ECPR, isolated sarcopenia and isolated myosteatosis were associated with worse outcomes than combined sarcopenia-myosteatosis. This apparently counterintuitive finding should be interpreted cautiously. Greater skeletal muscle mass during profound hypoperfusion may increase anaerobic metabolic burden, and subsequent reperfusion may amplify ischemia-reperfusion injury and systemic inflammatory stress.7,8 In addition, preserved muscle bulk on CT may not reflect metabolic resilience, because adverse metabolic features are not fully captured by gross muscle area alone. CT-derived fatty muscle composition predicts adverse outcomes,9 and inflammatory activation is linked to worse outcomes in cardiogenic shock.10 The higher prevalence of acute myocardial infarction in group 1 suggests a more abrupt and catastrophic clinical presentation. These mechanisms remain hypothesis-generating. Study limitations This retrospective, single-center study may limit generalizability, and selection bias is possible because CT was performed according to clinical indications and patient stability. CT timing also may have influenced attenuation-based phenotyping, particularly after ECMO initiation when fluid resuscitation and edema may affect measurements. No universally accepted pectoralis-specific thresholds at the T4 level exist for acutely ill VA-ECMO/ECPR patients; phenotype classification therefore relied on cohort-derived medians, although a sex-specific sensitivity analysis showed similar directional results. Long-term outcomes and multicenter validation remain needed. Conclusions In VA-ECMO, preserved muscle was paradoxically associated with higher mortality, likely reflecting greater metabolic burden during the hyperacute phase; within ECPR, discordant phenotypes conferred excess risk. Early identification of these high-risk phenotypes may refine patient selection and guide monitoring intensity. Funding Support and Author Disclosures This study was supported in part by National Taiwan University Hospital (115-N0119) and the Ministry of Science and Technology of Taiwan (NSTC 114-2314-B-002-303). The authors have reported that they have no relationships relevant to the contents of this paper to disclose.