Beyond ECOG: Assessment of physical function enhances risk stratification prior to CAR-T cell therapy

Abstract

Abstract Background The Eastern Cooperative Oncology Group performance status (ECOG PS) represents the standard for assessing health status prior to CAR-T cell treatment, but is limited by interobserver variability and low granularity. To address this limitation, we implemented a standardized and objective assessment based on two bedside physical fitness tests that evaluate functional capacity in patients with hematologic malignancies undergoing CAR T-cell therapy. This pilot study aims to improve patient stratification by analyzing associations of physical performance with CAR-T-related toxicities and clinical outcomes. Methods We prospectively evaluated 50 patients undergoing standard-of-care CD19 and BCMA CAR-T treatment for hematologic malignancies (including LBCL, MCL, B-ALL, and MM). In addition to ECOG PS, Barthel Index, and HCT-CI scoring, all patients underwent two bedside physical performance tests: i) 10-meter walking test (gait speed), and ii) 30-second sit-to-stand test. Patients completing both tests within reference values were classified as fit; those failing at least one were classified as unfit. Frailty was assessed using a modified Short Physical Performance Battery (SPPB). Clinical endpoints included nutritional indices (mGPS, PNI), treatment-related toxicities (CRS, ICANS, ICAHT), hospitalization duration, as well as progression-free survival (PFS) and overall survival (OS).Body composition analysis (BCA) was performed on pre-therapeutic PET/CT images using the Body-and-Organ-Analysis segmentation tool by the Ship-AI group, calculating sarcopenia index (muscle volume/bone volume) and fat index (total adipose tissue volume/bone volume). Kaplan-Meier estimates were used to examine survival outcomes. Results Of the 50 prospectively evaluated patients, 23 (46%) were classified as unfit and 27 (54%) as fit. The relative ECOG PS distribution showed no statistically significant difference between groups (p=0.27), but a trend toward higher ECOG PS in unfit: among unfit patients, 21.7% had ECOG 0, 60.9% ECOG 1, and 17.4% ECOG ≥2; in fit patients, the distribution was 22.2%, 74.1%, and 3.7%. This underscores the limited discriminatory power of ECOG PS alone in capturing the full spectrum of functional status. Fitness correlated significantly with the validated 5-item FRAIL scale (score 0-5; ρ = –0.46, p=0.0009), supporting its role in capturing functional vulnerability. SPPB (ρ = 0.36, p = 0.01) and Barthel Index (ρ = 0.37, p = 0.009) also correlated moderately with fitness, while HCT-CI showed no relevant association with functional metrics (p=0.48). Of interest, unfit patients were more likely to have lost weight (at least 5% reduction in BMI) in the 3 months prior to treatment. Unfit patients had higher pre-lymphodepletion Ferritin levels (605 vs. 383 ng/ml, p=0.03), and a trend toward elevated NT-proBNP (783 vs. 350 pg/mL, p=0.13), reflecting subclinical systemic inflammation and potential cardiac stress. They more frequently had longer hospital stays (median 18 vs. 11 days) and a trend toward increased grade ≥2 CRS (39.1% vs. 25.9%, p=0.5) and high-risk CAR-HEMATOTOX scores (26.1% vs 11.1%, p=0.06). On the other hand, the distribution of ICAHT and ICANS grades was comparable between both groups. Notably, fit patients achieved significantly higher CR rates (52% vs. 21.7%, p=0.04). BCA revealed a higher sarcopenia index in fit (mean ± SD: 4.03 ± 0.86) compared to unfit individuals (3.63 ± 0.56), reflecting greater muscle mass. On the other hand, fat index was comparable between groups (9.87 ± 7.27 vs. 9.87 ± 4.44). Univariable Cox models did not reveal statistically significant association between BCA and OS. Unfit patients trended toward lower PNI and higher mGPS, indicating poorer nutritional and inflammatory profiles. Conclusions In this prospective single-center pilot study, we established an objective physical performance testing and delineate its association with radiomic BCA and frailty assessment in CAR-T recipients. While ECOG PS remains a widely established clinical tool, the objective assessment provided additional granularity in capturing patient's functional capacity and demonstrated an improved identification of patients at risk for inferior outcomes with CAR-T therapy. These findings highlight the prognostic value of standardized functional metrics and support their integration into clinical trials and routine care to potentially enable more personalized treatment and rehabilitation measures.