Abstract A064: Dynamics in automatic CT based body composition and blood biomarkers in predicting mortality on immune therapy treated solid malignancy patients
Bibliographic record
Abstract
Abstract CT-derived body composition (BC) metrics are associated with mortality in cancer patients at baseline and during treatment. However, the relationship between CT-derived BC metrics and blood biomarkers, and their independent prognostic value, has been scarcely evaluated. This study assessed the correlation between BC metrics and blood test results, and whether longitudinal changes independently predict mortality in patients receiving immunotherapy for solid tumors. We included patients treated with immunotherapy for non-small cell lung cancer (NSCLC), melanoma, or renal cell carcinoma (RCC) between 2017 and 2024, who had baseline and follow-up CT scans. Patients could be recorded more than once as distinct treatment events if there was a treatment break of ≥180 days, or a new drug combination. BC was analyzed using fully automated AI software (CompoCT@), measuring skeletal muscle (SM), including healthy muscle (HM) and steatotic muscle (StM); subcutaneous layer (SCL), comprising subcutaneous fat (SCF) and subcutaneous edema (SCE); and visceral fat (VF) at the L3 vertebral level. BC indices (e.g. SMI) were calculated by dividing the corresponding area (cm2) by the patient’s height squared (m2). Laboratory data included albumin, LDH, CRP, hemoglobin (Hb), neutrophil-to-lymphocyte ratio (NLR) and white blood cell count (WBC). The cohort included 418 events from 392 patients (mean (median) age 66.2 (67) years; 66.5% male, 71.6% NSCLC, 14.2% melanoma, and 14.2% RCC). No significant correlations were observed between blood tests and CT BC metrics. Baseline values of albumin, Hb and NLR significantly correlated with mortality while baseline CT metrics did not. However, longitudinal percentage decrease in SMI (1/HR=25), HMI (1/HR=2.5) and SCFI (1/HR≈8.3) and increases in SCEI (HR=1.69), were all significantly associated with mortality (p<0.001). Changes in LDH, WBC, NLR, and Hb also correlated with mortality (HR=1.7;1.61;1.2;0.28 respectively), whereas changes in albumin levels did not. Three distinct multivariate models were constructed to evaluate the prognostic performance of different variable sets. The first model combined CT-derived BC metrics with blood biomarkers and demonstrated the highest predictive accuracy (concordance index [CI] = 0.78). The second model included only BC metrics (CI = 0.72), while the third relied solely on blood biomarkers (CI = 0.69). In the combined model, higher mortality was significantly associated with NSCLC diagnosis, longitudinal changes in CT-derived BC metrics, including %∆StMI, %∆HMI, and %∆SCFI, as well as baseline albumin values and changes in LDH levels (p < 0.05 for all). In patients with solid tumors receiving immunotherapy, longitudinal CT-based changes in muscle and fat were more predictive of mortality than traditional sarcopenia-related blood biomarkers. Opportunistic use of CT data, extracted via fully automated AI algorithms, may enhance clinical management decisions, by offering additive value to conventional blood tests related to muscle wasting and systemic inflammation. Citation Format: Shlomit Tamir, Hilla Vardi Behar, Ronen Tal, Ruth Tal Hasper, Mor Armoni, Hadar Pratt Aloni, Rotem Or Ad, Hillary Voet, Eli Atar, Ahuva Grubstein, Salomon Stemmer, Gal Markel. Dynamics in automatic CT based body composition and blood biomarkers in predicting mortality on immune therapy treated solid malignancy patients [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Artificial Intelligence and Machine Learning; 2025 Jul 10-12; Montreal, QC, Canada. Philadelphia (PA): AACR; Clin Cancer Res 2025;31(13_Suppl):Abstract nr A064.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.002 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.000 |
| Open science | 0.000 | 0.001 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".