Integrative transcriptomic profiling of tumor patients in surgical ICU: identifying prognostic immune signatures

Prognostic assessment of tumor patients in the surgical intensive care unit (SICU) remains challenging due to the complex interplay between systemic inflammation and immune dysfunction. In this study, we performed integrative transcriptomic profiling of peripheral blood mononuclear cells (PBMCs) from 112 SICU patients with solid tumors, aiming to identify immune-related signatures predictive of 30-day survival. High-throughput RNA sequencing quantified 19,832 expressed genes, while computational immune deconvolution using CIBERSORTx estimated relative fractions of 22 immune cell types. Differential expression analysis revealed 432 upregulated and 287 downregulated genes in non-survivors, with notable elevation of CXCL10 (27.5 ± 4.6 TPM vs. 12.3 ± 3.1 TPM in survivors, p = 0.002) and IL6 (19.4 ± 3.9 vs. 9.1 ± 2.7 TPM, p = 0.004), accompanied by reduced CD8A (12.3 ± 3.6 vs. 28.1 ± 4.2 TPM, p = 0.006) and GZMB (7.2 ± 1.9 vs. 15.4 ± 3.0 TPM, p = 0.008). Correspondingly, high-risk patients demonstrated decreased cytotoxic T lymphocyte fractions (12.3% ± 3.6% vs. 27.8% ± 4.2%, p = 2.6 × 10^-4) and NK cells (5.2% ± 1.3% vs. 11.1% ± 2.1%, p = 1.9 × 10^-4) with increased neutrophils (45.2% ± 5.1% vs. 31.7% ± 4.8%, p = 3.2 × 10^-4). A Prognostic Immune Score (PIS), constructed based on 12 key immune-related genes, achieved a concordance index of 0.81, effectively stratifying patients into high- and low-risk groups with 30-day survival rates of 42.5% and 86.1%, respectively (log-rank p < 0.001). Subgroup analyses confirmed that the PIS retained predictive accuracy across tumor types and APACHE II strata, underscoring its robustness. These findings demonstrate that integrative transcriptomic and immune profiling provides a quantitative framework for early risk stratification in SICU tumor patients and identifies actionable biomarkers for potential immunomodulatory interventions.