Neither the pan-immune-inflammation value nor the PILE score improved on APACHE-II or SOFA in a single-center cohort of critically ill older patients with acute myeloid leukemia

Background Older patients with acute myeloid leukemia who require intensive care admission have high mortality, and decisions regarding treatment escalation lack objective guidance. Inflammation-based indices derived from routine blood counts are widely reported in ambulatory oncology, but whether they hold any value once a patient becomes critically ill remains unestablished. We investigated whether the pan-immune-inflammation value (PIV) and the PILE score predict intensive care mortality and whether they add anything to the severity scores already in use. Methods In 75 patients aged 65 years or older admitted to a tertiary intensive care unit (ICU), both indices were computed at diagnosis and at ICU admission and compared with APACHE-II and SOFA scores. The primary endpoint was ICU mortality; overall survival was analyzed from diagnosis with delayed entry at ICU admission, and incremental value was assessed in nested logistic regression models. Results ICU mortality was 72.0%. APACHE-II and SOFA discriminated ICU mortality (area under the curve 0.778, 95% CI 0.646–0.892 and 0.758, 0.640–0.865), each significantly better than either PIV score on pairwise testing. Neither PIV score discriminated, and only the ICU-admission PILE score reached nominal significance (0.647; p = 0.037). Neither index added an increment after correction for multiple comparisons, and no other count-based index performed better. Measured from diagnosis, a higher baseline PILE score was associated with shorter overall survival after adjustment for ELN risk, comorbidity, treatment, and disease status (hazard ratio 1.95, 95% CI 1.33–2.85), although this was not specific to PILE and did not persist once survival was calculated from ICU admission. With 54 deaths, the smallest area under the curve separable from 0.5 with 80% power was 0.681. Conclusion An index calculated as the product of cellular counts is limited in its ability to stratify cytopenic patients: in nine patients, a neutrophil or monocyte count of zero drove PIV to zero, so the score classified the most complete marrow failure as the lowest inflammatory risk. Neither index improved upon existing ICU severity scores. Any prognostic role they hold more likely lies at diagnosis, while treatment intensity and the ceiling of care remain open questions.