Early On-Treatment Nutritional Change and Baseline Inflammation Stratify Outcomes in Patients Receiving Immune Checkpoint Inhibitors

Objectives: In patients receiving immune checkpoint inhibitors (ICIs), baseline nutritional status is frequently recorded, but what happens during treatment may be at least as informative. We examined whether the 3-month change in Mini Nutritional Assessment (ΔMNA)–Short Form (SF) is associated with survival, and how this relates to baseline systemic inflammation measured by modified Glasgow Prognostic Score (mGPS). Methods: We retrospectively screened institutional records and included 54 adults with advanced/metastatic solid tumors treated with ICIs who had MNA-SF documented at treatment start and again at approximately 3 months. ΔMNA was defined as MNA-SF (~3 months) minus baseline MNA-SF (higher values indicate improvement) and was categorized for Kaplan–Meier analyses as worsened (≤ −1), stable (= 0), or improved (≥ +1). Baseline mGPS was derived using CRP >5 mg/L and albumin <35 g/L. Overall survival (OS) and progression-free survival (PFS) were measured from ICI initiation. Cox regression assessed ΔMNA as a continuous variable, with prespecified adjustment for baseline mGPS, age, and sex.Results: Median follow-up was 317 days (IQR 170.5–457.3). Death occurred in 16/54 patients and a PFS event in 27/54. Median OS was 706 days (1-year OS 71.3%; 2-year OS 49.4%), and median PFS was 337 days (1-year PFS 48.5%; 2-year PFS 15.1%). Nutritional categories shifted toward better status at 3 months (baseline normal/risk/malnutrition 29/23/2 vs 35/17/2 at follow-up). In univariable Cox models, higher ΔMNA was associated with longer survival (PFS: HR 0.72, 95% CI 0.56–0.94; P=0.016; OS: HR 0.56, 95% CI 0.38–0.82; P=0.003). In prespecified multivariable models, baseline mGPS was independently associated with OS (HR 1.86, 95% CI 1.00–3.45; P=0.048). After adjustment, the association between ΔMNA and outcomes remained in the protective direction but did not meet conventional statistical significance (OS: HR 0.67, 95% CI 0.42–1.06; P=0.086; PFS: HR 0.72, 95% CI 0.56–1.05; P=0.094). In an exploratory lung cancer subgroup (n=36), ΔMNA was associated with improved PFS (HR 0.62; P=0.008) and OS (HR 0.59; P=0.018). These subgroup findings should be interpreted as hypothesis-generating.Conclusions: Baseline mGPS captured inflammatory risk at ICI start, while ΔMNA captured what happened to nutritional status over the first 3 months. In this cohort, nutritional improvement showed a consistent protective direction for OS and PFS even after accounting for baseline inflammation, and mGPS retained an independent association with OS. These findings support prospective studies that reassess nutrition during immunotherapy rather than relying on a single baseline measurement.