PP390 - DEFINING INFLAMMATION IN GLIM-BASED CANCER CACHEXIA: NON-LINEAR PROGNOSTIC ROLE OF CRP AND COMPARISON OF CANDIDATE CUTOFFS

Linked sessions

PP390

DEFINING INFLAMMATION IN GLIM-BASED CANCER CACHEXIA: NON-LINEAR PROGNOSTIC ROLE OF CRP AND COMPARISON OF CANDIDATE CUTOFFS

N. Mori1,*, K. Maeda2

1Department of Palliative and Supportive Medicine, Aichi Medical University, 2Nutrition Therapy Support Center, Aichi Medical University Hospital, Nagakute, Japan

 

Rationale: Systemic inflammation plays a central role in cancer cachexia, and C-reactive protein (CRP) is widely used. However, CRP is often treated as a binary variable using predefined cutoffs. We examined its prognostic impact without assuming linearity and evaluated candidate cutoffs within a GLIM-based framework.

Methods: This retrospective study included 254 patients with gastrointestinal cancers (191 events) from a previously established cohort. Cachexia was defined as ≥1 GLIM phenotypic criterion plus elevated CRP. CRP was analyzed as both a continuous variable and using predefined cutoffs (0.5–5.0 mg/dL). Cox models were adjusted for age, sex, performance status, edema, tumor site, and GLIM phenotypic status. Non-linearity was assessed using restricted cubic splines. Model performance was evaluated by AIC and Harrell’s C-index.

Results: CRP showed a non-linear association with survival in unadjusted analyses, with risk increasing gradually. Model performance varied by cutoff. A cutoff of 0.5 mg/dL showed the best fit (lowest AIC: 1732), whereas higher cutoffs showed slightly better discrimination (highest C-index: 0.731), although differences were small. No single cutoff consistently performed best.

Conclusion: CRP showed a non-linear relationship with prognosis, and its effect was not adequately captured by a single cutoff. Within a GLIM-based framework, CRP may be better interpreted as a continuous marker of inflammation rather than a dichotomous definition.

Disclosure of Interest: None declared