P539 - EVALUATION OF NUTRITIONAL ASSESSMENT TOOLS IN HEMATOLOGIC MALIGNANCIES AND PRELIMINARY DEVELOPMENT OF THE HMB-SGA

P539

EVALUATION OF NUTRITIONAL ASSESSMENT TOOLS IN HEMATOLOGIC MALIGNANCIES AND PRELIMINARY DEVELOPMENT OF THE HMB-SGA

C. Xinying^1,*

¹Beijing Shijitan Hospital, Beijing Shijitan Hospital, Capital Medical University, Beijing, China

Rationale: Hematologic malignancies pose a critical threat to global health, with their pathological progression intrinsically linked to metabolic dysregulation and nutrient depletion. Malnutrition exacerbates trajectories of adverse clinical outcomes while substantially diminishing survival quality. 

Methods: This prospective cohort study analyzed nutritional assessment data from 1,063 hematologic malignancy patients within the INSCOC registry. Methodological triangulation encompassed eight assessment systems. Predictive validity was interrogated through Kaplan-Meier survival stratification, Cox proportional hazards modeling, Time-dependent ROC Curve,Area Under the Curve (AUC), and concordance index (C-index) evaluation. 

Results: Methodological interrogation of eight nutritional assessment tools commenced with Cox proportional hazards regression (P<0.05 threshold), revealing showed limited prognostic discrimination across all systems through AUC analysis. Modified PG-SGA (mPG-SGA), and PG-SGA demonstrated marginal superiority, with AUC values of 0.561 and 0.550 respectively. Concordance index validation corroborated these findings, yielding C-statistics of 0.551 and 0.546 for the aforementioned instruments. The novel instrument achieved superior discriminative capacity (AUC=0.616; C-index=0.605), outperforming conventional metrics across all sensitivity analyses, thereby establishing enhanced prognostic stratification in hematologic malignancies.

Conclusion: Multidimensional validation revealed fundamentally inadequate predictive validity of eight conventional nutritional tools in hematologic malignancies (AUC<0.600, C-index<0.550), challenging their clinical utility. The HMB-SGA model  demonstrated enhanced prognostic capacity (AUC=0.616, C-index=0.605), laying methodological groundwork for developing malignancy-specific nutritional assessment paradigms.

Disclosure of Interest: None declared