Diagnostic Consistency of GLIM Criteria and PG-SGA for Malnutrition in Patients with Pancreatic Cancer

Feifei WangCong WangLiwei ZhangXiuli Li^*

• The First hospital of Hebei Medical University, Shijiazhuang, China

Objective: This study aimed to evaluate the diagnostic consistency between the Global Leadership Initiative on Malnutrition (GLIM) criteria and the Patient-Generated Subjective Global Assessment (PG-SGA) for identifying malnutrition in patients with pancreatic malignant tumors. Methods: A total of 108 pancreatic cancer patients from our hospital with a Nutritional Risk Screening (NRS) 2002 score ≥ 3 were enrolled. Demographic and clinical data were collected. Nutritional risk was assessed using NRS 2002, while malnutrition was evaluated by PG-SGA and GLIM criteria. The diagnostic consistency between GLIM and PG-SGA was analyzed using Cohen's kappa (κ) and prevalence-adjusted bias-adjusted kappa (PABAK), including a subgroup analysis of malnutrition severity. Pearson correlation examined the relationship between these tools and conventional nutritional indicators, while T-tests were used to compare functional outcomes (handgrip strength) and clinical outcomes (length of stay) between nutritional groups. Results: The mean NRS 2002 score was 3.37 ± 0.98, with 75.0% (81/108) of patients identified as being at nutritional risk. Significant differences in nutritional risk were observed based on age, Body Mass Index (BMI), tumor stage, and tumor size (P < 0.05). Malnutrition prevalence was 60.2% (65/108) according to GLIM criteria and 63.9% (69/108) according to PG-SGA. The consistency analysis between GLIM and PG-SGA yielded a Cohen's kappa value of 0.71 (P < 0.01), indicating substantial agreement. The PABAK was 0.78, confirming substantial agreement after adjusting for prevalence effects. The agreement for malnutrition severity (GLIM Stage 1/2 vs. PG-SGA Moderate/Severe) was moderate (κ = 0.58). Both GLIM and PG-SGA scores demonstrated significant positive correlations with arm circumference (AC), calf circumference (CC), BMI, serum albumin (Alb), and hemoglobin (Hb) levels (P < 0.05). Furthermore, malnutrition diagnosed by either GLIM or PG-SGA was significantly associated with lower handgrip strength and longer hospital stays (P < 0.01). Conclusion: Patients with pancreatic malignant tumors exhibit a high prevalence of nutritional risk and malnutrition. The GLIM criteria and PG-SGA demonstrate good consistency in diagnosing malnutrition in this patient population. Both tools effectively identify patients with functional deficits and poorer clinical outcomes, supporting the utility of GLIM as a practical assessment tool in clinical settings.