Sarcopenia as a prognostic factor in gastric cancer patients undergoing gastrectomy: Implications for postoperative and long-term outcomes

Table 2 Integrated prognostic models and composite indices in gastric cancer body-composition assessment

Model/index	Main components	Clinical meaning	Prognostic value	Main limitation
GLIM criteria	Phenotypic criteria, including weight loss, low BMI, reduced muscle mass; etiologic criteria, including reduced intake, inflammation, or disease burden	Standardized assessment of malnutrition with body-composition relevance	Identifies patients at higher risk of postoperative morbidity, poor recovery, and worse survival	Requires nutritional and clinical assessment; cut-offs and implementation may vary across populations
GNRI	Serum albumin and body weight relative to ideal body weight	Nutritional risk and systemic reserve	Associated with postoperative complications, treatment tolerance, and long-term survival	Does not directly measure muscle mass or muscle quality
PNI	Serum albumin and lymphocyte count	Combined nutritional and immune status	Predicts postoperative morbidity, recurrence risk, and overall survival	Influenced by inflammation, infection, hydration status, and non-cancer-related conditions
CXI	Skeletal muscle index × serum albumin/neutrophil-to-lymphocyte ratio	Integrates muscle mass, nutrition, and systemic inflammation	May improve risk stratification beyond isolated CT-derived or laboratory markers	Requires standardized CT assessment and laboratory timing
ALBI-related models	Serum albumin and bilirubin, alone or combined with sarcopenia/body-composition variables	Reflects hepatic and nutritional reserve	May help identify patients with impaired physiological reserve and poorer prognosis	More relevant when liver function has prognostic impact; should not be confused with liver-disease scores such as MELD or CTP
Radiomics models	CT texture features from tumor and/or body-composition compartments combined with clinical variables	Imaging-based prediction of recurrence and survival	Can improve individualized prediction beyond conventional staging alone	Requires software, technical standardization, and external validation
AI-based nomograms	Machine-learning or deep-learning features combined with clinicopathological, CT-derived, nutritional, and inflammatory variables	Multivariable personalized risk prediction	May predict postoperative complications, recurrence, and overall survival more accurately than single markers	Limited by retrospective datasets, black-box interpretability, and generalizability across centers
Integrated body-composition scores	SMI, SMD/myosteatosis, visceral fat, sarcopenic obesity, functional tests, and inflammatory/nutritional markers	Multidimensional assessment of host reserve	Captures the combined prognostic impact of muscle quantity, muscle quality, adiposity, and inflammation	Lack of universally accepted thresholds and standardized scoring systems

GLIM: Global leadership initiative in malnutrition; BMI: Body mass index; GNRI: Geriatric Nutritional Risk Index; PNI: Prognostic nutritional index; CXI: Cachexia index; CT: Computed tomography; ALBI: Albumin-bilirubin; MELD: Model for end-stage liver disease; CTP: Child-Turcotte-Pugh; AI: Artificial intelligence; SMI: Skeletal muscle index; SMD: Skeletal muscle density.