Effect of malnutrition risk on the association between skeletal muscle index and health-related quality of life and survival in upper gastrointestinal cancer

Introduction: Low skeletal muscle index (SMI) is an established prognostic indicator in cancer1, and may also be associated with poor health-related quality of life (HRQOL)2. Presence or risk of malnutrition is also associated with poor survival and HRQOL outcomes3,4, yet this confounding factor is rarely accounted for in studies investigating the consequences of low SMI. Malnutrition is prevalent in upper gastrointestinal (GI) cancer5, therefore this study examined the effect of malnutrition risk on the association between SMI at diagnosis of upper GI cancer, and both HRQOL at diagnosis and survival over the following year.
Methods: A prospective study was conducted using data from participants enrolled in a randomised controlled trial at diagnosis of oesophageal, gastric or pancreatic cancer6. Baseline data included: SMI from diagnostic computed tomography imaging, malnutrition risk (PG-SGA Short Form, PG-SGASF7), and HRQOL (EORTC-QLQ C30)8. Low SMI was identified using sex-specific thresholds9. The correlation between SMI and PG-SGASF scores was examined, and multiple linear regression was used to identify contributors to variation in HRQOL at baseline. Predictors of mortality within 12 months of diagnosis were investigated using Cox proportional hazards regression.
Results: The study included 105 treatment-naive patients with oesophageal (43%), gastric (20%), or pancreatic (37%) cancers. The cohort was 68% male, mean age 65.9 (SD±10.0) years. 56% of patients had low SM1, and 49% were at high risk of malnutrition (PG-SGASF score ≥9)10. SMI was weakly, negatively correlated with PG-SGASF score (r=-0.268, p=0.006), and some HRQOL domains (physical and role functioning, summary score11, all p<0.05). When adjusted for PG-SGASF score, SMI was not significantly associated with any measure of HRQOL at baseline. Higher PG-SGASF scores were independently associated with lower HRQOL across all scales. Advanced (unresectable) disease at diagnosis was independently associated with 12-month mortality (HR 2.643, 95% CI 1.041-6.711, p=0.041), adjusting for baseline PGSGASF score and cancer type; baseline SMI was not prognostic. In a subgroup analysis of those with resectable/borderline resectable disease (n=52), high PG-SGASF score (≥9) at diagnosis was the only variable associated with 12-month mortality (HR 4.539, 95% CI 1.012-20.352, p=0.048).
Conclusions: Malnutrition risk measured by the PG-SGASF was a stronger predictor of poorer HRQOL and survival than SMI, and should be included as a covariate in future studies investigating the effect of low SMI on these outcomes in people with cancer.