Abstract
BACKGROUND: CT measurements of psoas muscle density (PMD) and area (PMA) (with or without indexing to height-squared or body-surface-area) are used interchangeably as sarcopenia measures - it is unknown which best correlates with surgical risk. OBJECTIVES: 1.Determine the correlation between psoas muscle density, area, and indexed area; 2.Identify which psoas measures most strongly associated with surgical outcomes DESIGN: The University Hospital Geelong Colorectal database included all patients who underwent elective colorectal surgery from 2007 to 2014 (minimum five-years follow-up). Pre-operative CT scans were reviewed, psoas measures correlated with each other and with outcomes. SETTING: University Hospital Geelong is a regional referral hospital in Victoria, Australia. PARTICIPANTS: This database listed 552 patients, 120 were excluded as pre-operative CT-films were not accessible, leaving 432 patients included. EXPOSURE: Psoas muscle density, area, and area indexed by height-squared and body-surface-area. MEASUREMENTS: Pearson correlations investigated correlations between psoas muscle measures. Logistic regression and ROC-analysis investigated each psoas measures association with peri‑operative morbidity. Kaplan-Meier survival-analysis investigated the association of each psoas measure with long-term survival. RESULTS: Mean age was 70.4 years, 41 % were female. Psoas muscle density correlated poorly with area (R(2)=0.15). Unindexed psoas muscle area correlated well with area indexed by height-squared (R(2)=0.950) and body-surface-area (R(2)=0.938). Long-term survival was associated with psoas muscle density (HR1.515(95 %CI 1.062-2.161)) and area (HR1.886(95 %CI 1.322-2.692)). Increasing psoas muscle density (reduced sarcopenia) was associated with decreased major-complications (OR0.963(95 %CI 0.938-0.989)) and peri‑operative mortality (OR0.903(95 %CI 0.847-0.962)), with ROC-curve AUC=0.829 indicating an accurate test. There was no association between psoas muscle area and major-complications (OR1.000(95 %CI 1.000-1.000)), nor peri‑operative mortality (OR1.000(95 %CI 0.999-1.001)), with ROC-curves AUC=0.507-0.521. Indexed area measures were not associated with outcomes. CONCLUSIONS: Psoas muscle density and area did not correlate. Both were associated with long-term survival, but only density was associated major-complications and mortality. Indexing removed the correlation of area with long-term survival.