Validity of measuring psoas muscle mass index for assessing sarcopenia in patients with gynecological cancer

Abstract

Objective: The current study evaluated the performance of psoas muscle mass measurement for detecting low skeletal muscle mass quantity.

Methods: A sample of 82 consecutive patients with gynecological cancers was examined using computed tomography and dual energy X-ray absorptiometric scan before treatment. Skeletal muscle mass index was measured by dual energy X-ray absorptiometric scan and its cut-off value was set at 5.40 kg/m2 for detecting low skeletal muscle mass. Psoas muscle mass index was manually measured with cross-sectional computed tomography imaging at the level of L3 by six evaluators.

Results: Low skeletal muscle mass index was identified in 23 (28.0%) patients. Two-way analysis of variance confirmed a significant main effect of skeletal muscle mass index on mean psoas muscle mass index values (P < 0.0001). A receiver operating characteristic curve obtained from a total of 492 psoas muscle mass index data points gathered from six evaluators produced an area under the curve value of 0.697 (95% confidence interval 0.649-0.744) and a cut-off value of 3.52 cm2/m2, with sensitivity of 79.0% and specificity of 59.6%. Using the cut-off value, the kappa coefficient for evaluating diagnostic agreement between skeletal muscle mass index (low vs. normal) and psoas muscle mass index (low vs. normal) was 0.308 (95% confidence interval 0.225-0.392), suggesting poor agreement. Fleiss' kappa produced a coefficient of 0.418 (95% confidence interval 0.362-0.473), suggesting moderate agreement.

Conclusions: Although relevance between skeletal muscle mass index and psoas muscle mass index was confirmed, intensity of relevance between them was weak. Psoas muscle mass index measurement should be subordinated to skeletal muscle mass index measurement for detection of low skeletal muscle mass.

Keywords: gynecological cancer; interobserver variability; psoas muscle mass index; sarcopenia; skeletal muscle index.