RESUMEN
BACKGROUND: Computed tomography images acquired during routine cancer care provide an opportunity to determine body composition with accuracy and precision. Quantification of skeletal muscle is of interest owing to its association with clinical outcomes. However, the standards of precision testing considered mandatory in other areas of radiology are lacking from the literature in this area. We aim to describe the change in skeletal muscle over time at different anatomical levels using the precision error. METHODS: Thirty-eight male patients with squamous cell carcinoma of the head and neck were evaluated at two time points encompassing their treatment plan. Precision testing consisted of analyzing the cross-sectional area (CSA) of the skeletal muscle and total adipose tissue of 76 CT studies (38 images at baseline repeated twice and 38 follow-up images repeated twice) measured by a skilled observer. The % coefficient of variation (%CV), the root-mean-square standard deviation (RMS SD) and the corresponding 95% least significant change (LSC) were calculated for four anatomical levels: upper arm, thigh, chest and abdomen. RESULTS: The median time between scans was 223.6 (SD 31.2) days. Precision error (% CV) for total skeletal muscle cross sectional area was 0.86% for upper arm, 0.26% for thigh, 0.39% for chest and 0.63% for abdomen. The corresponding LSC values in upper arm, thigh, chest and abdomen were 2.4%, 0.7%, 1.1% and 1.8%, respectively. Based on the LSC for RMS SD, patients were classified in two categories according to muscle cross-sectional area: stable (i.e within LSC value) or gained and loss. To compare the four anatomical levels, the proportion of patients with muscle loss exceeding the LSC value was 74.3% for arm, 86.2% for thigh, 82.9% for chest and 76.3% for abdomen. For these same anatomic regions, the mean muscle loss for those patients classified below the LSC was 14.6% (SD 9.3), 13.4% (SD 7.8), 11.9% (SD 6.5) and 11.6% (SD 5.5), respectively. Only the loss of muscle area was significantly higher in thigh (p = 0.023), using L3 as the reference level. CONCLUSIONS: We recommend the uniform use of a standard precision test when reporting muscle change over time. LSC values vary from 0.7 to 2.4% depending on anatomic site; with the lowest precision error to detect change in the thigh. Based on this analysis, muscle wasting appears to be systemic and while present in limbs and trunk is significantly higher in the thigh than in the chest, abdomen or upper arm.
