Comparison of temporal muscle indices for survival prediction in patients with glioblastoma: Prognostic value, accuracy and reliability

Abstract Background Temporal muscle thickness (TMT) and cross-sectional area (TMA) are readily available estimates of whole-body muscle mass. Low TMT and TMA have been associated with increased mortality in various neurological diseases. Several measurement protocols for TMT and TMA exist, but their reproducibility, accuracy, and prognostic value have not been compared. We compared the published protocols in terms of these characteristics in a cohort of patients with glioblastoma. Methods We retrospectively identified patients with glioblastoma treated at Oulu University Hospital between January 2018 and December 2022. We compared the reproducibility, accuracy, and association with overall survival (OS) of9temporal muscle indices (4 protocols for TMT, 4 for TMA, and 1 for temporal muscle index (TMI; TMA/height2)) identified from the literature. Results One-hundred-and-ten patients with a mean age of 64 (SD 13) years were included. Sixty-eight (62%) were males, 83 (76%) underwent surgical resection, and 27 (25%) underwent a biopsy. One- and 2-year mortality rates were 66% and 87%, respectively. All 9 temporal muscle indices were positively associated with OS. The prognostic association was stronger among males. Overall accuracy for predicting 1-year survival ranged from 56% to 69%. Intra-rater reproducibility was almost perfect (0.88-0.98) for all indices, and inter-rater reproducibility ranged from fair to almost perfect (0.50-0.93). Conclusions All tested temporal muscle measurement protocols were associated with post-operative OS. Their dichotomized prognostic accuracy values for 1-year mortality were modest, but comparable. Reproducibility varied significantly. Although cut-off values differed between protocols, they mostly categorized the same patients to low muscle mass group.