Repeatability of Computerized Tomography-Based Anthropomorphic Measurements of Frailty in Patients With Pulmonary Fibrosis Undergoing Lung Transplantation

摘要

Purpose: To determine interreader and intrareader repeatability and correlations among measurements of computerized tomography-based anthropomorphic measurements in patients with pulmonary fibrosis undergoing lung transplantation. Methods: This was an institutional review board-approved, Health Insurance Portability and Accountability Act-compliant retrospective study of 23 randomly selected subjects (19 male and 4 female; median age = 69 years; range: 66-77 years) with idiopathic pulmonary fibrosis undergoing pulmonary transplantation, who had also undergone preoperative thoracoabdominal computerized tomography. Five readers of varying imaging experience independently performed the following cross-sectional area measurements at the inferior endplate of the L3 vertebral body: right and left psoas muscles, right and left paraspinal muscles, total abdominal musculature, and visceral and subcutaneous fat. The following measurements were obtained at the inferior endplate of T6: right and left paraspinal muscles with and without including the trapezius muscles and subcutaneous fat, Three readers repeated all measurements to assess intrareader repeatability. Results: Intrareader repeatability was nearly perfect (interclass correlation coefficients = 0.99, P < 0.001). Interreader agreement was excellent across all 5 readers (interclass correlation coefficients: 0.71-0.99, P < 0.001). Coefficients of variance between measures ranged from 3.2%-6.8% for abdominal measurements, but were higher for thoracic measurements, up to 23.9%. Correlation between total paraspinal and total psoas muscle area was strong (r~2 = 0.67, P < 0.001). Thoracic and abdominal musculature had a weaker correlation (r~2 = 0.35-0.38, P < 0.001). Conclusion: Measures of thoracic and abdominal muscle and fat area are highly repeatable in patients with pulmonary fibrosis undergoing lung transplantation. Measures of muscle area are strongly correlated among abdominal locations, but inversely correlated between abdominal and thoracic locations.