210. A Morphomic Index Is an Independent Predictor of Survival After Lung Cancer Resection
Brian Mitzman1, *Jules Lin2, Brian A. Derstine2, Michael J. Pienta2, Stewart C. Wang2, Sang Mee Lee3, *Mark K. Ferguson3
1NYU - Winthrop, Mineola, NY;2University of Michigan, Ann Arbor, MI; 3University of Chicago, Chicago, IL
Invited Discussant: *Katie S. Nason
Objective: Sarcopenia and BMI have been associated with acute outcomes after lung cancer surgery, but little is known about the relationship of anthropometric data to long-term survival. In addition, almost nothing is known about the association of CT morphomics other than core muscle area to survival. We assessed the relationship of CT morphomics to survival after lung cancer resection.
Methods: Patients at two academic medical centers who underwent lung cancer resection from 1995 to 2014 and who had preoperative CT scans suitable for morphometric analysis were evaluated. The relationship of CT morphomics to overall survival was assessed by log-rank tests and Cox proportional hazard models. Individual morphomics were not independently related to survival, nor were doublets of these measurements. A primary predictor (morphomic index) was developed as the sum of the three morphomic variables most closely related to survival (percentiles for dorsal muscle group area, vertebral trabecular bone density, and visceral fat area compared to a normal reference population). For purposes of analysis, the morphomic index was divided into terciles, and multivariate analyses for survival were adjusted for age, stage, PS, CAD, DLCO, and FEV1.
Results: We analyzed outcomes for 961 patients including 461 women (48%); the mean age was 66.2 ± 10.4 years. Median follow-up was 4.5 years; 456 patients had died at the time of last follow-up. The morphomic index was significantly associated with overall survival; the lowest tercile had the worst survival (log-rank test p=0.005 compared to terciles 2 and 3), whereas terciles 2 and 3 were similar (log-rank test p=0.93). The 2nd and 3rd tercile groups were combined, creating a binary variable for further analyses. Median overall survival was 5.8 years in the lowest tercile group and 7.5 years in the upper terciles group; 5-year survivals were 53% and 63%, respectively. The morphomic index was an independent predictor of survival (hazard ratio, 0.79; 95% confidence interval, 0.64 to 0.97; p-value=0.026; Table).
Conclusions: Unlike prior studies, a statistically significant association between core muscle cross-sectional area and survival was not identified in this dataset. However, by assessing morphometry more globally we identified a combination of metrics that was strongly associated with survival. This morphomic index, including core muscle area, bone density, and visceral fat area, is an independent predictor of survival in patients who undergo lung cancer resection. The index may help in calibrating patient expectations and in shared decision making regarding lung cancer surgery.
Cox Regression Model for Survival
0.64 to 0.97
1.11 to 1.38
1.07 to 1.81
2.04 to 3.30
0.41 to 0.83
1.11 to 1.76
0.88 to 0.98
0.89 to 0.99