Background: Predicting oncologic outcomes is essential for optimizing the treatment for patients with cancer. This review examines the feasibility of using Computed Tomography (CT) images of fat density as a prognostic factor in patients with cancer.
Methods: A systematic literature search was performed in PubMed, Embase and Cochrane up to March 2020. All studies that mentioned using subcutaneous or visceral adipose tissue (SAT and VAT, respectively) CT characteristics as a prognostic factor for patients with cancer were included. The primary endpoints were any disease-related outcomes in patients with cancer.
Results: After screening 1043 studies, ten studies reporting a total of 23 - ten for SAT and thirteen for VAT - comparisons on survival, tumor recurrence and postsurgical infection were included. All ten studies included different types of malignancy: six localized, two metastatic disease, and two both. Five different anatomic landmarks were used to uniformly measure fat density on CT: lumbar (L)4 (n = 4), L3 (n = 2), L4-L5 intervertebral space (n = 2), L5-S1 intervertebral space (n = 1), and the abdomen (n = 1). Overall, six of ten SAT comparisons (60%) and six of thirteen VAT comparisons (46%) reported a significant (p < .05) association of increased SAT or VAT density with an adverse outcome. All remaining nonsignificant comparisons, except one, deviated in the same direction of being predictive for adverse outcomes but failed to reach significance. The median hazard ratio (HR) for the nine SAT and thirteen VAT associations where HRs were given were 1.45 (95% confidence interval [CI] 1.01-1.97) and 1.90 (95% CI 1.12-2.74), respectively. The binomial sign test and Fisher's method both reported a significant association between both SAT and VAT and adverse outcomes.
Conclusion: This review may support the feasibility of using SAT or VAT on CT as a prognostic tool for patients with cancer in predicting adverse outcomes such as survival and tumor recurrence. Future research should standardize radiologic protocol in prospective homogeneous series of patients on each cancer diagnosis group in order to establish accurate parameters to help physicians use CT scan defined characteristics in clinical practice.