Original Article
Quantification of sarcomatoid differentiation in renal cell carcinoma on magnetic resonance imaging
Abstract
Background: Sarcomatoid differentiation in renal cell carcinoma (sRCC) is histologically characterized by anaplastic changes of renal cell carcinoma (RCC) subtypes, which has been associated with a poor prognosis. sRCC is managed more aggressively than RCC without sarcomatoid components, so pre-operative detection of sarcomatoid differentiation would significantly affect surgical management. The purpose of this study is to compare the quantification of sarcomatoid features in RCCs on pre-operative magnetic resonance imaging (MRI) to standard histological examination.
Methods: Patients who had nephrectomy at our institution between 2000 and 2015 with pathology proven RCC and pre-operative contrast enhanced MRI abdominal scans were retrospectively reviewed. A custom MATLAB routine calculated the portion of each manually segmented whole tumor with MRI signal suggestive of sarcomatoid involvement based on prior research (MRI%SARC). The primary endpoint compared MRI%SARC to percent sarcomatoid involvement estimated by histological examination (HIST%SARC) using Pearson correlation and Bland Altman analysis.
Results: A total of 17 patients with sRCC (10 males, age 60.3±11.1 years) and 17 consecutive control patients with clear cell RCC (ccRCC) without sarcomatoid components (10 males, age 64.5±7.6 years) were evaluated. Pearson correlation analysis revealed a strong association between MRI%SARC and HIST%SARC (r=0.782, P<0.001). Bland-Altman analysis demonstrated proportional bias, with a mean bias of 19.29 [95% confidence interval (CI): 9.79–28.79] and with 95% limits of agreement of −16.93 (95% CI: −33.38 to −0.48) to 55.51 (95% CI: 39.06−71.96), suggesting MRI%SARC underestimated values compared to HIST%SARC by 19%.
Conclusions: Multiparametric pre-operative MRI analysis to quantify sarcomatoid features in RCC correlates with standard histological examination but underestimates percent sarcomatoid involvement.
Methods: Patients who had nephrectomy at our institution between 2000 and 2015 with pathology proven RCC and pre-operative contrast enhanced MRI abdominal scans were retrospectively reviewed. A custom MATLAB routine calculated the portion of each manually segmented whole tumor with MRI signal suggestive of sarcomatoid involvement based on prior research (MRI%SARC). The primary endpoint compared MRI%SARC to percent sarcomatoid involvement estimated by histological examination (HIST%SARC) using Pearson correlation and Bland Altman analysis.
Results: A total of 17 patients with sRCC (10 males, age 60.3±11.1 years) and 17 consecutive control patients with clear cell RCC (ccRCC) without sarcomatoid components (10 males, age 64.5±7.6 years) were evaluated. Pearson correlation analysis revealed a strong association between MRI%SARC and HIST%SARC (r=0.782, P<0.001). Bland-Altman analysis demonstrated proportional bias, with a mean bias of 19.29 [95% confidence interval (CI): 9.79–28.79] and with 95% limits of agreement of −16.93 (95% CI: −33.38 to −0.48) to 55.51 (95% CI: 39.06−71.96), suggesting MRI%SARC underestimated values compared to HIST%SARC by 19%.
Conclusions: Multiparametric pre-operative MRI analysis to quantify sarcomatoid features in RCC correlates with standard histological examination but underestimates percent sarcomatoid involvement.
