Diagnostic Performance and Interreader Agreement of a Standardized MR Imaging Approach in the Prediction of Small Renal Mass Histology

Abstract

Purpose To determine the diagnostic performance and inter-reader agreement of a standardized diagnostic algorithm for determining the histologic type of small (<=4cm) renal masses (SRM) with multiparametric magnetic resonance imaging (MRI). Materials and Methods This single-center, retrospective, HIPAA-compliant, IRB-approved study included 103 patients with 109 SRM, resected between December 2011 and July 2015. The requirement for informed consent was waived. Pre-surgical renal MRIs were reviewed by 7 radiologists with diverse experience. Eleven MRI features were assessed and a standardized diagnostic algorithm used to determine the most likely histologic diagnosis, which was compared to histopathology after surgery. Inter-reader variability was tested with Cohen’s κ. Regression models using MRI features were used to predict the histopathologic diagnosis with 5% significance level. Results Clear-cell (ccRCC) and papillary type renal cell carcinomas (pRCC) were diagnosed with respective sensitivities of 85% (47/55) and 80% (20/25), and specificities of 76% (41/54) and 94% (79/84). Inter-reader agreement was moderate-to-substantial (ccRCC, κ=0.58; pRCC, κ=0.73). Signal intensity of the lesion on T2-weighted images (T2W) and degree of contrast enhancement during corticomedullary phase (CE) were independent predictors of ccRCC (T2W OR: 3.19 CI95%: [1.4, 7.1], p=0.003; CE OR: 4.45 [1.8, 10.8], p<0.001) and pRCC (CE OR: 0.053 [0.02, 0.2], p<0.001), both with substantial inter-reader agreement (T2W, κ=0.69; CE, κ=0.71). Lower performance was observed for chromophobe histology, oncocytomas, and minimal-fat angiomyolipomas, [ranges, sensitivity=14%(1/7)–67%(4/6), specificity=97%(100/103)–99%(101/102)], with fair-to-moderate inter-reader agreement (κ=0.23–0.43). Segmental enhancement inversion was an independent predictor of oncocytomas (OR: 16.21 [1.0, 275.4], p=0.049), with moderate inter-reader agreement (κ=0.49). Conclusion The proposed standardized MRI-based diagnostic algorithm had a diagnostic accuracy of 81% (88/109) and 91% (99/109) for the diagnosis of ccRCC and pRCC, respectively, while achieving moderate to substantial inter-reader agreement among 7 radiologists.

Diagnostic performance of prospectively assigned clear cell Likelihood scores (ccLS) in small renal masses at multiparametric magnetic resonance imaging

“…Each study was reviewed on a Picture Archiving and Communications System (PACS) workstation (iSite, Philips Healthcare, Best, Netherlands). As part of the clinical, structured report, a ccLS [14,15,18] was prospectively assigned for each renal tumor identified. If multiple, a separate score was assigned to each.…”

Renal and adrenal masses containing fat at MRI: Proposed nomenclature by the society of abdominal radiology disease‐focused panel on renal cell carcinoma

“…Studies investigating the algorithmic diagnosis of small renal masses with MRI have noted that the presence of a microscopic fat may distinguish clear‐cell RCCs from oncocytic neoplasms (eg, chromophobe RCC and oncocytoma), each of which is characteristically hyperintense on T 2 W imaging . Clear‐cell RCC also have been shown to be more avidly enhancing compared with oncocytic neoplasms . A minority of papillary tumors also may show microscopic fat, which has been speculated to be due to clear cell heterogeneity; such tumors may be distinguished from fat‐poor AML by their high signal intensity on T 2 w imaging (compared with typical papillary RCCs) .…”

Society of Abdominal Radiology disease-focused panel on renal cell carcinoma: update on past, current, and future goals

“…More recently, multiple institutions have reported single-center retrospective studies suggesting that CT and MR can be used to diagnose previously indeterminate specific benign (e.g., fat poor AML) and malignant (e.g., clear cell RCC) diagnoses without need of percutaneous biopsy [48, 49]. MRI is particularly well suited to subtype renal masses given the lack of ionizing radiation that complicates multiphase CT, and the multiparametric data sets obtained in standard clinical examinations.…”

“…For example, it has been shown that small enhancing renal masses with homogeneous low signal intensity on T2-weighted images exhibit slower growth (i.e., tumor doubling time > 2 years) and may be better candidates for active surveillance [16]. Others have explored whether an MRI-based likelihood score of clear cell RCC histology may better identify which patients are suitable for active treatment [49]. Though promising, much of this work has been single center with relatively small datasets, especially for uncommon histologic variants.…”