Moderated Poster

Poster, Podium & Video Sessions

MP08-03: Prediction of Histological Subtypes of Small Renal Masses: Striving for a Standardized MRI Diagnostic Algorithm

Friday, May 12
9:30 AM - 11:30 AM
Location: BCEC: Room 160

Presentation Authors: Fernando U. Kay*, Noah E. Canvasser, Yin Xi, Daniella F. Pinho, Daniel Costa, Alberto Diaz de Leon, Gaurav Khatri, John R. Leyendecker, Takeshi Yokoo, Aaron Lay, Nicholas Kavoussi, Ersin Koseoglu, Jeffrey A. Cadeddu, Ivan Pedrosa, Dallas, TX

Introduction: MRI may aid in the management of small renal masses (<=4 cm, SRM) by differentiating among histologic subtypes. However, standardized approaches to MRI interpretation and interobserver agreement data are lacking.

To assess the performance of a wide spectrum of MRI features for predicting the histologic diagnosis of SRM, and determine the interobserver agreement among multiple readers.



Methods: Retrospective HIPAA-compliant IRB-approved study including 109 patients with cT1a SMR and a pre-surgical MRI. Images were reviewed by 7 radiologists with body MRI training and 1-15 years of experience. The following characteristics were analyzed on non-contrast images: T2-weighted (T2W) signal intensity/texture, presence/absence of intravoxel or bulk fat, magnetic susceptibility, central scar, and hemorrhage. Features assessed on post-contrast images included contrast avidity, enhancement homogeneity, dynamic characteristics, and segmental enhancement inversion. Multivariate generalized linear mixed model analysis with logit link was used to identify independent subtype predictors, as confirmed by histopathology, with p < 0.05 considered significant. Pairwise weighted analysis was used to measure interobserver agreement.


Results: Clear cell renal cell carcinomas (ccRCC) represented 51% of the masses, papillary RCC (pRCC) 25%, chromophobe RCC (chrRCC) 6%, oncocytoma 6%, minimal-fat angiomyolipoma (mfAML) 6%, and others 9%. Table 1 includes values for the MRI features. ccRCC was predicted by signal intensity on T2W (high vs low, OR, odds ratio: 3.2 CI 95%: [1.4, 7.1], p < 0.001) and contrast avidity (avid vs. low, OR: 4.5 [1.8, 10.8], p < 0.0001), while pRCC was predicted by contrast avidity (low vs avid, OR: 0.05 [0.02, 0.2], p < 0.0001) on multivariate analysis. Segmental enhancement inversion was an independent predictor of oncocytoma (present vs absent, OR: 16.2 [1.0, 275.4], p < 0.05). None of the features were significant predictors of chrRCC or mfAML on the multivariate analysis.


Conclusions: Our data support the use of T2W signal intensity and contrast avidity as critical steps in the implementation of standardized MRI interpretation algorithms for predicting the histological subtype of SRM. Segmental enhancement inversion can be used as a feature for the diagnosis of oncocytomas.

Source Of Funding: Partially funded by grant 5RO1CA154475

Fernando U. Kay, MD, MBA

UT Southwestern Medical Center

Fernando Kay, MD MBA, radiologist at UT Southwestern Medical Center.
Completed medical school and radiology training at University of Sao Paulo, Brazil. Fellowship training in Body MRI at UT Southwestern Medical Center. Completed MBA at Insper, Sao Paulo, Brazil. Has published scholarly articles and book chapter in GU and cardiothoracic radiology.

Presentation(s):

Send Email for Fernando Kay


Assets

MP08-03: Prediction of Histological Subtypes of Small Renal Masses: Striving for a Standardized MRI Diagnostic Algorithm



Attendees who have favorited this

Send Email for Prediction of Histological Subtypes of Small Renal Masses: Striving for a Standardized MRI Diagnostic Algorithm