Presentation Authors: Marc A. Lubin*, Brooklyn, NY, Hari Haran Sudhan, Rachel Weil, Alp Tuna Beksac, Ashutosh Tewari, New York, NY
Introduction: Prostate specific antigen is a poor biomarker for prostate cancer (PCa), with a nearly 70% negative biopsy rate as well as overdiagnosis of clinically insignificant PCa. Both the 4Kscore test (4K) and multiparametric magnetic resonance imaging (MRI) have shown improved discrimination of clinically significant PCa (csPCa). Recently, Punnen et al showed that the combination of 4K and MRI (4K-MRI) outperforms either test individually in both area under the curve (AUC) and decision curve analyses. In contrast, their attempts at devising a practical algorithm for 4K-MRI yielded relatively weak sensitivities and negative predictive values (NPV). _x000D_
We present a validation of 4K-MRI, as well as an improved algorithm for the identification of csPCa while limiting unnecessary MRI&[prime]s.
Methods: Included is a retrospective analysis of patients considered at risk for PCa, and tested with 4K and MRI between 2016 and 2018. The primary outcome, csPCa, was defined as Gleason grade group (GG) â‰¥ 2 on biopsy. 4K and MRI were compared via receiver operating characteristic (ROC) curves and logistic regression, and a model combining the two was tested for accuracy and efficiency.
Results: The cohort consists of 138 patients. Multivariable regression analysis identified an association between csPCa and both 4K and MRI (Table 1). Analyzing our 4K-MRI model (Figure 1), a first test of either 4K â‰¥ 20% or MRI-PIRADS 3-5 yields equivalent sensitivity, 90%, though 4K&[prime]s specificity is dramatically better, 69% vs 49%. As a second test, applying PIRADS 3-5 to the positive 4K group (n=61), or 4K ≥ 20% to the positive PIRADS group (n=83) provides added value with a much-improved overall specificity (83%), while maintaining sensitivity (87%) and negative predictive accuracy (96%).
Conclusions: Patients with both 4K â‰¥ 20% and PIRADS â‰¥ 3 should proceed to biopsy. Alternative outcomes confidently predict not-csPCa, identifying 83% of negative biopsies that could be sparred. Finally, only 44% of the cohort (61/138) would require an MRI.