Differences in negative predictive value of prostate MRI based in men with suspected or known cancer / Diferenças no valor preditivo negativo da ressonância magnética da próstata baseada em homens com câncer suspeito ou conhecido

Abstract Objective: To compare the negative predictive value (NPV) of multiparametric MRI for Gleason score (GS) ≥ 3+4 cancer and evaluate predictors of these tumors in men with suspected disease and under active surveillance (AS). Materials and Methods: This retrospective study included 38 men with suspected prostate cancer and 38 under AS with scans assigned PI-RADS v2 scores 1 or 2 between May 2016 and September 2017. Biopsy results were no cancer, GS = 3+3, or GS ≥ 3+4. Pre-MRI PSA, gland volume, and PSA density were recorded. Chi-square, equality of proportions, and logistic regressions were used to analyze the data. Results: Intermediate to high-grade cancer was found in 12.8% (95% CI = 2.3-23.3) and 35.9% (95% CI = 20.8-50.9) of men with suspected cancer, and under AS (p = 0.02), respectively. The NPV for GS ≥ 3+4 were 87.2% (suspected cancer; 76.7-97.7) and 64.1% (AS; 49.0-79.2). In neither group PSA significantly predicted cancer grade (p = 0.75 and 0.63). Although it did not reach conventional statistical significance, PSA density was a good predictor of cancer grade in men with suspected disease (p = 0.06), but not under AS (p = 0.62). Conclusion: The NPV of multiparametric MRI for GS ≥ 3+4 is higher in men with suspected prostate cancer than in men under AS. PSA density ≤ 0.15 improved the prediction of intermediate to high-grade disease in patients without known cancer.

Resumo Objetivo: Comparar o valor preditivo negativo (VPN) da RM multiparamétrica da próstata para o diagnóstico de tumores escore de Gleason (EG) ≥ 3+4 e avaliar os preditores desses tumores em homens com suspeita de doença e nos sob vigilância ativa (VA). Materiais e Métodos: Este estudo retrospectivo incluiu 38 homens com suspeita de câncer de próstata e 38 em VA com RM, aos quais foram atribuídos escores PI-RADS v2 1 ou 2 entre maio de 2016 e setembro de 2017. Os resultados da biópsia foram ausência de câncer, câncer EG = 3+3 ou câncer EG ≥ 3+4. PSA pré-RM, volume da glândula e densidade de PSA foram anotados. Qui-quadrado, igualdade de proporções e regressões logísticas foram utilizados para analisar os dados. Resultados: Câncer de grau intermediário a alto grau foi encontrado em 12,8% (IC 95% = 2,3-23,3) e 35,9% (IC 95% = 20,8-50,9) dos homens com suspeita de câncer e nos sob VA (p = 0,02), respectivamente. O VPN para GS ≥ 3+4 foi 87,2% (suspeita de câncer; IC 95% = 76,7-97,7) e 64,1% (VA; IC 95% = 49,0-79,2). Em nenhum dos grupos o PSA previu significativamente o grau de câncer (p = 0,75 e 0,63. Embora não tenha alcançado o limiar de significância estatística usual, a densidade de PSA foi um bom preditor de grau de câncer em homens com suspeita de doença (p = 0,06), mas não sob VA (p = 0,62). Conclusão: O VPN da RM multiparamétrica para GS ≥ 3+4 é maior em homens com suspeita de câncer de próstata do que em homens sob VA. Uma densidade de PSA ≤ 0,15 melhorou a previsão de doença de grau intermediário a alto grau em pacientes sem diagnóstico prévio de câncer.

Detection of clinically significant prostate cancer with PI-RADS v2 scores, PSA density, and ADC values in regions with and without mpMRI visible lesions

ABSTRACT Purpose To determine if PSAD, PSADtz, and ADC values improve the accuracy of PI-RADS v2 and identify men whose concurrent systematic biopsy detects clinically significant cancer on areas without mpMRI visible lesions. Materials and methods Single reference-center, cross-sectional, retrospective study of consecutive men with suspected or known low to intermediate-risk prostate cancer who underwent 3T mpMRI and TRUS-MRI fusion biopsy from 07/15/2014 to 02/17/2018. Cluster-corrected logistic regression analyses were utilized to predict clinically significant prostate cancer (Gleason score ≥3+4) at targeted mpMRI lesions and on systematic biopsy. Results 538 men (median age=66 years, median PSA=7.0ng/mL) with 780mpMRI lesions were included. Clinically significant disease was diagnosed in 371 men. PI-RADS v2 scores of 3, 4, and 5 were clinically significant cancer in 8.0% (16/201), 22.8% (90/395), and 59.2% (109/184). ADC values, PSAD, and PI-RADS v2 scores were independent predictors of clinically significant cancer in targeted lesions (OR 2.25-8.78; P values <0.05; AUROC 0.84, 95% CI 0.81-0.87). Increases in PSAD were also associated with upgrade on systematic biopsy (OR 2.39-2.48; P values <0.05; AUROC 0.69, 95% CI 0.64-0.73). Conclusions ADC values and PSAD improve characterization of PI-RADS v2 score 4 or 5 lesions. Upgraded on systematic biopsy is slightly more likely with PSAD ≥0.15 and multiple small PI-RADS v2 score 3 or 4 lesions.