TL;DR
This study investigates the role of dynamic contrast enhancement (DCE) in prostate MRI for detecting clinically significant prostate cancer (csPCa). While DCE can help identify higher-risk lesions, the research suggests that biparametric MRI (bpMRI) performs comparably to multiparametric MRI (mpMRI) in nomogram-based risk prediction, making bpMRI a potentially more accessible alternative.
What Problem Does This Paper Address?
Prostate cancer screening relies heavily on MRI, particularly multiparametric MRI (mpMRI), to detect clinically significant prostate cancer (csPCa) and avoid unnecessary biopsies. However, mpMRI is expensive, time-consuming, and requires specialized resources, creating disparities in access to care. Biparametric MRI (bpMRI), which omits the dynamic contrast enhancement (DCE) sequence, offers a potentially more accessible and cost-effective alternative. The challenge lies in understanding if omitting DCE impacts diagnostic accuracy, especially for borderline lesions, and whether bpMRI can reliably replace mpMRI in clinical decision-making. This paper aims to quantify the added value of DCE and compare the performance of bpMRI and mpMRI nomograms for predicting csPCa.
Methods
The researchers conducted a retrospective analysis of biopsy-naïve patients who underwent pre-biopsy MRI. They evaluated the association of DCE with csPCa detection by comparing risk stratification based on bpMRI, mpMRI (using PI-RADS v2.1), and a fully DCE-stratified approach. Specifically, they analyzed lesions in the peripheral zone (PZ) and categorized them based on PI-RADS scores from diffusion-weighted imaging (DWI) and DCE.
To assess the overall utility of bpMRI versus mpMRI, they developed and validated nomograms (predictive models) for csPCa detection. These nomograms incorporated clinical factors, serum biomarkers (like Prostate Health Index - PHI, % free PSA, total PSA), and PI-RADS scores derived from both bpMRI and mpMRI approaches. The performance of these nomograms was compared using receiver operating characteristic (ROC) curves and decision curve analysis.
Datasets
The study utilized a retrospective cohort of biopsy-naïve patients who underwent pre-biopsy MRI between March 2018 and February 2022 from the Northwestern Medicine healthcare system.
- Initial Analysis Cohort: 1010 patients with the highest PI-RADS lesion in the peripheral zone (PZ).
- Nomogram Development Cohort: 1410 patients who received mpMRI for elevated prostate-specific antigen (PSA).
- Nomogram Validation Cohort: 353 patients.
The paper does not introduce a new benchmark dataset but rather uses existing patient data for analysis and validation.
Key Insights and Ideas
- DCE's Role in Lesion Stratification: While DCE alone might not be the sole determinant, its positivity is associated with a higher likelihood of csPCa, particularly for PI-RADS 3 and 4 lesions. This suggests DCE can refine risk assessment for specific lesion categories.
- bpMRI's Comparable Performance: The study's most significant finding is that bpMRI-based nomograms perform comparably to mpMRI-based nomograms in predicting csPCa. This implies that the added information from DCE might not be critical for overall risk prediction when using nomograms.
- Accessibility Advantage of bpMRI: The comparable performance of bpMRI, coupled with its reduced cost and time, makes it a strong contender for wider adoption, potentially improving access to prostate cancer screening.
- PI-RADS 3+ Lesions: The study highlights the importance of PI-RADS 3+ lesions (DWI 3 with positive DCE, upgraded to PI-RADS 4 by mpMRI) as being associated with higher csPCa rates compared to PI-RADS 3- lesions.
Key Results
- csPCa Detection Rates: csPCa detection rates varied significantly with PI-RADS scores and DCE status. For example, PI-RADS 3- lesions had a 19% csPCa rate, while PI-RADS 5+ lesions had a 92.3% rate.
- DCE Impact on PI-RADS 3: PI-RADS 3+ lesions (DWI 3 with positive DCE) were associated with significantly higher csPCa rates (OR 1.86) compared to PI-RADS 3- lesions (DWI 3 with negative DCE).
- Nomogram Performance: In both the development and validation cohorts, bpMRI nomograms showed no significant difference in performance compared to mpMRI nomograms. Area Under the Curve (AUC) values for both approaches were consistently high (ranging from 0.900 to 0.931).
- DCE Stratification Benefit: A fully DCE-stratified scoring system showed a marginal improvement in discrimination for patients with the highest PI-RADS lesions in the PZ compared to standard mpMRI PI-RADS.
Limitations and Future Work
- Retrospective Design: The study's retrospective nature is a limitation, and the authors acknowledge the need for prospective studies.
- Reader Experience: The study did not blind readers to the DCE series, and the authors note that reader experience can affect diagnostic performance. They suggest that healthcare systems adopting bpMRI will need robust quality control and patient recall protocols.
- External Validation: While the nomograms were validated on an independent institutional cohort, further external validation across diverse populations and reader groups is recommended to assess generalizability.
- DCE Utility in Specific Scenarios: The authors suggest that DCE might still hold utility for patients with intermediate risk of csPCa, those with unexplained persistent PSA elevations, or those undergoing active surveillance, as it could improve reader confidence and guide biopsy decisions in complex cases.
The authors propose that bpMRI can be a reasonable alternative to mpMRI and may be considered for wider adoption in the United States, especially given its potential to improve accessibility.
Summary updated