Could Baseline MRIs Reshape Prostate Cancer Risk Assessment?

Adding baseline MRI to conventional prostate cancer risk stratification could improve prognostic accuracy, potentially affecting active surveillance and treatment decisions for some patients, according to the investigators on a new trial.
The multicenter, real-world trial showed that men with low-risk or favorable intermediate-risk disease who had higher Prostate Imaging Reporting and Data System (PI-RADS) scores at baseline were more likely to be reclassified with more aggressive disease on a future biopsy, wrote lead author Kiran R. Nandalur, MD and colleagues. 
This means that without MRI, some cases of prostate cancer are being labeled as lower-risk than they actually are.
Writing in The Journal of Urology, the investigators noted that MRI is increasingly being used to choose patients who are appropriate for active surveillance instead of treatment, but related clinical data are scarce.
Although PI-RADS is the preferred metric for characterizing prostate tumors via MRI, “most previous studies on the prognostic implications of baseline PI-RADS score included smaller populations from academic centers, limited inclusion of clinical and pathologic data into models, and/or [are] ambiguous on the implications of PI-RADS score,” they wrote.
These knowledge gaps prompted the present study.
How Were Baseline MRI Findings Related to Prostate Cancer Disease Risk?
The dataset included 1491 men with prostate cancer that was diagnosed at 46 hospital-based, academic, or private practice urology groups. All had low-risk or favorable intermediate-risk disease and had undergone MRI within 6 months before or after initial biopsy, along with enrollment in active surveillance.
“A novel aspect of this study was that the MRIs were not read by dedicated prostate MRI experts at academic institutions, but rather a mix of community and academic radiologists,” said Nandalur, medical director of Corewell Health East Radiology, Royal Oak, Michigan, in an interview.
After traditional risk factors were accounted for, baseline PI-RADS (four or more lesions) was significantly associated with increased likelihood of biopsy reclassification to high-grade prostate cancer on surveillance biopsy (hazard ratio, 2.3; 95% CI 1.6-3.2; P < .001). 
“These patients with suspicious lesions on their initial MRI were more than twice as likely to have higher-grade disease within 5 years,” Nandalur noted. “This result was not only seen in the low-risk group but also in the favorable intermediate-risk group, which hasn’t been shown before.”
Grade group 2 vs 1 and increasing age were also associated with significantly increased risk for reclassification to a more aggressive cancer type.
How Might These Findings Improve Outcomes in Patients With Prostate Cancer?
Currently, 60%-70% of patients with low-risk disease choose active surveillance over immediate treatment, whereas 20% with favorable intermediate-risk disease choose active surveillance, according to Nandalur.
For low-risk patients, PI-RADS score is unlikely to change this decision, although surveillance intervals could be adjusted in accordance with risk. More notably, those with favorable intermediate-risk disease may benefit from considering PI-RADS score when choosing between active surveillance and immediate treatment.
“Most of the management strategies for prostate cancer are based on just your lab values and your pathology,” Nandalur said, “but this study shows that maybe we should start taking MRI into account — into the general paradigm of management of prostate cancer.”
Ideally, he added, prospective studies will confirm these findings, although such studies can be challenging to perform and similar data have historically been sufficient to reshape clinical practice.
“We are hoping that [baseline PI-RADS score] will be adopted into the NCCN [National Comprehensive Cancer Network] guidelines,” Dr. Nandalur said.
How Likely Are These Findings to Reshape Clinical Practice?
“The study’s large, multicenter cohort and its focus on the prognostic value of baseline MRI in active surveillance make it a crucial contribution to the field, providing evidence that can potentially refine patient management strategies in clinical practice,” said Ismail Baris Turkbey, MD, FSAR, head of MRI Section, Molecular Imaging Branch, National Cancer Institute, Rockville, Maryland, in a written comment.
“The findings from this study are likely to have a significant impact on clinical practice and potentially influence future guidelines in the management of localized prostate cancer, particularly in the context of active surveillance,” Turkbey said. “MRI, already a commonly used imaging modality in prostate cancer management, may become an even more integral part of the initial assessment and ongoing monitoring of patients with low or favorable-intermediate risk prostate cancer.”
Turkbey noted several strengths of the study.  
First, the size and the diversity of the cohort, along with the variety of treatment centers, support generalizability of findings. Second, the study pinpoints a “critical aspect” of active surveillance by uncovering the link between baseline MRI findings and later risk reclassification. Finally, the study also showed that increasing age was associated with higher likelihood of risk reclassification, “further emphasizing the need for personalized risk assessment” among these patients.
What Were Some Limitations of This Study?
“One important limitation is the lack of inter-reader agreement for PI-RADS evaluations for baseline MRIs,” Turkbey said. “Variation of PI-RADS is quite known, and centralized evaluations could have made this study stronger. Same applies for centralized quality evaluation of MRIs using The Prostate Imaging Quality (PI-QUAL) score. These items are difficult to do in a multicenter prospective data registry, and maybe authors will consider including these additional analyses in their future work.” 
How Does This New Approach to Prostate Cancer Risk Assessment Compare With Recent Advances in AI-Based Risk Assessment?
Over the past few years, artificial intelligence (AI)–assisted risk assessment in prostate cancer has been gaining increasing attention. Recently, for example, Artera, a self-styled “precision medicine company,” released the first AI tool to help patients choose between active surveillance and active treatment on the basis of analysis of digital pathology images. 
When asked to compare this approach with the methods used in the present study, Nandalur called the AI model “a step forward” but noted that it still relies on conventional risk criteria.
“Our data show imaging with MRI has independent prognostic information for prostate cancer patients considering active surveillance, over and above these traditional factors,” he said. “Moreover, this predictive ability of MRI was seen in low and favorable intermediate risk groups, so the additive value is broad.”
Still, he predicted that the future will not involve a binary choice, but a combination approach.
“The exciting aspect is that MRI results can eventually be added to this novel AI model and further improve prediction models for patients,” Nandalur said. “The combination of recent AI models and MRI will likely represent the future paradigm for prostate cancer patients considering active surveillance versus immediate treatment.”
The study was supported by Blue Cross and Blue Shield of Michigan. The investigators and Turkbey reported no conflicts of interest.
 
Send comments and news tips to [email protected].