Magnetic Resonance Imaging–Based Prostate Cancer Screening: Is High-Value Care Achieved or Does the Holy Grail Remain Elusive?

Boesen and colleagues¹ assess a novel diagnostic triage strategy for prostate cancer detection using biparametric magnetic resonance imaging (bpMRI). This strategy could improve value by reducing morbidity and mortality in prostate cancer, decreasing unnecessary and harmful prostate biopsies, curtailing overdiagnosis and overtreatment, and lowering health care costs—the Holy Grail of prostate cancer care. Their approach may mitigate physical, psychological, and financial harms of prostate cancer detection and treatment strategies that rely on widespread prostate-specific antigen (PSA) testing, blind systemic biopsies for men suspected of having prostate cancer based on PSA results and/or abnormal digital rectal examination results, and treatment of most identified cancers.

Higher-value strategies are needed for several reasons. Prostate cancer is common, potentially deadly, and costly. However, most screen-detected prostate cancers, even those defined as clinically significant prostate cancer, will remain asymptomatic if untreated—yet nearly all are treated. Diagnostic biopsies and treatments have harms that can have a negative impact on patients’ health. Furthermore, widespread PSA testing has increased the number of men undergoing evaluation, subsequently being diagnosed with prostate cancer, and being treated for prostate cancer.² The widening array of diagnostic and treatment options has complicated approaches without clearly improving health. Thus, the words of Willet Whitmore, MD, spoken more than 35 years ago remain even more relevant today: “Is cure possible? Is cure necessary? Is cure possible only when it is not necessary?”³ and “The current state of prostate cancer may not be good medicine but it sure is good business.”³

Boesen and colleagues¹ conducted a single-center prospective cohort study to assess the diagnostic accuracy of bpMRI. Biparametric MRI offers the advantage of a relatively short imaging time vs the widely used multiparametric MRI and does not require intravenous contrast administration. Their study included 1020 men having clinical suspicion of prostate cancer, based on a PSA level of 4 ng/mL or higher (to convert to micrograms per liter, multiply by 1.0) and/or abnormal digital rectal examination results, who had not previously undergone a biopsy. The MRI results were interpreted by 1 experienced radiologist using the Prostate Imaging Reporting and Data System version 2 criteria on a scale ranging from 1 (prostate cancer highly unlikely) to 5 (prostate cancer highly likely). Patients underwent standard systematic prostate biopsies under ultrasonography guidance by a physician blinded to bpMRI results. Additional biopsies were performed for lesions with a Prostate Imaging Reporting and Data System score of 3 to 5 and for suspicious areas based on ultrasonography findings. The reference standard was the combined biopsy results of standard systematic biopsies and bpMRI-targeted biopsies. Their primary definition of clinically significant prostate cancer was any core with a Gleason score of 4 + 3 prostate cancer or a maximum cancer-core length greater than 50% with a Gleason score of 3 + 4 prostate cancer. The bpMRI findings had a negative predictive value of 97% (95% CI, 95%-99%), would allow 305 of 1020 men (30%) to avoid unnecessary biopsies, and decrease insignificant prostate cancer diagnoses by 40% (173 vs 288 men) compared with standard biopsies. Boesen and colleagues recommended bpMRI as a risk stratification tool to exclude aggressive disease and unnecessary biopsies.

While methodologically rigorous and transparently reported, several issues deserve mention. First, Boesen and colleagues used a combination of bpMRI-targeted plus standard biopsies as the reference standard, rather than template-guided prostate mapping biopsy with 5-mm increments as used in the PROMIS study by Ahmed et al.⁴ Therefore, the reported diagnostic accuracy likely exaggerates the true diagnostic accuracy of bpMRI. Second, while the authors placed great emphasis on the sensitivity (98%) and negative predictive value (97%), it is important to consider the less impressive specificity (48%) and positive predictive values (56%). Based on the authors’ findings, we calculated that for every 1000 men who underwent bpMRI, 311 patients would be told they have a lesion concerning for prostate cancer, increasing their anxiety. Third, their results heavily depend on prostate cancer prevalence, which was extraordinarily high (40% rate of clinically significant disease). A patient more typical of US clinical practice (64-year-old white man with a PSA level of 6.4 ng/mL, a negative family history for prostate cancer, and negative digital rectal examination results) would only have a 9% preprobability of harboring high-risk prostate cancer based on a widely used risk calculator.⁵ In this situation, if 1000 at-risk men were tested, 559 men would have suspicious MRI findings, but a prostate biopsy would rule out high-risk disease in 471 of these men. Therefore, bpMRI is a poor diagnostic test for ruling in clinically significant prostate cancer. Fourth, because studies were interpreted by 1 experienced radiologist, the lack of information on generalizability or interobserver agreement is problematic given MRI interpretation variability.⁶ Fifth, the natural history of and outcomes due to interventions based on MRI-detected clinically significant disease are not known. Recent randomized treatment trials demonstrated low long-term prostate cancer mortality and no mortality differences with observation or PSA-based active monitoring vs radical intervention among men with low-risk disease determined by biopsies obtained without MRI.^2,7,8 Among these men, MRI evaluation would uncover and falsely label these low-risk tumors as clinically significant, increasing the reservoir of latent tumors classified as higher grade and undergoing unnecessary early radical interventions. Even higher-risk tumors detected in treatment trials likely have a worse natural history than MRI-detected cancers of similar histologic grade.

These issues as well as the need for further external validation of the findings of Boesen and colleagues and others with multiparametric MRI notwithstanding, the rise of modern MRI is a potentially positive development, offering at-risk men the opportunity to decrease unnecessary and harmful biopsies.^6,9 More importantly, it may lessen overdiagnosis and overtreatment. However, for MRI-based triaging strategies to be cost-effective, patients and physicians must be willing to change management based on MRI findings (ie, decrease additional tests and treatments) rather than using MRI as an add-on to identify, upgrade, and treat more patients with prostate cancer. Unfortunately, history demonstrates that strategies with similar operating characteristics to bpMRI and multiparametric MRI such as free PSA, PSA velocity or density, PSA isoforms, other urine and serum markers, and risk calculators have not decreased biopsies, radical interventions, or costs. The use of MRI will result in low-value rather than high-value care if patients with negative MRI findings, such as a lesion with a Prostate Imaging Reporting and Data System score of 2, still undergo a biopsy or active surveillance imaging and if all men with MRI-defined clinically significant disease undergo radical intervention.

The relative merits of an MRI-based triage system must be placed into context with the prostate cancer screening and treatment debate that involves balancing small to no long-term mortality benefits against the more frequent, early, and sometimes persistent testing and treatment harms. The recently published UK-based cluster-randomized screening trial,¹⁰ which failed to demonstrate a mortality decrease but led to increased detection of indolent cancers from 1-time PSA-based screening, as well as the PROTECT⁷ and PIVOT⁸ treatment trials showing no decrease in long-term overall or prostate cancer mortality in men treated with radical interventions further underscore this issue. Current US prostate cancer screening and treatment practice involving widespread screening, biopsies, and early treatment is low-value care. For many men, higher health care value would be achieved by screening and treating fewer men rather than screening widely with routine follow-up imaging triage techniques. However, given the high health impact and strong financial incentives surrounding prostate cancer, screening and treatment enthusiasm is likely to continue. While we are concerned that the Holy Grail remains elusive, wise implementation of less invasive strategies, such as described by Boesen and colleagues, would lead to high-value prostate cancer care.

Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2018 Wilt TJ et al. JAMA Network Open.

Corresponding Author: Timothy J. Wilt, MD, MPH, Minneapolis Veterans Affairs Health Care System, University of Minnesota School of Medicine, One Veterans Drive (111-0), Minneapolis, MN 55417 (tim.wilt@va.gov).

Conflict of Interest Disclosures: Dr Wilt is chairman of the Veterans Affairs/National Cancer Institute/Agency for Healthcare Research and Quality Cooperative Program Study 407, Prostate Cancer Intervention Versus Observation Trial (PIVOT), referenced herein. Dr Dahm is the coordinating editor of Cochrane Urology.