MRI Use After Prostatectomy Varied Dramatically by Hospital
Postoperative MRI ordering after prostatectomy for prostate cancer was far from uniform across seven hospitals in a multicenter Korean cohort. In a retrospective study of 3,754 patients, repeated MRI examinations clustered heavily at some institutions and remained sparse at others, even after accounting for biochemical recurrence and several clinicopathologic variables. The spread was striking: compared with Hospital 1, the average hazard ratios for repeated MRI use at Hospitals 2 through 7 ranged from 4.083 to 23.136, all with p values below 0.001.
One result stood out immediately: institutional practice patterns were a stronger driver of MRI use than tumor grade or pathologic T stage.
During a median follow-up of 936 days, with an interquartile range of 432 to 1,687 days, clinicians recorded 2,027 postoperative MRI examinations. The study used an Andersen-Gill counting-process Cox model with patient-clustered robust variance, a design well suited to recurrent events such as repeated imaging. A complementary Poisson model examined temporal patterns and helped confirm that the institutional signal persisted. Even with those analytic safeguards, the hospital effect remained large and statistically robust.
Biochemical Recurrence Increased Imaging, but Did Not Explain the Full Pattern
Biochemical recurrence, defined as postoperative prostate-specific antigen of at least 0.2 ng/mL, increased the likelihood of MRI use. The hazard ratio for imaging after biochemical recurrence was 1.565, with a 95% confidence interval of 1.371 to 1.786 and p below 0.001. Clinically, that fits the expected pathway: once PSA begins to rise, MRI becomes a plausible tool for localizing recurrent disease, assessing the prostate bed, or informing salvage treatment planning.
Yet the recurrence signal did not erase the institutional differences. Hospitals continued to diverge widely in how often they ordered MRI, and the magnitude of those differences shifted over time after surgery. The hospital-by-time interaction was significant, with p below 0.001, indicating that utilization was not only variable across centers but also dynamically variable across the postoperative course. In practical terms, the same patient profile could trigger very different imaging trajectories depending on where care was delivered.
Grade group and pathologic T stage did not show significant associations with repeated MRI use. That absence is important because it suggests that postoperative MRI was not being driven primarily by the pathology variables one might expect to anchor imaging decisions. In other words, clinicians were not simply imaging the highest-risk tumors more often; local custom and institutional thresholds appear to have played a major role.
Imaging Intensity Fell Sharply After Year Five
MRI use did not remain steady over long-term surveillance. The annual MRI rate dropped from 18.08 per 100 person-years in postoperative year 5 to 5.94 in year 6. That decline marks a clear inflection in follow-up behavior and suggests that clinicians either became less reliant on MRI as time passed or shifted toward less intensive surveillance once patients moved farther from surgery.
The timing of that decline matters because postoperative surveillance after prostatectomy is often front-loaded, especially when PSA kinetics raise concern for early recurrence. A pronounced reduction after year 5 may reflect a transition from active recurrence workup to more selective, symptom- or PSA-triggered imaging. It may also reflect institutional comfort with de-escalation once patients remain stable beyond the early recurrence window.
These patterns are best understood against the biology of prostate cancer recurrence. PSA rise after prostatectomy can precede radiographic findings, and MRI is often used to clarify local anatomy when salvage treatment is under consideration. Still, the study shows that imaging behavior is shaped as much by local policy and practice culture as by disease biology alone.
Overall Survival Did Not Track With Cumulative MRI Use
The exploratory survival analysis did not show a significant association between cumulative MRI use and overall survival. The hazard ratio was 0.782, with a 95% confidence interval of 0.564 to 1.086 and p value of 0.142. That finding should not be read as evidence that MRI lacks value; rather, it indicates that, in this dataset, more imaging did not translate into a measurable survival difference.
Several explanations are possible. More frequent MRI may simply identify recurrence earlier without changing the downstream biology of the disease. Imaging intensity may also be a marker of local practice patterns rather than a causal determinant of survival. The study was retrospective, so treatment selection, salvage therapy timing, and unmeasured clinical preferences could all influence the relationship between imaging and outcomes.
A second point is equally important: survival was not the endpoint that best captured the study’s main signal. The strongest result was utilization variability, not efficacy. The data describe how MRI is used after prostatectomy, where it concentrates, and how those patterns shift over time. They do not establish that more MRI is better.
Standardization Gaps Remain the Central Clinical Issue
Across seven hospitals, postoperative MRI use varied substantially, and measured clinical factors explained only part of that divergence. The study’s central message is not about the presence or absence of MRI itself, but about the lack of consistency in when it is ordered after prostatectomy. Large interhospital differences raise the possibility that patients receive markedly different follow-up intensity depending on local habits rather than a shared evidence-based threshold.
For urologists, oncologists, and radiologists, the finding points toward a familiar problem in postoperative cancer care: surveillance often evolves through institutional convention before it hardens into routine. MRI may be deployed early in some centers and sparingly in others, even for similar postoperative scenarios. Without clearer criteria, repeat imaging after prostatectomy risks becoming a reflection of geography rather than biology.
The study also leaves open a practical question that matters at the bedside: when should MRI be used after prostatectomy, and at what PSA threshold or recurrence pattern does it add the most value? The current data show recurrence-related escalation and a steep institutional gradient, but they stop short of defining an optimal imaging strategy. For now, the evidence argues for more standardized, evidence-based criteria rather than broader or more frequent scanning by default.