Multiparametric MRI offers significant potential for prostate cancer detection and surveillance. The existing literature recommends its use as a triage test and supports its accuracy in delineating clinically significant from insignificant prostatic disease [4, 5, 10, 11]. The PROMIS study previously demonstrated mpMRI-guided biopsy to be more accurate than TRUS biopsy alone in PCa detection, with sensitivities of 93% and 48% respectively [4]. It further indicated a negative predictive value of 89% for clinically significant PCa, defined as ISUP GG ≥ 2. Multicentre randomized controlled trials, including the PRECISION trial, have suggested that up to a quarter of prostate biopsies could safely be omitted if MRI is used as a triage tool [5, 12]. In addition, the use of MRI-targeted biopsy in conjunction with systematic biopsy has been shown to have superior diagnostic potential compared to systematic biopsy alone [6]. These overall benefits of mpMRI in the prostate cancer diagnostic pathway have been highlighted in a Cochrane meta-analysis [11].
Our data support the existing literature in recommending mpMRI as a tool to guide decision-making around prostate biopsy. In our prior published cohort only 2.6% of men with PI-RADS 2 changes on mpMRI were found to have significant cancer, compared to 53.6% and 91.7% of those with PI-RADS 4 or 5, respectively [9]. In the present series, while 15% of patients with a negative mpMRI revealed clinically significant cancer, cases of high-grade disease (defined as ISUP GG ≥ 4) were rare (4%). With recent studies highlighting low rates of csPCa on follow-up in patients with negative mpMRI, risk-stratification and PSA surveillance may be an appropriate alternative to routine biopsy in these patients [10, 13]. Avoiding unnecessary biopsy reduces the risk of procedure-related complications, including life-threatening sepsis requiring hospitalisation [14–16]. Lastly, while it has yet to be comprehensively investigated in Australia, several international studies have suggested through modelling that the mpMRI pathway may be equivalent in cost-effectiveness compared to systematic biopsy alone [17–19]. This is largely attributed to fewer biopsies and associated morbidity, leading to reduced overdiagnosis and overtreatment. As such, adherence to a policy that avoids biopsy in men with negative MRI may provide benefits to both patients and service providers.
This rationale formed the basis of the Australian Government’s decision to introduce a Medicare rebate for prostate MRI in July 2018 (see Online Resource 1)[8]. Yet despite the benefits, a unit recommendation to omit or defer prostate biopsy in men with negative MRI was not adopted by our clinicians. The proportion of patients undergoing TRUS with negative MRI actually increased, and not surprisingly there was no improvement in the cancer pickup rate. In fact, cancer pickup rates were better prior to the Medicare rebate and routine use of mpMRI.
One could speculate that this data reveals an ingrained precautionary bias of clinicians to err on the side of caution when dealing with potential cancer diagnoses, despite clinical, serological, and radiological findings. Such behaviour remains contrary to the numerous studies showing the value of mpMRI in detecting and localising PCa [4, 5, 11], and highlights the challenge of trying to change clinical practice. Alternatively, clinicians may remain unconvinced about the negative predictive value of mpMRI, despite high-quality evidence in the form of the PROMIS and PRECISION trials. We recall that these studies were performed at centers of excellence with formidable radiology facilities, personnel, and expertise. As variability in imaging at institutions can affect detection rates [20], concerns regarding the quality of imaging and reporting may have impacted adherence to unit recommendations. This would have been despite our institution previously confirming the accuracy of its mpMRI service during the post-Medicare rebate introductory phase [9]. Finally, the eligibility criteria for the Medicare rebate remain quite onerous, having been designed to limit mpMRI to those men with a reasonable likelihood of a cancer (see Online Resource 1). Perhaps clinicians have begun to overestimate their findings on digital rectal examination, again from an abundance of caution, thereby fast-tracking mpMRI and biopsy for their patients. Regardless of the cause, our data carries the uncomfortable implication that prostate mpMRI has been introduced into the Australian healthcare system, at considerable taxpayer expense, without any tangible benefit.
Our study carries several limitations. Firstly, a consequence of retrospective data collection is that clinical decision-making cannot always be accurately determined nor inferred. Secondly, the study was conducted at a single institution. Despite a multicultural demographic, it remains unclear whether any selection bias existed within our population. The inclusion of nineteen patients undergoing active surveillance alongside the predominantly biopsy-naïve cohort also adds to heterogeneity. It was deemed inappropriate to exclude this subset in this series, as clinicians may consider negative mpMRI as a reason to forego biopsy in the cohort of patients receiving surveillance. Furthermore, assessment of a population without exclusions assesses the real-world application of mpMRI in influencing clinical decision-making. Moreover, implementation of a multivariable regression model with six adjustment variables may be overfitted due to the low event rate (15.3%) for clinically significant prostate cancer. This limits meaningful interpretation of these results. These factors combined reduce the likelihood of acquiring statistically significant variables from this population. However, one must note that such heterogeneity and low likelihood of clinically significant PCa often underlie clinician bias against negative mpMRI findings. While it has not been demonstrated locally, bias against mpMRI reliability may stem from a clinician’s experience with imaging in community practice compared to the institutional setting [21].
In addition, we did not investigate the role of PSA kinetics nor PSA density in this study, despite the latter’s proposed utility in helping risk stratify patients with a negative mpMRI [22, 23]. PSA density was not routinely cited by our clinicians as an aid to their decision-making, and its inclusion would warrant exclusion of more commonly-cited collinear variables – PSA and prostate volume – from the multivariable analysis. PSA kinetics was also not investigated as it would contribute to further heterogeneity without enabling meaningful comparison given patients had a variable number of PSA measurements prior to decision-to-biopsy (e.g. some patients recorded limited PSA measurements prior to biopsy compared to other patients who received regular PSA surveillance spanning several years).
We were unable to determine the total number of patients referred for mpMRI, nor determine the differences in patients with negative mpMRI who did and did not undergo biopsy. In our two-tier public-private health system, patients have the option of using the public health system or seeing individual practitioners privately. Patients requiring mpMRI at our institution are referred to an external private imaging company (Lumus Pty Ltd) that has a contract with our public hospital system. As the data is coded by the private imaging company according to referring specialist rather than private/public source, we could not accurately differentiate whether each mpMRI had been referred by the public hospital outpatient clinic or by specialists reviewing patients in the private sector. Furthermore, as a public hospital provider, we were unable to access the data of patients seen in the private sector, or follow-up biopsies performed in private hospitals. Our specialists see patients across multiple public and private hospital settings and in different geographic locations, so our only insight into their practice is limited by what they display at our institution. Given that our public hospital biopsy numbers remained relatively constant over the extended study period, this population appeared to be a satisfactory cohort to investigate the question of adoption of mpMRI at the institutional level.
Lastly, although a single institution study ensures consistency in MRI quality, acquisition and reporting, variability in these factors across institutions may hinder the reproducibility of these results. Acknowledging these limitations, one must recall that our aim was to assess outcomes following the routine introduction of rebatable mpMRI and recommendations to omit or defer biopsy in men with negative studies. Decision-making at a broader level with regards to mpMRI and omission of biopsy, with stratification of biopsy-naïve and active surveillance cohorts, may be assessed with further population-based analyses.