ProstaPilot: A Comparative Study of Biparametric Magnetic Resonance Imaging Versus Prostate-specific Antigen as a Screening Test for Prostate Cancer

Take Home Message

Our results demonstrate that biparametric magnetic resonance imaging of the prostate is accurate in diagnosing clinically significant prostate cancer (csPC) even in a highly prescreened population. This strategy could reduce the number of biopsies needed and potentially yield higher csPC detection rates in comparison to screening via prostate-specific antigen measurement.

Abstract

Background and objective

Our aim was to compare detection rates for clinically significant prostate cancer (csPC) between biparametric magnetic resonance imaging (bpMRI) and prostate-specific antigen (PSA) screening pathways in the Czech population, which has a high prevalence of opportunistic PSA testing (45% in the target age group).

Methods

Between May 2022 and May 2023, 423 men aged 50–69 yr who were enrolled via volunteer-based recruitment, underwent both PSA testing and bpMRI of the prostate. Positive results were defined as PSA ≥3 μg/l or a Prostate Imaging-Reporting and Data System score of 4-5. Those with positive tests were referred for further evaluation, including an MRI-targeted biopsy and a systematic 12-core biopsy. csPC was defined as any cancer of International Society of Urological Pathology grade group ≥2.

Key findings and limitations

Of the 423 men, 35 (8.3%) had at least one positive screening test. The PSA-based pathway identified 25 men (5.9%), whereas the MRI-based pathway identified 16 (3.8%) with suspicion of PC, with potential to reduce the number of biopsies required by 36%. Subsequent biopsy revealed PC in seven men (1.7%) with a positive PSA test and 14 men (3.3%) with positive MRI findings. Nonsignificant PC and csPC rates were 1.0% (n = 4) and 0.7% (n = 3) for the PSA pathway, and 1.7% and 1.7% (n = 7) for the MRI pathway, respectively.

Conclusions and clinical implications

In a highly prescreened population, bpMRI may increase the csPC detection rate while reducing the number of biopsies needed in comparison to PSA-based diagnosis. However, bpMRI may also slightly increase the detection of grade group 1 cancers.

Patient summary

We compared a blood test for PSA (prostate-specific antigen) and an MRI scan (magnetic resonance imaging) for detection of prostate cancer in men aged 50–69 years. We found that MRI scans may reduce the number of biopsies that need to be performed for cancer diagnosis while increasing the detection rate for cancers that are significant.

1. Introduction

Prostate cancer (PC) screening has traditionally relied on prostate-specific antigen (PSA) testing followed by systematic prostate biopsy, as exemplified by outcomes from the ERSPC trial on screening for PC, which demonstrated a 25% reduction in PC mortality after 16 yr of follow-up for men who attended at least one screening round [1]. However, PSA-based pathways are associated with significant harms, including excessive biopsy rates, overdiagnosis, and overtreatment of nonsignificant PC (nsPC) [2,3].

Magnetic resonance imaging (MRI) is now commonly used in the prebiopsy setting and reduces biopsy rates while detecting more high-grade tumors and fewer nsPC cases in comparison to PSA testing [4,5]. This benefit has also been observed in MRI-based screening protocols [6]. MRI introduction into screening algorithms may reduce the harms associated with the PSA-based systematic biopsy approach to screening [3,7]. While these outcomes suggest that MRI may enhance the effectiveness of PC screening pathways, further research is needed to evaluate its role in modern, population-based screening strategies.

The aim of our study was to assess the role of biparametric MRI (bpMRI) in screening for clinically significant PC (csPC) by comparing detection rates between MRI-based and PSA-based pathways in the Czech population, which has a high prevalence of opportunistic PSA testing [8].

2. Patients and methods

ProstaPilot is a prospective study that evaluated a bpMRI protocol for csPC screening in men from the general population. All participants provided informed consent. The project was approved by the institutional ethics committee (reference 2022/1303/MOU), and the study protocol was registered on ClinicalTrials.gov (NCT05603351). The study protocol is included in the Supplementary material. The study followed the Standards for Reporting Diagnostic Accuracy Studies (STARD) guidelines [9].

2.1. Study recruitment

Between May 2022 and May 2023, we enrolled men aged 50–69 yr with life expectancy of at least 10 yr in the study. Recruitment was conducted by the Cancer Prevention Unit of Masaryk Memorial Cancer Institute in collaboration with local general practitioners via a volunteer-based process. All men attended planned cancer prevention visits and were offered participation in the study. Men with a history of hip replacement or a known BRCA1/BRCA2 mutation were excluded. Participants were required to confirm that they had not undergone prostate biopsy, prostate MRI, or a PSA test in the previous 2 yr.

2.2. Interventions

All patients underwent a PSA test and bpMRI of the prostate. Positive findings were defined as a PSA level ≥3 μg/l or identification of a Prostate Imaging-Reporting and Data System (PI-RADS) 4–5 lesion on MRI. If either test was positive, the patient was referred to a urologist for a digital rectal examination and prostate biopsy. PSA density (PSAD) was also calculated. For participants with PSA <3 μg/l, PSAD >0.15 μg/l/cm³ was also considered a positive result that triggered prostate biopsy.

2.2.1. MRI acquisition protocol and reporting

Each participant received a leaflet with instructions to follow before their MRI scan, which included avoidance of prostate-irritating activities for 2 d and food guidelines (Supplementary material). Patients were advised to empty their rectum before the scan, and if gas was detected during the initial MRI sequence, they were asked to repeat this to improve image clarity. All examinations in this study used a 3-T MRI scanner (Magnetom Prisma, Siemens). The bpMRI protocol included T2-weighted sequences in three planes, diffusion-weighted imaging with b values of 50, 800, and calculated 1400, and an apparent diffusion coefficient map, all in accordance with PI-RADS version 2.1 guidelines [10].

MRI scans were independently evaluated by two experienced uroradiologists according to PI-RADS version 2.1. PI-RADS scores of 4 or 5 were considered positive, and scores of 1–3 were considered negative. Disagreements were resolved by a third radiologist. At the time of study initiation, the radiologists involved (M.S., M.P., and J.K.) had 4 yr of experience in reporting prostate MRI and had reported results for between 400 and 1000 scans. Image quality was assessed by the first radiologist using a PI-QUAL scoring system [11]; a score from 1 to 4 was assigned, as a score of 5 was not possible without a dynamic contrast-enhanced sequence (Supplementary material).

2.2.2. Prostate biopsy

Men with positive PSA or bpMRI (PI-RADS 4–5) findings were referred for a prostate biopsy. All biopsy-eligible men underwent a transrectal ultrasound (TRUS)-guided systematic 12-core biopsy from the peripheral zone of the prostate. MRI-positive cases received an additional three-core MRI/TRUS fusion biopsy. All biopsy specimens were reviewed by an expert uropathologist.

2.3. Study endpoints

The primary objective was to assess the role of bpMRI in csPC screening for asymptomatic men. csPC was defined as International of Urological Pathology grade group (GG) ≥2, and nsPC as GG 1. The primary endpoints were the proportion of positive MRI findings (PI-RADS 4–5) in the general population of men aged 50–69 yr and PC detection rates in the MRI-based and PSA-based screening pathways.

The secondary endpoint was the degree of concordance between radiologists performing the MRI reviews. In addition, we retrospectively assessed whether the study cohort represented a true screening-naïve population by verifying the patients’ self-reported history of no PSA test in the previous 2 yr against their health insurance records, using data gathered from the National Institute of Health Information and Statistics.

2.4. Statistical analysis

A sample size estimate for the primary endpoint was calculated using PASS 13 software with the exact (Clopper-Pearson) confidence interval (CI) formula. On the basis of previous publications, we anticipated that 11% of participants would have positive findings. We calculated that a cohort of 450 men would provide a two-sided 95% CI with a width of 6 percentage points (approx. ±3 percentage points).

Basic descriptive statistics and frequency counts for positive PSA and MRI tests and their combinations were computed. The nonparametric McNemar test at the 0.05 significance level was used to compare the number of biopsies required according to the PSA or MRI strategy. In the case of the MRI test, Cohen’s κ coefficient was used to compare the evaluation concordance between two radiologists. The proportion of nsPC (GG 1) and csPC (GG ≥2) cases detected with each screening modality was also calculated. Statistical analysis was performed using SPSS version 28.0 (IBM Corporation, Armonk, NY, USA).

3. Results

A total of 463 men were enrolled in the study, of whom 423 met the inclusion criteria and were included in the final analysis (Fig. 1). Demographic data for the study population are presented in Table 1. Some 71% of the participants were aged 50–59 yr, and almost half had university education.

Fig. 1.

Diagram of the study inclusion process. GG = International Society of Urological Pathology grade group; MRI = magnetic resonance imaging; PSA = prostate-specific antigen.

Table 1.

Demographic data for the study cohort of 423 patients

Parameter	Result
Mean age, yr (median)	57.2 (56.2)
Age group, n (%)
50–54 yr	169 (40.0)
55–59 yr	133 (31.4)
60–64 yr	76 (18.0)
65–69 yr	45 (10.6)
Median PSA, μg/l (IQR)	0.87 (0.53–1.48)
PSA test within 2 yr before enrollment (%)	40
Mean prostate volume, cm3 (SD)	35.9 (16.7)
Education level, n (%)
Primary	17 (4.0)
Secondary	183 (43.3)
University	203 (48.0)
Unknown	20 (4.7)

IQR = interquartile range; PSA = prostate-specific antigen; SD = standard deviation.

3.1. PSA results

Median PSA was 0.87 μg/l (interquartile range 0.53–1.48), and mean PSA was 1.39 μg/l. A positive PSA result (≥3 μg/l) was observed for 25 participants (5.9%, 95% CI 4.0–8.6%). PSA <1 μg/l was observed for 238 participants (56.3%) and PSA 1–3 μg/l for 160 participants (37.8%). PSAD >0.15 μg/l/cm³ was observed in only three participants (0.7%), all of whom had PSA ≥3 μg/l.

3.2. bpMRI results

Analysis of the highest-grade lesion for each participant revealed PI-RADS scores of 1–3 for 407 men (96.2%; Table 2). Only 16/423 men (3.8%; 95% CI 2.3–6.1%) had a positive bpMRI finding (PI-RADS 4–5). On stratification by PSA, the rate of positive MRI findings was 8.6% (16/185 men) for the group with PSA ≥1 μg/l and 24.0% (6/25 men) for the group with PSA >3 μg/l. No positive MRI findings were observed for men with PSA <1 μg/l.

Table 2.

Distribution of PSA levels by MRI results

MRI result	Patients, n (%)
	PSA<1 μg/l	PSA 1–<3 μg/l	PSA ≥3 μg/l	Total
PI-RADS 1–3	238 (56.3)	150 (35.5)	19 (4.5)	407 (96.2)
PI-RADS 4–5	0 (0.0)	10 (2.4)	6 (1.4)	16 (3.8)
Total	238 (56.3)	160 (37.8)	25 (5.9)	423 (100)

MRI = magnetic resonance imaging; PSA = prostate-specific antigen; PI-RADS = Prostate Imaging-Reporting and Data System.

All MRI scans were double-read by two independent radiologists who were unaware of the PSA results. There was disagreement between the two radiologists for 11/423 cases regarding MRI assessment (positive vs negative). Cohen’s κ coefficient was 0.607, indicating moderate to strong concordance between the observers.

Quality assessment revealed suboptimal scores of PI-QUAL 1 for six examinations (1.4%) and PI-QUAL 2 for 18 (4.3%), and sufficient scores of PI-QUAL 3 in 75 cases (17.7%) and PI-QUAL 4 in 324 cases (76.6%).

3.3. PC detection rates via the PSA and MRI pathways

Out of 423 men, 35 (8.3%) had at least one positive screening test. PSA testing was positive (≥3 μg/l) in 25 men (5.9%). bpMRI was positive (PI-RADS 4–5) in 16 men (3.8%), which could reduce biopsies by 36% in comparison to PSA testing alone (p = 0.137). Six men had positive results for both tests (Table 3). Prostate biopsies were performed in 31 men, and four men declined. No adverse events related to MRI testing or prostate biopsy were observed. Prostate cancer was diagnosed in 17 cases (4.0%; Fig. 2), which was nsPC (GG 1) in ten men (2.4%) and csPC (GG ≥2) in seven (1.7%). Among the latter seven, GG 2 cancer was diagnosed in four men, GG 3 in one man, and GG 4 in two men.

Table 3.

Grade group distribution by test status for PSA and MRI

Test status	Negative	nsPC	csPC				Total
	biopsy	GG 1	GG 2	GG 3	GG 4	GG 5
PSA+ MRI−	12	3	0	0	0	0	19 a
PSA+ MRI+	2	1	2	1	0	0	6
PSA− MRI+	0	6	2	0	2	0	10
Total	14	10	4	1	2	0	31

MRI = magnetic resonance imaging; PSA = prostate-specific antigen; nsPC = nonsignificant prostate cancer; csPC = clinically significant prostate cancer; GG = International Society of Urological Pathology grade group.

^aFour men refused biopsy in the PSA⁺ MRI⁻ subgroup.

Fig. 2.

Patient flow through each step of the study. GG = International Society of Urological Pathology grade group; MRI = magnetic resonance imaging; PI-RADS = Prostate Imaging-Reporting and Data System; PSA = prostate-specific antigen.

In the MRI-positive group, all PCs were detected via targeted biopsy, with no additional csPC found on systematic biopsy. In the group with PSA ≥3 μg/l, systematic biopsy identified csPC in three men, while targeted biopsy added no further csPC cases. One PSA-positive patient had nsPC detected only via targeted biopsy.

The PSA pathway would diagnose PC in seven men (1.7%), with four nsPC and three csPC cases, but would miss four of seven csPC cases. Addition of PSAD ≥0.15 μg/l/cm³ to the PSA pathway would reduce positive findings and biopsies to three men (0.7%) but would miss six of seven csPC cases. Even lowering the PSAD cutoff to 0.10 μg/l/cm³ would still miss five of seven csPC cases detected via bpMRI (Supplementary Table 1).

The MRI pathway would diagnose PC in 14 men (3.3%), with seven cases each of nsPC and csPC (1.7%). Omission of bpMRI from the screening pathway would result in missing four of seven csPC cases and six of 10 nsPC cases. In the PSA-positive group (≥3 μg/l), bpMRI had sensitivity of 57.1% for PC and 100% for csPC, albeit based on a limited number of events. MRI missed only three nsPC cases. Restricting bpMRI to individuals with PSA ≥1 μg/l would avoid MRI in 238 men (56.3%) without missing PI-RADS 4–5 cases.

3.4. Recent PSA test contamination in the study group

Although an interval of 2 yr since the last PSA test was required, independent verification using a national registry revealed that 170 men (40.2%) had actually had a PSA test within 2 yr, and only 91 men (21.5%) had no prior PSA test. Documentation of a PSA test within 2 yr was found for eight of 16 men in the MRI-positive group and 17/25 in the PSA-positive group.

4. Discussion

To overcome the limitations of PSA testing, image-based screening is being pursued for detection of csPC in its earliest stages. When MRI is used for PC screening within the context of a defined PSA cutoff, it decreases the rate of prostate biopsies and reduces the detection of nsPC [6,7,12]. bpMRI has emerged as a more accessible alternative to multiparametric MRI, offering shorter examination times (∼15 min), lower costs, and elimination of contrast agents. Studies have shown that bpMRI is noninferior to multiparametric MRI and are evaluating its use as a primary screening test in general and high-risk populations [6,[12], [13], [14], [15]]. Our study demonstrates that bpMRI effectively doubled csPC detection in a screening population in comparison to PSA alone (seven vs three cases). bpMRI detected all csPC cases among men with PSA ≥3 μg/l. Notably, and consistent with other studies, more than half of the patients diagnosed with csPC had PSA <3 μg/l [6,12,14]. For example, in the IP1-PROSTAGRAM study, the csPC detection rate was 1.5% in the PSA pathway (PSA ≥3 μg/l and PI-RADS 4–5), compared to 2.7% in the MRI pathway without a PSA cutoff [6]. It remains to be evaluated whether delaying diagnosis of these tumors until PSA reaches ≥3 μg/l would affect the chance of being cured.

bpMRI is increasingly being integrated into population-based screening studies using biomarker preselection, as MRI alone is not an ideal primary screening test because of the low prevalence of high-grade cancer, which ranges from 0.2% to 4% in most studies. To enhance the efficiency of MRI in diagnosing csPC, further investigation of risk stratification via alternative PSA cutoffs or novel biomarkers is required. Evaluation of optimal PSA thresholds is ongoing [15,16]. In our study, all csPC cases had PSA ≥1 μg/l; however, application of this cutoff would still necessitate MRI in 44% of men. A PSA threshold of 1.8 ng/ml was evaluated in arm 3 of the Göteborg-2 study, which would trigger MRI for 18% of the participants [16]. PSAD appears to be ineffective for MRI preselection in PSA-positive men in a screening setting, as it may reduce csPC detection. However, some studies outside the screening setting suggest that PSAD may help in identifying patients with elevated PSA and negative bpMRI who might still require biopsy [17].

Another key consideration is the definition of MRI positivity, as screening programs must balance the csPC detection rate against the biopsy burden. Eldred-Evans et al [15] compared biopsy strategies using a threshold of PI-RADS 3–5 versus PI-RADS 4–5 in a cohort of men with PSA ≥1 ng/ml. Expanding the biopsy criterion to include PI-RADS 3 lesions increased the csPC detection rate from 2.5% to 2.7%, but also led to a 69% increase in the biopsy rate and greater detection of nsPC. The authors concluded that restricting biopsies to PI-RADS 4–5 lesions provides the best trade-off between minimizing unnecessary biopsies and maintaining a high csPC detection rate. Interestingly, we observed low positive MRI (3.8%) and PSA (5.9%) rates, probably because of the high prevalence of PSA prescreening in Czechia (45% in the target age group) [8]. A retrospective inquiry revealed that 40% of the participants had a PSA test within the previous 2 yr, despite an initial statement to the contrary. Although there are challenges in interpreting bpMRI in younger men (71% of our cohort were aged 50–59 yr), we demonstrated that PI-RADS 4–5 on bpMRI is accurate in a PSA prescreened population [18]. Notably, other MRI screening studies have reported a higher prevalence of positive MRI (10–21%) and PSA (3–18%) findings corresponding with lower levels of opportunistic PSA testing in those populations [6,[12], [13], [14]]. In addition, csPC detection rates in these studies varied significantly, from 1.6% in our study to 4.0% in the PROSTAGRAM study and even 9.6% in ReIMAGINE [19]. We found no reduction in the nsPC detection rate with the MRI-based approach, which diagnosed nsPC in seven men (1.7%), compared to four (1.0%) via the PSA-based strategy [16,20]. This may be because of the low event rate in our study and variability in definitions of csPC, which sometimes include high-volume GG 1 cancer [20]. Similar MRI screening studies have also reported comparable nsPC detection rates between MRI-based and PSA-based approaches [6,14], aligning with our findings.

Interestingly, our study also allowed a comparison between two biopsy approaches. We did not find that systematic biopsy provided additional value over MRI-targeted biopsy in detecting csPC, which is consistent with the Göteborg-2 study [21]. However, it is important to acknowledge that some studies suggest that relying solely on targeted biopsy might delay the diagnosis of certain csPC cases [22].

Our study has several limitations. The relatively small sample size and low event rate may limit the generalizability of our findings to a population-based screening setting. However, our cohort size is comparable to other MRI-based screening studies involving all-comers [6,12,14,19]. Another limitation is that participants with negative results for both tests did not undergo biopsy, which precluded calculation of the sensitivity and specificity for bpMRI. Moreover, biopsy was not completed for four PSA-positive patients. The cohort, comprising younger, more highly educated men interested in cancer prevention, may not fully represent the target population. The wide penetration of opportunistic PSA screening led to a low prevalence of positive MRI and PSA tests, a scenario that may not reflect the situation in countries with low prevalence of opportunistic screening. Lastly, as this was a single-site study conducted in an experienced diagnostic unit, the results may not be generalizable to settings with less MRI expertise.

5. Conclusions

Our study confirms that bpMRI may improve csPC detection while reducing biopsy rates in comparison to a PSA-based pathway, even in a highly prescreened population. Omission of bpMRI from the screening pathway would have resulted in lower csPC detection. However, for full bpMRI integration into multistep screening pathways, further research is needed to determine the optimal preselection criteria for participants.



Author contributions: Michal Staník had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.



Study concept and design: Staník, Standara, Křístek, Májek.

Acquisition of data: Standara, Miklánek, Pacal, Hrabec, Křístek.

Analysis and interpretation of data: Staník, Standara, Hejcmanová, Uher, Májek.

Drafting of the manuscript: Staník, Standara, Hejcmanová.

Critical revision of the manuscript for important intellectual content: Ngo, Hejduk, Uher, Křístek, Májek, Poprach.

Statistical analysis: Hejcmanová, Ngo, Uher.

Obtaining funding: Staník, Májek.

Administrative, technical, or material support: None.

Supervision: None.

Other: None.



Financial disclosures: Michal Staník certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: None.



Funding/Support and role of the sponsor: This work was supported by the Czech Ministry of Health under grant NU22-09-00539 to the Masaryk Memorial Cancer Institute (center 00209805).



Ethics considerations: Participants provided informed consent. The study was approved by the institutional ethics committee (approval reference 2022/1303/MOU).

Associate Editor: Roderick van den Bergh

Appendix A. Supplementary data

The following are the Supplementary data to this article: