Micro‐ultrasonography‐ vs MRI‐targeted biopsy in prostate cancer: Updated systematic review and meta‐analysis

Abstract

Objective

To refine prostate cancer (PCa) pathways by comparing micro‐ultrasonography‐targeted biopsy (micro‐US‐TBx) with magnetic resonance imaging (MRI)‐TBx for detecting clinically significant PCa (csPCa) and by evaluating diagnostic accuracy (sensitivity, specificity, positive predictive value, negative predictive value).

Methods

The Medical Literature Analysis and Retrieval System Online (MEDLINE; PubMed), Excerpta Medica dataBASE (EMBASE), and ClinicalTrials.gov were searched on 4 July 2025 and updated 5 August 2025, without language/date limits. Eligible randomised and observational studies compared micro‐US‐TBx vs MRI‐TBx (± systematic biopsy) in men with suspected PCa or on active surveillance; csPCa was Gleason Grade Group ≥2. Two reviewers screened/extracted per the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses 2020 guidelines; quality was assessed with the Quality Assessment of Diagnostic Accuracy Studies‐2 tool. Detection ratios (DRs) were pooled with random‐effects models; diagnostic accuracy used a bivariate random‐effects model from 2 × 2 tables. Certainty of evidence was rated with the Grading of Recommendations, Assessment, Development, and Evaluation framework.

Results

A total of 22 studies (micro‐US‐TBx = 5581; MRI‐TBx = 5937) were included. Micro‐US‐TBx detected 2037 csPCa cases and MRI‐TBx 2043. The pooled DR (micro‐US‐TBx vs MRI‐TBx) was 1.04 (95% confidence interval 0.95–1.15; I ² = 61.1%). Pooled sensitivity/specificity were 0.88/0.25 for micro‐US‐TBx and 0.81/0.17 for MRI‐TBx. Summary receiver operating characteristic curves overlapped (area under the curve: micro‐US, 0.658; MRI, 0.547). Meta‐regression found no significant modality effect on sensitivity (P = 0.192) or specificity (P = 0.257). Egger's test indicated no meaningful publication bias (P = 0.848).

Conclusion

The micro‐US‐TBx demonstrated csPCa detection and diagnostic accuracy comparable to MRI‐TBx. Given its real‐time imaging capability, accessibility, and potential workflow advantages, micro‐US‐TBx may represent a valuable complementary tool within contemporary PCa diagnostic pathways. Further cost‐effectiveness studies are warranted to better define its role in clinical practice.

Abbreviations

bpMRI

biparametric MRI

DR

detection ratio

GG

Grade Group

GRADE

Grading of Recommendations, Assessment, Development, and Evaluation; micro‐US, micro‐ultrasonography

mpMRI

multiparametric MRI

(cs)(ci)PCa

(clinically significant) (clinically insignificant) prostate cancer

PRI‐MUS

Prostate Risk Identification using Micro‐US

(N)(P)PV

(negative) (positive) predictive value

PRISMA

Preferred Reporting Items for Systematic Reviews and Meta‐Analyses

QUADAS‐2

Quality Assessment of Diagnostic Accuracy Studies‐2

RCT

randomised controlled trial

SBx

systematic biopsy

SROC

summary receiver operating characteristic

TBx

targeted biopsy

Introduction

Prostate cancer (PCa) affects millions of men worldwide each year. In high‐income regions, it is among the most prevalent solid malignancies, with prognosis varying according to age, ethnicity, genetic background, and disease stage [1, 2]. Clinical suspicion of PCa arises primarily from elevated PSA levels and/or abnormal DRE findings [3]; and is generally categorised into clinically significant PCa (csPCa), which requires curative interventions such as radical prostatectomy, and clinically insignificant PCa (ciPCa), which is typically managed with active surveillance guided by PSA monitoring [4]. In this context, there remains an unmet clinical need for diagnostic strategies that reliably detect csPCa while reducing overdiagnosis of indolent disease and avoiding unnecessary biopsies [5].

Multiparametric MRI (mpMRI) is a resource‐intensive diagnostic modality, which restricts its availability primarily to large centres. Nevertheless, it is widely used and demonstrates high sensitivity (86–92%) and specificity (60–83%) for detecting index lesions in PCa [6]. Current guidelines from the AUA [5] and the European Association of Urology [7] recommend TRUS‐guided biopsy as the standard approach for histological confirmation of PCa. When a suspicious lesion is identified on mpMRI, targeted prostate biopsy should be performed using either software‐based fusion of mpMRI with real‐time ultrasonography (US) images or cognitive targeting [8].

To address some of the limitations of mpMRI, a novel, low‐cost, high‐resolution US technology—micro‐US (micro‐US; ExactVu, Exact Imaging, Markham, Ontario, Canada)—has been developed. Operating at 29 MHz, substantially higher than conventional TRUS frequencies, micro‐US provides real‐time, high‐resolution imaging that enables direct visualisation of suspicious tissue. This capability holds promise for reducing reliance on pre‐biopsy MRI [9]. Lesions detected on micro‐US are scored using the Prostate Risk Identification using Micro‐US (PRI‐MUS) system, in which scores of 1–2 indicate low risk, 3 is equivocal, and 4–5 are considered highly suspicious for PCa [10].

The recently published OPTIMUM trial (ClinicalTrials.gov identifier: NCT05220501) [11], a large international randomised clinical study involving 678 men undergoing biopsy, demonstrated that micro‐US‐guided biopsy is non‐inferior to MRI‐fusion‐guided biopsy for the detection of csPCa.

Given the growing body of evidence supporting the integration of micro‐US into the diagnostic pathway for suspected csPCa, we conducted a systematic review of observational and randomised controlled trials (RCTs). We specifically compared micro‐US‐targeted biopsy (micro‐US‐TBx) with MRI‐TBx in terms of csPCa detection. The primary outcome was the detection rate of csPCa, while secondary outcomes included pooled diagnostic accuracy metrics (sensitivity, specificity, positive predictive value [PPV], and negative predictive value [NPV]).

The aim of this review was to compare the detection rate of csPCa between micro‐US‐TBx and MRI–TBx, and to pool diagnostic accuracy metrics.

Methods

Protocol Registration

This systematic review and meta‐analysis was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA) 2020 guidelines [12]. Thus, our systematic review was registered with the International Prospective Register of Systematic Reviews (PROSPERO) on 25 September 2025 (registration number CRD420251155775) [13].

Eligibility Criteria

We included observational studies and RCTs that compared the detection of csPCa, defined as Gleason Grade Group (GG) ≥2, in men with suspected PCa or those under active surveillance. Eligible studies evaluated outcomes from micro‐US‐guided biopsy vs MRI/conventional US‐guided biopsy. Clinical suspicion of PCa was defined as elevated PSA levels, abnormal DRE findings, and/or suspicious lesions on mpMRI. Additional eligibility criteria required the absence of contraindications to MRI or prostate biopsy, as well as the absence of a prior prostate MRI.

Information Sources

We systematically searched MEDLINE (PubMed), Excerpta Medica dataBASE (EMBASE), and ClinicalTrials.gov on 4 July 2025, and updated the search on 5 August 2025 (Fig. S1), with no language or publication date restrictions. Using both keywords and controlled vocabulary (Medical Subject Headings [MeSH]/Emtree), the core string was: (‘prostate cancer’ OR ‘prostate neoplasm’ OR ‘prostatic neoplasm’ OR ‘prostate adenocarcinoma’) AND (‘micro‐ultrasound’ OR ‘micro‐ultrasonography’ OR ‘micro ultrasound’ OR ‘micro‐ultrasound‐guided’ OR ‘micro‐US’) AND (‘magnetic resonance imaging’ OR ‘MRI’ OR ‘MR imaging’). Full database‐specific strategies are provided in the Appendix S1.

Selection Process

Two authors (J.M. and T.A.) independently selected the studies. In cases of disagreement between the two reviewers, a third author (C.S.) resolved discrepancies by consensus. For duplicate studies and multiple reports from the same cohort, we examined only (a) the largest cohort, (b) the one with the longest follow‐up, or (c) the most recent report.

Data Collection

Two authors (J.M. and T.A.) independently screened studies in three stages: (a) title review, (b) abstract review, and (c) full‐text assessment. Reference lists of eligible articles were also examined to identify additional relevant studies (snowballing/citation chaining). Disagreements between the two reviewers were resolved by consensus with a third author (L.S.). Literature screening and coding were managed using Rayyan software [14].

Data Items

Data were extracted into a piloted spreadsheet using Microsoft Excel 2019 (Microsoft Corporation, Redmond, WA, USA). The following information was collected: first author's name, article title, journal, year of publication, study design, type of guided biopsy, number of participants, study objectives, inclusion criteria, and variables related to PCa outcomes and technical characteristics. When relevant information was missing, attempts were made to contact the corresponding authors via e‐mail to obtain additional details.

Effect Measures

We investigated the main outcome by analysing detection risk measures and their corresponding 95% CIs for variables associated with the detection of csPCa. Accuracy measures, including sensitivity, specificity, PPV, and NPV, were also assessed.

Synthesis Methods

The meta‐analysis was independently conducted by two reviewers (J.M. and T.A.) using R version 4.3.2 (The R Foundation for Statistical Computing, Vienna, Austria) [15]. Discrepancies between their analyses were resolved through consensus with a third reviewer (C.S.).

Heterogeneity was assessed using Cochran's Q test [16] and the Higgins and Thompson I ² statistic [17]. A P < 0.05 was considered statistically significant for Cochran's Q test. I ² values were interpreted as follows: 0–<25% indicated negligible heterogeneity, 25–<40% moderate heterogeneity, and ≥40% substantial heterogeneity.

Micro‐US‐TBx was considered the index test, with MRI‐TBx as the comparator (reference standard). The index test produced binary outcomes (presence or absence of csPCa) at the patient level. Data were organised into 2 × 2 contingency tables to calculate sensitivity and specificity for each included study. Study‐level results were presented as forest plots and summary receiver operating characteristic (SROC) curves.

Summary estimates of sensitivity and specificity were obtained using a bivariate random‐effects model, with 95% CIs derived from the model along with pooled estimates. To further explore heterogeneity, a bivariate model with covariates was also applied.

Study Methodological Quality Assessment

We assessed the risk of bias and applicability concerns for each study using the Quality Assessment of Diagnostic Accuracy Studies‐2 (QUADAS‐2) tool. QUADAS‐2 evaluates four domains: patient selection, index test, reference standard, and flow and timing. Each domain is assessed for risk of bias, and the first three domains are also evaluated for applicability concerns [18]. Quality assessment was independently performed by two authors (L.A., B.M.), with disagreements resolved by a third reviewer (J.M.). In addition, we evaluated risk of bias across studies (publication bias) through visual inspection of contour‐enhanced funnel plot asymmetry and the Egger test [19].

Grading of Evidence, Data Synthesis and Statistical Analysis

We assessed the certainty of the evidence for csPCa detection rates and diagnostic accuracy (sensitivity, specificity, PPV, and NPV) comparing micro‐US‐TBx and MRI‐TBx using the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) framework. Two reviewers (J.M., T.S.) independently rated the GRADE domains—risk of bias, inconsistency, indirectness, imprecision, and publication bias—with disagreements resolved by consensus (Fig. S4). For each modality, the detection rate was defined as the proportion of participants diagnosed with the relevant Gleason GG of PCa among those who underwent that biopsy approach (in head‐to‐head studies, denominators corresponded to the number evaluated by each modality).

Classification of outcomes (true positives, false positives, true negatives, and false negatives) used to calculate diagnostic—and, where applicable, staging—accuracy was based on histopathological findings from prostate biopsy as the reference standard. Pooled sensitivity and specificity estimates, and hierarchical SROC curves were generated in R (version 4.5.1; R Foundation for Statistical Computing) using the ‘meta’ and ‘mada’ packages [20, 21]. Between‐study heterogeneity was quantified using the I ² statistic and assessed with Cochran's Q test. I ² was categorised as low (<25%), moderate (25–50%), or high (≥50%). Statistical significance was set at P < 0.05. All analyses were conducted in R (version 4.5.1; R Foundation for Statistical Computing).

Results

Study Selection

The literature search identified 283 records. After the removal of 114 duplicates, 169 records were screened by title and abstract. Of these, 140 were excluded for not meeting the study objectives. The full texts of the remaining 30 articles were assessed for eligibility, and 22 studies were ultimately included in the final analysis [11, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42]. The detailed study selection process is illustrated in the PRISMA flow diagram in Fig. S1. Of these, one was a RCT [11], while 21 were observational studies, comprising 10 retrospective and 11 prospective designs [22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42].

Study and Population Characteristics

Overall, 22 studies published between 2019 and 2025 were included, comprising 5581 patients who underwent micro‐US‐TBx and 5937 who underwent MRI‐TBx (Fig. S1) [11, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42]. The median (range) age was 66 (59–74) years, median (range) PSA level was 7.1 (5.39–11) ng/mL, and median (range) prostate volume was 50 (37–61.34) mL. Most patients were re‐biopsy (n = 5204), while 2909 were biopsy naïve. All had suspected csPCa, with 2037 cases detected by micro‐US‐TBx and 2043 by MRI‐TBx. Baseline characteristics are presented in Table 1 [11, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42], and the oncological characteristics of the included studies are summarised in Table 2. None of the included studies enrolled patients with a prior diagnosis of csPCa.

Table 1.

Study characteristics.

Study (n = 22)	Study period	Type of study	Compared techniques	Criteria for biopsy indication	No. patients	Age, years
Abouassaly et al., 2020 [22]	01/2018–08/2018	Prospective, single centre	Micro‐US‐TBx vs mpMRI‐TBx vs SBx	Patients underwent prostate biopsy	67	66 (IQR 59–69)
Albers et al., 2024 [23]	09/2021–07/2023	Retrospective, single centre	Micro‐US‐TBx vs MRI‐TRUS	Biopsy‐naive patients with clinical suspicion of prostate cancer who underwent a prostate biopsy with negative MRI	103	61.7 (SD 7.4)
Avolio et al., 2025 [24]	10/2017–04/2024	Prospective, single centre	Micro‐US‐TBx vs mpMRI‐TBx	Elevated PSA levels, abnormal DRE, and/or MRI findings at least 1 PI‐RADS ≥3	1423	65 (IQR 59–72)
Beatrici et al., 2024 [25]	10/2017–02/2023	Retrospective, single centre	Micro‐US‐TBx vs mpMRI‐TBx	Clinical examinations and PSA levels	304	66.0 (IQR 61–71)
Claros et al., 2020 [26]	02/2017–09/2018	Retrospective, single centre	Micro‐US‐TBx vs mpMRI‐TBx vs RBx	DRE and PSA values	269	68 (IQR 62–72)
Cornud et al., 2020 [27]	02/2019–07/2019	Prospective, single centre	Micro‐US‐TBx vs MRI‐TRUS	Elevated or rising PSA level and at least one focal abnormality with a bpMRI score ≥3	118	66 (SD 13)
Dagnino et al., 2024 [28]	10/2017–	Retrospective, single centre	Micro‐US‐TBx vs mpMRI‐TBx vs SBx	Biopsy naïve men with discordant lesions on micro US and mpMRI imaging, who underwent fusion prostate biopsy within 6 months of mpMRI	178	–
Dias et al., 2023 [29]	01/2021–06/2022	Retrospective, single centre	Micro‐US‐TBx vs mpMRI‐TBx vs SBx	Biopsy naïve, a total PSA <20 ng/mL and having a mpMRI performed prior to the micro‐US and biopsy	139	69 (IQR 64–73)
Eure et al., 2019 [30]	12/2016–12/2016	Prospective paired diagnostic	Micro‐US‐TBx vs mpMRI‐TBx vs SBx	Patients enrolled in the active surveillance program	9	66 (IQR 62–68)
Ghai et al., 2022 [31]	05/2019–09/2020	Prospective, single centre	Micro‐US‐TBx vs mpMRI‐TBx vs SBx	PSA level or abnormal DRE findings and no previous PBx	94	61 (IQR 57–68)
Hofbauer et al., 2022 [32]	01/2019–12/2019	Prospective multicentre	Micro‐US‐TBx vs mpMRI‐TBx	PSA value and/or suspicious DRE	203	66 (IQR 59–70)
Khajir et al., 2025 [33]	10/2021–01/2023	Retrospective, single centre	MRI‐micro‐US‐TBx vs MRI‐conventional US‐TBx vs SBx	PI‐RADS ≥2 and biopsy status (biopsy naïve, prior negative, or active surveillance), who underwent MRI‐conventional US fusion biopsy	899	67 (IQR 60–72)
Kinnaird et al., 2025 [11]	12/2021–09/2024	Randomised prospective multicentre	Micro‐US‐TBx vs micro‐US/mpMRI‐TBx vs MRI/conventional‐US‐TBx	Elevated PSA and/or abnormal DRE findings	802	–
Klotz et al., 2020 [34]	05/2019–09/2020	Multicentre, prospective	Micro‐US‐TBx vs mpMRI‐TBx vs SBx	Elevated PSA and/or abnormal DRE, with or without a suspicious mpMRI	1040	67 (IQR 61–72)
Lorusso et al., 2022 [35]	11/2018–05/2019	Retrospective, single centre	Micro‐US‐TBx vs mpMRI‐TBx vs whole mount sections	Patients with biopsy confirmed prostate cancer and scheduled for robot‐assisted radical prostatectomy	32	67 (range 52–72)
Lughezzani et al., 2021 [36]	10/2017–09/2019	Prospective, single centre	Micro‐US‐TBx vs mpMRI‐TBx vs SBx	Elevated PSA or abnormal DRE and PI‐RADS ≥3	320	65 (IQR 59–70)
Pereira‐Arias et al., 2019 [37]	02/2017–01/2018	Retrospective, single centre	Micro‐US vs mpMRI vs SBx	Elevated PSA or abnormal DRE	96	67 (SD 5.5)
Piccolini et al., 2024 [38]	10/2017–05/2023	Retrospective, single centre – Propensity score–matched comparison	Micro‐US TRBx vs TPBx + mpMRI	Elevated PSA, positive DRE, or mpMRI and PI‐RADS ≥3	1423	–
Pickersgill et al., 2025 [39]	02/2021–04/2023	Retrospective, single centre	Micro‐US‐TBx vs mpMRI‐TBx	Elevated PSA or abnormal DRE.	161	65.6 (SD 1.5)
Rojo et al., 2024 [40]	05/2021–06/2022	Prospective, multicentre	Micro‐US‐TBx vs mpMRI‐TBx	PSA ≥4 ng/mL and/or abnormal DRE and PI‐RADS ≥3	80	69.1 (SD 9.02)
Socarrás et al., 2020 [41]	02/2018–09/2019	Prospective, single centre	Micro‐US‐TBx vs mpMRI‐TBx vs SBx	Elevated PSA or suspicious DRE or PI‐RADS ≥3 in mpMRI	194	62 (IQR 58–68)
Wiemer et al., 2021 [42]	02/2018–12/2018	Prospective, single centre	Micro‐US‐TBx vs mpMRI‐TBx vs SBx	Clinical risk of PCa	159	70 (IQR 64–74)

PBx, prostate biopsy; RBx, random biopsies; TPBx, transperineal biopsy; TRBx, transrectal biopsy.

Table 2.

Oncologic Characteristics of the Included Studies.

Study (n = 22)	Biopsy Naive	PSAt (ng/mL)	DRE		Prostate Volume (cc)	PRI‐MUS (%)				PIRADS (%)
			+	−		1–2	3	4	5	1–2	3	4	5
Abouassaly, 2020 [22]	‐	5.39 (IQR: 4.1–8.7)	7	44	38 (IQR: 24–50)	‐	‐	‐	‐	‐	‐	‐	‐
Albers, 2024 [23]	103	‐	11	92	‐	37 (66%)	4 (7%)	3(5%)	12 (21%)	103 (100%)	‐	‐	‐
Avolio, 2025 [24]	533	7 (IQR: 5–9)	380	1043	50 (IQR: 36–67)	347 (24%)	147 (10%)	593 (42%)	336 (24%)	247 (17%)	241 (17%)	574 (40%)	361 (25%)
Beatrici, 2024 [25]		8.6 (IQR: 5.8–11.5)	58	206	52.6 (IQR: 49.5–77)	84 (27.6%)	27 (8.9%)	133 (43.8%)	60 (19.7%)	103 (33.9%)	57 (18%)	104 (34.2%)	40 (13.2%)
Claros, 2020 [26]	69	7.8 (IQR: 5.5–10.2)	‐	‐	48.5 (IQR: 36–65)	8 (17.4%)	2 (4.3%)	16 (34.8%)	20 (43.5%)	‐	21 (7.8%)	155 (57.6%)	82 (30.5%)
Cornud, 2020 [27]	79	11 (SD ± 19)	16	102	53 (SD: ± 26)	‐	‐	‐	‐	‐	‐	‐	‐
Dagnino, 2024 [28]	178	‐	21	92	‐	‐	31 (17.4%)	119 (66.9%)	28 (15.7%)	26 (14.6%)	36 (20,2%)	96 (54%)	20 (11.2%)
Dias, 2023 [29]	139	7.5 (IQR: 5.5–10.9)	27	112	47 (IQR: 33–65)	39 (28%)	48 (35%)	37 (27%)	15 (11%)	17 (12%)	27 (19%)	65 (47%)	30 (22%)
Eure, 2019 [30]	0	5.9 (IQR: 5.0–6.4)	2	7	37 (IQR: 33–44)	‐	‐	‐	‐	‐	‐	‐	‐
Ghai, 2022 [31]	94	6.37 (IQR: 5.1–8.4)	57	37	49 (IQR: 34–64)	‐	‐	‐	‐	‐	‐	‐	‐
Hofbauer, 2022 [32]	147	6.5 (IQR: 4.8–9.3)	31	172	48 (IQR: 37–63)	16 (7%)	56 (28%)	96 (47%)	35 (17%)	15 (7.5%)	37 (18%)	109 (54%)	41 (20%)
Khajir, 2025 [33]
ExactVu	261	6.70 (IQR: 4.8–9.5)	94	287	49 (IQR: 35–71)	‐	‐	‐	‐	24 (6%)	80 (19%)	198 (46%)	127 (30%)
Artemis	280	7.66 (IQR: 5.4–10.8)	120	346	55 (IQR: 40–77)	‐	‐	‐	‐	30 (6%)	95 (20%)	200 (43%)	145 (31%)
Kinnaird, 2025 [11]
Micro‐US	121	7.2 (IQR: 5.5–11)	25	64	‐	24 (20%)	21 (18%)	43 (36%)	24 (20%)	‐	‐	‐	‐
MRI/US	331	6.7 (IQR: 5.1–9.5)	81	196	‐	‐	‐	‐	‐	111 (48%)	42 (14%)	84 (27%)	70 (23%)
Klotz, 2020 [34]	407	7 (IQR: 5.1–10)	208	832	38 (IQR: 28–53)	166 (16%)	177 (17%)	686 (66%)		372 (36%)	124 (12%)	351 (34%)	193 (18%)
Lorusso, 2022 [35]	0	6.2 (R: 2.4–15.9)	7	9	51 (R: 35–98)	‐	‐	‐	‐	‐	‐	‐	‐
Lughezzani, 2020 [36]	200	7.3 (IQR: 5.2–9.9)	72	248	45 (IQR: 30–70)	65 (20.3%)	38 (11.9%)	159(49.7%)	58 (18.1%)	‐	56 (17.5%)	209 (65.3%)	55 (17.2%)
Pereira‐Arias, 2019 [37]	51	7.5 (R: 1.26‐34.9)	‐	‐	48 (R: 15–113)	‐	‐	‐	‐	‐	‐	‐	‐
Piccolini, 2024 [38]
Micro‐US	‐	6.6 (IQR: 5–9)	143	439	50 (IQR: 38–70)	‐	‐	‐	‐	‐	‐	‐	‐
mp‐MRI	‐	6.7 (IQR: 5–9)	140	372	50 (IQR: 35–68)	‐	‐	‐	‐	‐	‐	‐	‐
Pickersgill, 2025 [39]	115	7.6 (SD: ±0.6)	‐	‐	56.2 (SD: 30.3)	50 (31%)	111 (68.9%)			69 (43%)	16 (9.9%)	38 (23.6%)	38 (23.6%)
Rojo, 2024 [40]	‐	7.66 (SD: ±6.29)	‐	‐	61.34 (SD:± 31.7)	‐	17 (14.4%)	80 (67.8%)	21 (17.8%)	‐	33 (31.7%)	56 (53.8%)	15 (14.4%)
Socarrás, 2020 [41]	129	6.5 (IQR:4.7–9.2)	31	163	58.1 (IQR: 36.6–81.5)	‐	‐	‐	‐	‐	‐	‐	‐
Wiemer, 2021 [42]	79	7.59 (IQR: 5.8–11.5)	42	117	53 (IQR: 35.5–76.5)	16 (15%)	22 (14%)	74 (47%)	47 (30%)	7 (5%)	15 (9%)	74 (47%)	63 (40%)

DRE, digital rectal examination; IQR, interquartile range; PSA, prostate‐specific antigen; R, range; SD, standard deviation.

Imaging Acquisition Parameters for Micro‐US and MRI

Across all studies, suspicious lesions were defined as PI‐RADS ≥3 and PRI‐MUS ≥3, as both scores are used as common thresholds to define a positive, suspicious lesion that warrants a biopsy. The operators performing micro‐US‐TBx received appropriate training and most of the mpMRIs were evaluated by a radiologist experienced in mpMRI. Moreover, all mpMRIs were performed with a strength of 1.5 or 3 Tesla with or without endorectal coil placement and one study included biparametric MRI (bpMRI) [27]. The operator was blinded to the MRI results at the time of micro‐US‐TBx in 15 studies and unblinded in seven (Table S1).

Among the studies reporting PRI‐MUS, a total of 1811 patients were assessed. Of these, 449 (24.8%) had a PRI‐MUS score of 1–2, 639 (35.3%) a score of 3, 493 (27.2%) a score of 4, and 230 (12.7%) a score of 5. For PI‐RADS, data were available for 3113 patients. The distribution was 618 (19.9%) with a PI‐RADS score of 1–2, 588 (18.9%) with a score of 3, 1140 (36.6%) with a score of 4, and 767 (24.6%) with a score of 5. Notably, high‐grade lesions (scores 4–5) were more frequent with PI‐RADS (61.2%) than with PRI‐MUS (39.9%), whereas intermediate findings (score 3) were proportionally higher with PRI‐MUS (35.3%) compared to PI‐RADS (18.9%).

Methodological Quality of Studies

On QUADAS‐2, the patient selection domain was rated low risk of bias in 14/22 studies (63.6%; Fig. 2). The reference standard domain was low risk in 21/22 (95.4%), as histopathology served as the reference standard (targeted biopsy and/or radical prostatectomy were considered appropriate). The index test domain was judged low risk in 19/22 studies (86.3%). Likewise, the flow and timing domain was low risk in 19/22 (86.3%). No study raised concerns regarding applicability in any domain (Fig. S2).

Fig. 2.

Pooled sensitivity and specificity: MRI‐TBx vs Micro‐US‐TBx for csPCa (GG ≥2).

Despite the comparable diagnostic yields reported, our GRADE assessment classified the certainty of evidence as predominantly low. This is largely a reflection of the significant heterogeneity across the included studies—ranging from varying biopsy protocols to the lack of a consistent reference standard. Such variability limits the weight of these findings in the context of clinical guidelines; they should be viewed as a signal of potential rather than a mandate for practice change. To bridge this gap, future research must shift away from small, heterogeneous cohorts. We need multicentre trials with standardised training for operators, as the learning curve for micro‐US remains a critical confounder. Only by reducing this ‘user‐dependency’ and focusing on well‐defined patient groups can we determine if micro‐US‐TBx is a true successor to MRI‐TBx or a specialised niche tool.

Detection Rates

We included 19 studies comprising 5581 participants who underwent micro‐US‐TBx and 5937 who underwent MRI‐TBx. Micro‐US‐TBx detected 2037 cases of csPCa (GG ≥2), while MRI‐TBx detected 2043 cases. The pooled detection ratio (DR) for csPCa comparing micro‐US with MRI was 1.04 (95% CI 0.95–1.15; I ² = 61.1%; Fig. 1). Funnel plot inspection and Egger's test revealed no evidence of significant publication bias among the included studies (P = 0.848; Fig. S3).

Fig. 1.

Forest plot of detection rates: micro‐US‐TBx vs MRI‐TBx for csPCa (GG ≥2).

Micro‐US‐TBx Diagnostic Accuracy

The sensitivity and specificity of micro‐US‐TBx across the included studies (n = 10) ranged from 0.72 to 1.00 and 0.00 to 1.00, respectively. The pooled sensitivity was 0.88 (95% CI 0.80–0.93), while the pooled specificity was 0.25 (95% CI 0.19–0.33). Heterogeneity was substantial, with I ² values from the bivariate model ranging between 83.7% and 84.4% (Fig. 2).

The MRI Diagnostic Accuracy

The sensitivity and specificity of MRI‐TBx across the included studies (n = 13) ranged from 0.55 to 1.00 and 0.01 to 1.00, respectively. The pooled sensitivity was 0.81 (95% CI 0.74–0.86), and the pooled specificity was 0.17 (95% CI 0.10–0.27). Heterogeneity, assessed using the bivariate model, was high, with I ² values ranging from 88.3% to 95.8% (Fig. 2).

Diagnostic Accuracy Comparison

The 95% prediction regions of the SROC curves overlapped substantially, indicating similar performance between modalities. In a bivariate random‐effects meta‐regression with imaging modality as a covariate, model‐based areas under the curve were closely aligned (MRI, 0.547; micro‐US, 0.658). Modality effects were not significant for sensitivity (estimate = 0.09; 95% CI −0.04 to 0.22; P = 0.192) or specificity (estimate = 0.49; 95% CI −0.36 to 1.34; P = 0.257). These results indicate no statistically significant advantage of either technique (Fig. 3; Fig. S5).

Fig. 3.

Comparative SROC: micro‐US‐TBx vs MRI‐TBx.

Discussion

Summary of Results

This study represents the largest systematic review and meta‐analysis to date and the first to incorporate the OPTIMUM randomised clinical trial [11]. In this analysis of 22 studies, we compared the diagnostic performance and csPCa detection rates of micro‐US‐TBx and MRI‐TBx. The pooled results demonstrated comparable csPCa detection between the two modalities, supporting micro‐US‐TBx as a feasible and clinically meaningful alternative to MRI‐TBx.

Outcomes

This review synthesises head‐to‐head evidence comparing micro‐US‐TBx with MRI‐TBx for diagnosing csPCa. Across the included studies, DR, sensitivity, and specificity for csPCa did not differ significantly between the two modalities. Specifically within the randomised OPTIMUM trial [11], micro‐US‐TBx met pre‐specified non‐inferiority thresholds relative to MRI‐TBx. Given its comparable diagnostic accuracy and potential to streamline evaluations, micro‐US may reduce unnecessary interventions without compromising csPCa detection. Taken together, the comparable outcomes observed in prior prospective studies [43, 44], reinforced by the non‐inferiority established in the OPTIMUM trial [11], support the utility of micro‐US‐guided biopsy as a valid alternative to MRI–US fusion‐guided biopsy.

The Lancet Commission on Prostate Cancer [44, 45] has underscored the urgent need to reassess diagnostic pathways to enhance accuracy, reduce missed diagnoses of csPCa, and minimise the overdiagnosis and overtreatment of indolent disease. Given that ~4 million prostate MRIs are performed annually, this issue has significant global health implications. Incorporating novel strategies such as micro‐US‐TBx could alleviate scanner and workforce demands, broaden access to imaging, and even yield substantial cost savings [46].

While our analysis did not assess the cost‐effectiveness of micro‐US‐TBx for the detection of csPCa, the substantial findings—particularly from the OPTIMUM trial—suggest that micro‐US‐TBx warrants evaluation from a health economics perspective. Future studies should assess its viability as a complement to, or potentially a substitute for, MRI within the csPCa screening pathway.

Active surveillance is endorsed by major urological guidelines as the preferred strategy for managing patients with ciPCa [5, 7]. The role of micro‐US in this setting has been evaluated against mpMRI for detecting progression from GG 1 to GG ≥2, and a PRI‐MUS threshold ≥3 achieved 97% sensitivity for identifying csPCa [47]. In this study, no significant difference was observed between the two modalities in identifying GG ≥2 lesions; however, micro‐US using a PRI‐MUS threshold ≥3 demonstrated greater sensitivity than mpMRI with a PI‐RADS threshold ≥3 (86% vs 84%, respectively). These findings suggest that micro‐US may offer superior sensitivity in the active surveillance population, with additional advantages in terms of cost‐effectiveness and accessibility to be further investigated in future studies.

Recent studies have focused on optimising the detection of csPCa without increasing the diagnosis of ciPCa. The PRIME study (ClinicalTrials.gov identifier: NCT04571840) [48] demonstrated that bpMRI, a shorter and less resource‐intensive protocol, achieves comparable detection rates of csPCa to full mpMRI, while avoiding an increase in ciPCa diagnoses. Similarly, the OPTIMUM trial [11] showed that micro‐US‐guided biopsy was non‐inferior to MRI/conventional US fusion‐guided biopsy for detecting csPCa in biopsy naïve men.

This review synthesises contemporary evidence to inform recommendations on micro‐US‐TBx. Because micro‐US‐TBx is operator‐dependent and requires structured training, diagnostic performance varied across studies and correlated with operator experience. In 14 studies, urologists had completed formal training through the Exact Imaging programme, and in three studies operators reported >3 years of experience. Reader experience is a key determinant of mpMRI performance. Dedicated uroradiologists demonstrate higher inter‐reader agreement than non‐specialists, with the highest agreement seen for clearly malignant lesions (PI‐RADS 4–5) and in patients with elevated PSA density or high clinical suspicion [49]. In our review, nine studies reported radiologist experience ranging from 2 to 30 years, which may have contributed to inter‐reader variability across the included studies.

These learning‐curve effects should be considered when interpreting comparative accuracy estimates and when planning implementation. Although micro‐US enables real‐time targeting of suspicious lesions, it still demands dedicated equipment and skilled personnel, which may limit scalability for rapid assessments and routine surveillance in some settings. Where available, its high‐resolution, point‐of‐care imaging has the potential to streamline PCa diagnostic pathways and support earlier, more confident treatment decisions.

Strengths

Our study exhibits several strengths. First, in our head‐to‐head meta‐analysis, modality‐level meta‐regression (micro‐US‐TBx vs MRI‐TBx) showed no evidence of superiority for either technique in DR, sensitivity or specificity; the corresponding SROC curves largely overlapped. Second, our study updated the diagnostic accuracy performance between micro‐US‐TBx and MRI‐TBx with a large multicentre RCT [11] and 12 prospective studies [22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42] and the inclusion of a large sample of multiple centres ensure comprehensive heterogeneity. Additionally, the determination of the presence or absence of csPCa was based on most available histopathological evidence, and reduced the biased whole‐mount histopathology, which might be prone to sampling errors.

Limitations

Limitations include heterogeneity in biopsy triggers across studies. Most trials triaged men to TBx only when PI‐RADS ≥3, whereas some protocols mandated biopsies for all enrolled participants—an approach that diverges from contemporary practice. Such differences in verification strategies and disease spectrum can shift observed detection rates and test characteristics, limiting generalisability to routine care, where men with PI‐RADS 1–2 and favourable PSA density commonly forgo biopsy. Second, several studies mandated a concomitant systematic biopsy (SBx) in addition to image‐guided TBx. This design complicates estimation of the standalone (‘pure’) accuracy of micro‐US‐TBx and MRI‐TBx by introducing partial‐verification/incorporation bias, as disease ascertainment is augmented by SBx rather than the targeted approach alone. Practice patterns also vary geographically: some contemporary protocols omit SBx when performing TBx in selected patients, which limits direct comparability across studies and constrains generalisability to settings that do not routinely use SBx. Third, study personnel were not blinded to the assigned biopsy technique, introducing a risk of performance and assessment bias. Fourth, the cohorts combined biopsy naïve patients, men undergoing repeat biopsy, and individuals on active surveillance, creating heterogeneity in baseline oncological risk and disease prevalence that may affect accuracy estimates and limit generalisability. Although prior meta‐analyses [50] estimated mpMRI sensitivity at 0.90 (95% CI 0.87–0.93) and specificity at 0.64 (95% CI 0.50–0.76), our pooled estimates were lower—sensitivity 0.84 (95% CI 0.76–0.90) and specificity 0.40 (95% CI 0.27–0.55)—likely reflecting differences in reader expertise, patient spectrum, verification strategies, and threshold definitions. Other limitations include heterogeneity in reader expertise and the potential impact of learning‐curve effects related to PRI‐MUS scoring. As a relatively new and operator‐dependent imaging modality, micro‐US relies heavily on the examiner's experience for lesion classification and target selection, which may introduce inter‐operator variability and influence pooled accuracy estimates. In addition, heterogeneity in the analysed populations represents an important limitation. Most included studies did not report individual‐level data or subgroup‐specific outcomes for biopsy naïve patients, repeat‐biopsy patients, or men undergoing active surveillance, precluding the performance of subgroup analyses to further explore sources of heterogeneity.

Another limitation of this review is that biopsy histopathology was used as the reference standard in the included studies. As this reference is not independent of the index tests and does not represent the true disease status, diagnostic accuracy estimates should be interpreted with caution.

An important methodological consideration relates to inherent differences in image acquisition and interpretability between MRI‐TBx and micro‐US‐TBx. MRI datasets are digitally stored and can be reviewed retrospectively at different time points, allowing independent reassessment by multiple observers, secondary reads, and formal quality control procedures. This capacity enhances reproducibility analyses and mitigates single‐operator bias. Another limitation of this study is the potential for incorporation bias and partial verification bias, as several included studies [26, 28, 29, 30, 31, 32, 33, 34, 36, 37, 41, 42] combined SBx with TBx in their diagnostic protocols, as summarised in Table 1. Whenever possible, only data derived from TBx cores were extracted and included in the analysis. Results from SBx or random biopsies were excluded to ensure that the pooled estimates reflected the diagnostic performance of image‐guided TBx alone.

Conclusions

This systematic review and meta‐analysis provides an up‐to‐date evaluation of the diagnostic performance of micro‐US‐TBx compared with MRI‐TBx for the detection of csPCa. Our findings indicate that micro‐US‐TBx achieves detection rates and diagnostic accuracy comparable to those of MRI‐TBx, consistent with previous reports. Direct comparative analyses showed no statistically significant differences in diagnostic performance between the two modalities, suggesting similar effectiveness under comparable clinical conditions.

This review suggests that micro‐US‐TBx may represent a valuable complementary tool within contemporary PCa diagnostic pathways, given its high sensitivity, comparable DRs, and the ability to integrate systematic sampling with real‐time targeting of suspicious lesions during biopsy. Nevertheless, further cost‐effectiveness analyses are warranted to better define its role and optimal implementation in clinical practice.

Disclosure of Interests

The authors declare that they have no conflicts of interest relevant to this work. No financial, personal, or institutional relationships exist that could inappropriately influence or bias the content of this manuscript.

Supporting information

Appendix S1. Search strategy.

Table S1. Comparative imaging acquisition characteristics.

Fig. S1. The PRISMA 2020 flow diagram of the literature search and selection.

Fig. S2. The QUADAS‐2.

Fig. S3. Funnel plot of pooled DR.

Fig. S4. Certainty of evidence (GRADE) by outcome.

Fig. S5. Pooled accuracy (Violin Plot) comparing MicroUS‐TBx and MRI‐TBx.