The value of [¹⁸F]PSMA-1007 PET/CT combined with mpMRI in diagnosing prostate cancer with PI-RADS 3–4 lesions

Introduction

Prostate cancer (PCa) is one of the most prevalent malignancies and a leading cause of cancer-related mortality among men worldwide. The precise identification of clinically significant prostate cancer (csPCa) is paramount to guide appropriate clinical management and prevent overtreatment. According to the current Chinese Society of Clinical Oncology (CSCO) Prostate Cancer Diagnosis and Treatment Guidelines (2025), multi-parametric magnetic resonance imaging (mpMRI) is endorsed as the primary (Grade I) imaging modality for the detection and localization of prostate lesions. The associated Prostate Imaging Reporting and Data System (PI-RADS) provides a standardized framework for risk stratification (1). Despite its established role, the clinical interpretation of mpMRI remains challenging, particularly for equivocal lesions. PI-RADS 3 lesions exhibit considerable heterogeneity in their association with csPCa (2), whereas a non-negligible proportion (up to 26.9%) of PI-RADS 4 lesions are ultimately confirmed as benign upon biopsy (3). These limitations underscore the need for complementary scanning strategies to enhance diagnostic precision in this specific subgroup.

Prostate-specific membrane antigen (PSMA) has emerged as a pivotal target for molecular imaging in PCa. Although the diagnostic efficacy of gallium-labeled PSMA ([⁶⁸Ga]Ga-PSMA) positron emission tomography/computed tomography (PET/CT) for intraprostatic primary lesions has been extensively validated, recent attention has shifted toward fluorine-18 prostate-specific membrane antigen-1007 ([¹⁸F]PSMA-1007). [¹⁸F]PSMA-1007, a novel PSMA-targeted tracer, offers distinct advantages including a longer half-life, favorable lesion-to-background ratios (LBRs), and minimal renal excretion, thereby facilitating superior visualization of the prostatic bed and pelvic structures (4). Notably, the application of the PRIMARY scoring system on [¹⁸F]PSMA-1007 PET/CT has demonstrated higher sensitivity and accuracy compared to mpMRI, aligning with findings from [⁶⁸Ga]Ga-PSMA studies (5). This evidence supports the role of PSMA PET/CT as a valuable adjunct to mpMRI, capable of mitigating false-negative findings and reducing unnecessary invasive biopsies.

However, the clinical utility of [¹⁸F]PSMA-1007 PET/CT specifically for PI-RADS 3–4 lesions remains to be fully defined. Existing literature has predominantly focused on 2-(3-{1-carboxy-5-[(6-[¹⁸F]fluoro-pyridine-3-carbonyl)-amino]-pentyl}-ureido)-pentanedioic acid ([¹⁸F]DCFPyL) or [⁶⁸Ga]Ga-PSMA-11 in low-risk populations (6). Consequently, there is a paucity of studies evaluating the incremental value of [¹⁸F]PSMA-1007 in this diagnostically challenging cohort, particularly regarding the differentiation of benign prostatic hyperplasia (BPH) from malignancy and the interpretation of equivocal PSMA uptake.

To address this gap, this study retrospectively analyzed data from patients with PI-RADS 3–4 lesions who underwent [¹⁸F]PSMA-1007 PET/CT scanning at our institution. The primary objective was to evaluate the diagnostic efficacy of [¹⁸F]PSMA-1007 PET/CT for PCa, and compare its performance directly with mpMRI. Furthermore, we aimed to identify predictive factors for PCa in PI-RADS 3 and 4 subgroups and develop a predictive model to assist in clinical decision-making. We present this article in accordance with the STARD reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1-2844/rc).

Methods

Participants

The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the institutional ethics board of The First Affiliated Hospital of Soochow University (No. 2024244) and the requirement for individual consent for this retrospective analysis was waived.

Male patients who underwent [¹⁸F]PSMA-1007 PET/CT and had confirmed pathological results between 2 February 2024 and 30 September 2025 were included in this retrospective study. The exclusion criteria were as follows. Firstly, to ensure that both preoperative imaging modalities reflected the same tumor status, a strict 1-month cutoff between scans was adopted based on prior literature (7). Similarly, a <3-month interval for prostate-specific antigen (PSA) measurement was required to ensure the PSA value accurately reflected the tumor burden at the time of imaging. Secondly, since the aim of this study was to evaluate diagnostic efficacy in clinically equivocal lesions, we specifically focused on PI-RADS 3 and 4 categories, which represent a “gray zone” where clinical decision-making is most challenging (8). PI-RADS 5 lesions typically proceed directly to intervention, whereas PI-RADS 1–2 lesions are usually managed by follow-up. Thirdly, patients with neoadjuvant treatments were excluded because interventions can significantly alter tumor size, cellularity, and PSMA expression, potentially confounding the intrinsic diagnostic performance of the imaging modalities. Finally, patients lacking clinical data were excluded to ensure statistical integrity, a standard practice in retrospective imaging studies, and those with other malignancies were excluded to avoid false-positive PSMA uptake from non-prostate tumors (9).

A total of 135 patients with PI-RADS 3–4 scores were finally enrolled (Figure 1). Among them, 68 were diagnosed with BPH (age: 66.62±7.66 years), and 67 were diagnosed with PCa (age: 69.79±7.63 years). Serum PSA and free prostate-specific antigen (fPSA)/total prostate-specific antigen (tPSA) levels should be collected for all patients whenever possible.

Figure 1 The flowchart of patient selection. [¹⁸F]PSMA-1007, fluorine-18 prostate-specific membrane antigen-1007; mpMRI, multi-parametric magnetic resonance imaging; PET/CT, positron emission tomography/computed tomography; PI-RADS, Prostate Imaging Reporting and Data System; PSA, prostate-specific antigen.

Qualitative and quantitative mpMRI analysis

All mpMRI images were acquired using a 3.0T scanner (Magnetom Skyra, Siemens Healthineers, Erlangen, Germany) with a standard 18-channel phased-array body coil. The imaging protocol included T1-weighted imaging (T1WI), T2-weighted imaging (T2WI), and axial diffusion-weighted imaging (DWI) imaging calculated to b1,000. Images were interpreted by two board-certified radiologists with 3 and more than 5 years of experience in prostate imaging, respectively (10). In cases of discrepancy, a consensus interpretation was reached to determine the final PI-RADS score, in accordance with the PI-RADS version 2.1 (PI-RADS v2.1). Although the readers had access to patients’ clinical information such as PSA levels, they were blinded to other imaging examination results and the pathological diagnosis.

This study enrolled patients with PI-RADS 3 (equivocal) or 4 (likely malignant) scores. In clinical practice, the management of PI-RADS 3 lesions typically involves active surveillance or short-term follow-up rather than immediate invasive biopsy, reflecting the inherent diagnostic uncertainty. Therefore, the primary objective was to evaluate whether [¹⁸F]PSMA-1007 PET/CT could provide more definitive guidance in these cases. For comparative analysis, PI-RADS 3 patients were defined as negative, and PI-RADS 4 as the higher probability group (11,12).

Prostate volume (PV) was measured using the ellipsoid formula: PV = 0.52 × maximal transverse diameter × maximal anteroposterior diameter × maximal craniocaudal diameter (13). Consistent with PI-RADS v2.1 recommendations, the maximal anteroposterior and craniocaudal diameters were measured on the midsagittal T2-weighted image, whereas the maximal transverse diameter was placed on the axial T2-weighted image. Prostate-specific antigen density (PSAD) was calculated as the ratio of tPSA to PV (13).

[¹⁸F]PSMA-1007 PET/CT scanning acquisition and analysis

Scan findings from PET/CT were acquired using a Siemens Biograph Vision 600 hybrid scanner, with [¹⁸F]PSMA-1007 as the radioactive tracer. Patients were intravenously injected with 3.7 MBq/kg of [¹⁸F]PSMA-1007 without the preparation of fasting. Scanning was initiated approximately 90 minutes post-injection with the patient in the supine position. A low-dose unenhanced CT scan was performed from the skull base to the mid-thigh for anatomical localization and attenuation correction, using the following parameters: tube voltage of 120 kV, automatic tube current modulation, and a section thickness of 5 mm. This was followed by the PET scan that matched the CT section thickness. A three-dimensional (3D) acquisition model was employed for PET scanning with a field of view of 576 mm and a matrix of 144×144. The emission scan duration was 1.5 minutes per bed positions. Finally, PET images were reconstructed using an iterative reconstruction algorithm with CT attenuation correction.

Two nuclear medicine experts with 3 and 15 years of experience, respectively, delineated the regions of interest (ROIs) on the primary prostate lesions. Both reporters were blinded to the patients’ mpMRI results. The maximum standardized uptake value (SUV_max) was automatically measured using the workstation software. The LBR was calculated as the ratio of the SUV_max of the intraprostatic lesion to the SUV_max of normal liver parenchyma. Qualitative interpretation was performed according to the PRIMARY scoring system (14), which is defined as follows: Score 1 indicates no significant uptake pattern (normal finding); Score 2 indicates diffuse intense uptake in the transition zone or central zone (non-focal); Score 3 indicates focal intense uptake in the transition zone; Score 4 indicates focal intense uptake in the peripheral zone; and Score 5 indicates SUV_max ≥12.0 (regardless of anatomical location). For the purpose of this study, a PRIMARY score of ≤2 was considered negative, and a score of ≥3 was considered positive.

Pathological reference standard

Prostate biopsy was performed using an ultrasound-guided transperineal approach. For each suspicious lesion, three targeted biopsy cores were obtained, followed by a systematic 12-core biopsy to ensure comprehensive sampling of the prostate. Specimens were reviewed independently by two professional uropathologists, whose professional experience spanned 5 years and >15 years. Tumor differentiation was evaluated utilizing the Gleason score, aligning with the criteria established during the 2019 International Society of Urological Pathology (ISUP) consensus. The final ISUP grade was categorized as follows: (I) ISUP Grade Group 1 (GG1), Gleason score ≤6; (II) ISUP Grade Group 2 (GG2), Gleason score 3+4; (III) ISUP Grade Group 3 (GG3), Gleason score 4+3; (IV) ISUP Grade Group 4 (GG4), Gleason score 4+4; and (V) ISUP Grade Group 5 (GG5), Gleason score 9–10 (15).

Statistical analysis

Statistical analysis was performed using the software SPSS 26.0 (IBM Corp., Armonk, NY, USA). Initially, the proportion of missing data for each variable was calculated, and the pattern of missingness was evaluated using Little’s test of missing completely at random (MCAR). Missing values were subsequently handled by deletion or imputation based on the assessment results.

Continuous variables that were normally distributed are presented as mean ± standard deviation and compared using independent-samples t-tests. Non-normally distributed continuous data are summarized as median with interquartile range and compared using the Mann-Whitney U test. Categorical data are presented as frequencies (percentages).

Using pathological findings as the reference standard, diagnostic agreement was assessed with Cohen’s kappa statistic. The diagnostic performance [sensitivity, specificity, accuracy, positive predictive value (PPV), and negative predictive value (NPV)] of the two imaging modalities was calculated and compared using the paired McNemar test.

To identify independent predictors for the discrimination of PCa within different PI-RADS subgroups, a two-stage regression analysis was conducted. Given the limited sample size, a variable selection strategy was employed to prevent model overfitting. Variables with a P value <0.05 in the univariate analysis were subsequently included in the multivariate logistic regression model, adhering to the 10 events per variable (EPV) rule.

The diagnostic value of individual predictors and the combined multivariate model was evaluated by calculating the area under the receiver operating characteristic (ROC) curve (AUC). Differences between AUCs were compared using the DeLong test. The optimal diagnostic cut-off value was derived by maximizing the Youden index (J = sensitivity + specificity − 1) on the ROC curve.

All statistical tests were two-sided, and a P value <0.05 was considered statistically significant.

Results

General patient characteristics

A total of 135 patients with PI-RADS 3–4 lesions were included in this study. Among them, 3 patients (2.2%) had missing fPSA/tPSA ratio data. Little’s MCAR test demonstrated that the data were MCAR [χ²=2.069, degrees of freedom (df) =6, P=0.911], and pairwise deletion was employed for subsequent correlation analyses incorporating this variable.

Pathological results confirmed 68 cases of BPH and 67 cases of PCa. For PI-RADS score distribution, within the BPH group, 63.2% of patients had a PI-RADS score of 3 and 36.8% a score of 4 (Figure 2), whereas in the PCa group, the corresponding proportions were 26.9% (Figure 3) and 73.1%, respectively.

Figure 2 A PI-RADS 4 lesion on mpMRI was positive on [¹⁸F]PSMA-1007 PET/CT (SUV_max 7.0) and confirmed to be benign prostatic hyperplasia on biopsy. (A) T2-weighted image; (B) diffusion-weighted image; (C) attenuation correction CT; (D) [¹⁸F]PSMA-1007 PET/CT image. CT, computed tomography; [¹⁸F]PSMA-1007, fluorine-18 prostate-specific membrane antigen-1007; mpMRI, multi-parametric magnetic resonance imaging; PET, positron emission tomography; PI-RADS, Prostate Imaging Reporting and Data System; SUV_max, maximum standardized uptake value.

Figure 3 A positive lesion (SUV_max 50.3) on [¹⁸F]PSMA-1007 PET/CT was reported as PI-RADS 3 on mpMRI and confirmed to be prostate cancer on biopsy. (A) T2-weighted image; (B) diffusion-weighted image; (C) attenuation correction CT; (D) [¹⁸F]PSMA-1007 PET/CT image. CT, computed tomography; [¹⁸F]PSMA-1007, fluorine-18 prostate-specific membrane antigen-1007; mpMRI, multi-parametric magnetic resonance imaging; PET, positron emission tomography; PI-RADS, Prostate Imaging Reporting and Data System; SUV_max, maximum standardized uptake value.

Intergroup comparisons demonstrated statistically significant differences between the BPH and PCa groups with respect to age at diagnosis, PSA, fPSA/tPSA, PV, prostate- PSAD, lesion SUV_max, and LBR (all P values <0.05). Although the median PSA level was higher in the PCa group compared to the BPH group (10.338 vs. 8.195 ng/mL, P=0.023), this difference was not statistically significant following stratification of patients into subgroups according to clinically used PSA thresholds (<4, 4–10, >10 ng/mL). Detailed data are provided in Table 1.

Comparison of diagnostic efficacy between [¹⁸F]PSMA-1007 PET/CT and mpMRI

Using pathology as the reference standard, the diagnostic performance of [¹⁸F]PSMA-1007 PET/CT and mpMRI for PCa in patients with PI-RADS 3–4 lesions is summarized in Table 2. [¹⁸F]PSMA-1007 PET/CT demonstrated substantial agreement with pathological results [κ=0.660, 95% confidence interval (CI): 0.537–0.783], which was superior to the moderate agreement observed for mpMRI (κ=0.363, 95% CI: 0.206–0.520). In terms of key diagnostic metrics, [¹⁸F]PSMA-1007 PET/CT exhibited significantly higher sensitivity (95.5% vs. 73.1%) and accuracy (83.0% vs. 68.1%) than mpMRI (both P<0.05). Specificity, PPV, and NPV were also higher for [¹⁸F]PSMA-1007 PET/CT compared to mpMRI.

In the PI-RADS 3 subgroup, [¹⁸F]PSMA-1007 PET/CT yielded 15 false-positive cases, including 10 lesions located in the transition zone and 5 in the peripheral zone. One false-negative case was observed, which presented as diffuse uptake in the transition zone. Likewise, in the PI-RADS 4 subgroup, there were 5 false-positive cases, with 3 located in the transition zone and 2 in the peripheral zone. Two false-negative cases were identified, characterized by diffuse uptake in the transition zone or central zone.

Predictive model analysis in PI-RADS 3 and 4 subgroups

PI-RADS 3 subgroup

This subgroup comprised 61 patients, including 43 with BPH and 18 with PCa. Univariate analysis identified lesion SUV_max, and LBR as significant predictors for differentiating PCa from BPH (all P<0.05). The multivariate logistic regression model (Table 3) was statistically significant (χ²=22.562, df=2, P<0.001) and demonstrated good fit (χ²=6.499, df=8, P=0.591). Multivariate analysis revealed a positive trend between SUV_max and the risk of PCa, although this was not statistically significant [odds ratio (OR) = 1.261, 95% CI: 0.999–1.590, P=0.051].

The diagnostic performance of individual variables and the multiparameter combined model is presented in Figure 4 and Table 4. LBR alone (cutoff >0.868) achieved the highest AUC of 0.833 (95% CI: 0.724–0.943). However, DeLong analysis revealed no statistically significant difference in diagnostic performance when compared to the multiparametric combined model (incorporating SUV_max and LBR; AUC =0.806) or SUV_max alone (AUC =0.790; cutoff >12.825).

Figure 4 ROC analysis of individual and combined predictors for PCa in PI-RADS 3 patients. LBR, lesion-to-background ratio; PCa, prostate cancer; PI-RADS, Prostate Imaging Reporting and Data System; ROC, receiver operating characteristic; SUV_max, maximum standardized uptake value.

PI-RADS 4 subgroup

This subgroup consisted of 74 patients, comprising 25 with BPH and 49 with PCa. Univariate analysis (Table 5) identified PV, lesion SUV_max, and LBR as significant predictors of PCa (all P<0.05). The multivariate model, which incorporated these three variables, was statistically significant (χ²=22.715, df=3, P<0.001) and demonstrated a good fit (χ²=6.230, df=8, P=0.622). Multivariate analysis confirmed PV as an independent predictor of PCa (OR =0.976, 95% CI: 0.952–0.999, P=0.045).

The diagnostic performance of each parameter and the combined model is summarized in Table 6. The AUC was 0.708 for PV (optimal cutoff <18.708 cm³), 0.758 for SUV_max (optimal cutoff >10.565), 0.774 for LBR (optimal cutoff >1.198), and 0.808 for the three-parameter combined model (Figure 5). According to the DeLong test, the AUC of the combined model was significantly greater than that of PV alone (P<0.01) but did not differ significantly from the AUCs of SUV_max or LBR alone.

Figure 5 ROC analysis of individual and combined predictors for PCa in PI-RADS 4 patients. LBR, lesion-to-background ratio; PCa, prostate cancer; PI-RADS, Prostate Imaging Reporting and Data System; PV, prostate volume; ROC, receiver operating characteristic; SUV_max, maximum standardized uptake value.

Discussion

mpMRI, which integrates T2WI, DWI, and dynamic contrast-enhanced imaging, remains the Grade I recommended imaging modality in the CSCO Prostate Cancer Diagnosis and Treatment Guidelines (2025). However, in this study involving patients with equivocal PI-RADS 3–4 lesions, the PRIMARY scoring system applied to [¹⁸F]PSMA-1007 PET/CT demonstrated better diagnostic efficacy compared to mpMRI. Specifically, [¹⁸F]PSMA-1007 PET/CT achieved significantly higher sensitivity (95.5% vs. 73.1%, P<0.001) and accuracy (83.0% vs. 68.1%, P=0.009) for detecting PCa. Compared to a previous study on [⁶⁸Ga]Ga-PSMA PET/CT in PI-RADS 3 lesions, [¹⁸F]PSMA-1007 exhibited markedly improved sensitivity (95.5% vs. 69%) with comparable specificity (70.6% vs. 73%), which might highlight the robust diagnostic value of the [¹⁸F] tracer in this challenging subgroup (16).

The observed discrepancies between mpMRI and [¹⁸F]PSMA-1007 PET/CT across different clinical contexts can be attributed to their distinct imaging mechanisms. mpMRI relies on morphological and structural changes, whereas PSMA PET/CT targets the PSMA, reflecting the biological activity of the tumor. This distinction is critical for PI-RADS 3 lesions, where morphological features are often atypical; in such cases, PSMA PET/CT provides essential molecular information that enhances discriminatory power. Conversely, the diagnostic accuracy of PSMA PET/CT can be compromised by variable PSMA expression. For instance, ductal prostatic carcinoma, a rare histological subtype, frequently exhibits low or heterogeneous PSMA expression, leading to potential false-negative results (17). Although no cases of ductal carcinoma were identified in our cohort, this remains a pertinent limitation when interpreting PSMA PET findings. Additionally, non-specific uptake caused by inflammatory processes or BPH remains a source of false positives.

The diagnostic performance of these modalities also varies significantly across risk stratifications. In intermediate- to high-risk patients, [¹⁸F]PSMA-1007 PET/CT has been reported to surpass mpMRI in detecting intraprostatic primary lesions (sensitivity: 85% vs. 62%, P<0.001), though no significant differences were observed for extraprostatic extension or seminal vesicle invasion (18). In contrast, for low-risk patients (ISUP Grade 1), long-term follow-up results suggested that mpMRI yielded a higher false-positive rate compared to PSMA PET/CT (40% vs. 17.5%) (19). In the specific subgroup with PSA levels of 4–10 ng/mL, [¹⁸F]PSMA-1007 PET/CT demonstrated significantly higher specificity than mpMRI (85.7% vs. 35.7%), albeit with lower sensitivity (76.5% vs. 91.2%) (20). This trade-off may be partially explained by the high prevalence of PI-RADS ≥4 lesions (63.4%) in this subgroup. Furthermore, anatomical location plays an important role; mpMRI is known to have reduced sensitivity for anteriorly located tumors, whereas [¹⁸F]PSMA-1007 PET/CT may offer improved visualization in these regions due to its molecular targeting capability.

Despite these advantages, visual interpretation of imaging remains inherently subjective and susceptible to inter-observer variability. To overcome these limitations, the integration of radiomics and deep learning has emerged as a promising frontier. Recent evidence supports the utility of a combined deep learning model integrating mpMRI and [¹⁸F]PSMA-1007 PET/CT. Such a model has demonstrated good discrimination (AUC =0.803, 95% CI: 0.734–0.861) in predicting pathological upgrading in patients with PI-RADS ≥4 and ISUP ≤4 (21). More importantly, for predicting adverse pathological features (extraprostatic extension, seminal vesicle invasion, and positive surgical margins), the combined model achieved superior discrimination (AUC =0.93) compared to models relying solely on PET/CT (AUC =0.813) or mpMRI (AUC =0.865) (22). These findings underscore the synergistic potential of fusing anatomical and molecular data. Future studies should prioritize moving beyond subjective scoring systems toward these quantitative, multiparametric frameworks to standardize assessments and refine the prediction of tumor aggressiveness.

Although imaging characteristics and advanced models define the technical landscape, the clinical decision-making process for PI-RADS 3–4 lesions is multifactorial. Therefore, beyond imaging parameters, identifying reliable clinical predictors of malignancy is essential to further refine the diagnostic algorithm.

Our study demonstrates that the semi-quantitative parameters derived from [¹⁸F]PSMA-1007 PET/CT provide distinct diagnostic value in the PI-RADS 3–4 cohort. Notably, SUV_max exhibited a tendency to serve as an independent predictor for PI-RADS 3 lesions. In the broader context of PI-RADS 3–4 scores, however, both SUV_max and LBR were important predictive factors for the presence of PCa. It is important to acknowledge that SUV_max is influenced by multiple factors, including PSMA expression levels in the lesion (23), the radiotracer used, the type of acquisition equipment, individual physiological metabolism, and scanning acquisition time (24,25).

A key observation in this study is that the SUV_max cutoff values for diagnosing csPCa with [¹⁸F]PSMA-1007 (ranging from 11.13 to 13.13) in our PI-RADS 3–4 population were notably higher than those reported for [⁶⁸Ga]Ga-PSMA-11 (4.2–6.0) in previous studies (26,27). This discrepancy may be attributed to several reasons. First, differences in radiotracer pharmacokinetics exist, as [¹⁸F]PSMA-1007 is primarily excreted via the hepatobiliary system with a slow blood clearance rate, leading to relatively high background soft-tissue activity, which may render lesions with high PSMA expression more visually prominent. In contrast, [⁶⁸Ga]Ga-PSMA-11 is predominantly excreted renally. Second, variations in the study cohort likely contributed to the results. This study focused on a higher-risk cohort (PI-RADS 3–4), contrasting with the broader inclusion criteria (PI-RADS 1–5) in previous studies.

LBR can eliminate systematic bias arising from individual metabolic variations and reconstruction algorithms (28). In the PI-RADS 3 subgroup, LBR (AUC =0.833) demonstrated superior diagnostic performance compared to SUV_max (AUC =0.790). The lack of synergistic effect in the combined model suggests a strong correlation between the two variables, rendering SUV_max redundant in this context. The inclusion of SUV_max likely offered no complementary diagnostic value and may have reduced the model’s predictive power through multicollinearity.

Meanwhile, establishing subgroup-specific cutoff values may improve the generalizability and accuracy of these metrics. This involves stratifying patients based on lesion location, clinical risk, or imaging features to calculate distinct thresholds. For instance, Meng et al. calculated separate SUV_max cutoffs for each PI-RADS stratum and found that this stratified approach improved diagnostic accuracy (27). Similarly, our data suggest that the optimal SUV_max threshold is higher for PI-RADS 3 lesions (12.825; specificity 97.7%) than it is for PI-RADS 4 lesions (10.565; specificity 84%). This may reflect an intentional increase in specificity to mitigate false positives from benign conditions in lower-risk subgroups.

Notably, PV was identified as a negative predictor of PCa in the PI-RADS 3–4 cohort, yielding an AUC of less than 0.5. This inverse relationship suggests that as PV increases, the probability of BPH rises while the risk of PCa declines. Such a trend is consistent with the pathological distinction between the focal infiltrative nature of PCa and the diffuse proliferative pattern of BPH. Given the high prevalence of BPH in middle-aged and older men, relying solely on prostate size may lead to misdiagnosis. In theory, PSAD, which corrects for volume effects, should offer superior diagnostic value compared to PSA alone.

Multiparameter models that integrate multidimensional data represent a promising approach to overcoming the limitations of individual variables and improving diagnostic performance. Prior research has demonstrated, for example, that combining PSAD with [⁶⁸Ga]Ga-PSMA-11 SUV_max adds diagnostic value in evaluating PI-RADS 1–2 lesions (29). Furthermore, predictive models built on such parameters can effectively guide clinical decisions; one study reported that 57 patients with a model-derived probability >0.6 underwent biopsy-naïve radical prostatectomy, with pathology confirming the predictions in all cases (30,31). The parameters incorporated into such models have since expanded to include clinical indicators, and predictive tools based either on clinical data alone or combined with radiomics have shown accurate risk stratification (32-34). For PI-RADS 4 patients, the multiparametric approach proved superior to standalone mpMRI metrics, with the AUC improving significantly from 0.708 (using PV alone) to 0.808 (P<0.001). Conversely, when applied to the PI-RADS 3–4 group, the combined model did not show an advantage over PSMA PET/CT.

Our study has some limitations. Focusing exclusively on PI-RADS 3–4 lesions and classifying PI-RADS 3 as negative may introduce spectrum bias, which precludes direct comparison with studies involving the full PI-RADS spectrum and lowers the sensitivity of mpMRI. Besides, the conservative strategy for variable selection using univariate model risks omitting predictors that showed only marginal significance. Our study was performed in a single center without external validation and calibration or decision curve analysis for model evaluation, which could limit the generalizability of our findings. In addition, our analysis relied on conventional clinical metrics (e.g., PSA, SUV_max) and did not incorporate advanced quantitative features, such as radiomics signatures from PSMA PET/CT or mpMRI, potentially missing critical predictive information.

Future studies should validate diagnostic consistency across different PI-RADS strata and use advanced algorithms such as least absolute shrinkage and selection operator (LASSO) for feature selection in larger cohorts. Comprehensive evaluation, including calibration and external validation, is essential to enhance generalizability. Additionally, integrating radiomics into multiparametric models offers a promising approach to improve risk stratification in equivocal cases.

Conclusions

[¹⁸F]PSMA-1007 PET/CT demonstrates higher sensitivity and accuracy compared to mpMRI in patients with PI-RADS 3–4 lesions, offering a potential diagnostic strategy for this challenging cohort. Notably, for patients with PI-RADS 4 lesions who have negative biopsy histology but persistent clinical suspicion, [¹⁸F]PSMA-1007 PET/CT may provide diagnostic value by detecting lesions. However, the widespread clinical implementation of this modality is currently constrained by its high cost and limited availability in many regions. Therefore, PSMA PET/CT should be strategically considered when the diagnostic benefits outweigh these practical limitations. Future multicenter prospective studies are warranted to validate these findings and establish cost-effective clinical pathways.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the STARD reporting checklist. Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1-2844/rc

Data Sharing Statement: Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1-2844/dss

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1-2844/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the institutional ethics board of The First Affiliated Hospital of Soochow University (No. 2024244) and individual consent for this retrospective analysis was waived.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Yamaya N, Kimura K, Ichikawa R, Kawanishi M, Kawasaki Y, Higuchi S, Fukui K, Tsuchiya J, Kobayashi M, Yoshida S, Fujii Y, Tateishi U. Prospective evaluation of PI-RADSv2.1 using multiparametric and biparametric MRI for detecting clinically significant prostate cancer based on MRI/US fusion-guided biopsy. Jpn J Radiol 2025;43:472-82. [Crossref] [PubMed]
2. Maggi M, Panebianco V, Mosca A, Salciccia S, Gentilucci A, Di Pierro G, Busetto GM, Barchetti G, Campa R, Sperduti I, Del Giudice F, Sciarra A. Prostate Imaging Reporting and Data System 3 Category Cases at Multiparametric Magnetic Resonance for Prostate Cancer: A Systematic Review and Meta-analysis. Eur Urol Focus 2020;6:463-78. [Crossref] [PubMed]
3. Wang X, Liu W, Lei Y, Wu G, Lin F. Assessment of prostate imaging reporting and data system version 2.1 false-positive category 4 and 5 lesions in clinically significant prostate cancer. Abdom Radiol (NY) 2021;46:3410-7. [Crossref] [PubMed]
4. Yang J, Tang Y, Zhou C, Zhou M, Li J, Hu S. The use of 68 Ga-PSMA PET/CT to stratify patients with PI-RADS 3 lesions according to clinically significant prostate cancer risk. Prostate 2023;83:430-9. [Crossref] [PubMed]
5. Chen Y, Zhang X, Bi J, Wang J. Diagnostic Performance of PET/CT Using Different Radiotracers in Clinically Significant Primary Prostate Cancer: A Systematic Review and Network Meta-analysis. Acad Radiol 2025;32:4678-88. [Crossref] [PubMed]
6. Fu Y, Zhao M, Chen J, Wen Q, Chen B. Enhancing prostate cancer diagnosis and reducing unnecessary biopsies with [18F]DCFPyL PET/CT imaging in PI-RADS 3/4 patients. Sci Rep 2024;14:15525. [Crossref] [PubMed]
7. Yao F, Zhu D, Lin H, Miao C, Lin Q, Li T, Zhuang Y, Bian S, Yang Y, Lin J, Pan K. Deep learning algorithm assisting diagnosis of prostate cancer extracapsular extension based on [18F]PSMA-1007 PET/CT and multiparametric MRI: A multicenter study. Prostate Cancer Prostatic Dis 2026;29:440-8. [Crossref] [PubMed]
8. Fazekas T, Shim SR, Basile G, Baboudjian M, Kói T, Przydacz M, Abufaraj M, Ploussard G, Kasivisvanathan V, Rivas JG, Gandaglia G, Szarvas T, Schoots IG, van den Bergh RCN, Leapman MS, Nyirády P, Shariat SF, Rajwa P. Magnetic Resonance Imaging in Prostate Cancer Screening: A Systematic Review and Meta-Analysis. JAMA Oncol 2024;10:745-54. [Crossref] [PubMed]
9. Perry E, Talwar A, Sharma S, O’Connor D, Wong LM, Taubman K, Sutherland TR. Non-prostate cancer tumours: incidence on 18F-DCFPyL PSMA PET/CT and uptake characteristics in 1445 patients. Eur J Nucl Med Mol Imaging 2022;49:3277-88. [Crossref] [PubMed]
10. Pepe P, Candiano G, Pepe L, Pennisi M, Fraggetta F. mpMRI PI-RADS score 3 lesions diagnosed by reference vs affiliated radiological centers: Our experience in 950 cases. Arch Ital Urol Androl 2021;93:139-42. [Crossref] [PubMed]
11. Noujeim JP, Belahsen Y, Lefebvre Y, Lemort M, Deforche M, Sirtaine N, Martin R, Roumeguère T, Peltier A, Diamand R. Optimizing multiparametric magnetic resonance imaging-targeted biopsy and detection of clinically significant prostate cancer: the role of perilesional sampling. Prostate Cancer Prostatic Dis 2023;26:575-80. [Crossref] [PubMed]
12. Mourato FA, Schmitt LG, Mariussi M, Torri G, Altmayer S, Metser U, Veit-Haibach P, Brandão S, Purysko AS, Woo S, Dias AB. The PRIMARY Scoring System on PSMA PET for Clinically Significant Prostate Cancer Detection: A Systematic Review and Meta-Analysis. Clin Nucl Med 2026;51:406-15. [Crossref] [PubMed]
13. Turkbey B, Rosenkrantz AB, Haider MA, Padhani AR, Villeirs G, Macura KJ, Tempany CM, Choyke PL, Cornud F, Margolis DJ, Thoeny HC, Verma S, Barentsz J, Weinreb JC. Prostate Imaging Reporting and Data System Version 2.1: 2019 Update of Prostate Imaging Reporting and Data System Version 2. Eur Urol 2019;76:340-51. [Crossref] [PubMed]
14. Fendler WP, Eiber M, Beheshti M, Bomanji J, Calais J, Ceci F, et al. PSMA PET/CT: joint EANM procedure guideline/SNMMI procedure standard for prostate cancer imaging 2.0. Eur J Nucl Med Mol Imaging 2023;50:1466-86. [Crossref] [PubMed]
15. van Leenders GJLH, van der Kwast TH, Grignon DJ, Evans AJ, Kristiansen G, Kweldam CF, et al. The 2019 International Society of Urological Pathology (ISUP) Consensus Conference on Grading of Prostatic Carcinoma. Am J Surg Pathol 2020;44:e87-99. [Crossref] [PubMed]
16. Kawada T, Yanagisawa T, Rajwa P, Sari Motlagh R, Mostafaei H, Quhal F, Laukhtina E, Aydh A, König F, Pallauf M, Pradere B, Ceci F, Baltzer PAT, Hacker M, Rasul S, Karakiewicz PI, Araki M, Nasu Y, Shariat SF. Diagnostic Performance of Prostate-specific Membrane Antigen Positron Emission Tomography-targeted biopsy for Detection of Clinically Significant Prostate Cancer: A Systematic Review and Meta-analysis. Eur Urol Oncol 2022;5:390-400. [Crossref] [PubMed]
17. Pepe P, Pepe L, Curduman M, Pennisi M, Fraggetta F. Ductal prostate cancer staging: Role of PSMA PET/CT. Arch Ital Urol Androl 2024;96:12132. [Crossref] [PubMed]
18. Exterkate L, Hermsen R, Küsters-Vandevelde HVN, Prette JF, Baas DJH, Somford DM, van Basten JA. Head-to-Head Comparison of 18F-PSMA-1007 Positron Emission Tomography/Computed Tomography and Multiparametric Magnetic Resonance Imaging with Whole-mount Histopathology as Reference in Localisation and Staging of Primary Prostate Cancer. Eur Urol Oncol 2023;6:574-81. [Crossref] [PubMed]
19. Pepe P, Pepe L, Tamburo M, Marletta G, Savoca F, Pennisi M, Fraggetta F. 68Ga-PSMA PET/CT evaluation in men enrolled in prostate cancer Active Surveillance. Arch Ital Urol Androl 2023;95:11322. [Crossref] [PubMed]
20. Luo L, Wang R, Bai L, Shang J, Wang X, Chang R, Dong W, Li Y, Li Y, Liang H, Xie H, Duan X. The accuracy of fluorine 18-labelled prostate-specific membrane antigen PET/CT and MRI for diagnosis of prostate cancer in PSA grey zone. Br J Cancer 2025;132:253-8. [Crossref] [PubMed]
21. Xu J, Chen H, Chen L, Li T, Lin H, Bian S, Lin Q, Zhuang Y, Xue Y, Yang Y, Su X, Yao F. The predictive value of multiparametric MRI combined with (18)FPSMA-1007 PET/CT for the pathological upgrade in prostate cancer: a multicenter study. Eur J Nucl Med Mol Imaging 2025;52:4425-33. [Crossref] [PubMed]
22. Lin H, Yao F, Yi X, Yuan Y, Xu J, Chen L, Wang H, Zhuang Y, Lin Q, Xue Y, Yang Y, Pan Z. Prediction of adverse pathology in prostate cancer using a multimodal deep learning approach based on (18)FPSMA-1007 PET/CT and multiparametric MRI. Eur J Nucl Med Mol Imaging 2025;52:2814-25. [Crossref] [PubMed]
23. Ahmadi Bidakhvidi N, Gevaert T, De Schepper M, Baldewijns M, Havinga E, Deckers W, Laenen A, Devos G, Giesen A, Joniau S, Koole M, Everaerts W, Deroose CM, Goffin K. Comparison of PSMA immunohistochemistry scoring systems to parametric [18F]PSMA-1007 PET/MRI in primary prostate cancer. Eur J Nucl Med Mol Imaging 2025;52:766-78. [Crossref] [PubMed]
24. Adams MC, Turkington TG, Wilson JM, Wong TZ. A systematic review of the factors affecting accuracy of SUV measurements. AJR Am J Roentgenol 2010;195:310-20. [Crossref] [PubMed]
25. Nassour AJ, Jain A, Khanani H, Hui N, Thompson NJ, Sorensen B, Baskaranathan S, Bergersen P, Chalasani V, Dean T, Dias M, Wines M, Symons J, Tarlinton L, Woo H. Impact of Uptake Period on 18F-DCFPyL-PSMA PET/CT Maximum Standardised Uptake Value. Cancers (Basel) 2025;17:960. [Crossref] [PubMed]
26. Jiao J, Kang F, Zhang J, Quan Z, Wen W, Zhao X, Ma S, Wu P, Yang F, Guo W, Yang X, Yuan J, Shi Y, Wang J, Qin W. Establishment and prospective validation of an SUVmax cutoff value to discriminate clinically significant prostate cancer from benign prostate diseases in patients with suspected prostate cancer by 68Ga-PSMA PET/CT: a real-world study. Theranostics 2021;11:8396-411. [Crossref] [PubMed]
27. Meng X, Ma W, Zhang J, Quan Z, Zhang M, Ye J, Shu J, Ren J, Qin W, Kang F, Wang J. PI-RADS-Based Segmented Threshold of PSMA-PET SUVmax Is Better than Traditional Fixed Threshold for Diagnosing Clinically Significant Prostate Cancer Especially for PI-RADS 3 Lesions. Mol Imaging Biol 2023;25:887-96. [Crossref] [PubMed]
28. Zhao J, Hamm B, Brenner W, Makowski MR. Lesion-to-background ratio threshold value of SUVmax of simultaneous [68Ga]Ga-PSMA-11 PET/MRI imaging in patients with prostate cancer. Insights Imaging 2020;11:137. [Crossref] [PubMed]
29. Hu W, Guo S, Meng X, Wang F, Ma S, Zhang C, et al. Development and validation of a visual nomogram for predicting clinically significant prostate cancer in negative mpMRI using 68Ga-PSMA PET/CT. Sci Rep 2025;15:27453. [Crossref] [PubMed]
30. Wang C, Xie Q, Yuan L, Ni M, Zhuo D, Gao Y, Liu Y, Liu X, Ma Y, Xiao J, Tao T. Radical prostatectomy without prostate biopsy based on a noninvasive diagnostic strategy: a prospective single-center study. Prostate Cancer Prostatic Dis 2025;28:496-502. [Crossref] [PubMed]
31. Wang CM, Yuan L, Liu XH, Chen SQ, Wang HF, Dong QF, Zhang B, Huang MS, Zhang ZY, Xiao J, Tao T. Developing a diagnostic model for predicting prostate cancer: a retrospective study based on Chinese multicenter clinical data. Asian J Androl 2024;26:34-40. [Crossref] [PubMed]
32. Zamboglou C, Drendel V, Jilg CA, Rischke HC, Beck TI, Schultze-Seemann W, Krauss T, Mix M, Schiller F, Wetterauer U, Werner M, Langer M, Bock M, Meyer PT, Grosu AL. Comparison of 68Ga-HBED-CC PSMA-PET/CT and multiparametric MRI for gross tumour volume detection in patients with primary prostate cancer based on slice by slice comparison with histopathology. Theranostics 2017;7:228-37. [Crossref] [PubMed]
33. Li Y, Li J, Yang J, Xiao L, Zhou M, Cai Y, Rominger A, Shi K, Seifert R, Gao X, Tang Y, Hu S. Using a novel PSMA-PET and PSA-based model to enhance the diagnostic accuracy for clinically significant prostate cancer and avoid unnecessary biopsy in men with PI-RADS ≤3 MRI. Eur J Nucl Med Mol Imaging 2025;52:913-24. [Crossref] [PubMed]
34. Guo S, Zhang J, Wang Y, Jiao J, Li Z, Cui C, Chen J, Yang W, Ma S, Wu P, Jing Y, Wen W, Kang F, Wang J, Qin W. Avoiding unnecessary biopsy: the combination of PRIMARY score with prostate-specific antigen density for prostate biopsy decision. Prostate Cancer Prostatic Dis 2024;27:288-93. [Crossref] [PubMed]