A thorough investigation was carried out in the PubMed and Embase repositories to locate existing publications until February 2024.The search utilized the keywords ‘⁶⁸Ga-PSMA PET’, ‘mpMRI’, ‘PSMA’, and ‘Prostate Cancer’. Additional information can be found in Supplementary Table 1 . The identification of relevant articles was done through manual searches of the reference lists in the included studies.

For studies to be eligible for inclusion in this meta-analysis, they needed to evaluate the diagnostic accuracy of ⁶⁸Ga-PSMA PET and mpMRI in determining the local staging of prostate cancer. The included studies had to meet specific criteria: (1) population: patients with prostate cancer undergoing local staging; (2) interventions: use of ⁶⁸Ga-PSMA PET; (3) comparators: use of mpMRI; (4) outcomes: sensitivity and specificity; and (5) study type: retrospective studies and prospective studies.

Articles that were duplicates, lacked full texts, consisted of editorials, letters, case reports, reviews, meta-analyses, items with irrelevant titles or abstracts, and publications in languages other than English were excluded. Additionally, studies without enough data to calculate sensitivity or specificity of the imaging technique being studied were also excluded.

Two researchers assessed the studies’ quality using the QUADAS-2 tool (16, 17), which evaluates diagnostic performance in four main areas: patient selection, index test, reference standard, and flow and timing. Each study was categorized as having high, low, or unclear bias risk.

Data extraction for all included papers was carried out independently by two researchers. The data that were extracted included: (1) the author; (2) year of publication; (3) study characteristics including country, study design, analysis, outcome; (4) patient characteristics including number of patients, PSA level, mean/median age, Gleason score, reference standard; (5) technical characteristics including types of imaging tests, scanner modality for PET, scanner modality for mpMRI, radiotracer dose, image analysis, image analysis, TP, FP, FN, TN for ECE(PET), TP, FP, FN, TN for ECE(mpMRI), TP, FP, FN, TN for SVI(PET), TP, FP, FN, TN for SVI (mpMRI).

The DerSimonian and Laird method was utilized for random-effects meta-analysis to evaluate sensitivities and specificities, accounting for variability between studies. Sensitivities and specificities were subsequently converted using the Freeman-Tukey double inverse sine transformation to stabilize the variance of proportion data and improve the reliability of pooled estimates (18). Confidence intervals were determined using the Jackson technique. Heterogeneity within and across groups was assessed utilizing the Cochrane Q and I² statistics. If significant diversity was observed among the studies (P<0.10 or I² > 50%), a sensitivity analysis was performed to determine the reasons for the discrepancies.

Funnel plots and Egger’s test were used to evaluate publication bias. Statistical tests had to reach a significance level of P < 0.05. Statistical analysis and graphical representation were conducted using R software version 4.3.2.

The initial search found a total of 1138 published works. However, 277 studies were duplicates, and 850 were considered ineligible and removed from further review. After evaluating the full manuscripts of the remaining 11 studies, one was disqualified for missing critical data (TP, FP, FN, and TN). Ultimately, the meta-analysis included 10 studies (19–28) that evaluated the diagnostic efficacy of ⁶⁸Ga-PSMA PET and mpMRI. The article selection process, as detailed in the PRISMA flowchart (29), is depicted in Figure 1 .

The 10 qualifying studies included 505 patients diagnosed with prostate cancer, aged between 24 and 81. Out of these studies, 8 were retrospective (19, 21, 22, 24–28), while 2 were prospective (20, 23). All 10 studies used patient-based analyses, consistently referring to pathology as the standard for analysis. 8 studies (21–27) provided data on the sensitivity and specificity of two diagnostic tools for extracapsular extension. Furthermore, 8 studies included statistics on the sensitivity and specificity for seminal vesicle invasion (19–21, 23–26, 28). The details of the study and methodology for ⁶⁸Ga-PSMA PET and mpMRI were outlined in Tables 1 , 2 .

Figure 2 displays the potential bias in each study as determined by the QUADAS-2 tool. Eight studies were deemed ‘unclear’ in terms of patient selection due to a lack of information on the inclusion of consecutive patients, indicating potential selection bias. All 10 studies received an ‘unclear’ rating for the index test because visual grading thresholds could not be identified, raising concerns about detection bias. Regarding the reference standard, all 10 studies were also rated as ‘unclear’ due to uncertainty about whether the final diagnosis was independently made by two or more doctors, which may impact the reliability of the reference standard. Seven studies were classified as ‘unclear’ in terms of flow and timing criteria due to uncertainty about the correct timing between the diagnostic test and the gold standard evaluation, which could introduce variability in diagnostic accuracy. Overall, while these potential biases may have some impact, the quality of the included literature is generally acceptable.

The evaluation comprised 8 research studies, with a sensitivity of 0.56 (95% CI:0.41-0.71) in identifying ECE of prostate cancer through ⁶⁸Ga-PSMA PET, whereas mpMRI demonstrated an overall sensitivity of 0.57 (95% CI:0.43-0.71) ( Figure 3 ). There was no notable significant difference in sensitivity between ⁶⁸Ga-PSMA PET and mpMRI (P = 0.89) ( Figure 3 ).

The overall sensitivity of ⁶⁸Ga-PSMA PET and mpMRI demonstrated I² percentages of 67% and 56% correspondingly. Upon leave-one-out sensitivity analysis, the I² percentage for ⁶⁸Ga-PSMA PET dropped to 33% when Chen’s study was omitted, indicating it may have caused heterogeneity. Likewise, excluding Yilmaz’s or Muehlematter’s study led to reduced I² percentages for mpMRI, reaching 42% and 44% respectively, suggesting each study could have contributed to heterogeneity ( Supplementary Figures 1, 2 ).

The examination revealed that the specificity was 0.84 (95% CI:0.75-0.91), with mpMRI demonstrating a comparable specificity of 0.84 (95% CI:0.76-0.91). There was no notable difference in overall specificity between ⁶⁸Ga-PSMA PET and mpMRI (P = 0.93) ( Figure 4 ).

The specificity of PET and mpMRI demonstrated I² percentages of 57% and 55%, respectively. Upon further leave-one-out sensitivity analysis, it was found that excluding either Arslan’s or Celen’s research decreased the I² value for ⁶⁸Ga-PSMA PET to 34% and 40%, respectively, indicating they may be sources of heterogeneity. Similarly, the removal of Stasiak’s or Koseoglu’s study led to a reduction in I² values for mpMRI to 34% and 45%, respectively, highlighting their potential impact on heterogeneity. ( Supplementary Figures 3, 4 ).

The analysis of 8 research studies revealed that ⁶⁸Ga-PSMA PET had a sensitivity of 0.57 (95% CI:0.46-0.68) for detecting SVI in prostate cancer, while mpMRI had a sensitivity of 0.70 (95% CI:0.60-0.80), as shown in Figure 5 . There was no significant difference in sensitivity between the two imaging methods (P = 0.09) ( Figure 5 ).

The collective responsiveness of PET and mpMRI indicated I² values of 36% and 0%, respectively, indicating that both techniques exhibit acceptable heterogeneity in identifying SVI.

Analysis of 8 studies with 417 patients found that ⁶⁸Ga-PSMA PET had an specificity of 0.92 (95% CI:0.86-0.96) in detecting SVI in prostate cancer, while mpMRI had an specificity of 0.94 (95% CI:0.89-0.98), as illustrated in Figure 6 . The methods did not show any significant difference in specificity, with a P value of 0.57.

The overall specificity of ⁶⁸Ga-PSMA PET and mpMRI showed I² values of 62% and 55%, in that order. Upon leave-one-out sensitivity analysis, the I² value for ⁶⁸Ga-PSMA PET dropped to 0% when excluding Stasiak’s study, indicating it may be a significant factor contributing to heterogeneity. For mpMRI, after omitting Ucar’s study or Koseoglu’s study, the I² value reduced to 27% and 44%, respectively.( Supplementary Figures 5, 6 )

The information in Table 3 outlines the effectiveness of ⁶⁸Ga-PSMA PET/CT and mpMRI in detecting ECE in prostate cancer, with a collective sensitivity of 0.43 (95% CI:0.24-0.64) and 0.66 (95% CI:0.41-0.87) respectively. There was no significant difference in sensitivity between these methods (P = 0.17). Additionally, ⁶⁸Ga-PSMA PET/MRI demonstrated a combined sensitivity of 0.67 (95% CI:0.50-0.82), while mpMRI had a sensitivity of 0.51 (95% CI:0.40-0.62). This difference in sensitivity was also not significant (P = 0.12).

It is demonstrated that ⁶⁸Ga-PSMA PET/CT had a combined specificity of 0.78 (95% CI:0.56-0.95) for assessing ECE specificity, while mpMRI had a specificity of 0.75 (95% CI:0.66-0.83). The specificity difference between these techniques was not statistically significant (P = 0.78). Additionally, ⁶⁸Ga-PSMA PET/MRI exhibited a combined specificity of 0.87 (95% CI:0.80-0.92), compared to 0.90 (95% CI: 0.84-0.95) for mpMRI. This specificity difference was also not statistically significant (P = 0.41).

We found that ⁶⁸Ga-PSMA PET/CT had a sensitivity of 0.58 (95% CI:0.37-0.77) for detecting SVI in prostate cancer, while mpMRI had a sensitivity of 0.76 (95% CI:0.63-0.86). The difference in sensitivity was not considered statistically significant (P = 0.13). Additionally, the sensitivity of ⁶⁸Ga-PSMA PET/MRI was 0.68 (95% CI:0.49-0.86) compared to mpMRI’s 0.57 (95% CI:0.37-0.76), with no significant difference observed (P = 0.44).

Our analysis of SVI specificity revealed that ⁶⁸Ga-PSMA PET/CT and mpMRI both had a specificity of 0.91 (95% CI:0.80-0.98 and 0.83-0.97, respectively). The difference in specificity between these imaging techniques was not significant (P = 0.96). Additionally, ⁶⁸Ga-PSMA PET/MRI had a specificity of 0.93 (95% CI: 0.88-0.97) while mpMRI had a specificity of 0.97 (95% CI:0.93-0.99), with no statistically significant difference in specificity (P = 0.12). All data are presented in Table 3 .

The analysis of funnel plot asymmetry showed that there was no significant publication bias detected for the majority of outcomes, as all Egger’s test findings were above 0.05.Nevertheless, there were signs of bias in the publication of data regarding the sensitivity of ⁶⁸Ga-PSMA PET in identifying ECE and SVI, as shown by Egger’s test results of below 0.001 and 0.05, respectively (refer to Supplementary Figures 7-14 ).