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

Abstract

Background and Aims: The aims of this study were to establish a maximum standardized uptake value (SUV_max) cutoff to discriminate clinically significant prostate cancer (csPCa) from benign prostate disease (BPD) by ⁶⁸Ga-labeled prostate-specific membrane antigen (⁶⁸Ga-PSMA-11) positron emission tomography/computed tomography (PET/CT) in patients with suspected prostate cancer (PCa), and to perform a prospective real-world validation of this cutoff value. Methods: The study included a training cohort to identify an SUV_max cutoff value and a prospective real-world cohort to validate it. A retrospective analysis assessed 135 patients with suspected PCa in a large tertiary care hospital in China who underwent ⁶⁸Ga-PSMA-11 PET/CT. All patients were suspected of having PCa based on symptoms, digital rectal examination (DRE), total prostate-specific antigen (tPSA) level, and multiparameter magnetic resonance imaging (mpMRI). The ⁶⁸Ga-PSMA PET/CT results were evaluated using histopathological results from transrectal ultrasound-guided 12-core biopsy with necessary targeted biopsy as references. Patients with Gleason scores (GS) ≥7 from the biopsy results were diagnosed with csPCa, and patients with negative biopsy and follow-up results were diagnosed with BPD. Receiver operating characteristic (ROC) curve analysis was used to identify the optimal SUV_max cutoff value. The cutoff value was prospectively validated in 58 patients with suspected PCa. The diagnostic benefits of the cutoff value for clinical decision making were also evaluated. Results: According to ROC curve analysis, the most appropriate SUV_max cutoff value for discriminating csPCa from BPD was 5.30 (sensitivity, 85.85%; specificity, 86.21%; area under the curve [AUC], 0.893). The cutoff achieved a sensitivity of 83.33%, a specificity of 81.25%, a positive predictive value (PPV) of 92.11%, a negative predictive value (NPV) of 65.00%, and an accuracy of 82.76% in the prospective validation cohort. Metastases were used as an indicator to reduce false negative results in patients with SUV_max ≤ 5.30. In patients without metastases, an SUV_max value of 5.30 was also the best cutoff to diagnose localized csPCa (sensitivity, 80.43%; specificity, 86.21%; AUC, 0.852). The cutoff discriminated localized csPCa from BPD with a sensitivity of 76.19%, a specificity of 81.25%, a PPV of 84.21%, an NPV of 72.22%, and an accuracy of 78.38% in the prospective validation cohort. The cutoff, combined with metastases, achieved an accuracy of 89.12% in all patients, increasing accuracy by 8.29% and reducing equivocal results compared with manual reading. There was a strong correlation between SUV_max and PSMA expression (r_s = 0.831, P < 0.001) and a moderate correlation between SUV_max and GS (r_s = 0.509, P < 0.001). The PSMA expression and SUV_max values of patients with csPCa were significantly higher than those of patients with BPD (P < 0.001). Conclusion: We established and prospectively validated the best SUV_max cutoff value (5.30) for discriminating csPCa from BPD with high accuracy in patients with suspected PCa. 5.30 is an effective cutoff to discriminate csPCa patients with or without metastases. The cutoff may provide a potential tool for the precise identification of csPCa by ⁶⁸Ga-PSMA PET/CT, ensuring high accuracy and reducing equivocal results.

Keywords: PSMA PET/CT; SUVmax; benign prostate hypertrophy; cutoff; immunohistochemistry; prostate cancer.

Figure 1

Study design.

Figure 2

Scatter dot plots depicting H-score according to pathological diagnosis (A); SUV_max according to pathological diagnosis (B), H-score (C), intensity of PSMA staining (D), and percentage of PSMA-stained cells (E); and the percentage of stained cells according to intensity of staining (F). The vertical borders of the box represent the standard deviation, and the middle bar represents the mean value. The H-scores and SUV_max values of patients with lcsPCa and mcsPCa were significantly higher than those with BPD. The SUV_max values were higher in prostatic tissues with higher H-scores, intensities of PSMA staining, and percentages of stained cells than in tissues with lower values. The percentage of stained cells was higher in prostatic tissues with more intense PSMA staining. The detailed comparison data are shown in Tables S4A-C (*, P < 0.05; **, P < 0.01; ***, P < 0.001). SUV_max was closely positively correlated with PSMA expression, as validated by IHC staining (n = 193).

Figure 3

Example ⁶⁸Ga-PSMA PET/CT images and corresponding IHC and HE staining results showing that SUV_max is significantly correlated with PSMA expression in prostatic tissues. ⁶⁸Ga-PSMA PET/CT images (A, D, G, J, M), PSMA staining results (B, E, H, K, N), and HE staining results (C, F, I, L, O) of one patient with BPD (A-C) and four patients with PCa (D-O). The first patient (A, SUV_max = 2.90) was pathologically diagnosed with BPD (B, negative staining, 0% stained cells, tPSA = 19.09 ng/mL), while the remaining four patients (C, E, G, I; SUV_max = 3.00, 6.50, 14.50, 60.00, respectively) had pathologically proven PCa (D, F, H, J; GS = 6 (3 + 3), 6 (3 + 3), 7 (4 + 3), 8 (4 + 4); tPSA = 7.43, 19.09, 8.04, 936.10 ng/mL; percentage of stained cells = 15%, 35%, 57%, 95%; intensity of staining = 1, 2, 3, 4). The above representative patients were used to show the close correlation between PSMA expression and SUV_max. Detailed analysis of all patients is shown in Figure 2.

Figure 4

Correlations between GS and H-score (A, C) and SUV_max (B, D). The H-scores and SUV_max values of patients with GS = 7 PCa were significantly higher than those with GS = 6 PCa or BPD (GS=0). (*, P < 0.05; **, P < 0.01; ***, P < 0.001).

Figure 5

ROC curves for diagnosing patients with csPCa. (A, C) The SUV_max cutoff value of 5.30 yielded a sensitivity of 85.85% and a specificity of 86.21% in the training cohort (AUC = 0.893, Youden's index = 0.721). (B, D) The cutoff of 5.30 achieved a sensitivity of 83.33% and a specificity of 81.25% in the prospective validation cohort. The top and bottom ROC curves represent the upper and lower bounds of the 95% confidence interval of the middle bound, respectively.

Figure 6

Analysis of patients with SUV_max ≤ 5.30. (A) Pathological diagnoses of patients with SUV_max ≤ 5.30. (B) Percentage of patients with SUV_max ≤ 5.30 in each GS group. A higher percentage of patients with low GS had an SUV_max ≤ 5.30 than patients with high GS. (C) tPSA levels of patients with SUV_max ≤ 5.30. (D) ROC curves for diagnosing csPCa by tPSA. (E) Metastatic status of patients with SUV_max ≤ 5.30. (F) ROC curves for diagnosing patients with csPCa by metastatic status. The top and bottom ROC curves represent the upper and lower bounds of the 95% confidence interval of the middle bound, respectively.

Figure 7

ROC curves of patients with lcsPCa. (A, C) The SUV_max cutoff value of 5.30 yielded a sensitivity of 80.43% and a specificity of 86.21% in the training cohort (AUC = 0.852, Youden's index = 0.666). (B, D) The cutoff of 5.30 achieved a sensitivity of 76.19% and a specificity of 81.25% in the prospective validation cohort. The top and bottom ROC curves represent the upper and lower bounds of the 95% confidence interval of the middle bound, respectively.

Figure 8

Changes in clinical diagnosis based on ⁶⁸Ga-PSMA PET/CT between nuclear medicine expert experience and the combination of SUV_max cutoff and metastases. Compared with nuclear medicine expert experience, using the cutoff and metastases improved the diagnostic accuracy of clinical decision making with ⁶⁸Ga-PSMA PET/CT by 8.81% (17/193).