Detection of prostate cancer with ¹⁸F-DCFPyL PET/CT compared to final histopathology of radical prostatectomy specimens: is PSMA-targeted biopsy feasible? The DeTeCT trial

This was a prospective, non-randomised study in patients with diagnosed primary PCa. Pre-operative imaging results were compared to histopathology following RARP. All subjects signed informed-consent for the collection of their clinical data. The study has been approved by the ethical review board of the Amsterdam University Medical Centre (AUMC) (review number 2017.543). Patients were enrolled consecutively between May 2018-May 2019 in Amsterdam UMC, location VUmc.

Patients had histologically proven, intermediate or high-risk, PCa, for which they underwent RARP [3, 15]. Of all included patients, age, prostate volume, initial prostate-specific antigen (PSA)-level, clinical T-stage, pathological biopsy features (histopathological grade, number of cores with cancer) and European Association of Urology (EAU)-risk category were collected [3, 15]. A 12-segment anatomic mapping model of the prostate was used to localise and characterise the prostatic tumours, with 2 additional segments representing the seminal vesicles (pT3b) (Appendix-1 in ESM) [16].

Patients were staged with ¹⁸F-DCFPyL which was synthesised under Good Manufacturing Practices conditions, as described by Jansen et al. [17, 18]. PET images were acquired at a median of 118 min after injection of the radiotracer (interquartile range [IQR] 113–122 min) with a median dose of 313 MBq ¹⁸F-DCFPyL (IQR 299–324 MBq), and a median of 5.4 weeks (IQR 3.0–7.2) prior to surgery. Image-acquisitions were performed using a Philips Ingenuity TF (Philips Healthcare®, NL/USA)-PET/CT system. No diuretics were administered prior to the scan. The scan trajectory included mid-thighs to skull-base, with 4 min per bed position. All PET scans were combined with a diagnostic CT scan (110 mAs, 120 kV), without contrast-enhancement. Images were corrected for decay, scatter, random coincidences, and photon attenuation.

Images were reconstructed with a BLOB-based Ordered-Subsets Expectations Maximization algorithm (Philips, 3 iterations; 33 subsets) [19]). The default Ordered-Subsets Expectations Maximization with Time-of-Flight reconstruction was used. Reconstructions included both 4 mm for semi-quantification purposes, and 2 mm slices for visual interpretation (matrix size 144 × 144, slice thickness 4 mm; matrix size 288 × 288, slice thickness 2 mm).

Scan interpretation was performed blinded for the pathology results and other imaging by two nuclear medicine physicians (DO,GZ) with ample experience in ¹⁸F-DCFPyL-PET/CT reading (> 300 scans), in consensus. The readers used the 12-segment mapping model to demarcate the image-detected tumour extent (Appendix-1 in ESM) [20]. For all positive segments, the readers’ diagnostic confidence was evaluated using a five-point scale, alike the PSMA-RADS classification [21] (score 1–2 ‘benign’; 3 ‘equivocal’; 4–5 ‘likely PCa’). PSMA-RADS 4–5 were defined as suspicious for PCa (‘positive’), and were used for the final diagnostic accuracy analysis. Based on PET/CT-imaging, two segments per patient were indicated to be potentially targeted by prostate biopsy. These segments were selected based on both visual interpretation (location, size) and semi-quantification by determining the highest standardised uptake values (SUV_max) of the suspected lesions. Finally, the readers indicated if radiological extra-capsular extension (ECE, rT-stage 3a) or invasion into the seminal vesicles (rT3b) was suspected.

RARP specimens were processed according to clinical routine and the International Society of Urological Pathology (ISUP) guidelines [22]. All specimens were fixated in formaldehyde (10%) directly after surgery. The surface of the specimens was inked and the apex and base (bladder neck) were removed. The mid part of the specimen was cut perpendicular to the urethra in 4 mm slices. The apex and base parts were cut in a sagittal fashion. Histologic slices were produced after sectioning in quadrants. Blinded by PET/CT results, an experienced uro-pathologist (PV) reviewed all slices and delineated all tumour depositions on the 12-segment mapping model of the prostate and the 2 segments of the seminal vesicles (Appendix-1 in ESM). For all segments that contained tumour, a Gleason score and ISUP grade was provided and the presence of ECE (pT3a). Clinically significant PCa (csPCa) was defined as PCa with Gleason score ≥ 3 + 4 = 7 (ISUP grade ≥ 2) or any tumour with ≥ pT3a. The index lesion was defined as the largest lesion with the highest ISUP grade or stage.

The localisation of the detected prostate tumour by ¹⁸F-DCFPyL-PET/CT was matched to the histopathology results and the sensitivity, specificity, positive predicting value (PPV), and negative predicting value (NPV) were calculated on a segment basis. Correlation of ¹⁸F-DCFPyL-PET/CT with histopathology was considered if exactly the same segment was demarcated (total agreement). Since there are no anatomical landmarks to delineate the segments, artificial segmentation can occur, causing a mismatch between the PET/CT- and histopathological findings while both correspond with the same lesion. Therefore, a second analysis of diagnostic accuracy was performed, in which PET correlation was also considered if there was a discrepancy of up to 1 region in the coronal or sagittal plane (near-total agreement) [23, 24]. Reciever‐operating characteristic curves (ROC) and area under the curve (AUC) analysis were performed to explore the accuracy of PSMA-PET/CT in the detection of segments containing csPCa based on the 5-point scale. For the assessment of pathological tumour stage (pT), we investigated the accuracy of ¹⁸F-DCFPyL-PET/CT to differentiate locally advanced disease (> rT3a) from prostate-confined disease (rT2). Numerical variables were summarised with median values and interquartile ranges (IQR); categorical variables with proportions (%). To compare medians of non-parametric data, the Mann–Whitney-Wilcoxon test and the Kruskal-Wallis test were used (significance set at p < 0.05). Statistical analysis was performed with IBM® SPSS® Statistics for Windows®, version 26.

A total of 30 patients was included in this study, having a median initial PSA-level of 11.1 ng/ml (IQR 5.8–22.4). According to EAU guidelines, 10/30 (33.3%) patients had intermediate-risk PCa and 20/30 (66.6%) had high-risk PCa [15]. Pre-operative and post-operative characteristics of included patients are listed in Table 1.

All patients showed PSMA expression in the prostate. In 30 evaluated patients, 420 segments (12 prostate segments + 2 seminal vesicle segments per patient) could be used both for PET/CT and histopathological mapping evaluation. PCa was present in 129 of the 420 (30.7%) segments on histopathological examination, and csPCa was found in 122 of the 420 segments (29.0%) (median 3 segments per patient, IQR 2–5). The sensitivity, specificity, PPV and NPV of ¹⁸F-DCFPyL-PET/CT to detect csPCa per segment with total agreement was 61.4% (95%CI 52.2–70.0%), 88.3% (95%CI 83.9–91.6%), 68.1% (95%CI 58.5–76.6%), and 84.8% (95%CI 80.2–88.5%), respectively (Appendix-2 in ESM). For near-total agreement, the sensitivity, specificity, PPV and NPV of ¹⁸F-DCFPyL PET/CT to detect csPCa per segment was 84.4% (95%CI 76.5–90.1%), 97.0% (95%CI 94.1–98.5%), 92.0% (95%CI 84.9–96.0%), and 93.8% (95%CI 90.3–96.1%), respectively. The area under the curve (AUC) of ¹⁸F-DCFPyL-PET/CT was 0.78 (95%CI 0.73–0.84) for the total agreement scores, and 0.85 (95%CI 0.80–0.90) for the near-total agreement scores (Appendix-3 in ESM). True positive segments had a median SUV_max of 8.26 (IQR 5.25–11.40), which was significantly higher than the median SUV_max of false-positive segments of 4.06 (IQR 3.56–5.10) (p = 0.02). The median SUV_max of true positive segments did not correlate with ISUP grade groups (p = 0.95) (Appendix-4 in ESM).

The primary potential biopsy recommendation by the nuclear medicine physician harboured csPCa in 24/30 (80.0%) patients and detected the index PCa lesion in 23/30 (76.6%) patients. When both the primary and secondary recommended segments would potentially be targeted, ¹⁸F-DCFPyL-PET/CT revealed csPCa in 28/30 (93.3%) patients. Moreover, it pinned the index PCa lesion in 26/30 (86.7%) patients. An example of potential ¹⁸F-DCFPyL-guided biopsy recommendation and concurrent histopathological examination of the RARP specimen is shown in Fig. 1. Potential biopsy recommendation that matched the index PCa lesion had a median SUV_max 8.62 (IQR 6.41–12.62). The recommendation for potential biopsy from the nuclear physicians that matched csPCA had a median SUV_max of 8.55 (IQR 6.34–13.79), and was significantly higher than the recommended potential biopsy segments that did not contain csPCa (median SUV_max of 3.10 (IQR 2.86–3.87) (p = 0.02).

Final histopathological analysis revealed pT3a in 10/30 (33.3%) patients, and pT3b in 4/30 (13.3%) patients. The sensitivity, specificity, PPV and NPV of ¹⁸F-DCFPyL-PET/CT to detect locally advanced tumour growth (≥ pT3a) was 35.7% (95%CI 14.0–64.3%), 93.8% (95%CI 67.7–99.7%), 83.3% (95%CI 36.5–99.1%), and 62.5% (95%CI 40.8–80.4%), respectively (Appendix-5 in ESM). For the detection of pT3a sub-stage, the sensitivity, specificity, PPV and NPV of ¹⁸F-DCFPyL-PET/CT were 20.0% (95%CI 3.5–55.8), 100% (95%CI 80.0–100), 100.0% (95%CI 19.7–100), and 71.4% (95%CI 51.1–86.0), respectively. For the detection of pT3b sub-stage, the sensitivity, specificity, PPV and NPV of ¹⁸F-DCFPyL-PET/CT were 75.0% (95%CI 21.9–98.7), 92.3% (95%CI 73.4–98.7), 60.0% (95%CI 17.0–92.7), and 96.0% (95%CI 77.7–99.8), respectively.