Detection and Local Staging of Prostate Cancer by 68Ga-PSMA-PET/CT, Comparison with mpMRI and Histopathology

Introduction: 68Ga-PSMA-PET/CT has proven its value in prostate cancer with high positive predictive value for lymph node metastasis and superior detection of distant metastasis. There is growing evidence that 68Ga-PSMA-PET/CT has high sensitivity for detection of tumor lesions in the prostate as well. Studies thus far have mainly been performed in patients prior to prostatectomy. Aim of this study is to evaluate diagnostic accuracy in a mixed population of men with increased risk of prostate cancer and evaluate diagnostic possibilities with respect to extra-capsular extension and seminal vesicle invasion. Methods: The population consisted of a retrospectively included se-quential cohort of 69 patients with 68Ga-PSMA-PET/CT and mpMRI available. 68Ga-PSMA-PET/CT was re-evaluated by two readers blinded for mpMRI and clinical information. Likelihood of tumour presence, extra-prostatic extension and seminal vesicle invasion was scored on 5-point Likert scale and localized schematically. Results were compared with mpMRI. Available pa-thological outcome served as gold standard. SUVmax of index lesions was measured and correlated to index tumor Gleason grade. Results: Clinically significant prostate cancer (Gleason ≥ 3 + 4) was detected in 57 (83%) of 69 patients. Diagnostic accuracy was 89% for PET reader 1, 93% for PET reader 2 and 86% for mpMRI. Lesion concordance of 68Ga-PSMA-PET/CT and mpMRI was 97%. SUVmax of the index lesion correlated to Gleason grade. Sensitivity for extracapsular extension in the prostatectomy group was 62% for PET reader 1, 33% for PET reader 2 and 50% for mpMRI. Specificity was 62% for PET reader 1, 100% for PET reader 2 and 69% for mpMRI. Conclusion: Ga68-PSMA-PET shows high accuracy in the detection of tumor lesions in the prostate. Results on evaluating extra-capsular extension and seminal vesicle invasion are comparable to mpMRI. This study adds to the increasing evidence that 68Ga-PSMA-PET/CT is imperative in detection of prostate cancer prior to biopsy.


Introduction
Prostate cancer is a disease with high incidence and often has an indolent course.
Adequate early detection of more aggressive cancers is imperative when aiming for curation. Meanwhile preventing unnecessary invasive diagnostic methods is important as well. It is key to accurately detect the location of those lesions with a high grade of malignancy. In patients with increased risk for prostate cancer based on raised serum level of prostate specific antigen (PSA), standard transrectal ultrasound (TRUS) guided biopsies may not be representative and malignancy grade can be underestimated. Several studies have confirmed the additional value of multi parametric magnetic resonance imaging (mpMRI) guided biopsies and recent guidelines support mpMRI-based targeted biopsies in biopsy naïve patients [1] [2]. mpMRI-guided biopsy while omitting random biopsies will be at cost of missing 5% -10% of significant cancer [3] [4]. Sensitivity of mpMRI is lowest for lesions located ventrally in the prostate in the transition zone. This region is more difficult to evaluate on MRI than the peripheral zone and is traditionally not part of standardized random biopsies [5] [6]. Additionally, mpMRI positive lesions (PIRADS 4 or 5) lack specificity and are hard to differentiate from focal prostatitis, atrophy, scarring and benign hyperplasia nodules, resulting in a negative biopsy in 20% -40% of cases [2] [7]. More important, accurate acquisition and evaluation of mpMRI is a critical challenge and demands experience and frequent exposure [8]. With the increased demand for mpMRI's for cancer detection, an alternative diagnostic tool with straightforward accurate detection would be welcome. Moreover, an alternative to mpMRI would also be a solution for patients ineligible for an MRI because of metal implants or claustrophobia.
PSMA-PET/CT was initially indicated for detection of location of tumor activity in patients with biochemical recurrence [9]. High diagnostic accuracy for metastatic disease led to PSMA-PET/CT being advised for primary staging in

Study Population
The study was designed as a comparative study between the 68Ga-PSMA-PET/CT and mpMRI prostate performed for primary detection and staging in a retrospectively included cohort. Study population consisted of allpatients in whom a 68Ga-PSMA-PET/CT was performed between October 2016 and March 2019 and who were treated in the Meander Medical Centre Amersfoort. Ethical approval of the hospital board was given. The study was conducted according to the principles of the Declaration of Helsinki [27] and in accordance with the Medical Research Involving Human Subjects Act [28].
Inclusion criteria were availability of mpMRI prostate and 68Ga-PSMA-PET/CT obtained within a one year interval. Patients who already underwent treatment for prostate cancer before either the mpMRI or 68Ga-PSMA-PET/CT was performed were excluded. One patient showed bladder cancer after revision of biopsy histopathology and was excluded. 69 patients were available for analysis. In 68 patients, biopsy histopathology results of the prostate were available, one diagnosis was made based on a histopathological proven lymph node metastasis. In 31 patients prostatectomy was performed with additional definitive histopathological results available of the entire prostate and extra-prostatic extend. Advances in Molecular Imaging seminal vesicles and on focal intensity. MRI was performed on a Philips Achieva 1.5T after intravenous administration of 20 mg Buscopan. T1w, T2w, diffusion 0-50-400-800-1200 and dynamic contrast enhanced series after administration of 20 cc Gadolinium, 2.5 cc/sec were obtained. Evaluation was performed in clinical setting by one of two experienced radiologists according to PI-RADS-v2 protocol. Localization in the 12segment-model was done additionally for study purposes.

Histological Examination
Random biopsy was performed by TRUS, at least 4 cores left and right were acquired, withadditional cores depending on prostate size. In case of palpable or hypo-echogenic lesions, additional targeted biopsy was performed. mpMRI guided biopsies were performed in nine patients because of negative random biopsies and persistent suspicion of prostate cancer. Histopathological evaluation was done according to ISUP standard protocols for Gleason grading by experienced pathologists [29]. Clinically significant disease was defined as Gleason 3 + 4 of higher. In a subgroup of patient with prostatectomy, presence of extra-capsular extension and seminal vesicle invasion was registered. In case of discrepancy between results of biopsies and prostatectomy, results of the latter overruled earlier results.

Assessing Imaging and Histopathological Concordance
68Ga-PSMA-PET/CT was considered positive for lesions with Likert 4 and 5 scores. mpMRI was positive in case of PI-RADS 4 and 5 scores. Histopathological Gleasongrade ≥ 3 + 4 was considered as clinically significant prostate cancer.
Concordance of lesionlocation was defined as presence of a lesion (PI-RADS or Likert ≥ 3) in a least one corresponding segment. Discordance was defined as presence of a lesion (optionally a second lesion) with no identical segments ( Figure 1). Lesion absence in both 68Ga-PSMA-PET/CT and mpMRI was considered as concordance. Tumor volume on 68Ga-PSMA-PET/CT was compared to mpMRI by per patients registration whether more segment were involved on 68Ga-PSMA-PET/CT or on mpMRI (Figure 1).

Statistical Analysis
A quantitative description (continuous or ordinal scale) of different parameters was given (median or mean and standard deviation). Comparison of the sensitivity, specificity, positive and negative predictive values as well as accuracy of the diagnostic test was presented with histopathology as gold standard. Extracapsular extension and seminal vesicle invasion was compared to mpMRI for the whole group and an additional analysis was performed in a subgroup of patients that had a prostatectomy.
Correlation between SUV max and Gleason grade was determined by twosided spearman rho with correlation coefficient and p-value presented. Location concordance was set as a per-patient categorical variable (all matching locations: yes or no).
SUVmax of the most active lesion was correlated to Gleason grade ( Figure 2). SUVmean background had a range of 1.0 to 3.0. The correlation of SUVmax with Gleason grade was slightly better than the correlation of the ratio of SUVmax target lesion/SUV mean background with Gleason grade, r = 0.583 vs r = 0.518. Lowest SUVmax value for a clinically significant lesion (Gleason ≥ 3 + 4) was 4.2. Highest value for a lesion in a patient without prostate cancer was 4.   Prostatectomy was performed in 31 of 69 patients. 18 of those patients had ECE and six SVI. In two patients both 68Ga-PSMA-PET/CT and mpMRI scored SVI as positive. The others were falsely negative by interpreting Likert 3 as negative ( Figure 3). Results of evaluation of ECE and SVI are presented in Table 3.

Discussion
68Ga-PSMA-PET/CT detected more patients with clinically significant prostate cancer than mpMRI. Sensitivity for clinically significant cancer was slightly higher (reader 1 95% and reader 2 100%) for 68Ga-PSMA-PET/CT than for mpMRI (91%) with equal specificity. These results are in line with previous published studies. Kalapara [14]. Specificity in our study was 67%, this is lower than in the study of Donato  Earlier studies showed that approximately 5% -10% of clinically significant prostate cancer does not show high PSMA-receptor expression [32]. Our population included a patient with a Gleason 5 + 4 tumour with relatively low PSMAuptake. Despite low uptake, this tumour was indicated as target lesion by standing out against prostate background activity. Uprimny et al. showed that in 9% of patients the lesion did not stand out from prostate background [22]. These were all Gleason 6 or 7 graded tumours and PSA levels were <10 ng/ml. PSA levels of the three falsely negative patients in our study as evaluated by reader 1 on 68Ga-PSMA-PET/CT were 8.8; 3.9 and 8.9 ng/ml. Sensitivity of ECE was low for both 68Ga-PSMA-PET/CT and mpMRI. Sensitivity for ECE with mpMRI in our study was only 50%, comparable with the acknowledged low sensitivity 57% in a meta-analysis [33]. However, specificity in this meta-analysis was high (91%), while our specificity was 69%. Specificity for mpMRI may have been low because "likely" ECE was included as positive. There are a few studies that evaluated extra-prostatic extension on PSMA-PET/CT. Whole group results in our study for diagnosis SVI in 68Ga-PSMA-PET/CT were more robust than for ECE and showed acceptable inter-reader variation. Diagnosis was consistent with mpMRI results. Sensitivity and specificity cannot be reliable determined as numbers in the prostatectomy-subgroup were too small. This problem is difficult to overcome as was emphasised by others as patients with SVI will often not have a prostatectomy and as such no definitive histopathology is available [25] [26]. Most mpMRI studies encountered this problem as well [33].
Study results show convincing evidence of additional value of 68Ga-PSMA-PET/CT in detection of prostate cancer. For successful clinical use, PET guided biopsies are required. Several have shown that PET/CT-TRUS fused biopsies are technical feasible and these experiences are shared by our group [20] [35] [36] [37].
The PROMIS-trial has shown that using mpMRI to triage men might allow 27% of patients avoid a primary biopsy and diagnosis of 5% fewer clinically insignificant cancers [1]. Sensitivity was 93%. Results for PSMA-PET imply that sensitivity might even be better than with mpMRI. Donato et al. found clinically significant prostate cancer in 70% -80% of patient with negative mpMRI [17]. This could be at cost of finding more clinically insignificant cancer. Our study showed intense PSMA-uptake in two patients with Gleason 3 + 3 tumour, mpMRI was evaluated as PI-RADS 3 and 4. Uprimny et al. showed comparable results with a positive correlation between SUVmax and Gleason grade, but with several outliers, SUVmax values for Gleason 3 + 3 had a range up to 19.3 [22].

Conclusion
Blinded evaluation of 68Ga-PSMA-PET/CT showed higher sensitivity than mpMRI for the overall detection of primary prostate cancer in a clinical cohort of patients with and without prostate cancer. Evaluation of extra-prostatic extension and seminal vesicle invasion on 68Ga-PSMA-PET/CT is comparable to mpMRI. 68Ga-PSMA-PET/CT can probably replace mpMRI as initial diagnostic step, while PET-guided biopsies are feasible.