Seek and Find: Current Prospective Evidence for Prostate-specific Membrane Antigen Imaging to Detect Recurrent Prostate Cancer

Abstract

Context:

Men with biochemically relapsed prostate cancer face a clinical conundrum. Depending on the detected distribution of disease, treatment goals may range from cure with focal therapy to palliative with systemic therapy to expectant observation. Retrospective studies of prostate-specific membrane antigen (PSMA)-based imaging demonstrate higher disease detection rates than conventional imaging.

Objective:

This review focuses on available prospective evidence for diagnostic use of PSMA-based imaging to accurately restage recurrent prostate cancer and explores the potential clinical impact, near future uses, and challenges for PSMA-based imaging in this setting.

Evidence acquisition:

PubMed and EMBASE databases were searched for prospective studies with primary, secondary, or exploratory endpoints evaluating PSMA-based imaging for patients with recurrent prostate cancer published in English in the past 10 yr.

Evidence synthesis:

We reviewed 56 prospective studies evaluating the role of PSMA positron emission tomography (PET) in recurrent prostate cancer. These studies establish the diagnostic accuracy and safety of PSMA PET using the ⁶⁸Ga-PSMA-11 and ¹⁸F-DCFPyL radiotracers even at lower prostate-specific antigen (PSA) levels (0.5 ≤ PSA < 1.0 ng/ml: disease detection rate 51–78%). The use of PSMA PET has been shown to result in changes in management in up to two-thirds of patients.

Conclusions:

There is now higher-level regulatory-quality prospective evidence for PSMA-based imaging for the detection of recurrent prostate cancer. There is prospective evidence of superiority over cross-sectional imaging and bone scintigraphy, and resulting alterations in disease management course as a result of PSMA-based imaging.

Patient summary:

When prostate-specific antigen (PSA) level is rising after primary therapy, prostate-specific membrane antigen (PSMA) positron emission tomography (PET) is excellent at detecting and localizing prostate cancer, even at low PSA levels. Those who benefit best from treatment modifications based on PSMA PET findings are yet to be defined.

1. Introduction

Men with prostate cancer who have undergone definitive local therapy but who have a rising prostate-specific antigen (PSA) level face a significant clinical dilemma. Accurate disease localization is necessary to formulate a treatment plan based on an understanding of the distribution of disease. Men with locally recurrent disease are potentially curable, and salvage treatments alone or in combination with systemic therapy may be appropriate. Men with distant metastatic disease might be spared salvage treatments, and systemic approaches may more aptly address systemic disease. Until recently, few imaging modalities could accurately localize prostate cancer as the PSA began to rise. However, prostate-specific membrane antigen (PSMA)-directed positron emitting tomography (PET) imaging has created a new standard of diagnostic accuracy in the early detection and localization of PSA-producing recurrent disease.

Following completion of definitive treatment for clinically localized prostate cancer with either radical prostatectomy surgery or radical radiotherapy, guidelines recommend monitoring patients for recurrent disease using PSA [1,2]. Despite primary therapy with curative intent, up to one-third of patients will experience a biochemical recurrence and 50–95% of men with high-risk prostate cancer will experience recurrence [3]. Biochemical recurrence is defined as a PSA level of ≥0.2 ng/ml measured >6 wk after prostatectomy or a PSA level of ≥2 ng/ml rise above nadir following radiation therapy [4,5]. Using traditional or “conventional” imaging, recurrent disease may manifest in one of three ways as described in Prostate Cancer Working Group 3: as radiographically evident metastatic disease (metastatic disease, noncastrate), as disease restricted to the local/locoregional area around the prostate bed, or as disease that may not be visualized (rising PSA, noncastrate) [6]. Therapeutic interventions diverge based on delineation of one of the above-described clinical states; locoregional failures with disease confined to the prostate bed or pelvis would be considered for salvage therapies with the aim of disease eradication. Detection of more extensive disease might require palliative systemic therapy [7]. At present, conventional imaging undertaken in this setting includes computed tomography (CT) with sensitivity of approximately 54% for local recurrence detection, magnetic resonance imaging (MRI), and/or bone scintigraphy, approaches that are limited by both poor sensitivity and relatively poor specificity, particularly at low levels of PSA [8]. When the PSA level is >10 ng/ml, a CT scan and bone scan are usually sufficient to confirm the metastatic status [9]. In the absence of a reliable way to completely visualize and determine the extent or state of disease at the time of initial recurrence, clinical decision makers are compelled to rely on risk prediction nomograms and prognostic markers such as PSA doubling time and Gleason score. A prospective comparison between PSMA PET/CT and conventional imaging carried out prior to surgery in those with high-risk prostate cancer has shown 27% absolute superiority of PSMA PET over the combination of CT and bone scan for the primary outcome of the detection of prostate cancer in pelvic metastases or distant sites [10]. Imaging modalities that achieve a sensitive and accurate assessment of the recurrent disease represent key decision-making tools, or imaging biomarkers, to guide apposite therapeutic interventions, particularly at low PSA values, when patients are most likely to have curable disease [11].

The ⁶⁸Ga- or ¹⁸F-labeled ligands PSMA-11, PSMA-1007, and DCFPyL bind to PSMA and facilitate visualization of PSMA expressing prostate cancer using PET/CT or PET/MRI imaging. Each has already demonstrated diagnostic accuracy for biochemically recurrent prostate cancer in retrospective studies and in meta-analyses of retrospective studies [12–15]. However, in order to secure regulatory approval and furnish higher levels of evidence for clinicians to make decisions based on PSMA PET, prospective data have now been completed or are underway. Indeed, based on some of these data, ⁶⁸Ga-PSMA-11 has recently been approved by the US Food and Drug Administration (FDA) for PET imaging of patients with suspected prostate cancer metastasis who are potentially curable by surgery or radiation therapy, and for patients with suspected prostate cancer recurrence based on elevated PSA levels. Our objective is to review and summarize the outcome of a literature search for published prospective study data for PSMA PET detection rate, accuracy, influence on management, and when available, comparative performance with conventional imaging modalities for patients with recurrent prostate cancer following any modality of primary therapy.

2. Evidence acquisition

A comprehensive literature search was performed using PubMed and EMBASE databases (last accessed November 27, 2020) to identify relevant studies using the abstract search terms “PSMA” AND “prostate” AND “recurrent” AND “prospective”. This resulted in a review of 133 studies. For the purposes of this review, recurrent disease was considered to include patients who had prior radical treatment for primary prostate cancer with surgery or radiation, and participated in prospective PSMA imaging trials with one of the following three clinical disease states: rising PSA with no definitive evidence of metastatic disease on conventional imaging, evidence of localized recurrence, or evidence of oligometastatic disease defined as one to three lesions identified by conventional imaging.

Of the 133 studies screened, 74 original manuscripts describing prospective studies published in indexed and peer-reviewed journals and written in English between January 2010 and November 2020 were analyzed, and 48 were included for qualitative synthesis. Excluded were studies of prostate cancer outside of recurrent disease context (n = 13), retrospective studies (n = 16), pilot studies or preliminary results (n = 5), systematic reviews or review articles (n = 23), correspondence or editorials (n = 10), and other (n = 7), as shown in Figure 1. Studies of comparative imaging with non-PSMA radiotracers (n = 11) were excluded due to scope limitations. All the studies ultimately included for qualitative synthesis (n = 37) report prospectively collected data, and were either registered on a clinical trials database (n = 21) or had been approved by a local institutional review board. Data elements extracted from the manuscripts of included studies included the number of participants, available treatment history, mean PSA, overall detection rate, stratified detection rate, positive predictive value (PPV), and management change if reported.

Fig. 1 –

Flow diagram of the process for selecting papers for inclusion in review. PET = positron emission tomography; PSMA = prostate-specific membrane antigen.

3. Evidence synthesis

3.1. Design considerations

Interpreting the quality of these data requires an understanding of these trials’ challenges and the complexity of their designs. Biochemically relapsed patients rarely have disease amenable to biopsy given that they have either negative or equivocal standard imaging. As a result, composite endpoints are commonly employed, representing a combination of biopsy, comparisons with other imaging modalities, or observation of post-treatment PSA declines. The potential for bias in interpreting these complex endpoints can be mitigated by employing central readers blinded to the clinical circumstances of the patient in order to evaluate all imaging data independently, blinded adjudicators of the composite “truth standard,” and formal guidelines for resolving discrepant interpretations [16–20]. Typically, these techniques inform primary and secondary endpoints that include lesion detection rate, sensitivity, specificity, PPV, and negative predictive value. The prospective studies included in this review ranged from small, single-center studies to large multicenter trials that have formed the basis for regulatory approval. Clinician questionnaires are frequently employed to assess whether the imaging altered the clinical decision-making for the patient. At present, most of the prospective trials of PSMA-based imaging in biochemical recurrence focuses on three tracers: ⁶⁸Ga-PSMA-11, ¹⁸F-DCFPyL, and ¹⁸F-PSMA-1007.

3.2. Diagnostic performance of ⁶⁸Ga-PSMA-11

The development of ⁶⁸Ga-PSMA-11, also referred to as HBED, HBED-CC, PSMAHBED, Glu-urea-Lys(Ahx)-HBEDCC, and PSMA-HBED-CC, initially published in 2013, represented a significant breakthrough in the evolution of PSMA ligands [21]. This small molecule inhibitor showed both high PSMA receptor affinity and excellent tissue penetration into solid lesions such as bone, and offers a favorable tumor-to-background ratio [22]. A meta-analysis of ⁶⁸Ga-PSMA-11 PET retrospective studies on biochemical recurrence reported a positive detection rate of 0.63 (95% confidence interval [CI], 0.55–0.70) in those with a PSA value of <2.0 ng/ml and 0.94 (95% CI 0.91–0.96) for those with a PSA value of >2.0 ng/ml [23]. These studies have sometimes included men with biochemical relapse as well as overt metastatic disease, and so detection rates in a strictly selected population of men with biochemical recurrence had not been determined. Nonetheless, detection rates from these studies have largely been borne out in prospective studies summarized in Table 1 [24,25].

Table 1 –

Selected trials with primary endpoint of PSMA PET detection rate and/or diagnostic accuracy

	Total patients	Median PSA (ng/ml; range)	Primary endpoint	Primary therapy	Disease detection rate (%)	Detection rate by PSA stratification	Histopathological confirmation per patient	Composite confirmation per patient
68Ga-PSMA-11
Fendler et al (2019) [25] NCT02940262 NCT03353740	635	2.1(0.1–1154)	Diagnostie accuracy	RP = 41% RT = 27% RP + sRT	75	PSA <0.5 ng/ml:38% 0.5 ≤ PSA < 1.0: 57% 1.0 ≤ PSA < 2.0: 84% 2.0 ≤ PSA < 5.0: 86% PSA >5.0 ng/ml: 97%	73/87(84%) PPV = 0.84 (95% CI: 0.76–0.90)	200/217(92%) PPV = 0.92(95% CI: 0.880.95)
Hamed et al (2019) [26]	188	2.2(0.01–70)	Diagnostie accuracy	RP = 13% RT = 62% RP + RT = 25%	87.8	PSA <0.5 ng/ ml: 54% 0.5 ≤ PSA < 1.0: 71% 1.0 ≤ PSA < 2.0: 96% PSA ≥2.0 ng/ml: 99%	151/165 91.5%	14/165 8.4% Overall PPV = 100(95% CI: 97.79–100)
Lawhn-Heath et al (2019) [24] NCT026118 82	150	2.1(0.05–83.7)	Diagnostie accuracy	RP = 35% RT = 32% RP + RT = 26%	81.3	PSA <0.5 ng/ml: 55.6% PSA >5.0 ng/ml: 95.9%	43 had histopathological assessment	PPV = 90.6% Subset of 72 patients
Deandreis et al (2020) [31]	223	0.65(0.2–8.9)	Detection efficacy	RP = 97% RT = 3%	39.9	NA	17/223(7.6%)	NA
Zacho et al (2018) [32] EudraCT no.: 2014-005073-37	70	0.55(0.2–11.3)	Detection rate	RP = 91% RT = 9%		PSA <0.5 ng/ml: 22% PSA >5.0 ng/ml: 83%	43/150(29%)	29/150(19%)
Rousseau et al (2019) [44] NCT03443609	52	0.44(0.07–1.5)	Detection rate	RP = 54% RP + sRT = 46%	73.1	PSA <0.25 ng/ml: 58% 0.25 ≤ PSA < 0.69: 81% PSA >0.70–1.5: 82%	8/38(21%)	NA
Beheshti et al (2020) [30]	135	<1.0	Detection rate	RP = 100% RP + RT = 19% RP + ADT = 7%	49.6	PSA <0.2 ng/ml: 32% 0.2 ≤ PSA < 0.5: 45% 0.5 ≤ PSA < 1.0: 71%	NA	NA
Caroli et al (2018) [29]	314	0.83(0.003–80)	Detection rate	RP = 46% RT = 16% RP + sRT = 32% RP + ADT = 4% RP + sRT + ADT = 2%	62.7	PSA <0.2 ng/ml: 27% 0.2 ≤ PSA < 1.0: 47% 1.0 ≤ PSA < 2.0: 75% PSA ≥2.0 ng/ml: 95%	NA	NA
Dong et al (2020) [28]	51	1.02(0.21–31.98)	Detection rate	RP = 100%	66.7	PSA 0.2–0.49 ng/ml: 30% PSA 0.5–0.99 ng/ml: 60% PSA 1.0–3.99 ng/ml: 79% PSA >4.0 ng/ml: 00%	NA	NA
18F-DCFPyL
CONDOR Morris et al (2020) [19] NCT03739684	208	0.8(0.17–98.4)	Correctlocalization rate	RP = 49.5% RT = 15% RP + Rt = 36% Sys = 28%	59–66	PSA <0.5 ng/ml: 36% 0.5 ≤ PSA < 1.0: 51% 1.0 ≤ PSA < 2.0: 67% 2.0 ≤ PSA < 5.0: 85% PSA ≥5.0 ng/ml: 97%	NA	CLR = 84.8–87.0% Lower bound of 95% CI: >77% For 3 readers
OSPREY Rowe et al (2019) [20] Cohort B NCT029813 68	93	11.3(0.003–596.9)	Diagnostie accuracy	NA	78–91	NA	81.2–87.8% Lower bound of 95% CI: 72.9–80.4%	NA
Mena et al (2018) [64] NCT03181867	90	2.5(0.21–35.5)	Detection rate	RP = 42% RT = 30% RP + RT = 28%	77.8	0.2 ≤ PSA < 0.5: 48% 0.5 ≤ PSA < 1.0: 50% 1.0 ≤ PSA < 2.0: 89% PSA ≥2.0 ng/ml: 94%	28/70 (40%) PPV = 93.3% (95% CI: 77.699.2%)	PPV = 96.2% (95% CI: 86.399.7%)
Rousseau et al (2019) [44] NCT02899312	130	Mean = 5.2	Detection rate	RP = 72% RT = 28%	84.6	0.4 ≤ PSA < 0.5: 60% 0.5 ≤ PSA < 1.0: 78% 1.0 ≤ PSA < 2.0: 72% PSA ≥2.0 ng/ml: 92%	NA	NA
Song et al (2020) [43] NCT03501940	72	3.0(0.23–698)	Detection rate	RP = 58% RT = 42%	85	PSA <0.5 ng/ml: 50% 0.5 ≤ PSA < 1.0: 69% 1.0 ≤ PSA < 2.0: 100% 2.0 ≤ PSA < 5.0: 91% PSA >5.0 ng/ml: 96%	4/4(100%)	NA
Rowe et al (2020) [65] NCT02523924	31	0.4(0.2–28)	Detection rate	RP = 100%	67.7	PSA 0.2–1.0 ng/ml: 59% PSA >1.0 ng/ml: 89%	NA	NA
18F-PSMA-1007
Witkowsk a-Patena et al (2020) [46]	60	0.65(0.008–2.0)	Detection rate	RP = 80% RT = 20%	60	PSA <0.5 ng/ml: 39% 0.5 ≤ PSA < 1.0: 55% 1.0 ≤ PSA < 2.0: 100%	NA	NA

ADT = androgen deprivation therapy; 95% CI = 95% confidence interval; CLR = correct localization rate; NA = not available; PET = positron emission tomography; PPV = positive predictive value; PSA = prostate-specific antigen; PSMA = prostate-specific membrane antigen; RP = radical prostatectomy; RT = radiation therapy; sRT = salvage radiation therapy; Sys = at least one prior systemic therapy.

Findings from the first prospective trial of PSMA imaging in the setting of biochemical recurrence were reported by Fendler et al [25] in a multicenter, single-arm trial of ⁶⁸Ga-PSMA-11 that aimed to assess the accuracy of the tracer and report endpoints of detection rate, PPV, inter-reader reproducibility, and safety. For 635 patients with biochemical recurrence after primary therapy with either surgery (64%) or radiation (73%) and with a median PSA value of 2.1 ng/ml (range: 0.1–1154 ng/ml), there was at least one positive lesion detected by ⁶⁸Ga-PSMA-11 PET in 75% of participants by independent reads [25]. Detection rates in other prospective trials of ⁶⁸Ga-PSMA-11 PET have ranged through 87.8% (n = 188; median PSA = 2.2 ng/ml) [26] to 76% (n = 121; median PSA = 3.87 ng/ml) [27] to 66.7% (n = 51; median PSA = 1.02) [28] to 62.7% (n = 314; median PSA = 0.83 ng/ml) [29] to 49.6% (n = 135; PSA < 1.0 ng/ml) [30]. These studies were all undertaken in the context of prostate cancer biochemical recurrence, but examined populations with heterogeneous primary treatment histories and variable biological characteristics such as PSA doubling time, Gleason score, or tumor grade that may or may not influence detection rate.

Detection of PSMA uptake does not necessarily indicate prostate cancer (representing false positives), and so prospective PSMA imaging trials have sought to verify the accuracy of PSMA-avid lesions as actually representing prostate cancer. As mentioned earlier, however, histopathological correlation—the gold standard—is rarely retrievable in these patients. In the aforementioned Fendler et al’s [25] study, histopathological validation was possible only in 93 (14.6%) participants, yielding a PPV of 0.84 (95% CI 0.75–0.90 and 0.76–0.91) using a pathology reference standard. In the absence of complete histopathological specimen availability, confirmation of detection accuracy was achieved with a composite reference standard. Using a composite endpoint comprising pathology, other imaging, and post-treatment PSA decline, 200 of 217 PSMA PET–positive patients were true positives, giving a PPV of 0.92 (95% CI 0.88–0.95) [25]. The findings of this study have informed the US FDA approval of ⁶⁸Ga-PSMA-11 PET for the indication of biochemical recurrence, although availability is currently limited to two sites in California: University of California, LA (UCLA) and University of California, San Francisco (UCSF).

The relatively low rate of availability of material for histopathological validation in this trial is seen across the included prospective trials of ⁶⁸Ga-PSMA-11, where pathological validation of PSMA-detected disease was carried out in only a relatively small proportion of study participants (7.6–29%) [25,31,32]. One notable exception to this pattern is a study reported by Hamed et al [26] that confirmed histopathological diagnosis in 92% (n = 151) of participants. This study aimed to assess the detection rate and diagnostic accuracy of ⁶⁸Ga-PSMA-11 PET in patients with rising PSA after primary therapy. The study recruited 188 patients, the majority of whom had radiation (62%) as primary treatment, and there was a preimaging median PSA value of 2.2 ng/ml (range: 0.01–70 ng/ml). An overall detection rate of 87.8% is reported and notably, 130/165 PET-positive patients had extraprostatic lesions. Similar to the study by Fendler et al [25], a composite PPV is reported by Hamed et al [26] using histopathological verification in conjunction with clinical and imaging follow-up to give a PPV of 100 (95% CI 97.8–100).

Further underscoring the high degree of confidence one can have that a positive lesion on PSMA PET actually represents disease is a study by Lawhn-Heath et al [24], who reported a study of 150 men, all with biochemical recurrence and a median PSA level of 2.1 ng/ml, undergoing ⁶⁸Ga-PSMA-11 PET with a primary study endpoint to determine diagnostic accuracy for disease detection. The study demonstrated a PPV per patient of 90.6% for a subpopulation of 72 men with composite reference elements. Notably, PPV is perhaps the most relevant metric for clinicians who tend to act on positive findings in defining treatment plans. Fendler et al [33] published an analysis, from their prospective trial of 635 participants using ⁶⁸Ga-PSMA-11 PET, of findings that were discordant with the reference standard. Consensus reads were false positive in 17/217 (8%) patients with lesion validation. False positive interpretations occurred most often in this cohort in the context of suspected prostate bed relapse after radiotherapy (n = 11). False negatives are also important, but render decision-making no less reliable than standard imaging, which is characterized by false negatives.

Prospective studies of ⁶⁸Ga-PSMA-11, as well as the other tracers, demonstrate a significant increase in disease detection rates as PSA levels increase. In the Fendler et al’s [25] study, detection rates ranged from 38% at PSA <0.5 ng/ml to 84% at PSA <2.0 ng/ml to 97% when PSA was ≥5.0 ng/ml. Two prospective studies have carried out a receiver operator characteristic (ROC) curve analysis for PSA, plotting the true positive rate against the false positive rate. The area under an ROC curve is a measure of accuracy of a diagnostic test across all possible cutoff points. Hamed et al [26] report an area under the curve (AUC) of 0.96 and an optimal cutoff PSA value to predict a positive scan of 0.65 ng/ml. In another study by Caroli et al [29], with the purpose of determining a detection rate of ⁶⁸Ga-PSMA PET/CT, 314 patients who had a median PSA level of 0.83 ng/ml with a negative or equivocal choline scan reported a PSMA detection rate of 62.7%. This study demonstrated an AUC of 0.78 (95% CI 0.734–0.833) and a PSA level of 1.062 ng/ml as the optimal cutoff value, calculated to determine the value that gave the best combination of sensitivity and specificity for predicting positive ⁶⁸Ga-PSMA-11 PET [29]. Note that these PSA-based thresholds and cutoffs are intrinsically an oversimplification of the true nature of prostate cancer and its relationship with PSMA detection. PSMA expression, similar to PSA, can vary by histological differentiation, neuroendocrine content, androgen receptor signaling, prior treatment exposure, site of disease, or clinical state [34,35]. These factors are often not accounted for in such models.

The most clinically significant findings on PSMA PET that can impact patient management is that of disease outside of the prostate or prostate bed. Commensurate with this principle, most prospective studies have presented data on the anatomical distribution of detected prostate cancer recurrences. The pattern of PSMA-positive distribution seen in prospective studies so far has been consistent with recognized patterns of prostate cancer extraglandular spread, especially since these trials involve patients early in their disease course, with the most common site of recurrent disease identified by PSMA imaging being lymph nodes, as might be expected [36,37]. Notably, PSMA-positive lymph nodes frequently are smaller than the size defined by the Response Evaluation Criteria in Solid Tumors version 1.1 (RECIST 1.1) to be deemed pathological and so offers an advantage over conventional imaging, which at least partly relies on size to determine the presence of pathological lymph nodes. Earlier detection of isolated nodal recurrences may be amenable to treatments with salvage radiation or may facilitate extension of a planned radiation field to lymph node dissection[38]. PET using ⁶⁸Ga-PSMA-11 detects extrapelvic disease that may alter the management approach across a range of proportions of studied populations: between 8.6% of patients with a median PSA level of 0.2 (n = 70), [39] 40.6% of patients with a median PSA level of 0.83 (n = 314) [29], and 57.9% of patients with a median PSA level of 2.2 (n = 188) [26]. Given that PSMA PET allows sensitive visualization of both bone and soft tissue lesions in a single imaging study, it may potentially render the need for bone scans for staging of prostate cancer redundant in the future.

3.3. Diagnostic performance of F-18–labeled tracers

The ¹⁸F-labeled PSMA-targeted tracers have lower- energy emitted positrons that have shorter path lengths to annihilation in soft tissue, which may increase tumor-to-background resolution compared with ⁶⁸Ga-PSMA. Logistical advantages for ¹⁸F- labeled tracers are related to their generation in a cyclotron before distribution instead of requiring an on-site generator and their longer half- life for radionuclide decay (109 vs 68 min for ⁶⁸Ga) [40]. Of the ¹⁸F-labeled compounds, ¹⁸F-DCFPyl is most commonly studied. DCFPyL has a high affinity for prostate cancer and has been shown to detect sites of recurrent prostate cancer at PSA levels of <0.5 ng/ml [41,42]. Initial prospective studies of ¹⁸F-DCFPyL PET/CT report overall detection rates ranging from 79% to 85% [37,43]. In a single-academic-center prospective study evaluating the positivity rate of ¹⁸F-DCFPyL PET/CT, Song et al [43] report outcomes for 72 men with biochemical recurrence after prostatectomy (n = 42) or radiotherapy (n = 30), and a median PSA level of 3.0 ng/ml. PET/CT using ¹⁸F-DCFPyL demonstrated uptake in 85% of participants, with just under a third of patients presenting with a PSA level of <1.0 ng/ml [43]. Rousseau et al [44], in a trial of 130 men with an objective to evaluate the safety, sensitivity, and impact on patient management of ¹⁸F-DCFPyL in the settings of biochemical recurrence, reported that a mean PSA value of 5.2 ng/ml yielded a similar detection rate of 85%. Similar to the Ga-68 tracers, when stratified for PSA, ¹⁸F-DCFPyL detection rates are lower as PSA is lower.

Two large single-arm prospective trials of ¹⁸F-DCFPyL PET have provided prospective evidence of diagnostic imaging accuracy. Mena et al [37] reported a study with a primary endpoint of lesion detection rate, in which 90 patients with a median PSA level of 2.5 ng/ml had an overall detection rate of 77.8% who had lesion verification with histology (28/70; 40%) as well as composite imaging follow-up (13/70; 18.6%). The histology-based PPV was 93.3% (95% CI 77.6–99.2%) and the composite PPV was 96.2% (95% CI 86.3–99.7%). An ROC analysis to assess the discriminatory power of PSA for test sensitivity was carried out for patients who had undergone a radical prostatectomy. This analysis produced an AUC of 0.83 with an optimal PSA cutoff of 0.81 ng/ml to predict ¹⁸F-DCFPyL PET scan positivity [37]. In addition, the OSPREY prospective phase 2/3 trial evaluated the diagnostic performance of ¹⁸F-DCFPyL in a variety of clinical contexts, but contained that one cohort was specifically dedicated to recurrent prostate cancer (cohort B) that mandated evidence of disease amenable to biopsy [20]. This approach yielded, from the 93 evaluable patients who underwent biopsy, sensitivity among three readers ranging from 92.9% to 98.6% (lower bound of 95% CI 84.0–91.6%) and a PPV between 81.2% and 87.8% (lower bound of 95% CI 72.9–80.4%) with nine to 16 (12.2–18.8%) false positive results [20]. By organ system, DCFPyL performed well in bone (n = 51, sensitivity 96.8%, and PPV of 81.6%), nodes (n = 39, sensitivity 96.7%, and PPV 81%), and viscera (n = 3, sensitivity 100%, and PPV 90).

PET/CT using ¹⁸F-DCFPyL has also been tested rigorously in men with biochemically relapsed prostate cancer, with no radiographic evidence of disease by standard imaging in the CONDOR trial. The correct localization rate (CLR), a term that equates to PPV but with the added requirement of anatomical location matching, reflected the percentage of patients with an ¹⁸F-DCFPyL PET/CT–positive lesion that corresponded to the study’s composite truth standard. The CLR was the primary endpoint of this study. This study recruited 208 men who had been treated with primary therapy and who had relapsed biochemically. The study population had a median PSA level of 0.8 (0.2–98.4) ng/ml and underwent ¹⁸F-DCFPyL PET/CT imaging. PSMA-avid lesion(s) were identified in 59–66% of participants, who, by definition, had noninformative standard scans. The study reported a primary endpoint of CLR of 84.8–87.0% among three readers and a lower limit CI of >77% for all three readers [19]. This advances the evidence for ¹⁸F-DCFPyL PET/CT toward regulatory approval, and potentially presents clinicians and patients with the challenge of choosing the most appropriate PSMA imaging modality.

Compared with ⁶⁸Ga-PSMA-11, the radioligand ¹⁸F-PSMA-1007 has a longer half-life, better physical spatial resolution, and lower urinary excretion, all potential advantages for identifying smaller recurrent locoregional lesions that may be obscured by intense activity in the bladder of tracers with more intense urinary excretion. This characteristic of ¹⁸F-PSMA-1007 may facilitate greater detection sensitivity at PSA levels of <0.5 ng/ml, where recurrence is more likely to be locoregional. In a retrospective analysis of the diagnostic performance of ¹⁸F-PSMA-1007, 204 (81.3%) had evidence of recurrence, with a median PSA level of 1.2 ng/ml (range, 0.2–228 ng/ml). The detection rates for ¹⁸F-PSMA-1007 were also dependent on PSA levels, reporting a 61.5% (40/65) detection rate at a PSA level of 0.2–≤0.5 ng/ml increasing up to 94.0% for PSA ≥2 ng/ml [45]. In a small prospective study of 40 patients to determine the diagnostic performance of ¹⁸F-PSMA-1007, PET/CT was positive in 60% of patients with a median PSA level of 0.65 ng/ml. The rate is consistent with the detection rates at lower PSA values reported in the retrospective study. Follow-up according to a reference standard of histopathology or a composite of imaging or treatment response yielded 4/6 ¹⁸F-PSMA-1007–positive lesions verified as true positives, while all 34 ¹⁸F-PSMA-1007–negative lesions were confirmed true negatives. From this small verified cohort, a PPV of 66.7% and a negative predictive value of 100% are reported [46]. Based on tracer elimination data, it was expected that greater sensitivity for prostatic bed lesions might be revealed. However, only 9% of detected lesions in this study were in the prostatic bed, one of which underwent reference standard verification by histopathology. This relatively low proportion of detected disease in the prostate bed is consistent with rates reported in prospective studies of other tracers and may be related to the fact that 80% of this cohort had a radical prostatectomy. Overall, these prospective findings are promising for parallel performance with the previously described tracers, but likely a larger prospective trial will be needed to determine localization sensitivity and a more comprehensive PPV. A phase II diagnostic study comparing ¹⁸F-PSMA-1007 PET/CT with ¹⁸F-fluciclovine, in patients with early biochemical recurrence (PSA levels between 0.2 and 5.0 ng/ml) of prostate cancer, is currently recruiting (NCT04239742).

3.4. Implications for management

Prospective studies of PSMA PET radioligands have focused on the therapeutic consequences of undergoing PSMA PET imaging. Reviewed questionnaire-based studies, surveying referring clinician’s management intent, report that ⁶⁸Ga-PSMA-11 PET resulted in a change in management intent in 53–63% of patients. Roach et al [47] prospectively surveyed referring physicians at multiple institutions in Australia and recorded a change in management intent in 62% (192/312) of those with biochemical failure, with fewer patients consequently undergoing observation and more undertaking systemic therapy than prior to ⁶⁸Ga-PSMA-11 PET.

In a follow-up to the impact on management study by Roach et al [47], Emmett et al [48] carried out 3-yr freedom from progression (FFP) study on 260 of the participants who had prostate cancer recurrence and no contraindication to salvage radiation therapy. The results of the ⁶⁸Ga-PSMA-11 PET taken as part of the initial study on management change were shared with treating physicians, and subsequent management was recorded but not mandated by the follow-up study described by Emmett et al [48]. Of the 260 patients included, 186 underwent salvage radiation. Sites of radiation were as follows: prostatic bed only (38.2%), prostatic bed plus pelvic lymph nodes (49.4%), pelvic lymph nodes only, or metastases within or external to the pelvis (12.4%). Emmett et al [48] observed that 120/186 (64.5%) had FFP at 3 yr. Men with either a negative PSMA PET result or a scan positive for disease confined to the prostatic bed who underwent salvage radiation therapy had significantly higher 3-yr FFP rates than men with either pelvic nodal or distant metastatic disease on PSMA PET. The authors draw our attention to the potential for responder bias here. Calais et al [49] carried out a prospective survey of referring physicians to assess a secondary endpoint in a study of ⁶⁸Ga-PSMA-11 PET detection rates for patients with biochemical recurrence and found that, on follow-up, the intended management changes indicated after the ⁶⁸Ga-PSMA-11 PET/CT study were not implemented in 35% (35/101) of patients (NCT02940262).

In a further study by Hope et al (NCT02611882) [50] to measure the effect of PSMA imaging on intended management reported a ⁶⁸Ga-PSMA-11 PET detection rate of 82%, and pre- and postimaging surveys on intended management were returned by referring providers for 84% of patients. A change in management intent was observed in 59.6% (75/112). The most common change was a switch from systemic therapy to focal therapy [50], reflecting a shift in treatment goal from noncurative to curative intent with a more limited duration of therapy. It could also be argued that this shift in intent is more aptly described as a goal to eradicate gross evidence of disease and limit exposure to potentially toxic systemic therapy over the short term in order to delay the progression of metastatic disease. Bianchi et al (EudraCT 2015– 004589- 27 OsSC) [51] in a study to determine the impact on management reported that ⁶⁸Ga-PSMA-11 PET triggered a major change in therapeutic approach, as determined by the multidisciplinary team in 64% (177/276) of patients: 17.8% switched from palliative to curative intent, 7.2% from curative to palliative intent, and 23.9% from curative to surveillance intent. There was no change in management in 33.3%.

Similar to studies on ⁶⁸Ga-PSMA-11, prospective studies of ¹⁸F-DCFPyL PET/CT also assessed the impact of imaging on patient management. The CONDOR study reported that 63.9% of evaluable patients had a change in intended management after ¹⁸F-DCFPyL PET/CT [19,43]. For 21% (n = 43) of patients, the change in management intent was from noncurative systemic therapy to salvage local therapy, but similarly, higher sensitivity to metastatic disease means that management changes were also recorded as a change from salvage local therapy to systemic therapy in 28.3% (n = 58). Rousseau et al [44] reported a study (NCT02899312) that aimed to assess safety, sensitivity, and impact on patient management in 130 patients, of whom 55 (42%) had completed postscan assessments of changes in management. A change in treatment intent occurred in 65.5% of these patients and change in plans for systemic therapy occurred in 56.4%. Song et al [43] prospectively enrolled 72 men with biochemical recurrence after primary definitive treatment with prostatectomy (n = 42) or radiotherapy (n = 30). Of 72 patients, 43 (60%) had treatment changes after ¹⁸F-DCFPyL PET, as determined by the commencement of new treatment after imaging. In a specific study of men with recurrent prostate cancer following primary external beam radiotherapy or brachytherapy, Liu et al [52] describe a study in which ¹⁸F-DCFPyL PET alters the course of management. The primary endpoint of the trial (NCT02793284) was the detection rate of extraprostatic disease in this specific population of men for whom salvage treatment considerations at the point of identified local recurrence are diversifying and include radical prostatectomy, high-intensity focused ultrasound, brachytherapy, and cryosurgery. In a comparison of management plans for 79 men with a median PSA level of 4.8 ng/ml, before and after ¹⁸F-DCFPyL PET, potentially curative local salvage therapy was removed for six and added for 11 patients. This may reflect a confidence on the part of clinicians in the sensitivity of PSMA PET for detecting extraprostatic disease compared with conventional imaging. Taken overall, prospective studies surveying clinician treatment intent report that intended treatment approach before imaging evaluation was altered in roughly half to two-thirds of individuals.

The ultimate influence of these changes in management on a patient’s overall disease course and clinical outcome has not yet been defined well. For recurrence that manifests as low-volume metastatic disease, or oligometastatic disease, as detected on either conventional imaging or molecular imaging, incorporation of focal radiotherapy has been shown to delay progression of disease and the use of antiandrogen therapy [53,54]. However, both the definition and the management of oligometastatic disease, as defined by conventional imaging, in prostate cancer are unresolved. These uncertainties are broadened by the earlier detection of disease by PSMA PET. Although some clinicians act on these findings, further prospective clinical trials to optimize treatment paradigms are warranted [55]. Kneebone et al [56] carried out a prospective, single-center study of patients with oligometastatic prostate cancer undergoing stereotactic body radiotherapy (SBRT). The primary endpoint of biochemical failure was defined as a PSA level of nadir +0.2 ng/ml following SBRT. Fifty-seven patients were eligible with a median PSA level of 2.12 (0.15–8.9 ng/ml). There was a minimum follow-up of 6 mo following SBRT. The median biochemical disease-free survival for the cohort was 11 mo. In the 43 patients who had a biochemical failure, a repeat PSMA PET scan revealed no in-field failures, but distant recurrences occurred in most patients by 15 mo. Stefan et al [57] reported primary endpoints of overall survival, biochemical progression-free survival, and androgen deprivation therapy (ADT)-free survival for 86 patients with recurrent, oligometastatic prostate cancer identified by PSMA PET/CT over 8 yr and treated with image-guided radiotherapy of their metastases. After a median follow-up of 26 mo, the 3-yr overall survival and biochemical progression-free survival were 84% and 55%, respectively. The median time of ADT-free survival was 13.5 mo. Preliminary evidence for the benefit of radiation to PSMA-detected lesions in the oligometastatic setting suggests that this modality has facilitated a shift in therapeutic approach, but the ultimate impact of this shift on overall survival or indeed on intermediate endpoints such as metastasis-free survival have yet to be established fully [58,59]. As PSMA PET earns regulatory approval based on diagnostic performance and is therefore more available to clinicians to act on these findings, the question of whether such treatment changes lead to improved clinical outcomes must be the subject of future clinical trials. An ongoing phase III trial will recruit 193 patients to evaluate the success of salvage radiation therapy for recurrence of prostate cancer after prostatectomy with or without planning based on ⁶⁸Ga-PSMA-11 PET and a PSA level <1.0 ng/ml (NCT03582774) [60].

3.5. Comparative imaging

Conventional imaging with CT, bone scan, and MRI has been the primary tool for seeking the anatomical source of biochemical recurrence in the past. The prospective evidence summarized so far assessing ⁶⁸Ga-PSMA-11 and ¹⁸F-DCFPyL PET shows that the diagnostic performance of molecular PSMA-based imaging in terms of detection and accuracy is superior to that of conventional imaging across all studies, reporting high detection rates in those with negative conventional imaging scans. The performance of ⁶⁸Ga-PSMA-11 PET has also been prospectively directly compared with standard imaging [61,62] and whole-body MRI [58] for the investigation of biochemical recurrence after radical prostatectomy, and with pelvic MRI for those in whom salvage radiation therapy is being considered in this context [63]. In a large, multicenter, prospective study of PSMA-11 PET/CT imaging reported by McCarthy et al [62], patients with early biochemical relapse of prostate cancer (median PSA = 2.55 ng/ml) underwent both CT and bone scan as well as a ⁶⁸Ga-PSMA-11 PET scan within 8 wk of each other to determine differences in detection rate. Of the 199 patients who had negative standard imaging with CT and bone scan, 148 (74%) had detectable disease on PSMA imaging. Emmett et al [63] describes an absolute increase of 24% for the detection of extraprostatic fossa disease by pelvic MRI (7/88) versus PSMA (10/31) in men with high-risk features and biochemical failure after radical prostatectomy who were under consideration for salvage radiation therapy (NCT02131649). These findings highlight the magnitude of the difference in sensitivity between conventional and PSMA-based imaging, and highlight the potential divergence in decision-making that availability of PSMA imaging might prompt.

Similarly, in prospective studies that aimed to directly compare the performance of ¹⁸F-DCFPyL PET with conventional imaging, ¹⁸F-DCFPyL PET demonstrates a higher rate of detection in at least two prospective trials with a primary objective of comparison with conventional imaging [43,52]. Song et al [43] report that ¹⁸F-DCFPyL PET positivity was congruent with CT in only 33% of patients, MRI in 65% of patients, bone scan in 67% of patients, and lesion localization possible only on ¹⁸F-DCFPyL PET in 26 (36%) patients. For lesion detection in the prostate, ¹⁸F-DCFPyL PET detected the same proportion as conventional imaging, while showing increased sensitivity for regional and distant nodal recurrences and for oligometastatic disease [52]. Overall, prospective studies that report direct comparisons of the detection rates of conventional imaging with those of PSMA-based imaging affirm the superior detection rate of PSMA PET imaging. Beyond conventional imaging, prospective studies comparing PSMA with other metabolic tracers, including fluciclovine and choline, have been carried out, but these are beyond the scope of the current review.

The tracers discussed in this review, ⁶⁸Ga-PSMA-11, ¹⁸F-DCFPyL, and ¹⁸F-PSMA-1007, have not been compared with each other in any large prospective studies, and therefore superior detection ability for differing disease characteristics or localizations has not been demonstrated. Although the tracer with the optimal performance characteristics has yet to be defined for individual patients and their disease characteristics, the increasing availability of PSMA tracers assures that clinicians will have a set of diagnostic tools in the future, any one of which is superior to what is being used now as conventional imaging.

4. Conclusions

Prospective studies of PSMA-based imaging have, so far, verified detection rates reported in retrospective studies, establishing the diagnostic accuracy of PSMA-based imaging for the detection of recurrent prostate cancer, and affirmed the safety of these agents. Prospective studies of the radioligands ⁶⁸Ga-PSMA-11, ¹⁸F-DCFPyL, and ¹⁸F-PSMA-1007 with PET imaging have established the ability of these tracers to identify sites of recurrent disease accurately and safely, allowing discrimination between locoregional and distant recurrence of disease. Across studies, ⁶⁸Ga-PSMA-11 has a detection rate of 57–71% at a PSA level of 0.5–1.0 ng/ml, while ¹⁸F-DCFPyL across prospective studies reviewed here has a detection rate between 50% and 78%. These compounds have not been compared formally in a prospective setting to assess whether one performs better than another in set circumstances. Individually assessed, these have all shown similar detection rates with no sufficient evidence of superiority of one compound over the other. The detection of disease by these imaging techniques is high even at PSA levels as low as 0.5 ng/ml, and detection rates consistently increase as PSA increases. Regulatory approval for the use of ⁶⁸Ga-PSMA-11 PET has been obtained in the USA for patients with suspected prostate cancer recurrence based on elevated serum PSA levels and those with suspected metastasis who are candidates for initial definitive therapy. Further approvals are likely forthcoming for additional PSMA radioligands. So far, prospective studies of PSMA detection rates in those with recurrent prostate cancer have largely included patients with a mixed primary treatment history: radical prostatectomy, radical radiotherapy, and prostatectomy followed by radiation, with or without ADT. More granular understanding of the influence of treatment history and differing clinical states on PSMA expression and detection sensitivity could be gained by a prospective study of more singular patient populations.

Prospective studies report a shift in management intent in up to two-thirds of patients following imaging. Whether the consequent changes in management prompted by PSMA-based restaging results in meaningful differences in clinical outcomes remains to be determined. Incorporation of these imaging techniques into large prospective randomized trials of therapeutic interventions in the context of biochemical recurrence will allow assessment of the longitudinal impact on patient outcomes of these molecular imaging techniques and will likely be the focus of future prospective studies in this field, but will not be a requisite for the entry of PSMA PET into clinical practice as a diagnostic tool.