A Prospective Trial of ⁶⁸Ga-PSMA and ¹⁸F-FDG PET/CT in Nonmetastatic Prostate Cancer Patients with an Early PSA Progression During Castration

This article is featured in Highlights of This Issue, p. 4427

The development of next-generation imaging (NGI) has significantly changed the landscape of prostate cancer (1). In patients with biochemical recurrence (BCR), prostate-specific membrane antigen (PSMA) PET/CT is able to provide superior sensitivity to detect micrometastases and may lead to a change in treatment plan in nearly 50% of patients (2). The superiority of PSMA PET/CT has also been evidenced in nonmetastatic castration-resistant prostate cancer (nmCRPC). Fendler and colleagues recruited 200 patients with nmCRPC defined as no metastases on conventional imaging with prostate-specific antigen (PSA) levels >2 ng/mL who all received PSMA PET/CT (3). On the one hand, imaging analysis showed that 75% of patients had lymphatic or distant metastases disease (N+/M+) and 46% had multiple/disseminated metastases. On the other hand, 29% of patients had unifocal or oligometastatic disease who may be good candidates for oligometastases-directed therapy (OMDT; refs. 4, 5). Together, these results showed that the application of PSMA PET/CT has revolutionized the definition of nmCRPC previously defined by conventional imaging and provided new treatment opportunities in these patients.

While nmCRPC was a long-term disease using apalutamide, approximately 20% of nmCRPC patients developed metastases based upon conventional imaging within 1 year (6). Given the increasing potential role of OMDT, there are unmet needs to identify metastases as early as possible, when still in the oligometastatic setting. To address this, attempts have been made by incorporating dual-tracer PET/CT. A poor outcome was observed in patients with mCRPC with low PSMA expression or discordant fluorodeoxyglucose (FDG)-avid lesions (7, 8). Moreover, patients with mCRPC with fluorodihydrotestosterone (FDHT)-/FDG+ lesions were found to have a worse outcome than the other phenotypes and possibly resistant to androgen deprivation therapy (ADT; ref. 9). These studies suggest that when FDG-dominant disease is present in patients with mCRPC, outcomes are poor. However, whether similar results could be observed in men prior to documented metastasis (i.e., nmCRPC) remained unclear.

Therefore, by incorporating dual-tracer (⁶⁸Ga-PSMA and ¹⁸F-FDG) PET/CT, we performed the current trial to evaluate disease burden and determine whether heterogeneous disease (especially PSMA-FDG+ disease) exists in patients with an early PSA progression (≤2 ng/mL) during castration (Supplementary Fig. S1).

This study was an investigator-initiated, single-center trial and prospectively included 37 patients with an early PSA progression during castration (T < 50 ng/dL) and were confirmed as having high-risk [PSA doubling time (PSADT) ≤10 months] nonmetastatic disease by conventional imaging. The detailed inclusion and exclusion criteria are shown in Supplementary Table S1. All patients underwent ⁶⁸Ga-PSMA PET/CT and ¹⁸F-FDG PET/CT with a less than 5-day interval. To compare disease heterogeneity across different disease stages beyond the 37 patients in the prospective study, we further retrospectively included consecutive cohorts of patients who underwent dual-tracer PET/CT outside of the current trial during the same time period: BCR (n = 14), hormone-sensitive prostate cancer (HSPC; n = 18), as well as mCRPC (n = 9). All scans were reviewed and interpreted by three nuclear medicine specialists who were blinded to the patient-related medical data. The imaging specifics are shown in the Supplementary Methods. Because DNA damage repair (DDR) mutations may be associated with PSMA expression (10), germline DDR aberrations were tested in 29 patients of the prospective trial cohort (Supplementary Methods). The trial was performed after the approval of the Human Ethics Committee of Fudan University Shanghai Cancer Center and was conducted in accordance with the declaration of Helsinki (preregistered with ChiCTR 1900022634). Written informed consent was obtained from all prospectively included patients. Retrospective patients were included after obtaining Human Ethics Committee approval with waiver of informed consent.

The hypothesis was that the addition of FDG PET/CT to PSMA PET/CT would yield 20% (H₁ = 0.20) of PSMA-FDG+ disease in patients with an early PSA progression during castration. The sample size was estimated (power = 0.9, α = 0.05, H₀ = 0.05 and H₁ = 0.20; one-sided test resulting in 34 cases; with 10% inflation to 37 projected patients) by PASS 15 (NCSS Statistical Software.) prior to trial initiation.

Baseline characteristics as well as image findings were collected. Disease burden was classified into low burden (lymph node metastasis only or ≤3 bone metastases ± lymph node metastasis regardless of the location and no visceral metastasis) and high burden (≥4 bone metastases regardless of the location or any visceral metastasis). For the heterogeneity analysis, lesions were classified into PSMA+FDG± lesions (PSMA+/FDG- and PSMA+/FDG+) and PSMA-FDG+ lesions according to the tracer uptake status. Specifically, positive FDG and PSMA uptake was defined as focal avidity greater than the background in the mediastinal blood pool when excluding physiologic uptake or other important pitfalls of the two radiotracers (11, 12). PSADT was calculated using the three consecutive rising PSA levels before enrollment (13, 14). OMDT eligibility was analyzed by an experienced radiation oncologist according to the criteria of the SABR-COMET trial (Supplementary Table S2; ref. 5). The endpoint of time-to-systemic therapy was reached when patients were shifted to an additional life-prolonging systemic therapy (e.g., enzalutamide, chemotherapy, olaparib etc.) from the current used ADT regimen.

Categorical data were shown as frequencies and percentages, and continuous data were shown as medians and interquartile ranges (IQR). Logistic regression was performed to estimate the OR for N+/M+ disease on PET/CT in univariable analysis. Wilson score method was used to calculate the 95% confidence intervals (CI) for standard error. The χ² test was applied to analyze the difference in lesion distribution. The Mann–Whitney test was used to compare the mean values of PSMA SUV_max in PSMA+FDG± lesions versus PSMA-FDG+ lesions. Kaplan–Meier method was used to analyze time-to-systemic therapy. All other statistical analyses were performed using SPSS 22.0 (IBM Corp.), with a two-sided P < 0.05 considered statistically significant.

A total of 37 patients with an early PSA progression during castration (T < 50 ng/dL) were prospectively included. All patients had no medical records of metastatic disease and were confirmed as nonmetastatic by conventional imaging before the enrollment. The median age was 67 years (IQR: 65–73 years), and 73% (27/37) of patients had Gleason grade group ≥4 disease (Supplementary Tables S3 and S4). The median duration of ADT was 23 months, with a median PSA level before enrollment of 0.57 ng/mL (IQR: 0.24–1.24 ng/mL). The median PSADT was 3.34 months (IQR: 1.73–6.50 months). All patients had undergone radical prostatectomy, and 8 patients were treated with postoperative prostate radiotherapy. Two patients had grade 1 adverse events (headache N = 1, nausea N = 1) with no grade 2 or 3 adverse event observed (Supplementary Table S5).

Overall, 114 lesions were detected in 29 of the 37 patients by either PSMA PET/CT, FDG PET/CT, or both. The positive rate of PSMA PET/CT was 70% (26/37), of which 2 were local recurrence in the prostate bed. Sixty-five percent (24/37) of patients were shown to have N+/M+ disease. Of these patients, 17 patients and 7 patients were diagnosed with low burden and high burden metastatic disease, respectively. The positive rate was 58% (N = 14/24) and 92% (N = 12/13) with PSA of 0.2–1 ng/mL and 1–2 ng/mL, respectively. FDG PET/CT was positive in 49% (18/37) of the patients, and 1 had local recurrent disease. Forty-six percent (17/37) of the patients had N+/M+ disease. Among these patients, 13 and 4 were diagnosed with low burden and high burden metastatic disease, respectively (Fig. 1; Table 1).

We then performed an exploratory analysis to select predictive factors for N+/M+ disease on either PSMA PET/CT, FDG PET/CT, or both. In the univariable analysis, a short PSADT (≤6 months) was associated with N+/M+ disease on PSMA PET/CT (P = 0.009, OR = 8.000), while a high Gleason grade group (≥4) disease was the predictor of N+/M+ disease on FDG PET/CT (P = 0.022, OR = 13.091). A combination of Gleason grade group ≥4 and PSADT ≤6 months was strongly associated with N+/M+ disease on PSMA and FDG PET/CT (P = 0.030, OR = 6.800; Table 2). For those patients with both risk factors, 89% (17/19) had metastatic disease missed by conventional imaging modalities.

Using a head-to-head comparison of PSMA uptake and FDG uptake, we were able to classify every lesion into the PSMA+FDG± group or the PSMA-FDG+ group. Of the 114 lesions detected in 29 patients, 81 were PSMA+FDG± and 33 lesions were PSMA-FDG+ (Fig. 2). The mean SUV_max for PSMA positive lesions was 17.79 (95% CI: 9.85–25.72) and 5.03 (95% CI: 4.12–5.93) for FDG positive lesions (Supplementary Fig. S2). At the patient level, 9 patients (24%; 95% CI: 10%–39%) showed at least one new PSMA-FDG+ lesions (Fig. 1). Among them, 7 and 2 had low burden and high burden metastatic disease, respectively. For staging, the addition of FDG PET/CT was able to increase the N+/M+ detection rate from 65% to 73% compared with PSMA PET/CT alone (Fig. 1; Table 1).

The comparison of lesion distribution after categorization into bone, lymph nodes, prostate bed, and visceral lesions showed a quite similar general distribution between PSMA+FDG± lesions and PSMA-FDG+ lesions. However, within the bone lesions, the distribution of PSMA-FDG+ lesions was significantly different as compared with that of PSMA+FDG± lesions (P = 0.022). Specifically, there was a higher proportion of PSMA-FDG+ lesions than PSMA+FDG± lesions located on the axial skeleton (Fig. 2). For lymph node lesions, there was a higher proportion of PSMA-FDG+ lesions than PSMA+FDG± lesions in the pelvis, although no significant difference was observed (P = 0.363).

We reviewed these lesions and further classified their treatability according to the enrollment criteria of the SABR-COMET trial (5). At the patient level, 19 patients (51% of all patients) with 51 lesions could be enrolled for OMDT using the two tracers, while 6 (32%) of them with 10 PSMA-FDG+ lesions could not be precisely targeted using PSMA PET/CT alone (Supplementary Table S6).

We further retrospectively included consecutive BCR (n = 14), HSPC (n = 18), and mCRPC (n = 9) cohorts outside of the trial during the same inclusion period. By incorporating FDG PET/CT, PSMA-FDG+ disease was rarely detected in the hormone-sensitive cohorts, but frequently found in the castration-resistant cohorts (6% vs. 33%; Table 3).

Except for noncompliance in 1 of the 9 patients with PSMA-FDG+ lesions, we validated these patients with composite validation (Supplementary Table S7). We showed a high true positive rate by pathologic confirmation (n = 2; Fig. 3), response or progression on imaging (n = 4; Fig. 3; Supplementary Fig. S3) and decline in PSA level after OMDT (n = 2) in 8 of the 9 patients. With a median follow-up of 7.5 months (95% CI, 6.8–8.3 months), 14 of the 37 turned to an additional systemic therapy. Patients with PSMA-FDG+ disease seems to have a shorter time to systemic therapy, but no statistical significance was observed (P = 0.360, HR = 1.635; Supplementary Fig. S4). Further long-time follow-up is needed to suggest the outcome.

Using dual-tracer PET/CT (⁶⁸Ga-PSMA and ¹⁸F-FDG), we prospectively evaluated the disease burden and tumor heterogeneity in patients with an early PSA progression (0.2–2 ng/mL) during castration (T < 50 ng/dL) and negative on traditional imaging. Consistent with prior studies (15), many of these patients despite the low PSA levels had metastatic disease on PSMA PET/CT. Importantly, by further incorporating FDG PET/CT, we identified 3 metastatic patients from 11 patients who were negative on PSMA PET/CT and found that 24% patients had at least one FDG+/PSMA- lesion. On exploratory univariable analysis, Gleason grade group, but not PSADT, predicted N+/M+ disease on FDG PET/CT. Using combined PSMA and FDG PET/CT, N+/M+ disease was detected in 73% of these “nonmetastatic” patients, suggesting a high prevalence of metastatic disease even at a very low PSA level during castration. Specifically, we found that 51% patients would have been eligible for OMDT using the enrollment criteria of the SABR-COMET trial (3). These results highlighted the value of dual-tracer PET/CT and possible early treatment stratification in this group of patients.

While the presence of heterogeneous disease has been evidenced in patients with very late-stage mCRPC (7, 8), its prevalence has not been evaluated in other stages. In this study, we evaluated the prevalence of PSMA-FDG+ disease of patients in our prospective trial with that of the HSPC and BCR cohorts. Although the retrospective comparing cohorts might induce selection bias and weaken the evidence level, we found a higher rate (24%) of PSMA-FDG+ disease in patients with an early PSA progression during castration that was rarely observed in the BCR (0%) and HSPC cohorts (6%). The identification of PSMA-FDG+ disease is important and may provide potential clinical utility in the following aspects. First, PSMA-FDG+ may represent a variant with a distinct prognosis and treatment response. In the study done by Hofman and colleagues, the 16 patients with PSMA-negative or FDG discordant excluded from Lu-177 therapy experienced a poor outcome (median OS, 2.5 months; ref. 8), while the included patients showed a better response to Lu-177 therapy when compared with Heck and colleagues' study without FDG PET screening (PSA decline of ≥50%: 57% vs. 38%; ref. 16). Although, PSMA-FDG+ disease was excluded from Hofman and colleagues' study, patients PSMA+FDG+ disease was included and subsequently treated. The recent study by their group showed the FDG-positive tumor volume (HR = 2.6, 95% CI: 1.4–4.8) was a prognostic biomarker for survival in patients undergoing Lu-177 therapy (17). However, these are all non-head-to-head evidences, and whether PSMA-FDG+ disease harbored a poor prognosis should be further evaluated. In this study, although not statistically significant, a shorter time to systemic therapy was suggested for PSMA-FDG+ disease with a preliminary follow-up. Moreover, we observed one patient with both PSMA+FDG± and PSMA-FDG+ lesions having only PSMA-FDG+ progression after treated with OMDT (Fig. 3D–F; comparing case in Supplementary Fig. S3). Second, metastatic lesions with a high FDG uptake have been shown to be associated with a shorter time to hormonal treatment failure, suggesting that these lesions might have a poor response to ADT (18). From a theoretical perspective, it is plausible that these FDG-avid lesions are more likely to respond to chemotherapy (19). Finally, because PSADT is the only clinical parameter validated to predict metastasis in nmCRPC, PSMA-FDG+ lesions that are strongly associated with Gleason grade group may be omitted. Therefore, taking both PSADT and Gleason grade group into consideration may better identify aggressive disease in patients with early nmCRPC and allow for better treatment stratification.

OMDT is a promising approach for low-volume metastatic disease. Cumulative evidence from randomized phase II trials suggested that the ablation of all detectable metastatic lesions was most likely to benefit patients with improved PFS and OS (20–22). In our study, 73% of patients who failed definitive local treatment (radical prostatectomy with or without radiotherapy) had N+/M+ disease even at a low PSA level during castration. According to the enrollment criteria of SABR-COMET trial, 19 (51%) patients were eligible for OMDT. This rate was higher than that of patients with a PSA level ≥2 ng/mL (unifocal and oligometastatic disease, 29%; ref. 3), suggesting that patients with an early PSA progression after castration may be strong candidates for NGI and OMDT trials. Moreover, the addition of FDG PET/CT further identified an additional 8% metastatic patients and showed that 24% patients had at least one PSMA-FDG+ lesion that was missed by PSMA PET/CT. According to the STOMP trial, nearly 20% of patients who were randomized to the SBRT arm failed in the first 3 months (20). It is highly possible that these aggressive patients had undetected lesions missed by using single-tracer (choline) PET/CT. In addition, the published studies (Supplementary Table S8) that used SBRT in metastatic patients all adopted single nuclear imaging as the primary modality although disease heterogeneity might vary from HSPC to CRPC. These results highlight the importance of applying dual-tracer PET/CT to refine the enrollment of OMDT trials and ensure that heterogeneous metastases are treated. However, whether the introduction of dual-tracer PET/CT in OMDT trials could be translated into an improved survival benefit should be further evaluated. Moreover, for the cross-sectional nature, further longitudinal trials are warranted to sub-stratify oligometastatic disease and may better stratify patients' prognosis and treatment strategy (23).

This study suggested a role for dual-tracer PET/CT in patients with an early PSA progression during castration. Its application may not only identify aggressive disease but also better select the candidates for OMDT trials. Although this study was a preregistered and prospectively enrolled trial that included a series of homogeneous patients, it was not without its limitations. The first limitation lies in the small sample size. Although we achieved the primary outcome to detect >20% PSMA-FDG+ disease in the rarely studied population, secondary findings still needed to be validated in confirmatory series. Second, for the retrospective nature of the comparing cohorts (BCR, HSPC and mCRPC), the results are subject to selection bias and should be interpreted with caution. Moreover, although one experience radiation oncologist was included to analyze the eligibility for OMDT, bias of interpretation might be induced. Therefore, further OMDT trials using dual-tracer PET scans should include joint assessment for eligibility to reduce potential bias. Finally, a pathologic conformation of each PSMA-FDG+ lesion could not be obtained due to ethical and practical restrains. Nonetheless, except for noncompliance in 1 patient, we showed a high true positive rate by the composite validation in 8 of the 9 patients with PSMA-FDG+ lesions. The prognosis of these PSMA-FDG+ disease detected by dual-tracer PET/CT in this early disease phase should be further evaluated with long-term follow-up.

In conclusion, the dual-tracer PET/CT detected 114 lesions in 78% of patients with an early PSA progression during castration. The heterogeneity analysis showed that 33 PSMA-FDG+ lesions were detected in 24% of patients. A short PSADT (≤6 months) was associated with N+/M+ disease on PSMA PET/CT, while a high Gleason grade group predicted positive findings on FDG PET/CT. Compared with HSPC, a higher prevalence of PSMA-FDG+ disease was observed in patients with an early PSA progression during castration.

B. Wang reports grants from Fudan University Shanghai Cancer Center Fund during the conduct of the study. Y. Zhu reports grants from National Nature Science Foundation of China, Shanghai Rising Star Program, Fudan University Shanghai Cancer Center Fund, and Shanghai "Rising Stars of Medical Talent" Youth Development Program during the conduct of the study. No potential conflicts of interest were disclosed by the other authors.

B. Wang: Conceptualization, data curation, formal analysis, methodology, writing-original draft, writing-review and editing. C. Liu: Data curation, methodology, writing-review and editing. Y. Wei: Data curation, software, formal analysis, writing-review and editing. J. Meng: Formal analysis. Y. Zhang: Data curation. H. Gan: Data curation, investigation. X.-P. Xu: Data curation, investigation. F. Wan: data curation, software, methodology. J. Pan: Data curation, software, investigation. X. Ma: data curation, investigation. S. Hu: Data curation, investigation. S. Freedland: Writing-review and editing. S. Song: Conceptualization, resources, data curation. D. Ye: Conceptualization, resources, supervision, funding acquisition, writing-review and editing. Y. Zhu: Conceptualization, resources, data curation, formal analysis, funding acquisition, methodology, writing-original draft, writing-review and editing.

The study was supported by National Nature Science Foundation of China (grant no. 81972375), Shanghai Rising Star Program (grant no. 16QA1401100), Fudan University Shanghai Cancer Center Fund (grant nos. YJJQ201802 and YJQN201923) and Shanghai “Rising Stars of Medical Talent” Youth Development Program (grant no. 2018ZY).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.