Impact of long-term androgen deprivation therapy on PSMA ligand PET/CT in patients with castration-sensitive prostate cancer

The detection of recurrent prostate cancer (PC) has always been challenging using conventional imaging modalities such as computed tomography (CT) and magnetic resonance imaging. Consequently, there has been a need for new imaging tools with improved sensitivity and specificity in diagnosis. In this context, nuclear imaging targeting prostate-specific membrane antigen (PSMA) has received increased attention. PSMA is a transmembrane protein which is overexpressed in the majority of aggressive prostatic adenocarcinomas [1‐3]. In May 2011, the PSMA ligand ⁶⁸Ga-PSMA-11 was introduced for the clinical imaging of PC at our institution. Since then, positron emission tomography (PET) using ⁶⁸Ga-PSMA-11 and alternative PSMA ligands has been regarded as a significant step forward in the diagnosis of recurrent PC. The first studies indicated that this novel method is superior to alternative imaging modalities used for the detection of recurrent PC [4‐6]. Later studies including larger patient cohorts confirmed the high sensitivity and specificity of ⁶⁸Ga-PSMA-11 PET/CT [7‐10].

One of the most important questions in PSMA imaging concerns the role of androgen deprivation therapy (ADT). It is known from preclinical experiments that short-term ADT can quantitatively increase PSMA expression in castration-sensitive PC cells [11, 12]. This effect was shown with analogues of the luteinizing hormone-releasing hormone (LHRH) as well as with antiandrogen agents such as bicalutamide and enzalutamide. In addition, Hope et al. demonstrated in one patient, that PSMA ligand uptake in PC lesions significantly increased within 4 weeks of ADT with bicalutamide [13]. On the other hand, the effects of long-term ADT on PSMA expression and tumour visibility in PSMA ligand PET/CT have not yet been investigated in detail. Two clinical studies including larger patient cohorts (n = 317 and 1,007) demonstrated that patients during ongoing ADT significantly more often had a pathological ⁶⁸Ga-PSMA-11 PET/CT scan [6, 7]. The authors considered that this association could have been a result of either a possible biological impact of ADT (increase of PSMA expression similar to the findings of the preclinical experiments with short-term ADT mentioned above) or the fact that patients with a pathological scan had more advanced tumour stages which had led to the initiation of ADT.

The aim of the current evaluation was to directly analyse the clinical impact of long-term ADT on tumour visibility and tumour detection on ⁶⁸Ga-PSMA-11 PET/CT. To the best knowledge of the authors, no systematic studies have been published including patients who were scanned with PSMA ligand PET/CT before and after starting ADT.

The average duration of ADT before the second PSMA PET/CT scan was 229 ± 89 days (range 42–369 days, median 230 days). All patients had at least one lesion characteristic of PC on the initial PET/CT scan (without ADT). Overall, 31 PC lesions were visible in ten patients on the PET/CT scan before initiation of ADT. During ongoing and effective ADT, as demonstrated by a decreasing PSA level, 14 of the 31 lesions (45%) were still visible in eight of the ten patients (80%) on the second PET/CT scan (Table 1). The lower number of tumour lesions during ongoing ADT was statistically significant (p = 0.0199). Amongst all lesions characteristic of PC, 19 were lymph node metastases (eight still visible during ongoing ADT), ten were bone metastases (five still visible during ongoing ADT), one was a soft tissue metastasis (lung, no longer visible during ongoing ADT), and one was a primary tumour (also still visible during ongoing ADT). Even though the overall number of lesions was low, the statistical analysis showed that lymph node metastases were more likely to become no longer visible than bone metastases (p = 0.0499).

During ongoing ADT, SUVmean increased in four lesions (12.9%), decreased in 22 (71%) and remained stable in five (16.1%; Fig. 2). SUVmax increased in five lesions (16.1%), decreased in 23 (74.2%) and remained stable in three (9.7%; Fig. 3). The decrease in both SUVmean and SUVmax was statistically significant (p < 0.0001). Figures 4, 5 and 6 present examples of the negative impact of continuous ADT on tumour visibility on PSMA ligand PET/CT.

Between PET-1 (without ADT) and PET-2 (during ongoing ADT) the SUVmean to tumour volume ratio increased in six lesions (19.4%) and decreased in 23 (74.2%; Fig. 2). In two lesions, the ratio was not calculable because of a complete morphological response to the ADT. The same results were found for the SUVmax to tumour volume ratio (Fig. 3). The decrease in the SUV to tumour volume ratio during ongoing ADT was also statistically significant (p = 0.0009 for SUVmean and p = 0.0007 for SUVmax).

PSA levels showed a significant decrease in all patients (p = 0.0001; Fig. 7). Considering a PSA level of 0.1 ng/ml or lower as a complete response, six patients, patients 4–6 and 8–10 (Fig. 7) with an ADT duration of 184–369 days (average of 236 days), fulfilled this criterion. In these patients, seven of 21 lesions (33.3%) in four patients were still visible during ongoing ADT (Figs. 8 and 9). A complete imaging response (negative PSMA PET/CT) was observed in two of the six patients with a complete PSA response (patients 4 and 10).

One of the most important clinical questions in PSMA-based imaging concerns the role of ADT. It is known from preclinical research that short-term ADT with LHRH analogues as well as androgen receptor blockers such as bicalutamide and enzalutamide can quantitatively increase PSMA expression in PC cells [11‐13]. If this could also be observed in the clinical setting in patients receiving long-term ADT, it would have a substantial positive impact on PSMA imaging as well as radioligand therapy with PSMA ligands. Multivariate analyses including larger patient cohorts with PSMA ligand PET/CT published so far [6, 7] suggest that ⁶⁸Ga-PSMA-11 PET/CT significantly more often shows pathological findings in patients receiving ADT. All these preclinical and clinical reports have tended to convey the impression that ADT increases tracer uptake and thereby leads to improved sensitivity of PSMA-11 PET/CT. This impression is reinforced by the fact that even PSMA ligand PET/CT guidelines suggest that ADT does not need to be paused prior to scanning [17]. However, the results of the current study demonstrate for the first time that long-term and effective ADT significantly decreases tracer uptake on PSMA ligand PET/CT: only 45% of the lesions were still visible during ongoing ADT.

Considering the preclinical data mentioned above, no principal differences were detected between the androgen-deprivation agents administered to our patients and those agents included in the preclinical studies. In some of the studies, the dose of ADT was higher than normal doses administered to patients, though: Evans et al. [11] administered 10 mg/kg of enzalutamide to mice (dose in patients usually 160 mg/day), and Hope et al. [13] administered 10 mg/kg/day of apalutamide to mice (dose in patients 240 mg/day). On the other hand, Hope et al. used 50 mg/day of bicalutamide in the single patient reported, which is in the recommended range for treatment of PC (50–150 mg/day) [13].

The most significant difference between our patients and the studies discussed above is the duration of ADT: while the preclinical studies as well as the single patient of Hope et al. received ADT for 2–30 days [11‐13, 18], our patient cohort was treated for an average of 229 days. The shortest duration of ADT in our cohort was 6 weeks (patient 7) which is the closest to the 4 weeks of ADT in the patient of Hope et al. [13]. Although in patient 7 a significantly negative impact of ADT on the results of the PET was shown as well, we strongly believe that the duration of ADT is a key factor in the impact of ADT on PSMA expression. ADT of short duration may increase PSMA expression while long-term ADT probably has the opposite effect. The latter assumption has been observed in a preclinical study by Liu et al. [18]. However, more longitudinal clinical studies are necessary to analyse this issue.

The intensity of tracer uptake in relation to lesion volume was also analysed in this study. This represents a volume-independent estimation of PSMA expression. This analysis frequently demonstrated a decrease in uptake during ongoing ADT, counteracting any remaining assumption that long-term ADT does not negatively affect PSMA expression on PC cells. Overall, we assume that long-term androgen deprivation results in apoptosis of androgen-sensitive PC cells, an effect that was demonstrated by the 1966 Nobel Prize winner Charles Brenton Huggins. In general, the effect of the reduction in tumour volume and the decrease in PSA values caused by apoptosis is accompanied by a negative impact on tumour detection by imaging. The higher probability of a pathological PSMA PET/CT scan in patients receiving ADT found in previous studies was most likely caused by bias in the selection of patients with more advanced tumour stages which would have led to initiation of ADT.

Currently the most accepted indication for PSMA ligand PET/CT is biochemical relapse of PC. Depending on tumour spread, tumour detection can guide locoregional approaches or systemic therapies. In such a setting the highest possible sensitivity is mandatory and physicians as well as patients would rather accept short-term ADT or possibly pausing ADT to optimize the diagnostic accuracy of PSMA ligand PET/CT. Considering the mainly preclinical data discussed above, short-term ADT may eventually increase PSMA expression on PC cells, thereby increasing the sensitivity of PSMA ligand imaging. The findings of all preclinical studies discussed here as well as in the single patient of Hope et al. suggest that there is a functional coupling between PSMA and androgen receptors. However, these studies do not provide any clinical evidence. Therefore, further evaluations are necessary to determine if there is any real clinical coupling between PSMA and androgen receptors. Furthermore, it is well known that LHRH analogues cause an increase in the whole metabolic activity of PC cells. In this context, an increase in uptake of PSMA ligands seems possible and cannot be excluded. However, LHRH analogues as well as androgen receptor blockers such as bicalutamide caused a quantitative increase in PSMA expression in the studies discussed above [11‐13].

Previous evaluations have shown that about one third of patients referred for PSMA ligand PET/CT are receiving ADT [6, 7]. We assume that a proportion of them are referred to confirm the presence of recurrent disease. However, according to our results, PSMA PET/CT delineates only a fraction of the tumour disease in patients receiving long-term ADT with a continuous PSA response. Therefore, these scans are not suitable for subsequent therapy planning. Instead, PSMA ligand PET/CT in patients receiving ADT can be used for response assessment. For example, early detection of hormone-resistant lesions can open up new possibilities for individual treatment strategies. This suggestion is potentially supported by our results which showed that in patients with complete PSA remission, 33% of the lesions were still visible (Figs. 8 and 9). This implies that ADT can cause some “cosmetic effects” on PSA levels while viable tumour is still present. It may be that the 12.9% of lesions showing an increase in tracer uptake on PET-2 or those lesions still visible during ongoing ADT despite (complete or incomplete) PSA remission might correlate with those cell clones that become castration-resistant first.

A future strategy could be to perform PSMA ligand PET/CT despite a sufficient PSA response. If the scans show remaining viable (PSMA-positive) tumour lesions, ADT could be augmented by novel androgen signalling inhibitors or taxanes. Otherwise, ADT could be continued without change. Such a strategy would refine and individualize the recommendations derived from the CHAARTED, LATITUDE and STAMPEDE trials, which showed a remarkable survival benefit for primarily metastasized patients by adding therapy during the castration-sensitive stage [19‐21]. If the critical cell population, from which future tumour progression is derived, is characterized by continuing PSMA expression during ongoing ADT, PSMA-based radioligand therapy, which has already shown remarkable antitumour activity in the last-line setting [22‐26], might also be a powerful therapeutic partner for ADT.

Although controversial from a therapeutic point of view, pausing long-term ADT could help increase the sensitivity of PSMA ligand PET/CT with the aim of visualizing the maximum possible extent of disease. The still-unanswered question is how long does ADT need to be paused to increase the sensitivity of PSMA ligand imaging. To our knowledge, no data are available on how long PC cells need to achieve their natural PSMA status after termination of ADT or whether discontinuing ADT might even cause a temporary PSMA rebound. It is also unclear how PSA status is associated with the degree of PSMA expression. It is known from preclinical studies that PSA and PSMA promoters are regulated in an opposite manner with short-term ADT leading to a decrease in PSA levels and an increase in PSMA expression [12]. On the other hand, it remains a matter of speculation if the full PSMA expression status and therefore the maximum possible tumour visibility on PSMA ligand imaging will be achieved if PSA levels rise to pre-ADT levels.

In six of ten morphologically growing lesions the SUV decreased during ongoing ADT. This observation demonstrates that changes in SUV can occur independently of lesion volume. At least regarding bone metastases, it is known that their volumes can increase after different therapies despite a good PSA response. However, this increase in volume is caused by sclerosis and must not be interpreted as tumour progression. Consequently, the size of bone metastases was excluded from the Response Evaluation Criteria In Solid Tumors (RECIST). According to our experience, we believe that PSMA ligand uptake in tumour lesions represents a more accurate therapy response criterion than lesion size.

The limitations of this study include the low number of patients who could be included in the current analysis despite a large database of more than 1,700 patients, the wide range of ADT duration and the retrospective nature of the evaluation.

Currently, patients with biochemical relapse should be referred for PSMA PET before starting ADT. Future studies are needed to show the impact of short-term ADT in the clinical setting as well as of alterations in PSMA expression after pausing ADT.