Predictive factors for response to salvage stereotactic body radiotherapy in oligorecurrent prostate cancer limited to lymph nodes: a single institution experience

Prostate cancer (PCa) patients diagnosed with biochemical recurrence and limited metastatic disease are conventionally treated with androgen deprivation therapy (ADT). Recently, stereotactic body radiation therapy (SBRT) has emerged as a promising approach for oligometastatic PCa recurrence. This strategy might have the potential to defer disease progression, thereby lengthening the time till initiation of hormonal therapy, and thus improving quality of life.

Results from a recent prospective phase II study (STOMP trial) suggest that metastasis-directed therapy (surgery or SBRT) of up to three extracranial metastases (any N1 or M1) benefited patients by postponing the start of ADT when compared to surveillance (median ADT-free survival of 21 months and 13 months, respectively) [1]. In a pilot phase I/II study (POPSTAR trial), SBRT delivered to 1–3 metastatic lesions (bone or LN) resulted in a 2-year ADT-free survival rate of 48% among 22 patients who did not have ADT at the time of SBRT [2]. Results from the promising ORIOLE phase II trial (52 patients randomized 2: 1 to SBRT or observation, combined with exhaustive translational research) are not expected before 2022 [3].

Retrospective series published are quite heterogeneous regarding the location of treated metastases (including various amounts of bone, nodal or visceral locations), the irradiated volume for patients with nodal disease (involved-nodes only, or associated to a prophylactic irradiation of the pelvic lymph drainage), and the initial radical PCa treatment (mixing radical prostatectomy (RP) ± radiotherapy (RT) with “RT only” or brachytherapy cases) [4].

The aim of this single-institution study was to investigate a homogenous population of exclusively ADT-naïve prostate cancer patients following RP ± salvage RT, presenting with one to three metachronous LN metastases on 18F-Choline PET/CT (CholPET/CT) and treated by SBRT. Endpoints of interest were biochemical response (BR), time to biochemical recurrence (TBR), and time interval between SBRT and ADT start (TADT).

To the best of our knowledge, this is one of the largest retrospective series on LN recurrence treated with ablative SBRT exclusively in ADT-naïve prostate cancer patients. A biochemical response was observed in 52% of patients. After a median follow-up of 18 months, 68% patients were free of ADT, and only one of 36 treated LN (2.8%) required a second SBRT for in-field progression.

Jereczek-Fossa et al. recently reported on a cohort of 94 patients treated with salvage SBRT with a similar follow-up of 18.5 months [13]; initial treatment of prostate cancer was RP (42%), RP + RT (36%), RT alone (20%) and brachytherapy (2%), with or without ADT. In our series, PSA response, stabilization and progression after SBRT were observed in 68, 8 and 24% of cases, respectively. Our results replicate the 67, 16 and 18% reported by the colleagues from Milan, using similar biochemical response criteria. During the study period, ADT was started in 32% of our patients and in 36% in Milan [13]. Interestingly, for those receiving ADT after SBRT during the study period, treatment could be deferred by at least 1 year in 63% of our cases but only in 38% by the Milan group. Potential explanations for a later ADT start in our institution might include differences in ADT prescription practices or in tumour biology (8% M1a LN against 39.5% in Milan), but might just reflect higher median dose prescriptions (3 × 12 Gy in our center versus 3 × 8 Gy in Milan), corresponding to a twice higher biological equivalent dose (BED) with an α/β = 2 [14, 15] for prostate cancer (BED₂ of 252 Gy versus 120 Gy). The latter BED might not be ablative for all LN metastases, and would provide a reasonable explanation for the 10% rate of in-field progression observed by Jereczek-Fossa et al.

In prostate cancer, conventional imaging has very poor accuracy for identifying LN metastases. In a meta-analysis including 24 studies, sensitivity was 42% for CT and 39% for MRI, specificity 82% for both [16]. Integration of CholPET/CT into routine evaluation of PCa patients improved both staging accuracy and RT planning [17, 18]. The detection rate of CholPET/CT is higher in cases of high PSA velocity (> 5 μg/l/yr) and short PSA doubling time (< 2 mo or 3 mo). Detection rate falls to < 30% for PSA levels < 1 μg/l, but rises > 50% for PSA levels > 2 μg/l [19]. Kitajima et al. reported a 75% detection rate for a PSA level between 4 and 10 μg/l [20].

Prostate-specific membrane antigen (PSMA) PET was not available at the time of our study. PSMA PET outperforms CholPET/CT for the localization of disease sites at biochemical recurrence, particularly in small lymph nodes and at lower PSA levels (< 1 μg/l). [21]. Many PSMA PET positive lesions present actually without an anatomic correlation on CT or MRI, indicating that focal treatment can be initiated at an early stage of metastatic disease [21].

It remains unclear if PSMA PET instead of CholPET/CT guided salvage treatment would have had a major impact on our results. PSA levels before SBRT were quite high in our series (median PSA: 3.53 μg/l); it is therefore likely that most positive LNs were correctly identified on CholPET/CT.

This elevated PSA before SBRT rises also the issue of delayed referral to CholPET/CT, with the advantage to better select patients with true oligo-recurrent disease for SBRT, but at the cost of exposing other patients to a higher risk of delayed treatment for more advanced progressive disease. This selection bias appears to discriminate elderly patients, as there are only two octogenerians in our series (8%). We hypothesize that 80+ year patients are mostly followed up with PSA and CT scans, and receive ADT instead of a metastasis-directed therapy at the time of nodal progression.

We found that salvage SBRT for isolated LN recurrent PCa provides a twice longer biochemical control (21.8 months vs. 10.8 months) when delivered to patients presenting with larger LN metastases (≥ 14 mm). On the other hand, the common clinico-pathological risk factors (T3, Gleason 8–10, PSA doubling time < 3 months) didn’t lead to a shorter TBR (Additional file 1: Table S2). At first glance, such results appear counterintuitive, as one would expect larger LN to reflect a higher tumor burden and a more advanced stage of disease, leading to a shorter TBR. These patients with 1–3 LN macrometastases, despite high-risk features at the time of RP, exhibit a phenotype of slow lymphatic spreading, with only a few LN metastases growing up over time without developing multiple new lesions (true oligometastatic disease). Given the retrospective nature of the study design, we hypothesize that our data result from a selection bias, selecting tumors that didn’t develop yet multiple mechanisms of immune evasion. Identifying the underlying gene expression pattern in such patients would allow in the future for a better selection of candidates to metastasis-directed SBRT, and for an earlier start of treatment (PSA < 1.0 μg/l and LN size < 1 cm). MicroRNA profiling shows substantial potential to be used as prognostic biomarker for individuals with oligometastatic prostate cancer [22‐25].

The main strength of this study relies on quite restrictive inclusion criteria, leading to a series of 25 consecutive patients with a limited nodal recurrence (1–3 LN) after RP, in the absence of any prior or concomitant ADT. Regarding treatment, SBRT was always delivered at ablative doses (median dose: 36 Gy/ 3 fractions [range: 30–45 Gy]).

Our series has several limitations. The number of included patients was low, limiting the validity of subgroup analysis; observed results could have occurred merely by chance. Imaging and PSA levels during follow-up after SBRT were not obtained at predefined time points. We didn’t stage our patients with PSMA-PET, which has been shown in multiple studies to be superior to CholPET/CT imaging in the setting of biochemical relapse [26]. Further drawbacks include the lack of confirmatory biopsy for PET-positive LN, the absence of radiological response assessment after SBRT, and the lack of a control group. Time to ADT might also have been subscriber biased due to the lack of predefined prescription criteria. Finally, our results might not apply to patients with nodal metastases occurring with concomitant local recurrence (prostate bed) and/or distant (bone) metastases, nor to patients with castrate-resistant prostate cancer presenting with oligo-progressive disease.

Last, for patients experiencing isolated metastatic LN recurrence after primary treatment, there are currently no consensus guidelines on which strategy to follow. Besides treating only positive LN with SBRT, which could be considered in one sense as “just targeting the tip of the iceberg”, more comprehensive strategies might be applied as well, like salvage radiation of the whole pelvic lymph node basin combined with a simultaneous integrated boost (SIB) to PET-positive LNs, using either CholPET/CT [27] or PSMA PET/CT guided RT [28]. Extended salvage lymph node dissection (sLND) might be a valid option for selected patients, but remains technically challenging: sLND alone is usually not associated with a durable biochemical response; 50–80% of patients will receive some kind of adjuvant treatment right after sLND (ADT or RT) regardless post-operative PSA levels [29]. In an interesting retrospective analysis of 93 patients, adjuvant RT after sLND was shown to improve significantly the 5 year- relapse-free survival from 15.4% in the surgery only group to 34.3% in the group with additional RT (P = 0.0122) [30]. Prospective randomized trials are warranted to confirm these encouraging results.

The present work underlines the possible role of ablative SBRT in the management of patients affected by LN oligo-recurrent disease. In ADT-naïve prostate cancer patients with 1–3 LN metastases after definitive RP, SBRT can offer a good control of asymptomatic disease. After a median follow-up of 18 months, 68% of patients remained free from ADT. Our study is the first to date attempting to define an optimal LN size threshold prior to SBRT. Additional studies are required before the results should be used in usual clinical settings.