Sentinel Node Procedure to Select Clinically Localized Prostate Cancer Patients with Occult Nodal Metastases for Whole Pelvis Radiotherapy

Take Home Message In clinically localized high-risk prostate cancer patients, sentinel lymph node biopsy–based selection of men with occult nodal metastases for whole pelvis radiotherapy is associated with favorable oncological outcomes as compared with imaging-based prostate-only radiotherapy.


Introduction
Presence of lymph node (LN) metastases is considered an important prognostic factor in prostate cancer (PCa), as these are associated with a higher likelihood of disease progression and dissemination [1]. Currently, the European Association of Urology guidelines recommend prostateonly radiotherapy (PORT) combined with androgen deprivation therapy (ADT) for clinically node-negative (cN0) PCa patients, regardless of the estimated risk of nodal metastases [2]. Whole pelvis radiotherapy (WPRT) has shown encouraging results in pathologically node-positive PCa patients [3,4]. Recently, also prophylactic WPRT was proved to provide a significant survival benefit over PORT in cN0, high-risk PCa patients [5]. However, WPRT comes with increased toxicity [6][7][8]. Accurate identification of men who in fact harbor nodal metastases is necessary to select patients who are likely to benefit from WPRT. Conventional imaging (ie, computed tomography [CT] or magnetic resonance imaging [MRI]) techniques have insufficient sensitivity to detect nodal (micro)metastases [9]. However, novel molecular imaging approaches (ie, prostate-specific membrane antigen [PSMA] positron emission tomography [PET]/CT) also fail to detect nodal metastases <3 mm [10,11].
Extended pelvic lymph node dissection (ePLND)-the gold standard for nodal staging in clinically localized PCa-has been used as a staging tool for WPRT [3,4]. However, ePLND has been associated with increased morbidity, and its template does not include aberrant lymphatic draining sites of the prostate [12,13]. The ability to identify the location of nodal metastases based on the lymphatic drainage of the primary tumor has led to the exploration of the sentinel lymph node biopsy (SLNB). In PCa surgery, SLNB-directed dissections have yielded a diagnostic accuracy comparable with that of ePLND, but with lower complication rates [14,15]. Critically, SLNB helps identify aberrant drainage outside the standard ePLND template, which is seen in up to a third of the prostatic sentinel nodes (SNs) [13,16]. The objective of this study was to evaluate whether SN sampling in cN0 patients with an increased risk of nodal metastases, followed by selection of pN1 patients for WPRT, improved the oncological outcomes as compared with conventional imaging-based PORT.

Study design and patient population
This retrospective cohort study included cN0 PCa patients with a >5% Briganti et al [17]

Follow-up and outcomes
Prostate-specific antigen (PSA) levels were evaluated every 4 mo during the first 3 yr after radiotherapy and twice a year thereafter. Biochemical recurrence (BCR) was defined as a PSA nadir plus 2 ng/ml in accordance with the Phoenix definition [20]. In case of a BCR or symptomatic dis-

Statistical analysis
To compare continuous variables between treatment groups (ie, SLNB and non-SLNB groups), an unpaired T test or a Mann-Whitney U nonparametric test was used. A chi-square test or a Fisher's exact test was performed to compare discrete variables. Since patients were not randomly assigned to both treatment groups, we performed a propensity score analysis based on the inverse probability of treatment weighting (IPTW) [21]. Propensity scores were generated using a multivariable logistic regression adjusting for the following variables: age, cT stage,

Results
A total of 528 eligible patients were retrospectively included ( Fig. 1

Complications and toxicity
The 90-d complication rates of the SLNB procedure are reported in Table 3. High-grade complications (Clavien-Dindo3) occurred in 11 patients (4.2%). No grade 4 complication was observed.
Overall, 64 (12.1%) and 159 (30.1%) patients experienced grade 2 or 3 gastrointestinal (GI) or genitourinary (GU) toxicity, respectively (Table 4). No grade 4 or 5 toxicities were observed. Patients receiving WPRT had significantly higher GI and GU toxicity rates than those receiving PORT. Compared with the non-SLNB group, more patients in the SNdirected PORT arm experienced mild-to-moderate GI (18.3% vs 1.9%) and GU (39.8% vs 16.6%) toxicities, but no statistically significant difference was observed in the overall grade 3 toxicity.

3.5.
Patterns of radiological recurrence Patterns of radiological recurrence are demonstrated in Supplementary  in 18 patients [41.9%]), followed by regional LNMs (39 patients, 7.4%) and bone metastases (31 patients, 5.9%). A regional LNM was observed in ten patients (3.8%) in the SLNB group (two patients treated with WPRT and eight patients treated with PORT) and 29 patients (10.9%) in the non-SLNB group. The regional LN recurrences in two patients treated with WPRT occurred outside the WPRT field (ie, outside the region that received a high or an elective dose).

Discussion
To our knowledge, this is the first PSW study that demonstrates favorable oncological outcomes for SLNB-based selection of histologically node-positive PCa patients for WPRT. These findings were compared with (conventional) imaging-directed PORT in cN0 patients with an increased risk of nodal metastases. When corrected for baseline characteristics, SLNB-guided radiotherapy was associated with improved BCRFS and RRFS. These improved oncological outcomes could be attributed to the additional pelvic irradiation and the longer course of ADT for pN1 patients. It can be assumed that patients in the non-SLNB group with clinically occult nodal metastases would also benefit from a similar therapeutic approach. The difference in 5-and 7-yr survival in the PORT group may be explained by the testosterone recovery period after ADT. A testosterone recovery period of >2 yr after hormone therapy has been reported [22], and a longer duration of hormone therapy and older age are significantly associated with a prolonged recovery interval [23]. It is plausible that testosterone levels were still recovering 5 yr after the start of radiotherapy, resulting in a later onset of recurrences.  Our initial efforts to document the oncological outcomes of SLNB-guided radiotherapy were the results of a singlearm study that compared the outcomes of SLNB-guided radiotherapy versus Kattan nomogram-predicted BCR rates [24]. The results in the present study come from an extended SLNB cohort (including patients up until 2018 instead of 2016) with longer follow-up (71 vs 52 mo). Here, we found an adjusted 5-yr BCRFS rate of 81.9% in 85 pN1 patients who received WPRT. Previous literature on WPRT in pN1 patients using ePLND as a staging tool has shown a 5-yr BCRFS rate of 65-67% [3,4]. Hence, application of WPRT in pN1 patients staged using SLNB procedures provides favorable results, while omitting the morbidity associated with ePLND [5].
Randomized trials (ie, RTOG 9413 and GETUG-01) on prophylactic elective WPRT failed to show a survival benefit for WPRT compared with PORT [25,26]. However, the more recent POP-RT trial including (very) high-risk cN0 PCa patients has shown BCRFS and disease-free survival benefits of WPRT combined with long-term use of ADT [5]. As such, it could be that the inclusion of patients with a low risk of nodal metastases, short ADT duration, and relatively low radiation doses has diluted the benefit of WPRT in earlier trials. The unadjusted survival rates of our non-SLNB cohort are comparable with the outcomes of the PORT group in the POP-RT trial. However, the fact that the 5-yr BCRFS (95%) of the WPRT group in the POP-RT study far exceeded the outcomes of WPRT (82%), not only in our population, but also in WPRT populations in the literature (65-67%) [3,4], can be explained by three facts. First, lifetime androgen deprivation was achieved by surgical castration in 14.5% of WPRT patients in the POP-RT trial. Second, PSMA PET staging was performed in 80% of the POP-RT patients. Third, the WPRT population in the POP-RT trial included only cN0 patients, whereas the WPRT patients in our and the aforementioned studies included only pN1 patients.
Our grade 2 toxicity rates are higher than those reported elsewhere [6][7][8]. The use of different toxicity grading systems and variability in the documentation, interpretation, and scoring of toxicity may explain this difference. The POP-RT study reports significantly higher late GU toxicity (grade 2) for elective WPRT (17.7%) than PORT (7.5%, p = 0.03) [5]. Higher GI or GU toxicity rates after WPRT in both our study and previous literature stress the importance of adequate patient selection [6][7][8]. In that sense, our approach helped select patients with a pathologically negative SN (63%) for PORT as treatment rather than WPRT. Although mild-to-moderate GI and GU toxicity was higher in the SLNB-directed PORT arm than in the non-SLNB group-surgery in the small pelvis may contribute to increased toxicity, overall high-grade toxicity did not differ between the two arms. We believe that the high-grade com- plication rate of 4% of the SLNB procedure justifies its use to select only pN1 patients for WPRT, and to avoid pelvic irradiation and its toxicity in pN0 patients.
An important observation was that in the current study, none of the patients treated with WPRT had an LN recurrence inside the radiotherapy field. In line with previous studies, this suggests that WPRT is effective in preventing in-field nodal recurrences [4,27]. Moreover, regional LN recurrence rates were lower in the non-SLNB group than in the SLNB group (4.5% vs 10.9%). It is plausible that a subset of non-SLNB patients had clinically occult pelvic nodal metastases at primary staging that were detected during follow-up, resulting in a higher rate of regional nodal recurrences.
In our cohort, the surgical SLNB procedure had an overall 90-d complication rate of 21.8%. This is markedly lower than the up to 51% complication rate reported for ePLND [12], but higher than the 9% complication rate reported previously for SLNB procedures [24]. However, when comparing the rate of severe complications (Clavien-Dindo grade 3), the rate of 5% reported previously for SLNB procedures [24] is comparable with the rate of 4.2% observed in our cohort.
One of the main limitations of our study is the bias inherent to its retrospective design in addition to the missing data and loss of follow-up, since many patients went back to their referring hospital following treatment. In addition, all patients were treated in two hospitals that share a radiotherapy facility, which might have introduced a center effect bias.
Since imaging to detect radiological recurrences was performed as part of routine clinical care, the scans may have been performed at different time points, resulting in inconsistent imaging intervals. We attempted to control for bias with PSW analyses. Unfortunately, PSW analyses cannot exclude a selection bias or bias by unknown variables, and hence cannot replace a randomized controlled trial. The use of different tracers for the SLNB procedure introduced heterogeneity into our study. However, the pN1 rate did not differ significantly between patients who received SLNB with Tc-nanocolloid and those who received it with ICG-Tc-nanocolloid (36.2% vs 37.8%, p = 0.897). Some patients in our cohort were also included in (ongoing) radiotherapy trials and therefore received nonstandard (hypofractionated) radiotherapy dosing schemes (n = 60; 11.4%). For various reasons, 12 out 97 (12.4%) pN1 patients received PORT instead of WPRT. It should be noted that higher BCR and radiological recurrence rates in pN1 patients treated with PORT than in those treated with WPRT suggest a benefit from WPRT in this population. In addition, the diagnostic value of SLNB in PCa has yet to be validated in a randomized trial. High diagnostic accuracy is necessary to distinguish between pN0 and pN1 patients. The high sensitivity and negative predictive value of SLNB [15], combined with the low rate of regional nodal recurrences (4.5%) in our SLNB pN0 population, suggest that the majority of pN0 SLNB patients were truly node negative. Assuming that scoring more severe toxicities would overcome the limita-  tions of retrospective data collection, we scored only grade 2 toxicities. Lastly, given the period of inclusion, the majority of the patients were staged using conventional imaging methods and only 40 patients (7.5%) were staged primarily using PSMA PET. Of these patients, 31 underwent SLNB and 12 (38.7%) had pN1 disease. As the median metastasis size of 2 mm in our cohort lies below the reported 3 mm detection limit of PSMA PET and PSMA-based intraoperative radioguid-ance techniques [28][29][30], it is questionable whether the use of PSMA PET or intraoperative PSMA radioguidance would have impacted the nodal detection rate and treatment allocation given the low sensitivity of PSMA PET for nodal metastases <3 mm [10,11]. The diagnostic value of the SLNB procedure in PCa patients with localized disease on PSMA PET/CT will be evaluated in a future study.

Conclusions
In conclusion, SLNB-based selection of pN1 patients for WPRT is associated with favorable oncological outcomes as compared with imaging-based PORT in cN0 PCa patients. The safety profile of this treatment option is acceptable with a low rate of high-grade complications. By applying SLNB procedures as a means to select pN1 patients, the use of WPRT could be limited to patients that actually benefitted from the procedure. The lack of clear guideline recommendations on the use of WPRT in primary PCa [2] and the overall promising results from this study can be valid arguments for a randomized controlled trial comparing SLNB-guided radiotherapy versus imaging-based radiotherapy.
Author contributions: Hilda A. de Barros had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: de Barros, P.J. van