Robotic retroperitoneal lymph node dissection for testicular cancer at a national referral centre

Abstract Objectives We aim to determine if robot‐assisted retroperitoneal lymph node dissection (R‐RPLND) can be performed as a safe option to open RPLND in selected patients with metastatic germ cell cancer. Patients and methods This population‐based prospective study was performed at a one of two national referral centres for RPLND in Sweden. All patients referred during January 2017–March 2021 were screened for possible inclusion. R‐RPLND was performed using the Da Vinci Xi surgical system. Perioperative parameters, postoperative complications (Clavien–Dindo), final pathology, preservation of antegrade ejaculation and relapse rates were evaluated. Classifiers for selecting patients to open versus robotic RPLND were analysed by logistic regression modelling. The median follow‐up was 23 months. Results Of 87 patients referred, 29 were selected for R‐RPLND, 19 in a post‐chemotherapy setting. In median, retroperitoneal tumour diameter was 18 mm, BMI 24 kg/m2, operative time 433 min, estimated blood loss 50 ml and length of stay 3 days. One patient underwent open conversion due to failure to progress. Four patients had Clavien–Dindo grade 3 complications, of which three were chylous‐related. No in‐field recurrences occurred during follow‐up. Conclusion This population‐based study suggests that R‐RPLND can be safely performed in at least one third of patients referred for an RPLND. A relatively high rate of lymph‐leakage may represent a potential drawback. Tumour size may be the most important discriminator when deciding on robotic versus open RPLND. Further studies with longer follow‐up are needed to validate the results.


| INTRODUCTION
As a part of the National Cancer Strategy in Sweden, several processes have been on-going during the last decade to improve the quality of cancer care and to provide more equal access to advanced treatments. 1 In 2013, the Swedish Government and the Swedish Association of Local Authorities and Regions agreed on a plan to promote a national concentration of 10 highly specialized cancer treatments. One of the 10 identified treatments was RPLND in GCT patients. Following an application procedure, two centres were growth that require an RPLND. In addition, primary RPLND in lowvolume metastatic seminoumatous GCT (SGCT) is currently being evaluated in two trials. 4,5 A concentration of RPLND cases to SUH brought opportunities for research and technology development. At the time, only a limited number of small R-RPLND case series had been published. However, the short-term results were promising, and R-RPLND was considered feasible in selected patients. [6][7][8] SUH has a long experience of robotassisted surgery including 400 urologic robotic procedures yearly.
After ethical approval, we started a prospective R-RPLND study in 2017. The rationale was to prospectively explore if R-RPLND could be implemented safely without compromising the oncologic efficacy.
The primary aim was to evaluate perioperative and postoperative outcome of R-RPLND including complications and relapse rates. The secondary aim was to provide guidance on how to select patients to R-RPLND.

| Overview
All Swedish GCT patients considered for an RPLND are discussed in a weekly national MDT conference where urologists, oncologists, radiologists and pathologists are represented. Initial (pre-chemotherapy) imaging and post-chemotherapy imaging are presented. When needed, testicular or nodal biopsy histology are demonstrated by the attending pathologist. After decision to proceed with an RPLND, patients are referred to one of the two NRCs. In the present study, we screened all patients referred to SUH from 1 January 2017 until 1 March 2021 for possible inclusion in the R-RPLND study. Patients with limited nodal disease (<50 mm diameter) and no suspicion of tumour-infiltration of major vessels were considered for robotic surgery. Potential risks and benefits were discussed with the patient, and written informed consent was obtained. To be able to analyse selection mechanisms, comparisons with the O-RPLND patients referred during the same time period were performed. The study was approved by the regional ethics committee (Dnr 418-17 2017-06-29).

| Setting
According to the SWENOTECA treatment-protocol, a postchemotherapy RPLND (PC-RPLND) is indicated in NSGCT patients with a residual mass measuring ≥10 mm in largest transverse diameter after completion of chemotherapy. Patients with late marker-negative relapses (>2 years after initial successful treatment for metastatic disease) are also candidates for RPLND. A unilateral template restricted to the primary landing zone of the affected testicle is an option for tumours measuring 10-49 mm, whereas tumours ≥50 mm should prompt bilateral templates. In addition, location of enlarged nodes must be considered before deciding template to make sure that all areas with enlarged nodes on pre-chemotherapy and postchemotherapy radiology are excised. A primary (without inductionchemotherapy) RPLND (P-RPLND) is recommended in patients with: we changed to a 20-degree supine Trendelenburg position, with four 8-mm robot-ports placed in a linear configuration infra-umbilically and one 12-mm assistant port as described by others. 7,11 No re-docking was needed in any of the cases. Clavien-Dindo classification system, (complication grade 3a or higher considered 'major'). All patients received continuous oncological follow-up according to the SWENOTECA cancer care programme 9 and were monitored using the national quality register (NQR) for testicular cancer. 13

| Secondary outcomes
No explicit predefined criteria were used to select patients to robotic/ open RPLND other that the general principle that high volume residual disease probably is more suitable for open surgery whereas small residuals without vascular infiltration can be suited for robotics. The plan was to explore the selection mechanism in retrospect to provide guidance for future use.

| Analysis
Descriptive data was presented as frequencies and median (inter quartile range, IQR). Follow-up time was calculated from R-RPLND date until the last clinical follow-up date according to the medical charts and NQR. OT, EBL and LOS were compared between different template R-RPLNDs, and between O-RPLND and R-RPLND using the Mann-Whitney U-test. The exploration of selection mechanisms was done by means of logistic regression models of the type of surgery actually carried out, using retroperitoneal tumour diameter, risk-group, induction-chemotherapy (yes/no), tumour histology, ASA, BMI (<30 or ≥30), and age as explanatory variables. Area under the ROC (receiver operating characteristics) curves (AUC) was calculated, models compared by means of likelihood ratio tests, and cut-offs explored. No adjustment for multiple comparisons was performed.

| RESULTS
Of 87 patients admitted for an RPLND at SUH during the study period, 29(33%) were selected for robotic surgery. R-RPLND patients had a more favourable disease and smaller retroperitoneal tumours both pre-chemotherapy and pre-RPLND than O-RPLND patients (good risk in 93% vs. 59%; median residual tumour size 18 mm vs. 28 mm). There were more SGCT patients among the R-RPLND cases, whereas almost all O-RPLND patients had NSGCT or advanced extragonadal tumours. All patients had abdominal nodal involvement with at least one node ≥10 mm at time of surgery (CS ≥IIA) ( Table 1).
Of the 29 robot-cases, two with SGCT were scheduled for a lumpectomy only and therefore excluded from the separate analysis of the R-RPLND cases. The total R-RPLND study population comprised 27 patients undergoing a unilateral (n = 23, 85%), or a full bilateral (n = 4, 15%) template resection. Clinical details are presented in Table 2. Regarding primary tumour pathology, 19 patients had NSGCT, four had SGCT, two had teratoma and one had malignant transformation to adenocarcinoma. One patient had a primary extragonadal biopsy-verified SGCT and no testicular tumour.
The most common surgical indication among the R-RPLND patients was a residual mass after completion of chemotherapy for NSGCT (n = 18, 67%). Of the remaining patients, five had a late relapse after median (IQR) 6.3 years (5.6-6.9) following initial chemotherapy (n = 3, 11%) or initial surveillance (n = 2, 7.4%). Four CS IIA patients underwent a P-RPLND; two due to marker-negative SGCT, one due to teratoma only in the testicle and a growing paraaortic cystic mass, and one due to malignant transformation of teratoma in the testicle and an enlarged aortocaval node. Both recovered quickly afterwards. There were no Clavien-Dindo grade 4 or 5 complications among the R-RPLND cases ( Table 2).

| Pathological, oncological and functional outcome in R-RPLND
Pathology of the resected specimen are presented in Table 3. Positive nodes were detected in 17 cases (63%); 12 had teratoma (44%) and   c Referrals from outside the regional catchment-area. d Intra-regional referrals. e Abdominal stage at time of staging following diagnosis or recurrence.  Benign/necrosis 1 9 10

| Selection of patients
Pathological N stage It is well-known that primary resections have a lower complication rate than post-chemotherapy resections. The latter are more complex due to the chemotherapy-induced desmoplastic reaction. In this study of mainly small volume disease in the post-chemotherapy setting, we observed four major complications. Three were in postchemotherapy patients and chylous-related, of which two were surgically managed. This incidence of chylous ascites was higher than anticipated, although small in number (11% vs. 6 role. 17 In our study, we used Hem-o-Lok clip ligation meticulously.
Yet, when the two chyle-leak cases were explored, a diffuse leak from a tangle of clips near the renal hilum was observed. Whether the clips prevented the tissue from adhering, or whether major lymphatics had been left unsealed remains unknown. Percutaneous lymphatic embolization has become increasingly used to treat postoperative lymphleakage. [18][19][20] To date, those facilities are not readily available at our institution.
Apart from chylous ascites, another concern that has been raised regarding R-RPLND has been unusual patterns of disease recurrence. 21 We noted two out-of-field recurrences, although not in unusual places. Hence, we have no reason to believe that this was inherent to the robotic technique but longer follow-up is needed to monitor areas of potential recurrences.
The main advantages with minimally invasive RPLND are perhaps the short LOS and the low EBL, and possibly also the preservation of antegrade ejaculation. In a recent review of R-RPLND including eight series with >10 patients in each, the reported weighted means regarding EBL was 132 ml, and LOS 2 days. 14 This is in comparison with our results (median EBL 118 ml and LOS 3 days), given the differences in patient characteristics among open and robotic cases. Previous reports on preserved ejaculatory function vary, from 85% to 100% of cases, 6,16,17,22,23 to somewhat lower rates, 67%-81%. 24,25 We were unable to assess ejaculatory function in 11 patients, but 80% of the remaining R-RPLND patients reported antegrade function.
While striving towards decreasing overtreatment and reducing therapy-related side effects, it is important to recall that we lack diagnostic tools to tell whether a residual mass contains cancer, teratoma or fibrosis. We know from large O-RPLND studies of postchemotherapy NSGCT patients that the rate of teratoma in resected specimen is 40%, and viable cancer is 11%-17%. 26 It is important to recall that we selected small-volume infrahilar disease to R-RPLND, as opposed to clearly unsuitable cases with large-volume disease growing diffusely close to great vessels, renal hilum, bowel, and vertebrae. To further evaluate the selection of patients to either approach, and to possibly suggest a future selection model for clinical use, R-RPLND should be restricted to high volume centres with expertise in open RPLND and robotic surgery so that more data can be prospectively collected.
The strength of this study was the prospective design and the reasonably large patient volumes that comes with a national commission. The drawback of concentrating patients to NRCs is that it might be difficult to follow patients over time. It is possible that complications were caught to a higher degree in patients residing in the Gothenburg area. However, all patients are presented again at the national MDT conference when the pathology report is available, and this MDT serves as an additional follow-up of post-discharge complications. The NQR's high coverage also adds to our belief that all significant surgical complications and recurrences were detected. Another weakness was the lack of validated assessment tools for patient reported outcome measures including evaluation of ejaculatory function, especially with respect to the distinction between a nervesparing procedure and a unilateral template resection.

| CONCLUSION
According to these population-based results, at least one third of all RPLNDs in GCT patients may be performed as a robotic procedure at a high-volume centre without jeopardizing oncological safety. R-RPLND has the potential to decrease the burden of treatment-related side-effects although a higher than anticipated chyle-leak may be a concern. How to best select patients to open versus robot-assisted RPLND needs further evaluation and analysis with longer follow-up, but our results indicate that retroperitoneal tumour size may be the single most important determinant.

ACKNOWLEDGMENTS
The study was financed by grants from the Swedish State under the agreement between the Swedish government and the county councils, the ALF Agreement (ALFGBG-960563).