Introduction

Testicular germ cell tumors (GCT) occur mostly in young patients, and are highly curable, even in case of metastatic spread [1,2,3]. The management requires multidisciplinary collaboration and high expertise. In case of retroperitoneal lymph node metastasis, upfront cisplatin-based chemotherapy is mostly proposed, with three or four cycles of BEP according to IGCCCG risk group. Subsequently, post-chemotherapy retroperitoneal lymph node dissection (RPLND) is indicated for supracentimetric residual mass (non-seminoma GCT) or more than 3 cm hypermetabolic masses (seminoma GCT).

Robot-assisted surgery has become widely used for radical prostatectomy [4], partial nephrectomy [5], and is increasing also for radical cystectomy [6]. For urologists with extensive experience in robotic surgery, RPLND may represent the next challenge, keeping in mind that no compromise regarding the extent of surgery due to the surgical approach could be accepted [7].

Logically, robot-assisted RPLND started initially for stage I diseases [8], avoiding the potential difficulties of chemotherapy-induced modifications of the tissues. However, most of RPLNDs in Europe are nowadays planned in a post-chemotherapy setting [9], considering the superiority of adjuvant chemotherapy upon primary RPLND [10], the chemosensitivity of seminomas, and the rareness of stage IIA non-seminoma germ cell tumor with negative markers. Post-chemotherapy RPLND remains a demanding surgery, and open RPLND is undoubtedly the standard [3]. Evidently, PC-RARPLND will not replace open PC-RPLND for all cases, as the great extent of the disease frequently precludes a minimally invasive approach [11]. However, in selected cases, PC-RARPLND is proposed in more and more expert centers to improve perioperative outcomes [12].

Oncological concerns have been previously published regarding PC-RARPLND [13], suggesting that pneuperitoneum could alter the natural history of surgical relapses, describing paracolic recurrences or peritoneal carcinomatosis. Nevertheless, recent reports showed favorable oncological outcomes, without unusual recurrences [14,15,16]. Moreover, surgical safety is an important aspect to be questioned, since operating close to the large vessels could be associated with major bleeding complications, delicate to control in a robot-assisted laparoscopic setting.

Until now, no prospective trial has been published in this setting, and small retrospective series only started to emerge recently. Therefore, there is an unmet need for a comprehensive and contemporary analysis of a large PC-RARPLND series. Thus, the aim of this study was to describe our multicenter experience in patients managed with R-RPLND regarding perioperative and early oncological outcomes.

Methods

Study population

All patients who underwent post-chemotherapy robotic retroperitoneal lymph node dissection for metastatic germ cell tumor in four tertiary care centers from 2011 to August 2021, were retrospectively included. Patients with previous RPLNDs were excluded. RARPLNDs were classified “post-chemotherapy” when patients classically finished chemotherapy 4–6 weeks before surgery, or when patients had a history of successful first-line chemotherapy for stage II disease, and relapsed thereafter (considering that tissues had been significantly modified by the exposure of multiple cycles of chemotherapy in this setting). All patients were staged with CT scans before and after chemotherapy, and serum tumor markers (α-fetoprotein, human chorionic gonadotropin, and lactate dehydrogenase). Before surgery, all patients were discussed at multidisciplinary meetings including urologists, radiologists, medical oncologists, and radiation therapists. Institutional review board approved the study (ROBOTESTIS-IPC-2021-039). Follow-up was performed according to international guidelines: serum tumor markers every 3 months and CT scans every 6 months during the first 2 years, with reduced frequency thereafter.

Surgical technique

Every procedure was done by experienced robotic surgeons (FB, JCB, GR, ND, JBB, GV, JW). Patient positioning (decubitus dorsal or decubitus lateral) and port placement depended on surgeon’s preference. The extension of the templates was decided by each surgeon and depended on the size and location of the mass(es) in imaging. Postoperative drainage depended on surgeon’s preference.

Outcomes of interest

Preoperative data [histology of tumor at orchiectomy, history of chemotherapy, initial stage of the disease or relapse, location and size of the mass(es)], perioperative (length of surgery, estimated blood loss, intraoperative complications), and postoperative data (complications, pathologic results, relapse) were collected and analyzed. Regarding pathological results, patients were divided into three exclusive categories: necrosis/fibrosis only, teratoma (with or without necrosis/fibrosis, but without viable tumor), and viable tumor (whether concomitant teratoma or not). Postoperative complications were assessed using the Clavien–Dindo classification.

Statistical analysis

Descriptive statistics included frequencies and proportions for categorical variables. Means, medians, and ranges were reported for continuously coded variables. The statistical significance of differences in medians and proportions was evaluated with the Kruskal–Wallis and Chi-square tests. Univariable and multivariable logistic regression models tested the relationship between complications (all complications and severe complications) and several variables, namely age, body mass index (BMI), time between orchiectomy and lymph node dissection, type of preoperative chemotherapy, number of cycles of chemotherapy, type of lymph node dissection, intraoperative blood loss, intraoperative patient positioning, operative time, and histological type. They were included in the multivariable models if significantly associated with the outcome in the univariable analysis.

For all statistical analyses, R software environment for statistical computing and graphics (version 3.4.3) was used. All tests were two sided with a level of significance set at p < 0.05.

Results

Patient characteristics

Overall, 66 patients, with a median age of 34 years, were included (Table 1). The majority of patients had non-seminoma germ cell tumors (59 patients, 89%).

Table 1 Patient characteristics before robot-assisted retroperitoneal lymph node dissection (RARPLND) (66 patients)
Full size table

The majority of patients had lymph node metastasis at the initial diagnosis of the disease (53 patients, 80%) and 13 patients (20%) had retroperitoneal lymph node relapse after initial stage I disease, or after successful first-line chemotherapy for stage II/III disease (metachronous relapse, median time of the relapse: 9 months). The proportions of patients with good, intermediate, and poor IGCCCG prognostic classification were, respectively, 65%, 27%, and 7.5%.

Surgeries and postoperative results

The median size of the largest retroperitoneal mass was 25 mm in great axis; six patients (9.1%) had a retroperitoneal mass of more than 50 mm (Table 1). The location of largest mass was mainly pre or latero aortic (38 patients, 58%).

Median operative time was 200 min [108–248] and estimated blood loss was 50 mL [IQR 50–150] (Table 2). Most patients had a modified template RARPLND (45 patients, 68%). No additional procedures (such as organ resection, or aorta/vena cava resection or reconstruction) were needed.

Regarding intraoperative complications, four patients (6.0%) had a vascular injury (two vena cava, two aorta), only one had an estimated blood loss of more than 500 mL, and no one had blood transfusion. Three patients (4.5%) had a conversion to open surgery: one for active bleeding, and two for difficulty of dissection (one was a post-chemotherapy seminoma, the other one a primary retroperitoneal tumor).

Regarding postoperative complications, seven patients (11%) had a postoperative ileus, five during hospitalization and two after. Five patients (7.6%) had lymphatic complications (symptomatic lymphocele needing paracentesis or drainage, lymphatic flow through trocar orifice); all of these complications occurred after hospitalization. Six patients (9.1%) needed re-hospitalization.

Table 2 Characteristics of surgery and postoperative results (66 patients)
Full size table

Three patients (3.5%) had a IIIa Clavien–Dindo complication (percutaneous drainage of lymphocele), there was no IIIb or more Clavien–Dindo complications.

In univariable analyses predicting intraoperative and early postoperative complications within 30 days (14 events), blood loss (Odds ratio [OR]: 1.01, 95% confidence interval [95% CI] 1.0–1.01, p = 0.02), lateral intraoperative patient positioning toward the right side (OR: 0.2, 95% CI 0.03–0.7, p = 0.03), and operative time (OR: 1.01, 05% CI 1.0–1.01, p = 0.02) reached independent predictor status. None of the other tested variables, namely age, BMI, time between orchiectomy and lymph node dissection, type of preoperative chemotherapy, number of cycles of chemotherapy, type of lymph node dissection, and histological type reached independent predictor status.

In multivariable analyses (including blood loss, patient positioning, operative time) predicting intraoperative and early postoperative complications within 30 days, no factor reached independent predictor status.

In univariable analyses predicting severe (Clavien–Dindo ≥ IIIa) intraoperative and early postoperative complications within 30 days (3 events), no variable reached independent predictor status.

Pathology and early oncologic outcomes

Median number of lymph nodes in pathological reports was 10 [6–17]. Teratoma, necrosis, and viable tumor was found in 53%, 39%, and 7.6% of the cases, respectively.

With a median follow-up of 16 months, two patients had a relapse: one in the retrocrural space associated with reascending serum tumor markers and equivocal lung nodules months after PC-RARPLND, managed with second-line chemotherapy, with complete response; the other patient had multiple lung metastases 5 months after surgery and died 1 month later.

Discussion

Since there is a gap in the literature for a thorough and updated examination of a large PC-RARPLND series, the purpose of our study was to outline our multicenter experience with PC-RARPLND in terms of perioperative and early oncological outcomes. Our study showed that PC-RARPLND for selected cases was a feasible procedure with acceptable complication rates. In addition, no adverse or unusual oncological events occurred.

In the light of our results, PC-RARPLND should still only be performed by highly experienced robotic surgeons knowing that major vessel injury could occur and need to be managed with efficacy and safety [17]. A 6.0% rate of vessel injury was observed, but only one patient had significant bleeding with the need to an open conversion. This underlines the added value of the robotic procedure in selected cases as it provides important advantages such as short length of stay, low estimated blood loss, and small skin incisions.

The majority of patients had a modified template resection. This proportion is a consequence of a stringent selection of favorable patients for the robotic approach. A full bilaterally retroperitoneal lymph node resection is not mandatory for selected patients as described by Heidenreich et al. [18, 19]. The safety of modified template resection was recently validated by Gerdtsson et al. [20].

Our results are in accordance to the recent and growing data regarding PC-RARPLND, which are all retrospective series. Regarding intraoperative and postoperative outcomes, Fankhauser et al. reported a 3% conversion rate to open surgery, and a blood transfusion rate of 3% in the largest cohort so far reported [14]. Regarding oncological outcomes, recurrence rates ranged from 0 to 16.1% [14, 15, 21, 22].

Several limitations of our study need to be mentioned. First, our study is based on a retrospective analysis with all of its inherent limitations leading to selection and data collection bias. We cannot exclude that some adverse events had been missed, especially small complications resulting in outpatient visits if patients had been re-admitted to other facilities not visible to the surgical team. Second, the template of RARPLND was chosen by each surgeon and varied between patients and centers. Additionally, information regarding antegrade ejaculation was not available and follow-up was only intermediate, as centers only recently started to perform this procedure robotically. Lastly, a comparison to a control group treated with open surgery was not available.

Conclusion

In this multicenter study, we saw that for selected cases and in expert hands, PC-RARPLND seemed to be a feasible and reasonably safe procedure, with no adverse oncological results. Prospective evaluation or ideally a comparative randomized trial versus open surgery should be performed to have more consistent data regarding the safety and oncologic efficacy of this procedure.