Surgical navigation for targeted retroperitoneal lymph-node removal: a randomised, controlled, phase 3 trial

Abstract

Background: Metastatic retroperitoneal lymph node dissection (LND) for nodal recurrence is applied for a variety of cancers, such as urological, gynaecological and rectal cancer. Precise localisation and resection of these lymph nodes (LNs) during surgery can be challenging, especially after previous radiotherapy or surgery. The objective of this study was to assess the added value of surgical navigation for targeted LND in the retroperitoneum.

Methods: We performed an open-label randomised, controlled, phase 3 trial at the Netherlands Cancer Institute, Amsterdam. Eligible participants were over 18 years of age, scheduled for targeted retroperitoneal LND by laparotomy, with removal of one or more suspected (targeted) LN(s) as assessed by diagnostic imaging. Patients were randomised (1:1) between conventional LND and LND using surgical navigation, by means of a minimisation method stratified for tumour origin (urological, colorectal and other). For the surgical navigation, a digital 3D model of the patients' anatomy was created from diagnostic CT scans, including delineation of the targeted LN(s). The 3D model was linked to the patients' position in the operation room. Using an electromagnetic tracking system, with a sterile tracked pointer, the actual position of the pointer was shown in the 3D model, enabling the surgeon to localize the targeted LN(s). The primary outcome of the study was the percentage of successful procedures. Success was defined as no residual target LN(s) visible on postoperative CT imaging. This study was registered with ClinicalTrials.gov, NCT05867095.

Findings: From January 2017 to December 2020, 69 participants were included in the study, 35 (51%) in the conventional arm and 34 (49%) in the navigation arm. Four patients were not evaluable and excluded from further analysis; three in the conventional arm (patients withdraw from study participation), one in the navigation arm (discontinued surgery, misclassified diagnosis). According to intention-to-treat analysis, 50% (16/32) of the surgical procedures was successful in the conventional arm, versus 85% (28/33) in the surgical navigation arm (one-tailed p = 0.0028, 90% CI: 14%-56%). Using the Clavien-Dindo classification, the overall complication rate was comparable between the conventional arm and the navigation arm. Surgeons judged the surgical navigation setup as valuable, the median preference score to use surgical navigation was 3.7 (3.3-4.0) (scale 1-5), and the median system usability score was 75 (70-85) (scale 0-100).

Interpretation: Surgical navigation allows for significantly better localisation and removal of target LN(s) in the retroperitoneum.

Funding: This research was supported by the KWF-Alpe d'HuZes (NKI 2014-6596) and by an institutional grant of The Dutch Cancer Society and of the Dutch Ministry of Health, Welfare and Sport.

Keywords: 3D model for surgical localisation; Electromagnetic tracking; Retroperitoneal lymph node dissection; Surgical navigation.

Fig. 1

Consort flow diagram, with n = number of patients and LN = number of eligible target lymph nodes. Note that one patient can have one or more target LN(s). The intention-to-treat analysis includes all consented patients undergoing surgery for the correct diagnosis, the per-protocol analysis excluded clinical deviations, i.e. technical failures and non-resections of the target LN(s), as decided by intention during surgery.

Fig. 2

Schematic overview of live surgical navigation. Left: The surgical perspective during navigation, locating the target LN using the tracked pointer (black). Middle: OR screen displaying preoperative contrast enhanced CT scan with the location of the tracked pointer (white cross), and the patient's digital 3D model with the pointer (black). The 3D model is color-coded as follows; bone (yellow-white), urinary track (yellow), veins (blue), arteries (red) and target LN (green). Right: Close-up visualization of the 3D model and tracked pointer.

Fig. 3

Left: Intention-to-treat analysis showing a significant difference (one-tailed p = 0.0028) in success between the conventional arm (50%, 16/32) and the navigation arm (85%, 28/33), in which success is defined per surgical procedure as removal of all target LN(s) in the patient. Right: The probability of successful LN removal of individual target LN(s) as predicted by the generalized linear mixed model, which was 68% for the conventional arm and 93% for the navigation arm (p = 0.0019).

Fig. 4

Per-protocol analysis excluding clinical deviations, i.e. technical failures and non-resections (n = 5) of the target lymph node (LN) as decided by intention during surgery. Left: Significant difference (one-tailed p < 0.0001) in success between of the conventional arm (52%, 16/31) and the navigation arm (97%, 30/31), in which success is defined per surgical procedure as removal of all target LN in the patient. Right: The probability of successful LN removal of the individual target LN(s) as predicted by the generalized linear mixed model, which was 69% for the conventional arm and 99% for the navigation arm (p = 0.0014).

Fig. 5

Compiled visualization of the per-protocol analysis of all successful removed LN (yellow) and not removed LN (red) in the conventional arm (left) and the navigation arm (right).

Fig. 6

Surgeons' perspective on using surgical navigation (n = 30) expressed in a median (IQR) preference score to use the technique (Likert scale) and usability of the system (SUS).