Retroperitoneal Lymph Node Dissection in Colorectal Cancer with Lymph Node Metastasis: A Systematic Review

Simple Summary Retroperitoneal lymph node metastasis (RPLNM) occurs in up to 6% of colorectal cancer (CRC) patients. In general, there is no consensus on the treatment paradigm or optimal management of retroperitoneal lymph node dissection (RPLND) in CRC patients, necessitating a systematic review of the literature to evaluate preoperative imaging modalities, perioperative chemotherapy and radiotherapy regimens, and oncological outcomes of RPLND in CRC. Nineteen studies of 541 patients were included. Based on this systematic review and analysis, RPLND is a feasible treatment option with limited morbidity and possible oncological benefit for both synchronous and metachronous RPLNM in CRC. Future prospective clinical trials are required in order to establish further evidence for RPLND in the context of RPLNM in CRC. Abstract The benefits and prognosis of RPLND in CRC have not yet been fully established. This systematic review aimed to evaluate the outcomes for CRC patients with RPLNM undergoing RPLND. A literature search of MEDLINE, EMBASE, EMCare, and CINAHL identified studies from between January 1990 and June 2022 that reported data on clinical outcomes for patients who underwent RPLND for RPLNM in CRC. The following primary outcome measures were derived: postoperative morbidity, disease free-survival (DFS), overall survival (OS), and re-recurrence. Nineteen studies with a total of 541 patients were included. Three hundred and sixty-three patients (67.1%) had synchronous RPLNM and 178 patients (32.9%) had metachronous RPLNM. Perioperative chemotherapy was administered in 496 (91.7%) patients. The median DFS was 8.6–38.0 months and 5-year DFS was 24.4% (10.0–60.5%). The median OS was 25.0–83.0 months and 5-year OS was 47.0% (15.0–87.5%). RPLND is a feasible treatment option with limited morbidity and possible oncological benefit for both synchronous and metachronous RPLNM in CRC. Further prospective clinical trials are required to establish a better evidence base for RPLND in the context of RPLNM in CRC and to understand the timing of RPLND in a multimodality pathway in order to optimise treatment outcomes for this group of patients.


Introduction
Colorectal cancer (CRC) is the second most common cause of cancer-related deaths worldwide [1,2]. Approximately 20% of CRC patients present with metastases at initial diagnosis, and nearly 50% of patients develop metastases, contributing to the high mortality relate to: (1) RPLND in patients presenting with synchronous or metachronous RPLNM in CRC; (2) RPLND for isolated RPLNM or combined with other CRC metastatic lesions; (3) pathologically positive or negative RPLNM; or (4) reporting of either postoperative morbidity or oncological outcomes.
Studies were excluded if they were: (1) articles published in a non-English language or in a book; (2) letters to the editor, case reports, or conference abstracts; (3) lacking relevant morbidity or oncological outcomes.

Data Extraction
The titles and abstracts were assessed by two independent authors (M.G.F. and M.A.) against the inclusion and exclusion criteria in order to arrive at a final list of articles. Any disagreement was resolved by a third independent reviewer (C.K.). Each included manuscript was read to determine ultimate inclusion in the final analysis. From the manuscripts, the following information was extracted: author names, title, year of publication, country, study centre, study design, age, gender, primary tumour location, tumour differentiation, TNM staging and R0 resection for primary tumour, and chemotherapy and RT regimen. Results were stratified by three main groups: (1) synchronous RPLNM; (2) metachronous RPLNM; and (3) synchronous and metachronous RPLNM (presented as combined outcome data in studies). Preoperative imaging modality (computed tomography, CT; magnetic resonance imaging, MRI; positron emission tomography, PET; lymph node biopsy) was recorded to define the extent of RPLNM, information on patient selection criteria, anatomical boundaries of RPLND, number of lymph nodes harvested, and lymph node yield. Outcome measures were hospital length of stay, postoperative morbidity (Clavien-Dindo classification [22,23]), postoperative 90-day mortality, DFS, OS, re-recurrence rate and site. The median, 3-year, and 5-year DFS and OS were recorded where available.

Quality Assessment of Studies
The quality of all observational studies was assessed using the Newcastle-Ottawa Scale [24]. This was calculated by examining three factors: method of patient selection, comparability of the study groups, and number of outcomes reported. The full score was nine stars, and studies that had a score of seven stars or more were considered high quality. All studies were rated independently by two authors (M.G.F. and M.A.), with any differences resolved by consensus.

Statistical Analysis
The studies were assessed for information regarding the number of patients that underwent RPLND and their associated morbidity and mortality. The data were recorded in summary tables and divided into three groups: (1) synchronous RPLNM; (2) metachronous RPLNM; and (3) synchronous and metachronous RPLNM. The mean, median, range, and standard deviation were reported where applicable. The median value of the 3-year DFS and OS and the 5-year DFS and OS were calculated.

Chemotherapy and Radiotherapy Regimens
Perioperative chemotherapy was administered in all studies except for one study by Bowne et al. [33] (Table 2). Neoadjuvant chemotherapy was used less frequently compared with adjuvant chemotherapy, with 128 patients (23.7%) versus 368 patients (68.0%). The most widely used chemotherapeutic agent was 5-fluorouracil (5-FU), with several other studies reporting the application of biological agents such as bevacizumab and cetuximab. RT was used less commonly when compared to chemotherapy, with nine studies [16,17,26,30,31,[35][36][37]39] reporting either neoadjuvant (34 patients, 6.3%) or adjuvant (29 patients, 5.4%) RT. RT administration was declared in a smaller proportion of studies on patients with synchronous RPLNM in comparison to the other two groups. The precise RT regimen was only reported in three studies [20,35,37], where the cumulative RT doses ranged from 45 to 55.4 Gy. Table 2. Chemotherapy and radiotherapy details for pre-and post-retroperitoneal lymph node dissection in colorectal cancer.

Discussion
The optimal treatment strategy for RPLNM in CRC is not yet fully established due to the low incidence of isolated RPLNM, differing views on the classification of RPLNM, and the limited evidence on associated survival outcomes and morbidity of surgery. We have presented the largest systematic review of 19 studies, comprising a total of 541 patients, in order to suggest practical recommendations for the multimodality treatment of RPLNM in CRC.
Across the 19 studies, preoperative CT imaging was routinely used followed by PET-CT imaging. Short-axis diameter > 5 mm combined with an irregular shape was the most common radiological CT criteria for RPLNM with positive fluorodeoxyglucose uptake on PET. The main patient selection factors observed included patient's fitness for surgery, MDT discussion, clinically suspected RPLNM below the renal veins amenable to resection, and responsiveness to previous chemotherapy/RT. Perioperative chemotherapy was the preferred treatment modality, with RT only sparingly applied. Synchronous RPLND was more commonly undertaken, with the vast majority performing dissection below the renal veins. Overall, there was a reported postoperative morbidity rate of 23.3%, with a 90-day mortality rate of only 0.02% for RPLND.
It has previously been shown that the addition of PET-CT to CT imaging can improve the negative predictive value for RPLNM in cases without suspicious features on CT and the positive predictive value in cases with suspicious features on CT [41]. In light of the significant morbidity associated with RPLND, ascertaining the lymph node status of patients with suspected RPLNMs to the greatest accuracy prior to surgical decisionmaking is important. MRI has been found to be non-inferior to CT in detecting RPLNMs in multiple prospective studies in testicular germ cell tumours [42,43]; however, further studies are required in order to elaborate whether MRI has a defined role in RPLNMs in CRC. Regarding neoadjuvant/adjuvant therapy, it has been demonstrated that definitive chemotherapy in RPLNM can be an effective salvage treatment [44]. It must be noted here that Lee et al. [45] have concluded, following treatment of 52 patients with isolated RPLNM, that both upfront RT and deferred RT are potentially effective treatment strategies.
There have been four prior systematic reviews assessing the management (surgery ± non-surgery) of RPLNM in CRC. Ho et al. [46] included 110 patients who underwent RPLND, with a median DFS of 17-21 months and median OS of 34-44 months. At the time, they were only able to identify two case series [16,33] that included more than five patients who had surgery. Wong et al. [19] studied a 370-patient population consisting of a surgical and non-surgical group with median OS of 34-40 months and 3-14 months, respectively. Sasaki et al. [47] included 227 patients, both surgical and non-surgical groups, with the 3-year OS ranging from 60% to 100% and a median OS of 34-80 months for patients who had RPLND. The median OS was 14-42 months in the non-surgical group. Similar to the other two reviews, case reports and case series (five patients or less) were included in the analysis. Zizzo et al. [1] studied a patient population of 161 patients from nine studies with pathologically confirmed CRC isolated RPLNM who underwent RPLND. They identified a 5-year DFS rate of 0% to 60.5%, a 5-year OS rate of 53.4% to 87.5%, and a re-recurrence rate of 43.8% to 100%. In our present review of 19 studies, 541 patients underwent RPLND for isolated RPLNM (pathologically positive plus negative) with or without combined CRC metastatic lesions. A 5-year DFS rate of 10.0% to 65.0% was identified, along with a 5-year OS rate of 15.0% to 87.5% and a re-recurrence rate of 27.4% to 100%.
Regarding the surgical approach of RPLND, the dissection is often performed in the area with the following boundaries: renal hilum, bifurcation of the iliac artery, ureters bilaterally, and iliopsoas muscle [29]. This technique is challenging, particularly at the superior border, which can be a potential source of in-field recurrence. Care must be taken to identify and preserve the ureters as well as the sympathetic chain. The anatomical relationships of the lumbar vessels and sympathetic nerves in the infrarenal retroperitoneum are essential in performing a successful nerve-sparing bilateral RPLND to reduce postoperative co-morbidities and preserve nerve function [48]. Open RPLND has long been considered the standard of care. Laparoscopic RPLND, performed in three of the included studies [26,30,31], can be a technically challenging procedure requiring significant experience in laparoscopic dissection to safely access the lymph nodes posterior to the great vessels. As expertise and technology grows, in selected patients the robotic approach may provide improved ability to dissect behind the great vessels and control major bleeding more easily [49].
An emerging intraoperative technique in detecting lymph node metastases in CRC is near-infrared fluorescence (NIRF) imaging with the intention of avoiding a full (bilateral template) RPLND and its associated co-morbidities [50]. Indocyanine green (ICG) can provide visual assessment of blood vessels, blood flow, and lymph node road mapping [51,52]. Park et al. [53] used the NIRF technique for D3 lymphadenectomy in right-sided colon cancer; ICG was injected around the tumour for visualisation of lymphatic channels and lymph nodes. The number of atypical and harvested lymph nodes was significantly higher in the NIRF group compared with the conventional group. Various studies have reported on fluorescence imaging with ICG for the detection of occult CRC liver metastases with sensitivity exceeding 94% [54][55][56][57]. Furthermore, when fluorescence imaging was added to conventional imaging, extra metastases were found and resected in 20 out of 148 patients (13.5%) [58]. The potential benefits of ICG have been confirmed in other related surgical procedures, for example, in robotic-assisted radical prostatectomy with extended pelvic lymph node dissection [59]. Intraoperative confirmation and localisation of lymph nodes using fluorescents may have a role to play in increasing the accuracy of positive lymph node excision [60], and could potentially improve functional and oncological outcomes.
The presence of extracapsular invasion of resected lymph nodes for CRC has been proven to be associated with increased and earlier recurrence after CRC resection [61]. The currently included studies evaluated RPLNs on quantitative and anatomical aspects (i.e., number of positive RPLNs and their location with respect to other retroperitoneal structures). Thus, there is a need for studies to present other histological indicators, such as extracapsular invasion or extension, in order to allow for evaluation of the prognostic significance and better inform those patients who would benefit from earlier adjuvant therapy. A current phase II RCT (NCT03725254) is comparing radical surgery and retroperitoneal lymphadenectomy to radical chemoradiotherapy only for retroperitoneal lymph node recurrence of CRC, and is expected to be completed by October 2024.

Strengths and Limitations
This systematic review and analysis of the multimodality treatment and associated outcomes of RPLNM in CRC included a relatively large sample size of patients undergoing synchronous and metachronous RPLND to support practical recommendations on optimal surgical management.
The major limitation of this systematic review is the heterogeneity of the studies, which renders it more difficult to make firm evidence-based conclusions. The patients included were heterogenous (including isolated RPLNM with or without combined CRC metastases); certain studies reported a systematic approach to RPLND in the form of removing an entire lymph node basin, while others indicated that only grossly involved nodes were surgically removed. There were variability and differences in the patient selection process and perioperative modalities, with certain patients proceeding immediately to surgery. In general, most of the studies were single-centre retrospective studies lacking a control group, with only one RCT included.
In addition, information such as primary CRC histology, specific chemotherapy and RT regimens, and number of lymph nodes harvested were not reported in certain studies. The diagnostic criteria of CT imaging were only specified in six studies, and studies presenting sensitivity or specificity on diagnostic methods were not included. Survival rates in patients without surgery were not presented. Furthermore, the included articles ranged from 2001 to 2021, meaning that during this timeframe there have been changes and improvements in diagnostic methods, surgical techniques, and chemotherapy and RT regimens, with a growing emphasis on MDT management.

Implications for Multimodality Treatment of RPLNM in CRC
Based on the available evidence, practical recommendations for the surgical management of RPLNM in CRC can be proposed: all patients with suspected RPLNM should undergo PET-CT imaging prior to surgery; MRI imaging can be used in selected cases if felt necessary following MDT discussion, or where PET-CT imaging is not available; the indication and timing of chemotherapy and RT should be decided through MDT input in specialised centres; and perioperative chemotherapy may be considered to identify patients who are likely to benefit from synchronous or metachronous RPLND or to potentially improve DFS. However, the exact role of RT in the management of RPLNM is less clear based on the current evidence, and we have been unable to clearly identify whether synchronous or metachronous RPLND has superior oncological outcomes. RPLND above the renal vessels is generally avoided, highlighted in this systematic review by the patient selection process for surgery and by the observation that only six patients underwent type A RPLND. It is typically associated with high morbidity, and RPLNMs are often unresectable due to involvement of the coeliac axis, the root of the superior mesenteric artery, and adjacent organs such as the pancreas, stomach, duodenum, and renal hilum [34]. Furthermore, in these circumstances it is difficult to achieve adequate margins with irradiation.

Conclusions
RPLNM in CRC is typically considered to have a poor prognosis, with high rates of morbidity and recurrence. Based on the findings of this systematic review, it can be concluded that RPLND is a feasible treatment option with limited morbidity and possible oncological benefit for both synchronous and metachronous RPLNM in CRC. There is clearly a need for further prospective clinical trials in order to establish a better evidence base for RPLND in the context of RPLNM in CRC and to understand the timing of RPLND in a multimodality treatment pathway in order to optimise treatment outcomes for this group of patients.

Supplementary Materials:
The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/cancers15020455/s1, Table S1: Search strategy for the systematic review on the treatment of retroperitoneal lymph node dissection in colorectal cancer across the four databases (MEDLINE, EMBASE, EMCare and CINAHL), January 1990-June 2022; Table S2: Newcastle-Ottawa Scale scoring; Table S3: Patient selection factors affecting decision to proceed with retroperitoneal lymph node dissection in colorectal cancer; Table S4: Postoperative morbidity following retroperitoneal lymph node dissection in colorectal cancer; Table S5: Sites of re-recurrence following retroperitoneal lymph node dissection in colorectal cancer.