Introduction

Lateral pelvic lymph node (LPLN) metastases are present in 14–20% of all rectal cancer cases [1,2,3,4,5,6]. According to some studies, there is LPLN involvement in the majority of patients with a locoregional recurrence, while other studies emphasize the role of residual mesorectal fat with cancer cells or the omission of a rectal washout as cause of recurrence [3,4,5, 7,8,9]. The LPLN compartment includes the common, internal and external iliac and the obturator artery lymph nodes [10].

Different treatment strategies have been adopted worldwide for rectal cancer patients with LPLN suspicious for metastases (LPLNSM), and optimal management remains controversial. According to European and American guidelines, standard treatment consists of neoadjuvant chemoradiotherapy (nCRT) followed by a total mesorectal excision (TME) [11,12,13,14]. By most of these guidelines, the possibility of a (selective) lateral pelvic lymph node dissection (LPLD) is mentioned without discussing its necessity in case of LPLNSM that are persistent or responsive after nCRT [12,13,14]. The French guidelines for the treatment of rectal cancer advise against LPLD and recommend a lymph node excision in case of LPLNSM and positioning of a fiducial marker to help further irradiation targeting [11]. Hesitation to perform LPLD might be based on a presumed increase in morbidity. However, in the East, removal of LPLN is considered an important extension, since its addition to an oncologic mesorectal excision in primary low/mid rectal cancer patients without LPLNSM was reported to reduce the local recurrence rate [15]. The Japanese Society for Cancer of the Colon and Rectum (JSCCR) guidelines recommend TME with LPLD as standard treatment for patients whose lower tumor border is located distal to the peritoneal reflection and whose tumor has invaded beyond the muscularis propria [16]. For this subgroup of patients without nCRT, LPLN metastases were reported in 20% [16].

Differences exist between Western and Eastern approach to rectal carcinoma with LPLNSM. Whether a LPLD should be performed during TME depends on the risk of metastases after nCRT. The aim of this review study is to determine the percentage of LPLN metastases in patients with rectal carcinoma and LPLNSM on pretreatment imaging, treated with nCRT.

Materials and methods

Studies that reported the pathological outcomes after TME with LPLD after nCRT for LPLNSM on pretreatment imaging were reviewed. The search strategy is detailed in supplementary Table 1 in the appendix. Randomized, prospective and retrospective cohort studies were eligible for inclusion. The electronic databases MEDLINE, EMBASE, Web of Science and Cochrane library were searched from inception to the 17th of May 2018. Reference lists of the included studies were cross-checked for additional papers. Studies identified by the searches were sifted based on title, abstract and full text by a single reviewer (Fig. 1). Two reviewers (YA, ARW) assessed the full texts of the remaining studies independently. This review was conducted in accordance with PRISMA guidelines [17]. The quality of evidence was rated by means of GRADE methodology for the primary outcome [18, 19]. An evidence table and summary of findings table was produced. Authors were contacted by electronic mail in the event of any doubt. This review has been registered on PROSPERO international register of systematic reviews with registration number: CRD42018091327.

Fig. 1
figure 1

Study flow diagram

Full size image

The primary outcome was the percentage of pathologically confirmed LPLN metastases in rectal cancer patients with LPLNSM on initial imaging, treated with nCRT. Data extracted included author, year of publication, type of study, number of patients, number of LPLD in case the number of patients was not reported, type of imaging modality, author-defined cutoff criteria for LPLNSM before and after nCRT, nCRT regimen (Supplementary Table 2), interval between nCRT and surgery (Supplementary Table 2), percentage of persistent/responsive nodes after nCRT and percentage of patients/LPLD with pathologically confirmed LPLN metastases for both persistent and responsive LPLNSM.

Results

The literature search (Supplementary Table 1) identified 5892 initial articles, of which 1112 duplicates were removed, and 4696 articles were excluded on the basis of title or abstract (Fig. 1). Eleven studies fulfilled all of the selection criteria after full-text review of the remaining 84 articles [20,21,22,23,24,25,26,27,28,29,30]. These included two prospective [20, 21] and nine retrospective cohort studies [22,23,24,25,26,27,28,29,30]. The quality of evidence of the four comparative [26,27,28, 30] and seven non-comparative cohort studies [20,21,22,23,24,25, 29] was low because of the observational nature. However, the level of evidence was upgraded to moderate, since the chance at bias with a dichotomous outcome measure, as the one used in our study, is little. Ranging from eight to 77 patients per study, a total of 462 patients were included. An overview of the type of studies, number of patients, their pathology outcome and imaging modalities can be found in Table 1. Cutoff criteria and node aspects for the assessment of LPLNSM before and after nCRT were reported by the majority of studies (Table 2, 3). Radiologic cutoff criteria for the evaluation of LPLN differed between studies as noted in the tables (Table 2, 3). The LPLN pathology outcomes of responsive and persistent LPLNSM after nCRT are displayed in Tables 4 and 5, respectively. Two of the eleven included studies described their results on pathological outcome of the LPLN based on the number of excisions [20, 21], meaning that bilateral dissection was counted as two excisions. The rest of the studies described the pathological outcome of the LPLN based on the number of patients. Pathologically confirmed metastases occurred in a range from 39.0 to 71.1% of all LPLD [20, 21]. In the other nine studies, pathologically confirmed LPLN occurred in a range from 25.0 to 61.1% of all patients had after uni- or bilateral LPLD. Only five [22, 23, 26, 27, 29] and seven studies [22,23,24,25,26,27, 29] provided information about patients with responsive and persistent nodes, respectively. Patients with responsive LPLNSM turned out to have LPLN metastases in a range from 0 to 20.4%. Patients with persistent LPLNSM turned out to have LPLN metastases in a range from 25.0 to 83.3%.

Table 1 Characteristics of studies and pathological results of LPLNSM
Full size table
Table 2 Cutoff criteria for the assessment LPLNSM before nCRTx
Full size table
Table 3 Cutoff criteria for the assessment LPLNSM after nCRTx
Full size table
Table 4 Results of patients with responsive LPLNSM after nCRTx
Full size table
Table 5 Results of patients with persistent LPLNSM after nCRTx
Full size table

Discussion

To our knowledge, this is the first review study on pathologically confirmed LPLN metastases in rectal cancer patients with LPLNSM, treated with nCRT. LPLN metastases occurred in a large amount of patients with LPLNSM on initial imaging and patients with persistent nodes after nCRT. Even in LPLN that are considered responsive on restaging, in a significant amount of patients pathologically confirmed metastases are reported. These data suggest that LPLD seems to be justified in case of LPLNSM that are non-responsive to nCRT and should even be considered in LPLN that are responsive. This study should alert surgeons that are currently executing diverging policies.

The main limitation of retrospective studies investigating long-term outcomes for treatment of patients with LPLNSM is a lack of histological verification of the nature of LPLN. Brown et al. retrospectively reviewed baseline MRI images of rectal cancer patients that underwent a TME or abdominoperineal resection with or without neoadjuvant treatment for a minimum of 5-year follow-up [31]. A high suspicion of LPLNSM was issued in case of the presence of mixed signal intensity and/or an irregular border of the nodal capsule, irrespective of nodal size. MRI detected LPLNSM was neither biopsied nor evaluated by LPLD. The authors reported a 5-year disease-free survival of 42 and 71%, respectively, for 38 patients with MRI-identified LPLNSM at baseline and 287 patients without suspicious nodal characteristics. Kusters et al. conducted a study that compared the Dutch and Japanese approach for low rectal cancer: TME with/without neoadjuvant RTx compared to extended surgery consisting of LPLD and an oncologic rectal resection without any (neo) adjuvant treatment in the majority of cases, respectively [32]. The authors reported good local control for both approaches, as compared to TME alone, and therefore, the added value of the application of uni- or bilateral LPLD on the prevention of lateral recurrence was questioned [32]. Kim et al. conducted a retrospective study specifically on long-term outcomes such as local recurrence and overall survival. The authors analyzed 580 patients who underwent nCRT and TME. The results suggest that a subgroup with responsive nodes after nCRT may not benefit from LPLD and a subgroup with persistent LPLNSM after nCRT may benefit from LPLD [33]. The current study did not analyze any long-term outcomes, which is a limitation of this study. However, long-term outcomes rather describe the indirect effect than the direct effect of nCRT for clinically positive LPLN, since other oncological factors might also contribute to recurrence and survival.

Besides oncologic outcomes, the morbidity associated with LPLD should also be taken into account such as urogenital and bowel dysfunction and motor dysfunction of the lower extremity [34,35,36]. Such morbidity has been reported [36,37,38]. However, a recent RCT did not show any significant differences in postoperative morbidity by adding LPLD to TME despite a longer operation time and a higher amount of blood loss [39]. Nonetheless, Asian patients were treated in this trial, and therefore, morbidity associated with LPLD might be higher in a European population.

An accurate restaging after nCRT is of paramount importance. Promising techniques to improve the accuracy of nodal (re)staging include diffusion-weighted imaging (DWI), lymph node-specific contrast agents and PET-CT. Van Heeswijk et al. compared clinical rectal lymph node positivity according to DWI after nCRT with the number of metastatic nodes at histopathology or long-term follow-up [40]. They analyzed 90 rectal cancer patients including 71 patients with a yN0 status and 19 patients with a yN-positive status. To differentiate between yN0 and yN-positive status, the positive predictive value was 24% and the negative predictive value was 100%, concluding that absence of nodes at DWI is a reliable predictor of yN0 status after nCRT. An earlier study from the same group demonstrated a high negative predictive value for prediction of nodal status (0.95 and 0.85 for expert and general radiologist, respectively) in case of MRI with a lymph node-specific contrast agent [41]. Ishihara et al. evaluated the diagnostic value of PET-CT in 34 patients treated with nCRT and reported a high degree of accuracy for the prediction of metastatic LPLN [42]. Most recent studies suggest nodal size to be one of the most important risk factors for node positivity. Ogura et al. performed a multicentre pooled analysis of patients with low, locally advanced rectal cancer from 12 hospitals in seven Eastern and Western countries [43]. LPLD was performed in 142 patients in five hospitals, which resulted in 35 patients (24.6%) with pathologically positive LPLN. After multivariable analysis, LPLN with a short axis equal to or greater than 7 mm resulted in a significantly higher risk of lateral local recurrence (HR 2060; P = 0.045) compared with LPLN of less than 7 mm.

Up till now, to our knowledge, no results of RCT’s have been reported yet, investigating the value of LPLD in case of LPLNSM. Wei et al. are currently enrolling patients with LPLNSM in what they state to be the first clinical trial randomizing patients to either nCRT with TME alone or nCRT with TME combined with LPLD [44]. The randomization of patients in this study is executed prior to nCRT and irrespective of restaging, which implies that in the TME alone group there might be patients with persistent nodes that are not dissected. Although this study investigates a policy that currently is being executed, in light of the results of the present study, this study raises ethical concerns, since a significant number of patients with LPLNSM on pretreatment imaging turned out to have metastatic nodes. These nodes would be left untreated by this study.

It must be noted that four of the studies in this review only included patients with persistent LPLNSM after nCRT [20, 21, 24, 25]. Therefore, the percentage of LPLN metastases in these studies is expected to be higher than the percentage of LPLN metastases in the other included studies that also performed LPLD on patients with responsive LPLN after nCRT. This is supported by the data of this review showing LPLN metastases in a range from 25.0 to 83.3% of patients with persistent nodes and in range from 0 to 20.4% of patients with responsive nodes.

This current study has several limitations. First of all, variation in the use of imaging modalities to diagnose LPLNSM has possibly contributed to the interpretation of the measured values. A study of the JSCCR mentions that using MRI for identifying the cutoff value is presumably the best imaging modality, due to its simplicity and good interobserver agreement [45]. The use of other imaging modalities like transrectal ultrasound in the studies that were included in this systematic review could have negatively influenced the comparability. Secondly, cutoff criteria and node aspects for determining LPLNSM on imaging modalities were not reported by all studies and different cutoff values were used (Table 2, 3, 4, 5). The current review included studies with different cutoff values for lymph nodes to be considered ‘suspicious’. Therefore, staging accuracy of the majority of studies seems suboptimal and a selection bias cannot totally be excluded. Thirdly, the amount of time in between nCRT completion and surgery, the dose of radiation therapy and the type of CRT regimen could be of influence on the extent of downstaging and pathologic complete response (pCR) rate by analogy with the response of the primary rectal tumor itself. RCTs support surgical resection of the primary rectal tumor more than 6–8 weeks after nCRT due to more downstaging and a higher pCR rate, although a survival benefit is unproven [46]. Additionally, results from studies investigating the dose of radiation therapy and the influence of a neoadjuvant regimen incorporating induction chemotherapy on the responsiveness of LPLNSM are of interest.

Conclusions

In conclusion, in a large amount of patients with LPLNSM on initial imaging, the LPLD pathology report indicates lymphatic metastases after nCRT. Even LPLNSM that are responsive on restaging scans show significant rates of pathologically confirmed metastases. The wide range of pathologically confirmed LPLN metastases illustrates the heterogeneity among studies with a worrisome lower limit. Based on these findings, in patients with LPLNSM on pretreatment imaging, the potential for LPLN metastases after nCRT must be emphasized. Although the review is limited by small, mostly retrospective series and varying methods of LPLN staging, the reported high rates of pathologically confirmed LPLN metastases, a dichotomous outcome measure with a low chance at bias by itself, are worrisome. In the present era with increasing centralization of rectal cancer care, the complexity of LPLD does not seem an impediment.