Introduction

There is no discussion that for patients with rectal carcinoma, regional lymph node metastases (RLNM) are indicators of an increased risk of distant metastases. In contrast, the independent influence of RLNM on locoregional recurrence (LR) is still in discussion. In this context, it has to be differentiated according to type and quality of surgery (conventional or total mesorectal excision (TME) surgery), and the use of multimodal therapy has to be considered. In the following, ‘TME surgery’ is used for following the concept of total mesorectal excision for carcinomas of the middle and lower third and of partial mesorectal excision (PME) for carcinomas of the upper third, in each case with sharp dissection in the pelvis [1–3].

The subject of the association between the pretherapeutic regional lymph node status and the risk of LR directly influences the selection of patients for neoadjuvant treatment. Thus, this topic has practical clinical relevance, especially in the present time of general demand for a more selective and individualised indication for neoadjuvant treatment, preventing under- as well as overtreatment [4–10]. In the following, the actual data of literature including new data from the Erlangen Registry of Colorectal Carcinoma (ERCRC) [11] will be analysed.

Preliminary remarks and definitions

In the description of the pretherapeutic regional lymph node status according to the rules of the TNM classification [12], it is distinguished between a clinical and a pathological classification (cN/pN). The clinical assessment of the pretherapeutic lymph node status is associated with uncertainty. The real situation is described by the results of the histopathological examination of the resection specimens, however, only in case of primary surgery. Following neoadjuvant treatment, the pathological examination of the regional lymph nodes describes the situation following neoadjuvant treatment (ypN); thus, in case of neoadjuvant treatment, the pretherapeutic lymph node status can be assessed only clinically (cN), but not pathologically.

Between Western Countries and Japan, there are differences in the anatomic subdivision of the rectum. In Japan, ‘lower rectum’ includes all tumours below the peritoneal reflection, corresponding to tumours of the middle and low third of the rectum, about <11–12 cm from the anal margin. In the Western countries, ‘lower or low rectum’ is used only for tumours of the lower third of the rectum (lower margin <6–7.5 cm from the anal margin).

Whether for analysis of the association between the pretherapeutic regional lymph node status and the risk of LR only patients with primary surgery or also patients with neoadjuvant treatment can be included depends on the reliability of the clinical diagnosis of RLNM. Thus, in the following, at first, the accuracy of the present imaging methods concerning regional lymph nodes will be discussed.

In the lymphatic spread of the rectal carcinoma, it has to be further distinguished between the predominant spread within the mesorectum upwards and the spread to lateral pelvic nodes. The latter will be discussed in a separate section.

Possibilities of clinical assessment of pretherapeutic regional lymph node status

With regard to the possibilities of pretherapeutic diagnosis of RLNM, it has to be distinguished between the lymph nodes in the mesorectum and the lateral pelvic lymph nodes.

The present clinical assessment of mesorectal lymph nodes is associated with considerable uncertainty [5, 8, 13, 14]. According to two meta-analyses [13, 15] and three systematic literature reviews [16–18], no significant differences between EUS, CT, and MRI with respect to staging the nodal status have been found (Table 1). However, in these reviews, exact information about the proportion of patients with neoadjuvant treatment and about the use of endorectal coil for MRI are partly missing.

Table 1 Pretherapeutic assessment of regional mesorectal lymph node status by endorectal ultrasonography (EUS), computed tomography (CT) and magnetic resonance imaging (MRI)
Full size table

High-resolution MRI employing pelvic phase-array coils is the best modality to evaluate the relation between tumour and fascia mesorectalis (the later plane of surgery and circumferential resection margin). Thus, today, MRI can be considered as the established modality of choice for pre-operative staging and selection for neoadjuvant treatment [5, 8, 19, 20]. Near details of present day efficiency of MRI in the assessment of the pretherapeutic mesorectal lymph node status can be seen in Table 2. The pooled results are disappointing: positive predictive value only 63%, accuracy only 71%. A similar value for accuracy (380/502 = 76%) is reported from the National Cancer Center Hospital, Tokyo [21]. In this retrospective study (1988–2002, 817 patients) the lymph node status has been assessed by ‘CT and/or MRI’ and included for patients at or below the peritoneal reflection the lateral pelvic lymph nodes.

Table 2 Pretherapeutic diagnosis of involved regional mesorectal lymph nodes (cN1,2) by high-resolution MRI employing pelvic phased-array coils/review of the literature 1999–2009
Full size table

Summarising, it has to be stated that a reliable assessment of the pretherapeutic mesorectal regional lymph node status is not possible by the present imaging modalities [13, 14, 22].

An improvement in lymph node staging by the contrast agent ultrasmall super paramagnetic iron oxide-enhanced MRI was expected in future [8, 22–24]. However, this contrast medium is no more available at present. The results of the multi-slice (multi-detector row) CT (MDCT) seem to be similar to MRI [25–28]; however, the available data are limited.

The diagnosis of four or more involved mesorectal lymph nodes (cN2) is a further problem. Table 3 shows the results of four relevant studies (2000–2006, n = 362). The pooled positive predictive value is only 60%. Forty percent of the positive findings are false, i.e., the results are near to purely coincidental values. Thus, the pretherapeutic MRI findings of cN2 cannot be considered as basis of an indication for neoadjuvant therapy.

Table 3 Pre-operative MRI-diagnosis of more than three involved regional mesorectal lymph nodes (cN2)/review of literature 1999–2009
Full size table

The possibilities of clinical assessment of lateral pelvic lymph nodes could be studied only in Japan where patients with T3 and T4 rectal carcinomas below the peritoneal reflection (designated as lower rectal carcinoma according to the Japanese Society for Cancer of the Colon and Rectum, 1997) [29] have been treated by primary TME surgery with extended lateral pelvic lymphadenectomy (D3 lymphadenectomy) [3]. In a comparison between mesorectal and lateral pelvic lymph nodes [30], the results of MRI proved to be better for lateral than for mesorectal nodes: positive predictive value 56% (10/18) versus 46% (12/26), negative predictive value 91% (10/11) versus 81% (22/27), false positive findings 15% (8/53) versus 26% (14/53) and overall accuracy 83% (44/53) versus 64% (34/53). Only the last difference is statistically significant (p = 0.046).

In this context, it has to be emphasised that an involvement of lateral pelvic lymph nodes is in most cases associated with positive mesorectal lymph nodes: Sugihara et al. [31]: 103/129 (80%) and Ueno et al. [32]: 31/41 (76%).

Regional lymph node metastasis and the risk of locoregional recurrence

The following deals with the problem whether absence or presence of regional lymph node metastasis (N0 or N1,2) has an independent influence on the risk of LR (the impact of the number of involved lymph nodes is discussed in the following section). In this respect, it is essential to distinguish between conventional and optimised and quality-assured TME surgery.

As shown above, the pretherapeutic clinical assessment of the regional lymph node status is uncertain. Thus, a reliable analysis of the association between pretherapeutic regional lymph node status and the risk of LR is possible only by using the histopathological assessment of the pretherapeutic regional lymph node status, i.e., the pN status after primary surgery (without neoadjuvant treatment) describing the pretherapeutic situation.

Following conventional surgery, the different frequency of LR for pN0 and pN1,2 patients is proven for a long time already [33–39].

For the actual optimised TME surgery, a correlation between pN and LR seems unlikely for theoretical reasons. The TME surgery is characterised by the complete removal of the lymphatic drainage upwards as a package limited by fascias, thus, regional lymph nodes—except, lateral pelvic nodes (see below)—do not remain as origin of LR.

For patients treated by primary surgery following the concept of TME, two publications include relevant multivariate analyses dealing with the influence of pN0 and pN1,2, respectively, on the risk of LR. In both studies, the LR rates for pN0 and pN1,2 were not different. In the largest study [40] (n = 686), the crude LR rate for all patients was 7%, the hazard ratios compared to pN0 = 1.0 were for pN1 1.5 (95% CI, 0.8–3.1) and for pN2 1.4 (95% CI, 0.6–3.4; p = 0.49). In the study of Ross et al. [41] (212 patients after exclusion of 17 patients with distant metastasis, in 11 patients = 5.2% neoadjuvant radiotherapy), the crude LR rates were 16/112 = 14% for pN0 and 10/100 = 10% for pN2 (p = 0.41).

Other multivariate analyses of the association between pN status and LR in patients treated by primary TME surgery refer to selected subgroups, e.g., stage II patients treated by anterior resection alone [42] or pT1 and 2 patients only [43] or patients treated by intersphincteric resection only [44] and are, for the general question of the indication for neoadjuvant treatment, not relevant. The same applies to the study of Eriksen et al. [45] because the multivariate analysis includes only 1,676 (67.6%) of 2,480 pT3 patients, namely, those with known circumferential resection margin (CRM) status; thus, the results cannot be considered as representative for all pT3 patients. The multivariate study of Bufalari et al. [46] cannot be considered relevant because of the very limited number of patients (n = 73).

Recently Fujita et al. [21] reported the treatment results for 817 patients with rectosigmoid, rectal and anal carcinomas, clinical stage II and III. A multivariate analysis performed for pretherapeutic clinical factors only showed the pretherapeutic clinical assessment of regional lymph nodes (including lateral pelvic nodes in rectal carcinomas at and below the peritoneal reflection; cN+) as statistically significant factor affecting LR. This study is questionable because of insufficient technique in carrying out the residual tumour (R) classification (only one block histologically examined), missing data on the circumferential resection margin (CRM) status, missing differentiation between rectosigmoid, rectal and anal canal carcinomas in the multivariate analysis and the limitation of the multivariate analysis to clinical factors only. Thus, for the problem under discussion this study cannot be considered.

Summarising, it can be stated that there are no reliable data indicating that in patients treated by optimised TME surgery, any involvement of regional lymph is associated with an increased LR rate and requires neoadjuvant or adjuvant radiotherapy.

It has to be pointed out that following neoadjuvant treatment patients with RLNM (ypN+) represent a negative selection of the pretherapeutically lymph node positive patients, namely no (or not sufficiently) responding patients. For these patients, the proportion of pretherapeutically lymph node positive patients (cN+) remains unknown because the clinical assessment of lymph node metastasis is not sufficiently reliable (see previous discussion). The statistically significant correlation between ypN and LR after neoadjuvant radiochemotherapy (Rödel et al. [47]; n = 344) and neoadjuvant long-course radiotherapy (Larsen et al. [48]; n = 204), respectively, is not relevant for the question of indications for neoadjuvant treatment because from this correlation cannot be concluded to an association between the pretherapeutic lymph node status and the LR rate.

Number of involved regional lymph nodes and the risk of locoregional recurrence

A different prognosis, depending on the number of involved regional lymph nodes (1–3 versus ≥4), has already been described in the mid of the last century [49, 50] and was the reason for the subdivision of lymph node positive patients in the Dukes’ stages C1 and C2 and the (p)N1 and (p)N2 categories of the TNM classification. Moran et al. [51] confirmed the higher LR rates in cases of pN2 with data on conventional surgery without neoadjuvant treatment.

In the registry of the Norwegian Rectal Cancer Project [40], the risk of LR in patients treated by primary TME surgery was for pN1 and pN2 nearly identical (see previous discussion). In contrast, in the multivariate analysis of Ross et al. [41], a significant increase of the LR rate depending on the number of involved regional lymph nodes has been reported. However, further details, especially with regard to the cut-off value, are missing.

In a univariate analysis, Cecil et al. [52] demonstrated that following (low) anterior resection according to the principles of TME surgery and ‘only minimal’ use of neoadjuvant radiotherapy, the crude LR in patients with more than eight involved regional lymph nodes increases significantly in comparison with patients with no lymph node metastasis and 1–8 involved nodes: 3/13(23%) versus 13/457(2.8%; p = 0.008).

Because of these inconsistent data, the relationship between number of involved regional lymph nodes and the risk of locoregional recurrence was analysed on the basis of the data of the ERCRC from a time period with use of primary TME surgery (without neoadjuvant therapy) in the vast majority of patients, i.e., January 1, 1986 to January 31, 1995. Inclusion criteria, patients’ characteristics, LR rates and the results of uni- and multivariate analysis of risk factors for LR are presented in Table 4. While there is no statistically significant difference between pN0 and pN1, there are highly significant differences (p < 0.001) between pN1 and pN2 in the multivariate analysis too. Correspondingly, in view of locoregional recurrence risks, a subdivision between pN0,1 and pN2 is preferable because this subdivision results in the highest difference of the locoregional recurrence rates: 10.3% versus 33.5% (p < 0.001).

Table 4 Dependence of locoregional recurrence risk on the regional lymph node status in rectal carcinoma patients treated by primary TME surgery in curative intent. Data from the Erlangen Registry of Colorectal Carcinoma (ERCRC)
Full size table

From the data presented in Table 4, one may conclude that in case of extensive lymph node metastasis (involvement of four or more lymph nodes, N2) the risk of locoregional recurrence is about 30% and thus, a neoadjuvant radio-/radiochemotherapy is indicated. However, it has to be taken into consideration that in such patients, the prognosis is determined in the first place by the high risk of distant metastasis. Moreover, there is the problem of pretherapeutic diagnosis of N2 by imaging. As shown in Table 3, the positive predictive value is at present only 60%. Thus, the pretherapeutic clinical finding cN2 cannot be considered as certain basis for an indication for neoadjuvant radiotherapy. However, in this situation, the patient is to be included in the decision (see further discussion).

Problems of lateral pelvic lymph nodes

The predominant and typical lymphatic drainage of the rectum is directed upwards in the mesorectum to the nodes along the superior rectal and inferior mesenteric arteries. In addition, there is a lateral lymphatic drainage to the lateral lymph node compartment. The latter has been studied predominantly in Japan where an extended lateral pelvic lymph node dissection (pelvic side wall dissection, D3 dissection) has been performed routinely for years. According to the JSCCR [29] the lateral nodes include: (1) middle rectal root nodes (lateral to the pelvic nerve plexus); (2) internal iliac nodes; (3) obturator nodes; (4) median and lateral sacral nodes; (5) external iliac nodes; (6) common iliac nodes and (7) aortic bifurcation nodes. According to the TNM classification [12], involvement of the stations 1–4 is classified as regional lymph node metastasis, of stations 5–7 as distant metastasis.

In the Japanese literature 1998–2008, the frequency of lateral lymph node metastasis in rectal carcinoma is reported as about 10% (Mori et al. [53]: 40/906 = 4.4%; Takahashi et al. [54]: 57/632 = 9.0%; Koda et al. [55]: 35/265 = 13.2% and Sugihara et al. [31]: 129/930 = 13.8%, pooled value 261/2733 = 9.5%, 95% CI 8.4–10.7%). Not included in these figures are cases with immunohistochemical findings of tumour cells only [56–58] because the prognostic relevance of such findings is not proven.

The most important risk factors for lateral node metastases are the tumour site (tumours below the peritoneal reflection, especially tumours with a distal margin ≤4 cm [59] or ≤6.5 cm from anal margin [54] and involvement of mesorectal nodes [31, 54, 60]. Other factors are direct involvement of the circumferential resection margin by tumour [58, 61], G3,4 [59], cT3,4/pT2-4, tumour diameter ≥4 cm and female gender [31].

For rectal carcinomas ≤8 cm from the anal margin, a score for the estimation of the risk of lateral lymph node metastasis has been published by Ueno et al. [59]. Three risk factors are considered: distal margin of tumour 4 cm or less from the anal margin, histological type other than adenocarcinoma G1 or 2 and involvement of mesorectal lymph nodes. The frequencies of lateral lymph node metastasis are 3% (1/34) for zero, 8% (10/120) for one, 33% (24/73) for two and 60% (6/10) for three risk factors, respectively.

In patients with cT3,4 carcinomas below the peritoneal reflection, since many years in Japan, an extended lateral pelvic lymph node dissection has been performed; since the early 1990s, with pelvic autonomic nerve preservation [53, 62–66].

This Japanese strategy found no acceptance in western countries because

  • in comparison with TME surgery, the benefit of additional extended lateral pelvic lymph node dissection for survival could not be proven;

  • increased long-term urinary and sexual dysfunction;

  • longer operation time and

  • increased blood loss [5, 67–71].

Today, in Europe, for low cT3,4 carcinomas neoadjuvant radio-/radiochemotherapy generally is used [9, 72–77]. This is increasingly accepted as an alternative to extended lateral pelvic node dissection in Japan too [21, 55, 59, 78–80].

Implications for the indication to neoadjuvant therapy

For a long time, regional lymph node metastases have been considered as independent factor not only for distant metastases, but also for LR. This has changed in the era of TME surgery. Today, in case of careful and quality-assured TME for middle and lower rectal carcinoma and PME for upper rectal carcinoma, an increased risk for LR cannot be assumed for each patient with regional lymph node metastasis.

Frequently, all patients with clinically positive regional lymph nodes are treated with neoadjuvant radio-/radiochemotherapy as proposed by many actual guidelines. In the USA, it has been recommended to treat T1,2 tumours with clinically diagnosed regional lymph node metastasis by primary surgery [81]. This has been accepted in the last actualisation of the German guidelines [76] too.

Today, increasingly, the indication for neoadjuvant treatment is based on MRI. In this case, the regional lymph node status is usually not considered [45, 72, 75, 77, 82–84].

Recent data, limited so far (Table 4), show that involvement of four or more regional lymph nodes (N2) is an independent risk factor for LR in TME surgery too. Thus, an indication for neoadjuvant radiotherapy may be concluded. This is in agreement with the assessment of N2 as a poor prognostic factor and the use of neoadjuvant chemoradiation for such patients [8, 23, 73, 74, 85]. Future clinical trials should be based on a more differentiated indication of neoadjuvant treatment in stage III diseases with detailed documentation of lymph node findings in the imaging diagnosis.

However, there is the problem that with today’s available imaging methods, the involvement of four or more regional lymph nodes can preoperatively be diagnosed only with 60% certainty (Table 3). In this case, neoadjuvant radiotherapy is in 40% overtreatment associated with considerable late adverse effects. Thus, this situation is a typical example for the requirement to discuss risks and benefits with the patient [86, 87]. A differentiated shared discussion in the sense of an informed consent should take place in any such case.