Introduction

Although contemporary management of colorectal cancer (CRC) is multimodal, surgical resection remains the mainstay of curative treatment [1]. In addition to the degree of tumour invasion, the number of involved lymph nodes also defines the UICC stage, which is a crucial prognostic indicator and guides the use of adjuvant chemotherapy [2]. Hence, adequate lymph node retrieval is essential for the pathological evaluation of resected specimens. Since the introduction of the AJCC/UICC staging manuals in 1997 [3], which require a minimum of 12 lymph nodes for the adequate staging of CRC, various studies have addressed the impact of different factors on the lymph node count.

Independent of other factors, the depth of colonic wall penetration, reflected in the T category, directly correlates to the number of nodes retrieved [4]. Tumours that are larger or that have a higher T category are thought to provoke a greater antigenic immune and inflammatory response within regional lymph nodes leading to increased prominence and easier retrieval. It also seems probable that macroscopically aggressive tumours are resected and examined with greater diligence [5].

Furthermore, the site of the disease is related to the number of retrieved lymph nodes. Several studies have demonstrated that more lymph nodes are retrieved from right-sided than from left-sided cancers [5,6,7,8]. Although the reasons for this discrepancy are not entirely clear, differences in the anatomy of the blood and lymphatic drainage are hypothesised to be responsible [9]. Other possible explanations are location-specific molecular mechanisms: proximal carcinomas are more commonly microsatellite unstable, BRAF-mutated and express the CIMP phenotype, whilst chromosomal instability is more common in distal tumours [10]. Although the association between tumour location and the underlying molecular pathways is well established, various studies reported that adequate lymph node count (a minimum of 12 lymph nodes) is independently associated with microsatellite instability (MSI) [11,12,13].

In order to investigate the association between MSI and the lymph node count as well as the frequency with which it was adequate in resected specimens, we examined data from 1196 patients with CRC enrolled in a population-based case-control study.

Materials and methods

Study population

The patients investigated were participating in the DACHS study (DACHS: Darmkrebs, Chancen der Verhütung durch Screening), a population-based case-control study of CRC with patient follow-up, which has already been described in detail in previous publications [14, 15]. Patients were recruited from all 22 hospitals of the study region (the Rhine-Neckar region in southwest Germany), were 30 years or older with a histologically confirmed first diagnosis of primary colorectal cancer between 2003 and 2007, were physically as well as mentally fit to take part and able to communicate in German. Of this population, 1196 patients (those with available data on lymph node count, CRC category and microsatellite status) were included in our analysis. The characteristics of the study population are illustrated in Table 1.

Table 1 Clinical and pathological characteristics of the patient cohort
Full size table

Data collection

Patient information was acquired in standardised personal interviews after surgery, during hospitalisation or at home after discharge. Clinical data (discharge letters, pathology reports of the cases and endoscopy reports) was also collected.

Definition of proximal and distal colorectal cancer

Tumours oral to the left (splenic) colic flexure were defined as being proximal, those aboral as distal. This well-established cut-off was chosen based on the differing embryological origins of the proximal and distal segments of the colon and since it is the boundary between regions supplied by the superior and inferior mesenteric arteries.

Collection and processing of tumour samples

Formalin-fixed paraffin-embedded (FFPE) colorectal tumour samples were collected from the pathology departments of participating institutions and transferred to the tissue bank of the National Center for Tumour Diseases (NCT) in Heidelberg.

After microscopic identification of areas with the highest tumour cell concentration (comprised of at least 70% tumour cells), DNA was isolated from the tumour tissue blocks. DNA isolation was performed using a commercial DNA Extraction Kit (DNeasy; Qiagen) following the manufacturer’s protocols. Microsatellite instability was identified in 1196 cases using a mononucleotide marker panel (BAT25, BAT26 and CAT25). The marker panel used is able to differentiate MSI-high from non-MSI-high tumours with a sensitivity of 98.2%, a specificity of 100% and in 100% agreement on MSI-high tumours with the National Cancer Institute/International Collaborative Group on HNPCC (NCI/ICG-HNPCC) marker panel (BAT25, BAT26, D17S250, D2S123 and D5S346) for the evaluation of microsatellite instability in colorectal cancer [16, 17].

Lymph node examination

The number of retrieved lymph nodes was taken from pathology reports and discharge letters which were available from all participating institutions. Tumour processing (gross room dissection and histopathological evaluation) was performed by five different pathological institutes, which were not aware of the specimens’ inclusion in a study at the time of assessment, reflecting performance in routine practice in the study region. Following the, at that time contemporary, UICC TNM Classification Guidelines 6th edition [18], when fewer than 12 lymph nodes were histologically assessable after primary grossing, mesenteric fat was routinely re-examined. No fat clearance techniques were routinely used during lymph node assessment.

Statistical analysis

Descriptive analyses were carried out on the clinical and pathological characteristics of patients enrolled in the study. MSI/MSS subgroups were defined by differing clinicopathological characteristics of the tumours and tested for heterogeneity with respect to whether lymph node retrieval was adequate (≥12 assessable nodes retrieved) (I) and the total number of lymph nodes (II).

P values were determined using either (I) chi-square test (if one cell had <5 observations in the 2 × 2 table, Fisher’s exact test was used) or (II) non-parametric two-sample median test (Wilcoxon rank-sum test, two-sided p value).

All analyses were performed with SAS, software version 9.2 (SAS Institute, Cary, NC). Tests for statistical significance were two-sided and defined by p < 0.05.

Overview of studies reporting an association between MSI status and lymph node count

Studies reporting on the association of MSI status and lymph node count in CRC were identified by PubMed search up until Jan. 15, 2017 using “lymph node yield/count/harvest/number,” “MSI,” “colon cancer” and “colorectal cancer” as keywords. Manuscripts providing data concerning MSI status, the median lymph node count, mean lymph node count and/or the frequency of adequate (≥12) lymph node count were chosen for comparison.

Results

Lymph node count and MSI/MSS status

In the complete cohort of 1196 CRCs, a median of 16 and a mean of 17 lymph nodes were found. An adequate retrieval of at least 12 lymph nodes was seen in 81% of cases (Table 2).

Table 2 Association between lymph node count and microsatellite status according to clinicopathological characteristics
Full size table

The study cohort comprised 113 MSI and 1083 MSS tumours. Adequate lymph node count was reported for 91% of the MSI and 80% of the MSS cancers, respectively (p = 0.004). However, after stratification according to UICC stage, an association between adequate lymph node count and MSI status was only observed for UICC stage I (pT1/pT2, pN0, pM0) cancers (p 0.004).

When restricted to the proximal colon, no statistically significant difference in the frequency of adequate lymph node retrieval was observed between MSI and MSS cancers (89 versus 87%, p = 0.576). Further subclassification of proximal colon cancers by UICC stage showed a higher frequency of adequate (≥12) lymph node retrieval in MSI tumours only among stage I CRCs (p: 0.010).

Moreover, significant differences in the median and mean lymph node counts between MSI and MSS were restricted to UICC stages I-II and to T categories I-III. Within the proximal colon, MSI was, in general, associated with higher lymph node count (p 0.009). Subclassification based on UICC stage showed this association to be limited to stage I tumours.

Overview of studies reporting an association between MSI status and lymph node count

Our PubMed search found five manuscripts meeting the stated requirements (Table 3). Two publications by Ogino et al. [19, 20] were not used for comparison, in spite of giving data on lymph node number and MSI status, since the association with lymph node count was not directly addressed. The included investigations originated from four different countries and were published between 2009 and 2013, representing geographical diversity and scientific discourse. Except for the manuscript by Eveno et al. [13], a prospective study, the investigations were retrospective. However, one study population included a prospective patient series enrolled in a sentinel node project.

Table 3 Comparison with previous articles on the association of lymph node retrieval and MSI status
Full size table

Our study cohort was comprised of 1196 patients including 113 with microsatellite unstable CRC (9.4%). The largest previous study, conducted by Belt et al. [12], included more than 300 cases of CRCs. While Eveno et al. [13] and MacQuarrie et al. [21] detected MSI in 12–13% of their cohorts, the ratio of MSI cases reported by Belt et al. (19.6%), Søreide et al. [22] (27%) and Berg et al. [11] (33%) were much higher.

The minimum adequate lymph node count of 12, as required by the current AJCC/UICC staging manuals [23], was achieved in over 80% of our study cohort. Since the first publication by Søreide et al. [22] in 2009, the frequency of adequate (≥12) lymph node retrieval has risen from 36 to 67% in the 2013 publication by Berg et al. [11]. In the Dutch study by Belt et al. [12], a minimum lymph node count of more than 10 was achieved in 40% of cases. The other studies did not give data on minimum lymph node count compliance.

Discussion

A series of studies have shown that more lymph nodes are obtained in resected CRC specimens from the proximal than from the distal colon [5, 7, 24, 25]. Although the association between proximal tumour location and underlying MSI is well known [26, 27], some authors still argue against an explanation for differing lymph node count based strictly on anatomy. These reports link an adequate lymph node count to the influence of microsatellite instability [11,12,13].

In order to study the role of MSI on adequate as well as total lymph node count, we examined data from 1196 CRC patients (UICC stages I-IV), who were diagnosed between 2003 and 2007, following current standards for lymph node retrieval and assessment in pathology. A 12 lymph node minimum was used as the cut-off for the distinction between cancers with adequate and inadequate lymph node retrieval.

As in previous reports, our data showed a statistically significant difference (p = 0.004) in the number of lymph nodes retrieved in CRC with MSI versus MSS tumours, with an adequate lymph node retrieval of 91 and 80% for MSI and MSS cancers, respectively. However, a location-specific analysis of lymph node count in the proximal colon, where most MSI tumours (84% in this study) are found, showed no statistically significant difference (p = 0.576) in adequate lymph node count between MSI and MSS colon carcinoma. A further comparison of lymph node count in the distal colon and rectum lacked statistical significance due to low prevalence of MSI tumours in those locations.

In a detailed assessment, an association between MSI and adequate lymph node count was found for stage I cancers (p = 0.004). In a further subclassification of proximal colon cancers by UICC stage, this association was also restricted to stage I cancers (p 0.010). The frequency of adequate lymph node count was similar in MSI and MSS cancers of the proximal colon among UICC stages II-IV. Thus, in contrast to previous reports, we saw a significant association between MSI and adequate (AJCC/UICC minimum of 12 assessable lymph nodes) lymph node count only in UICC stage I, proximal cancers.

In order to study the association between MSI and total lymph node count, independently from whether the total was adequate, we also compared median and mean lymph node retrieval of MSI/MSS colorectal cancers. Here, we found a significantly higher lymph node count in MSI cancers of UICC stages I-II as well as T categories I-III. The difference in lymph node count, however, was again restricted to UICC stage I (pT1/T2, pN0) when analysis was restricted to the proximal colon.

The influence of MSI on higher lymph node count in the early stage of colorectal tumour development suggests underlying mechanisms of tumour biology. It is possible that tumour immunogenicity of early-stage MSI-H cancers leads to an increased immune reaction and enlargement of the lymph nodes, facilitating detection. This concept of tumour immunogenicity seems consistent with the characteristic “Crohn’s-like” lymphoid reaction accompanying CRC with microsatellite instability. The generation of frame shift mutations in coding microsatellites (coding MSI) leads to the production of neo-peptides in MSI-H cancers which are presented at the cell surface via MHC molecules [28]. This is believed to stimulate the immune system and is a possible explanation for the remarkably strong lymphocytic infiltration which is a characteristic feature of this cancer type [29].

The absence of an association between MSI and higher lymph node count in UICC stages II-IV could indicate a decreasing relevance of this MSI-related, immunogenic effect on the number of retrieved lymph nodes. In higher UICC stages, tumour-induced tissue destruction may cause a greater inflammatory response. This would also explain the higher number of lymph nodes found in UICC stages II-IV compared to stage I and indicate that the influence of MSI on lymph node count is overshadowed in more deeply invasive cancers.

Our results are in line with the findings in stage III CRC reported by MacQuarrie et al., which indicated no significant relationship between MSI-H status and a higher number of retrieved lymph nodes [21]. The study was restricted to stage III CRC and thus could not assess the influence of MSI-H on earlier stages, where we have identified an association. Moreover, a recent publication by Rössler et al., which investigated the relationship between lymph node size and lymph node involvement, supports our findings. Similar to the results of our study, the mean size of lymph nodes in MSI cancers was reported as being significantly larger than in MSS cancers. However, when the analysis was restricted to proximal tumours, microsatellite instability was not associated with lymph node size [30].

Søreide et al. and Berg et al. found, in their studies [11, 22], MSI-H to be associated with a higher frequency of adequate lymph node retrieval [11, 22]. However, in the whole study population, this was achieved only in 36% [22] and in 67% [11] of cases, respectively, which does not comply with current quality standards. Moreover, both Norwegian studies report a very high percentage of MSI in CRC (27% [22] and 33% [11]): more than double the average reported frequency of MSI in CRC [31]. While geographical/epidemiological differences in molecular biology seem unlikely to be responsible, a sample selection bias due to the ex vivo sentinel lymph node mapping performed in this prospective sentinel lymph node study seems possible.

Another investigation by Eveno et al. also found an association between MSI and lymph node count in the examination of 135 patients with CRC stage I and II [13]. However, the strength of this conclusion is limited by the small number of MSI tumours in this prospective study (n = 11). More CRC patients (n = 332) with a higher number of microsatellite unstable tumours (n = 65) were included in the study conducted by Belt et al., which also saw an association between MSI phenotype and a greater frequency of “high” (defined in the study as >10 retrieved nodes) lymph node count [12]. Interestingly, this relationship was strongest in stage III tumours, contrasting to previous studies. However, the results are limited by the generally low lymph node counts in the investigation. In only 40.1% of the cases was a “high” lymph node count reported.

Reviewing the presented data, we found no significant association between MSI and adequate lymph node count (minimum of 12 assessable lymph nodes retrieved) in CRC, in contrast to previous reports by different groups. However, MSI seems to lead to a higher mean and median lymph node count in early UICC stages. The discrepancy between our results and previous studies is possibly due to higher lymph node counts. In comparison to the aforementioned studies, the 12 lymph node minimum for adequacy was achieved in over 80% of the total study cohort while in previous analyses, this ranged from 36 to 67%. This is also represented by a significantly higher median and mean number of retrieved lymph nodes.

In conclusion, our data refutes the claim that MSI-driven molecular mechanisms are the determining factor in lymph node retrieval in CRC. Based on improved lymph node retrieval, we saw differences in lymph node count between the proximal and distal colon, while the impact of MSI on the median and mean lymph node count was restricted to early UICC stages. Moreover, a location-specific analysis detected higher mean lymph node count only in stage I (pT1, pT2) microsatellite unstable CRC of the proximal colon. Thus, our data shows that the observed associations between MSI-H cancers and lymph node count are predominantly attributable to their preferential location in the proximal colon, where the number of lymph nodes is generally higher than in the distal colon and the rectum. Our results further suggest that the immunogenicity of MSI-H has an impact on the median and mean number of retrievable lymph nodes in early-stage CRC, where the immunologic response due to tissue destruction is low enough to not overshadow this effect.