The clinicopathologic significance of extranodal tumor extension in locally advanced (pT3) colorectal adenocarcinoma and its association with the loss of E-cadherin expression

Abstract

We investigated the clinicopathologic significance of extranodal tumor extension (ENTE) in locally advanced and prognostically inhomogeneous pT3 (pathologic T3) colorectal adenocarcinomas with regional lymph node metastasis. ENTE is defined as an interruption of the nodal capsule by tumor cells with extranodal growth. ENTE was observed in 46.3% of pT3 colorectal adenocarcinomas and was significantly associated with vascular invasion (P = 0.037, chi-square test), tumor deposit (P = 0.004, chi-square test) and high pN (pathologic N) stage (P = 0.002, chi-square test). An immunohistochemical study revealed that the loss of E-cadherin was significantly associated with ENTE (OR, 2.265; 95% CI, 1.008-5.086; P = 0.048). Kaplan-Meier survival analyses showed a significant difference between ENTE (+) and ENTE (-) groups for both cancer-specific survival (CSS) and recurrence-free survival (RFS) (P = 0.004 and P = 0.020, respectively, log-rank test). In the pN1a (single lymph node metastasis) subgroup, CSS and RFS were significantly shorter in patients with ENTE (P = 0.001 and P < 0.001, respectively, log-rank test). Comparing CSS and RFS according to pN stages and ENTE status, the survival curves of the pN1 group with ENTE were similar to those of the pN2 group without ENTE. ENTE is a useful prognostic factor for pT3 colorectal adenocarcinomas with regional lymph node metastasis, especially depending on the pN stages. The loss of E-cadherin expression may be an indicator of ENTE. Therefore, ENTE in colorectal adenocarcinoma should be considered in pN staging systems in the future.

Introduction

Colorectal cancer (CRC) is one of the most common cancers in the United States and South Korea [1,2]. The histopathologic stage at the diagnosis is a well-known prognostic factor and is used as an indicator for additional treatment [3]. The AJCC TNM staging classification is widely used to evaluate the patients’ prognosis and to determinate adjuvant therapy in CRC [4]. Although the majority of patients are classified with pT3 disease, in which the tumor invades beyond the proper muscle but without perforation of the visceral peritoneum and without invasion of adjacent organs, these patients constitute a heterogeneous population with an inhomogeneous prognosis [5]. This inhomogeneity complicates the prediction of patient survival and the determination of adjuvant treatment [6].

The number of metastatic lymph nodes is one of the most important prognostic factors in all cancers [3]. In the 8th colorectal TNM classification system, the pN (pathologic N) category is subdivided into pN1a, pN1b, pN2a, and pN2b according to the number of metastatic lymph nodes, and also pN1c, which indicates the additional presence of tumor deposits in the pericolorectal soft tissue [7]. CRC patients with regional lymph node metastasis show heterogeneous survival rates. Nevertheless, little attention has been paid to the histopathologic features of metastatic lymph nodes [8].

Recently, the presence of extranodal tumor extension (ENTE) in metastatic lymph nodes has been reported as a poor prognostic factor in various human malignancies [9-12]. Several studies have demonstrated the prognostic significance of ENTE in CRC [13-19]. The ability of cancer cells to invade through the fibrous nodal capsule in an immunologically hostile environment reflects the invasiveness of the primary tumor [9]. One of the well-known proteins associated with the invasiveness of cancer cells is the epithelial cell adhesion molecule, E-cadherin, which also acts as a poor prognostic factor in CRC [13].

In this study, we investigated the clinicopathologic significance of ENTE in locally advanced and prognostically inhomogeneous pT3 colorectal adenocarcinomas with regional lymph node metastasis and further evaluated its association with E-cadherin expression.

Materials and methods

Patients and specimens

From January 2005 to December 2010, 421 patients underwent curative resection for colorectal adenocarcinoma at Hanyang University Hospital. Patients who had received neoadjuvant chemotherapy or radiotherapy and those with fewer than 12 lymph nodes or recurrent colorectal cancer were excluded from this study. Among the 421 patients, 261 were diagnosed with pT3 CRC, of whom 134 showed regional lymph node metastasis and were included in this study. We reviewed medical records to determine the clinical characteristics of age, gender, follow-up interval, adjuvant treatment, survival, and recurrence status. Two pathologists (Kim H and Paik SS) reviewed the pathology reports and hematoxylin and eosin-stained slides to define the clinicopathologic characteristics and confirm the diagnosis. TNM staging and other pathologic factors were determined according to the 8th AJCC TNM staging classification and a protocol for the examination of specimens from patients with primary carcinoma of the colon and rectum [4,14]. Pathologic data included gross type, tumor size, histologic grade, vascular invasion, perineural invasion, tumor budding, tumor deposit, pN stage, the state of distant metastasis, and microsatellite instability (MSI) status. ENTE was defined as an extension of cancer cells through the lymph node capsule into the perinodal soft tissue [8]. The representative microscopic pictures of ENTE are shown in Figure 1. All the slides of the resected lymph nodes were reviewed to determine the status of ENTE. This study was approved by the Institutional Review Board of Hanyang University Hospital (HYUH 2015-05-023).

Figure 1.

Representative microscopic pictures of lymph node metastasis without extranodal tumor extension (A) and with extranodal tumor extension (B) (Hematoxylin-eosin stain, original magnification, × 50).

Microsatellite instability status

The state of microsatellite instability (MSI) was assessed by using a panel of five National Cancer Institute workshop-recommended consensus microsatellite markers (BAT25, BAT26, D2S123, D17S250, and D5S346) [15]. MSI-High was defined as cases showing a shifting of microsatellites at two or more markers. A shifting of microsatellites at one marker was defined as MSI-Low. Non-MSI-High includes microsatellite stable and MSI-Low.

Tissue microarray construction and immunohistochemical staining

The tissue microarrays (TMAs) were constructed using a manual tissue microarrayer (Quick Ray Set, Unitama, Seoul, South Korea) as previously described [16]. The hematoxylin and eosin-stained slides were reviewed by two pathologists (Kim H. and Paik SS) to mark the representative area of the tumor. We punched a tissue cylinder with a 2 mm diameter from a previously marked lesion of each donor block and transferred it to the recipient block (Quick Ray Set, Unitama, Seoul, South Korea). Each tissue microarray was made up of 5 × 10 samples. E-cadherin expression was determined by immunohistochemical staining. We used a mouse monoclonal E-cadherin antibody (clone 36B5, Novocastra, Newcastle, UK) at 1:100 dilution. 4 μm sections were obtained from a tissue microarray block using a Leica microtome and transferred to adhesive coated slides and deparaffinized. The staining was performed using the Bond-max automated immunohistochemistry (IHC)/in situ hybridization stainer (Leica Biosystems, Nussloch, Germany) according to the manufacturer’s protocol. Before staining, the heat-induced epitope retrieval was performed in a Bond epitope retrieval solution. Endogenous peroxidase activity was blocked using 0.3% hydrogen peroxide. The primary antibody was incubated for 30 minutes and the slides were incubated with a postprimary reagent for 15 minutes at room temperature. The reactions were developed using a Bond polymer refine detection kit and followed by color development with 3,3-diaminobenzidine tetrahydrochloride as a chromogen.

Interpretation of immunohistochemical staining

The E-cadherin expressions were evaluated semiquantitatively by two independent pathologists (Kim H. and Paik SS) who were blinded to the patients’ information. We categorized the membranous E-cadherin expressions in terms of both staining intensity and extent. Staining intensity was graded as negative (0), weak (1), moderate (2), and strong (3), and the staining extent was graded as 0% (0), 1-10% (1), 11-50% (2), 51-80% (3), and 81-100% (4). The final staining score was calculated by multiplying the intensity and the extent. A cutoff value of 2 was adopted as previously described [17,18] and the cases were finally classified as either E-cadherin loss (score 0-1) or intact (score 2-12). The representative photomicrographs of E-cadherin expression are shown in Figure 2.

Figure 2.

Representative microscopic pictures of E-cadherin expression in colorectal adenocarcinoma. Loss of expression (A) and intact membranous expression (B) (Original magnification, × 200).

Statistical analysis

The statistical analysis was performed using SPSS software version 21 (IBM Corp., Armonk, NY, USA). Chi-square and Mann-Whitney U tests were used to examine the associations among ENTE and the clinicopathologic factors of age, gender, tumor size, macroscopic growth pattern, histologic grade, vascular invasion, perineural invasion, tumor budding, tumor deposit, pN stage, and the status of distant metastasis. The association of ENTE with E-cadherin was assessed by multivariate logistic regression analyses, controlling for covariates including vascular invasion, perineural invasion, tumor budding and tumor deposit. Cancer-specific survival (CSS) was defined as the time interval between the date of surgery and the date of death due to CRC. Recurrence-free survival (RFS) was defined as the time interval between the date of surgery and the date of disease progression, relapse, or death from CRC. The Kaplan-Meier method with a log-rank test and a multivariate Cox proportional hazard regression model were used for the survival analyses. P < 0.05 was considered statistically significant.

Results

Baseline characteristics of pT3 colorectal adenocarcinomas with regional lymph node metastasis

The clinicopathologic characteristics are summarized in Table 1. The median follow-up was 103 (± 50.1) months, the mean age at surgery was 64.1 (± 10.5) years, and 80 (59.7%) patients were men, with a male: female ratio of 1.48:1. The pathological evaluation revealed that 4 tumors (3.0%) were histologic grade 1, 53 (39.6%) were grade 2, 67 (50.0%) were grade 3, and 10 (7.5%) were grade 4. Concerning the pN stage, 32 (23.9%) were pN1a, 33 (24.6%) were pN1b, 34 (25.4%) were pN2a, and 35 (26.1%) were pN2b. Distant metastasis was identified in 10 (7.5%) patients. Among the 134 cases, 127 (94.8%) were non-MSI-High and 7 (5.2%) were MSI-High. For treatment, 39 (29.1%) patients received curative surgical resection only, and 95 (70.9%) patients received curative surgical resection and postoperative chemotherapy (83 cases) or combined postoperative chemo-radiotherapy (12 cases).

Table 1.

Baseline characteristics of pT3 colorectal adenocarcinoma patients with regional lymph node metastasis (n = 134)

Factors	Value (%)
Number of patients	134 (100%)
Mean age at surgery (years)	64.1 (± 10.5)
Median survival (months)	103.0 (± 50.1)
Sex
Female	54 (40.3%)
Male	80 (59.7%)
Tumor location
Cecum & ascending	24 (17.9%)
Transverse	5 (3.7%)
Descending & rectosigmoid	105 (78.4%)
Histologic grade
Grade 1	4 (3.0%)
Grade 2	53 (39.6%)
Grade 3	67 (50.0%)
Grade 4	10 (7.5%)
pN stage
1a	32 (23.9%)
1b	33 (24.6%)
2a	34 (25.4%)
2b	35 (26.1%)
pM stage
0	124 (92.5%)
1a	8 (6.0%)
1b	2 (1.5%)
Microsatellite instability status
Non MSI-High (MSS, MSI-Low)	127 (94.8%)
MSI-High	7 (5.2%)
Treatment
Surgery	39 (29.1%)
Surgery + Adjuvant therapy	95 (70.9%)
Mean number of removed LNs*	31.1 (± 15.9)
Mean number of metastatic LNs*	5.9 (± 6.8)

^*LNs, lymph nodes.

The correlations between ENTE status and the clinicopathologic features

Of the 134 patients, 62 (46.3%) cases showed ENTE and 72 (53.7%) showed no ENTE in the metastatic lymph node(s). We evaluated the correlations between ENTE status and the clinicopathologic factors to assess the significance of ENTE in pT3 colorectal adenocarcinomas with regional lymph node metastasis. The results are summarized in Table 2. ENTE was observed more frequently in tumors with vascular invasion (P = 0.037, chi-square test), tumor deposit (P = 0.004, chi-square test), and high pN stage (P = 0.002, chi-square test). However, there was no correlation with age, gender, tumor size, gross type, histologic grade, perineural invasion, tumor budding, distant metastasis, or MSI status.

Table 2.

Correlations between ENTE and clinicopathologic factors in pT3 colorectal adenocarcinomas with regional lymph node metastasis (n = 134)

Factors	n	ENTE		P-value
		Negative (%) (n = 72)	Positive (%) (n = 62)
Age (year)				0.728†
Mean (± SD)	134	64.49 (± 9.49)	63.68 (± 11.56)
Gender				0.483
Male	80	41 (51.2%)	39 (48.8%)
Female	54	31 (57.4%)	23 (42.6%)
Tumor size				0.872†
Mean (± SD)	134	5.57 (± 1.97)	5.52 (± 1.76)
Gross type				0.375
Polypoid	4	3 (75.0%)	1 (25.0%)
Ulcerofungating	53	31 (58.5%)	22 (41.5%)
Ulceroinfiltrative	77	38 (49.4%)	39 (50.6%)
Histologic grade				0.237
Grade 1 & 2	57	34 (59.6%)	23 (40.4%)
Grade 3 & 4	77	38 (49.4%)	39 (50.6%)
Vascular invasion				0.037
Absent	98	58 (59.2%)	40 (40.8%)
Present	36	14 (38.9%)	22 (61.1%)
Perineural invasion				0.123
Absent	39	25 (64.1%)	14 (35.9%)
Present	95	47 (49.5%)	48 (50.5%)
Tumor budding				0.150
Low grade	63	38 (60.3%)	25 (39.7%)
High grade	71	34 (47.9%)	37 (52.1%)
Tumor deposit				0.004
Absent	96	59 (61.5%)	37 (38.5%)
Present	38	13 (34.2%)	25 (65.8%)
pN stage				0.002
pN1a & 1b	65	44 (67.7%)	21 (32.3%)
pN2a & 2b	69	28 (40.6%)	41 (59.4%)
Distant metastasis				0.806
Absent	124	67 (54.0%)	57 (46.0%)
Present	10	5 (50.0%)	5 (50.0%)
MSI status				0.553
Non MSI-H	127	69 (54.3%)	58 (45.7%)
MSI-H	7	3 (42.9%)	4 (57.1%)

SD: standard deviation;

^†Mann Whitney U test.

The correlation between ENTE status and E-cadherin expression

In multivariate logistic regression analyses, the loss of E-cadherin was significantly associated with ENTE (OR, 2.265; 95% CI, 1.008-5.086; P = 0.048, Table 3), independent of other clinicopathologic factors including vascular invasion, perineural invasion, tumor budding, and tumor deposit. In addition to the loss of E-cadherin expression, vascular invasion was also significantly associated with ENTE.

Table 3.

Multivariate logistic regression analysis to assess the association between ENTE and the loss of E-cadherin expression

Predictor for ENTE	Multivariate OR (95% CI)	P-value
E-cadherin expression (intact vs. loss)	2.265 (1.008-5.086)	0.048
Covariates
Vascular invasion (absent vs. present)	2.475 (1.002-6.111)	0.049
Perineural invasion (absent vs. present)	1.059 (0.414-2.708)	0.904
Tumor budding (low grade vs. high grade)	1.612 (0.726-3.581)	0.241
Tumor deposit (absent vs. present)	1.818 (0.765-4.320)	0.176

Influence of ENTE status on patient survival

There was a significant difference between the ENTE (+) and ENTE (-) groups in both CSS and RFS (P = 0.004 and P = 0.020, respectively, log-rank test) (Figure 3A and 3B). Univariate survival analyses revealed that ENTE (P = 0.006 and P = 0.024, respectively), vascular invasion (P < 0.001 and P = 0.001, respectively), tumor deposit (P = 0.001 and P = 0.007, respectively) and high pN stage (P = 0.007 and P = 0.016, respectively) were associated with CSS and RFS. In the multivariate analyses, vascular invasion (P = 0.001), tumor deposit (P = 0.015), and high pN stage (P = 0.037) were independent poor prognostic factors in CSS, and vascular invasion (P = 0.005) was an independent poor prognostic factor in RFS (Table 4).

Figure 3.

Survival analyses according to the ENTE status. There was a significant difference between pT3 with ENTE and pT3 without ENTE groups in both CSS (A) and RFS (B). In the subgroup survival analyses according to pN stages and ENTE status, there were significant differences between ENTE (+) and ENTE (-) groups in both CSS (C) and RFS (D) in the pN1a subgroup analyses. In the survival analyses according to the number of metastatic lymph nodes, there were significant survival differences between four groups and the survival curves of the pN1 group with ENTE were similar to those of pN2 group without ENTE in both CSS (E) and RFS (F).

Table 4.

Univariate and multivariate analyses in pT3 colorectal adenocarcinomas with regional lymph node metastasis (n = 134)

Variables	Univariate analysis		Multivariate analysis
	HR (95% CI)	P-value	HR (95% CI)	P-value
Cancer-specific survival
ENTE (absent vs. present)	2.130 (1.240-3.658)	0.006	1.461 (0.827-2.579)	0.191
Vascular invasion (absent vs. present)	2.783 (1.622-4.773)	< 0.001	2.503 (1.446-4.332)	0.001
Tumor deposit (absent vs. present)	2.450 (1.431-4.194)	0.001	1.980 (1.142-3.431)	0.015
pN stage (N1a & N1b vs. N2a & N2b)	2.318 (1.225-3.729)	0.007	1.840 (1.039-3.260)	0.037
Recurrence-free survival
ENTE (absent vs. present)	2.019 (1.099-3.711)	0.024	1.396 (0.734-2.654)	0.309
Vascular invasion (absent vs. present)	2.307 (1.256-4.237)	0.001	2.430 (1.309-4.513)	0.005
Tumor deposit (absent vs. present)	2.579 (1.339-4.967)	0.007	1.863 (1.996-3.486)	0.052
pN stage (N1a & N1b vs. N2a & N2b)	2.170 (1.157-4.071)	0.016	1.886 (0.992-3.587)	0.053

HR, hazard ratio; CI, confidence interval.

Comparison of survival rates according to ENTE status and pN stage subgroups

To investigate the prognostic significance of ENTE according to pN stages, subgroup analyses were performed. In the pN1a subgroup (single lymph node metastasis), both CSS and RFS were significantly shorter in cases with ENTE (P = 0.001 and P < 0.001, log-rank test) (Figure 3C and 3D). According to the number of metastatic lymph nodes (pN1: three or less regional lymph nodes are positive; pN2: four or more regional lymph nodes are positive) and ENTE status, we divided all patients into four groups: pN1 without ENTE; pN1 with ENTE; pN2 without ENTE; pN2 with ENTE. There were significant survival differences between four groups in both CSS and RFS (P = 0.002 and P = 0.024, respectively, log-rank test) and N2 with the ENTE group showed the worst survival outcome. Interestingly, the survival curves of the pN1 group with ENTE were similar to those of the pN2 group without ENTE (Figure 3E and 3F).

Discussion

In this study, we demonstrated the prognostic role of ENTE in pT3 colorectal adenocarcinomas with regional lymph node metastasis. ENTE was observed more frequently in tumors with vascular invasion, tumor deposit, and high pN stage. There were significant differences between the ENTE (+) and ENTE (-) groups in both CSS and RFS. Significantly, ENTE was a prognostic indicator of poor CSS and RFS in the pN1a category. Interestingly, the survival curves of pN1 with the ENTE group were similar to those of pN2 without the ENTE group.

In various cancers, ENTE has been documented as a poor prognostic factor [9-12,19]. Wind et al. reported that gastrointestinal malignancies with ENTE showed a worse prognosis [9]. Their systematic review emphasized the importance of ENTE as an important prognostic factor and recommended that pathologists and clinicians should be aware of this histologic feature [9]. Moritani et al. examined the impact of ENTE on the prognosis of patients with papillary thyroid carcinoma. The distant recurrence rate was higher, and the 10-year disease-specific survival rate was significantly lower in patients with ENTE compared to those without ENTE [10]. Ahn et al. demonstrated that ENTE was a predictor of poor outcome for both RFS and CSS in a meta-analysis of bladder cancer [11]. Shigematsu et al. showed that ENTE at the sentinel lymph node was an independent predictor and a poor prognostic factor for breast cancer patients [12]. In the study by Alakus et al., ENTE was associated with higher pN stage, pM stage, UICC stage, and a poor survival rate in patients with gastric cancer [19].

Several studies have described the prognostic value of ENTE in CRC [8,20-23]. Veronese et al. showed that ENTE was associated with higher pT stage, higher histologic grade, an increased risk of mortality, and the recurrence of disease in a meta-analysis of 13 studies [8]. Kumuta et al. revealed that ENTE was a poor prognostic factor of colorectal adenocarcinoma. However, they didn’t reveal significant associations between ENTE and the clinicopathologic factors, and there was no statistically significant difference in recurrence or survival rate between the pN staging groups [20]. Yano et al. demonstrated that ENTE was a more important prognostic factor than the other factors such as pT stage, lymphovascular invasion, and the number of metastatic lymph nodes, especially in Duke stage C colorectal adenocarcinoma [21]. Puppa et al. focused on pericolonic tumor deposits and ENTE and revealed ENTE as a poor prognostic factor of borderline significance in stage III colorectal adenocarcinoma [22].

Recently, the prognostic role of ENTE was investigated in a stage subgroup of colorectal cancer patients [24-26]. Kim et al. revealed that ENTE was independently and significantly associated with poor disease-free survival (DFS) and overall survival (OS) rates for stage III colorectal cancer [17]. Lee et al. demonstrated that ENTE acted as a risk factor for recurrence and was associated with poor DFS in stage III colorectal cancer patients [24,25]. Yamano et al. assessed the prognostic significance of tumor deposits and ENTE in colorectal cancer patients with pT3-4, pN0-2, M0 stage. ENTE was associated with pT stages, pN stages, and the LN ratio. ENTE was a significant independent adverse prognostic factor for RFS in rectal cancer [26].

Herein, we demonstrated that both CSS and DFS were significantly associated with ENTE in pT3 colorectal adenocarcinomas with regional lymph node metastasis. Significantly, ENTE acts as a significant prognostic factor in the pN1a staging category; therefore, ENTE should be mentioned in the pathology report even if only one metastatic lymph node is identified. Furthermore, the survival curves of the pN1 group with extranodal tumor extension showed similar features to those of pN2 group without extranodal tumor extension. Our results suggest that ENTE status in pT3 colorectal adenocarcinomas with regional lymph node metastasis is an important prognostic factor, and ENTE status should be considered in pN staging system revision in the future.

E-cadherin is a transmembrane protein that facilitates the assembly of intercellular junctions which are necessary for epithelial cell adherence [27]. An in vitro study has shown that the downregulation of E-cadherin expression in colon cancer cells promotes tumor cell migration and invasion [28]. These results support the concept of the epithelial-mesenchymal transition and may provide evidence for the association of the loss of E-cadherin expression with ENTE of the CRC. Our results showed that the loss of E-cadherin expression was closely associated with ENTE in pT3 colorectal adenocarcinomas with regional lymph node metastasis.

In conclusion, ENTE is an important prognostic factor for pT3 colorectal adenocarcinomas with regional lymph node metastasis, and the loss of E-cadherin expression is an indicator of ENTE. The ENTE status should be considered in the pN staging system and strictly mentioned in the pathology report.

Disclosure of conflict of interest

None.