Questioning the staging of tumor deposits of colorectal cancer in the eighth edition of the TNM classification: validation by prognosis

Abstract

Tumor deposits (TD) of colorectal cancer (CRC) in the 8^th edition of TNM classification (TNM 8^th) were staged as N1c, but the number of TDs was ignored. The aim of this study was to analyze the association of TD with CRC and verify the rationale for TD staging in the TNM 8^th. A total of 517 patients with CRC, surgically treated from Aug. 2013 to Dec. 2017, were retrospectively reviewed. Univariate and multivariate analyses were used to observe the correlations between clinicopathologic features and TD. The reasonability of TD staging in TNM 8^th was validated by prognostic analysis. The occurrence of TD in CRC was 11.2% (154/1375). Multivariate analysis indicated that T stage (P<0.001, OR = 2.026 and 5.380, 95% CI: 0.917-4.474 and 2.229-12.981 for T3 and T4 respectively) and TNM stage (P<0.001, OR = 9.051 and 16.305, 95% CI: 2.055-39.857 and 3.780-70.323 for TNM II and TNM III respectively) were independent risk factors for TD status. Only degree of differentiation (P = 0.020, OR = 0.197, 95% CI: 0.050-0.774 for poorly differentiated) was an independent risk factor for number of TDs. Survival analysis showed that patients with TD exhibited a significantly worse prognosis compared to patients without TD (P<0.001), and a significant prognostic difference was found among groups TD = 1, TD = 2/3 and TD≥4 (all P<0.05). Patients in T3-4aN1a/1b had a worse prognosis than patients in T3-4aN1c did, although both groups were classified as TNM IIIB (P = 0.022). TD was an adverse indicator in CRC, and the varying number of TDs represented a classified prognostic factor in CRC. TD staging in the TNM 8^th might not be reasonable as it ignores the status and the number of TDs.

Introduction

Colorectal cancer (CRC) is one of the most common cancers with high morbidity and mortality [1,2]. The International Union Against Cancer (UICC)/American Joint Committee on Cancer (AJCC) TNM classification is used worldwide to evaluate the characteristics and prognosis of malignant tumors, and is considered the best reference in clinical decision-making.

Contents of TNM classifications such as tumor deposits (TD) are constantly being revised. TD were first reported by Gabriel in 1935 [3], but scholars generally still remained unclear about TD until 60 years later when TD were brought into TNM 5^th edition. In this edition, TD were divided into T stage regarded as continuous spread of primary tumors or N stage defined as a lymph node metastasis (LNM) based on whether its diameter up to 3 mm [4]. As support was lacking from clinical evidence and substantial data, this definition was amended in the TNM 6^th which abandoned the “3 mm rule” and applied contour features of tumor foci, which was called the “contour rule”, as a basis to classify TD: tumor foci in pericolic or perirectal fat with a smooth contour of a lymph node but without evidence of LNM was regarded as LNM and classified as an N stage; otherwise it was staged in T stage (T3), regarded as continuous spread of primary tumor [5,6]. The TNM 7^th edition, which renewed this definition of TD, treated the aforementioned tumor foci with a smooth contour as a lymph node that was replaced by tumor cells entirely, and regarded only the tumor foci with irregular contours and without residual lymph nodes, vascular and neural tissues as TD [7]. Meanwhile, a major change in TD staging was stated in the TNM 7^th edition, which classified patients with TD but without LNM as N1c. TNM 7^th also emphasized TD’s adverse influence on prognosis for patients with CRC, consistent with current research [8-10]. TNM 8^th had no clear distinction on the definition and the staging of TD compared with TNM 7^th [11]. With constantupdating of the TNM classification, there is not yet an unified consensus on TD. Although this has been revised three times, the definition ofstaging of TD in the TNM classification is still controversial. This study aims to explore the associations between TD and clinicopathologic features and prognosis for CRC, and we also validate the rationale of the TD staging in the TNM 7^th/8^th through prognosis.

Materials and methods

Patients

All the clinical information from patients with CRC who had underwent curative colorectalectomy from August 2013 to December 2017 at the Department of gastrointestinal surgery, Affiliated Hospital of Southwest Medical University were retrospectively investigated. The inclusion criteria were patients: (I) with complete medical records; (II) who had accepted a whole intravenous chemotherapy cycle. Excluded from this study were patients: (I) who had confirmed metastasis by auxiliary examination; (II) with other obvious diseases influencing prognosis; (III) who had accepted neoadjuvant therapy; (IV) who did not die from CRC; (V) who had multiple primary cancers or who did not have primary CRC; or (VI) who were lost to follow-up. Patients in this study were staged by using the TNM 8^th. This study was conducted in accordance with the declaration of Helsinki. This study was conducted with approval from the Ethics Committee of the Affiliated Hospital of Southwest Medical University, China. Written informed consent was obtained from all participants.

Data collection

The following information was acquired by electronic medical records: gender, age, levels of carcinoembryonic antigen (CEA), levels of carbohydrate antigen 199 (CA199), operation date, tumor location, tumor size, and postoperative pathology reports. Survival status (survival or death), date, and cause of death (if applicable) were acquired by follow-up. Follow-up was done from patients’ surgery date to December 31th, 2018. Follow-up period ranged from 12 months to 64 months.

Definition of TD

According to the definition of TD in the TNM 8^th, TD were considered as isolated tumor foci in pericolic or perirectal fat, discontinuous spread, and without recognizable lymph node, vascular, or nervous tissues of primary tumor (Figure 1).

Figure 1.

Two TDs in pericolic fat tissue of colonic mucinous carcinoma. TD was defined as tumor foci of CRC located in pericolic or perirectal fat discontinuous with primary tumor. A. TD in pericolic fat tissue, 20×; B. TD in pericolic fat tissue, 40×. TD, tumor deposits.

Statistical analysis

The clinicopathologic features in the tumor deposits positive (TDP) group and tumor deposits negative (TDN) group were analyzed by Chi-square test, and groups of different number of TDs were analyzed by means of Kruskal Wallis test [12] or Wilcoxon rank sum test [13]. All the clinicopathologic features with P<0.1 were entered into logistic regression model for multivariate analysis to determine independent risk factors. Survival analysis was performed by Kaplan-Meier method and Log-rank test. Statistical analyses were conducted by SPSS 23.0 software (SPSS, Chicago, IL). P<0.05 was regarded as significant.

Results

Patient characteristics

A total of 1375 patients’ electronic medical records were retrieved, in which TD was found in 154 patients (11.2%), confirmed by postoperative pathology. In 517 out of 1357 patients that were enrolled to our study after being filtered by the inclusion and exclusion criteria, there were 91 cases (17.6%) in the TDP group and 426 cases (82.3%) in TDN group. Clinicopathologic features in different status and number of TDs are shown in Table 1.

Table 1.

Clinicopathologic features in different status and different number of TDs

Clinicopathologic features	All patients				Patients with TD
	TDN	TDP	x2	P	TD = 1	TD = 2	TD = 3	TD≥5	x2 or Z	P
	n = 426	n = 91			n = 30	n = 27	n = 15	n = 19
Gender			2.666	0.102					-0.128	0.898
Male	264	48			17	13	7	11
Female	162	43			13	14	8	8
Age, years			0.578	0.447					-0.128	0.898
≤60	220	43			13	14	8	8
>60	206	48			17	13	7	11
Tumor location			0.467	0.494					-0.742	0.458
Colon	116	28			8	9	3	8
Rectum	310	63			22	18	12	11
Histological type			0.018	0.893					-0.102	0.919
Ulcer	306	66			24	15	13	14
Protruding	120	25			6	12	2	5
Pathological category			3.470	0.062					-1.675	0.094
Papillary or tubular adenocarcinoma	377	74			28	19	12	15
Mucinous or signet-ring cell carcinoma	49	17			2	8	3	4
Differentiation degree			7.808	0.020					16.968	<0.001
Well	147	24			11	8	1	4
Moderate	240	50			18	17	9	6
Poor	39	17			1	2	5	9
T stage*			37.116	<0.001					-2.317	0.021
2	136	9			21	19	8	7
3	227	46			9	8	7	12
4	63	36
N stage			16.316	<0.001					12.516	0.002
0	256	34			16	10	4	4
1	104	32			13	7	6	6
2	66	25			1	10	5	9
TNM stage*			26.471	<0.001					-2.457	0.014
I	102	2			16	10	4	4
II	154	32			14	17	11	15
III	170	57
CEA, 6 ng/ml			21.205	<0.001					-2.026	0.039
+	164	59			15	18	12	14
-	262	32			15	9	3	5
CA199, 37 u/ml			24.199	<0.001					-1.064	0.287
+	55	31			10	13	9	8
-	371	60			20	14	6	11
Vascular invasion			19.137	<0.001					-2.403	0.016
+	41	24			5	5	5	9
-	385	67			25	22	10	10
Neural invasion			9.522	0.002					-1.453	0.151
+	47	21			3	9	4	5
-	379	70			27	18	11	14
Tumor size, cm			1.507	0.220					-0.275	0.784
<5	261	62			20	18	11	13
≥5	165	29			10	9	4	6

^*Subcategories of clinicopathologic features in the group of patients with TDs were conflated.

TDN, tumor deposits negative; TDP, tumor deposits positive; CEA, carcinoembryonic antigen; CA199, carbohydrate antigen 199. Bold = P<0.05.

Of note was that some subcategories of clinicopathologic features in the TDP group were merged with other subcategories because their size was too small to show meaningful results. For instance, group TNM I stage (1 case, 0 cases, 1 case and 0 cases in groups TD = 1 to 4 respectively) was merged with TNM II stage as well as T2 stage and T3 stage. It was necessary for us to divide the TDP group into four subgroups according to the number of TDs (TD = 1, TD = 2, TD = 3 and TD≥4) for studying the relationships between the clinicopathologic features and the number of TDs. The reasons for that were: (I) small size of group TD≥4 (8 cases in TD = 4, 5 cases in TD = 5, only 2 cases in TD = 6) on account of a small size of TDP group; (II) censored data in all subgroups of TDP group. It would be meaningless to attempt statistics if more subgroups of TD≥4 group were split.

TD and clinicopathologic features

A univariate analysis indicated that there were significant correlations between the occurrence of TDs and differentiation degree as well as T stage, N stage, TNM stage, CEA levels, CA199 levels, vascular invasion, and neural invasion (P = 0.020 for differentiation degree, P = 0.001 for neural invasion, and P<0.001 for all others; Table 1). T stage (P<0.001, OR = 2.026 and 5.380, 95% CI: 0.917-4.474 and 2.229-12.981 for T3 and T4 respectively) and TNM stage (P<0.001, OR = 9.051 and 16.305, 95% CI: 2.055-39.857 and 3.780-70.323 for TNM II and TNM III respectively) were independent risk factors for the occurrence of TD (Table 2). In our study, however, CA199 level was found to be a protective factor (P = 0.012, OR = 0.447, 95% CI: 1.183-3.880) for the occurrence of TD (Table 2), and this opposite result needs to be verified with large sample in the future. Meanwhile, differentiation degree, T stage, N stage, TNM stage, CEA levels and vascular invasion but CA199 levels and neural invasion were significantly associated to the number of TDs (P<0.01 for differentiation degree, P = 0.021 for T stage, P = 0.002 for N stage, P = 0.014 for TNM stage, P = 0.039 for CEA levels, P = 0.016 for vascular invasion, and P>0.05 for both CA199 and neural invasion; Table 1). Of those, only differentiation degree (P = 0.020, OR = 0.197, 95% CI: 0.050-0.774 for poorly differentiated) was an independent risk factor for the number of TDs (Table 2). No significant difference was found among gender, age, tumor location, histological type or tumor size with the status and the number of TDs (all P>0.05 in Table 1).

Table 2.

Multivariate logistic regression analyses between clinicopathologic features and different status and different number of TDs

Different status of TDs				Different number of TDs
Variable	OR	OR (95% CI)	P	Variable	OR	OR (95% CI)	P
Differentiation degree
well*	1	-	0.994	well*	1	-	-
moderate	0.981	0.544-1.769	0.949	moderate	0.914	0.353-2.373	0.855
poor	0.951	0.374-2.421	0.917	poor	0.197	0.050-0.774	0.020
T stage
T2*	1	-	0.000	T2-3*	1	-	-
T3	2.026	0.917-4.474	0.081	T4	1.699	0.710-4.063	0.234
T4	5.380	2.229-12.981	0.000
N stage
N0*	1	-	0.166	N0*	1	-	-
N1	1.779	0.973-3.254	0.061	N1	1.242	0.482-3.200	0.654
N2	1.187	0.558-2.526	0.657	N2	2.924	0.983-8.697	0.054
TNM stage
I*	1	-	0.000	I-II*	1	-	-
II	9.051	2.055-39.857	0.004	III	-	-	-
III	16.305	3.780-70.323	0.000
Pathologic category	0.726	0.348-1.516	0.394		0.925	0.309-2.770	0.889
CEA levels	0.650	0.941-2.741	0.131		2.054	0.858-4.923	0.106
CA199 levels	0.447	1.183-3.880	0.012		-	-	-
Vascular invasion	0.576	0.961-3.761	0.129		1.452	0.559-3.770	0.444
Neural invasion	0.908	0.542-2.062	0.786		-	-	-

^*Reference variables.

Papillary or tubular adenocarcinoma in pathologic category, CEA negative, CA199 negative and vascular invasion negative were reference variables ineach group. Bold = P<0.05.

Prognostic value of TD in CRC

In all patients, the median survival time in the TDN group was 45 months (95% CI: 40.002-49.998) and it was 25 months (95% CI: 21.394-28.606) in TDP group. As for survival, the 1-, 2-, 3-, 4- and 5-year cumulative survival rates in the TDN group were all higher than those in the TDP group, and the 1- to 5-year cumulative survival rates in the TDN group were from 0.925 to 0.273 (57 months) compared with from 0.846 to 0.114 (55 months) in the TDP group (Table 3). Kaplan-Meier curves also showed that the TDP group had a significantly worse prognosis than the TDN group (P<0.001 in Figure 2A). For exploring the prognostic roles of different number of TDs, we entered the number of TDs as an independent variable and our study showed that there was a significant prognostic difference among the groups with varying number of TDs. The median survival time and 1- to 5-year cumulative survival rates were lower as the number of TDs increased. The median survival time was 42 months (95% CI: 32.479-51.521) and the 1- to 5-year cumulative survival rates were from 0.967 to 0.266 (55 months) in group TD = 1, and these were 25 months (95% CI: 17.303-32.697) and were from 0.852 to 0.1 (48 months) in group TD = 2, 21 months (95% CI: 7.114-34.886) and from 0.733 to 0 (39 months) in group TD = 3, 11 months (95% CI: 8.214-13.786) and from 0.414 to 0 (26 months) in group TD≥4 (Table 3). In addition, all of the patients with TD≥2 died within 5 years. From Kaplan-Meier curves we observed that the best prognosis was found in group TD = 1, compared to the other three groups with TD>1 (all P<0.05). Groups TD = 2 and TD = 3 had significantly better prognosis than TD≥4 (both P<0.05). However, no significant difference on prognosis was found between groups TD = 2 and TD = 3 (P = 0.186 in Figure 2B). But when we merged groups TD = 2 and TD = 3, a significant prognostic difference emerged among the groups TD = 1, TD = 2/3, TD≥4 (all P<0.05 in Figure 2C).

Table 3.

Survival outcomes of different status and different number of TDs

Category	All patients		Patients with TD
	TDN	TDP	TD = 1	TD = 2	TD = 3	TD≥4
Median survival time (95% CI) (month)	45 (40.002-49.998)	25 (21.394-28.606)	42 (32.479-51.521)	25 (17.303-32.697)	21 (7.114-34.886)	11 (8.214-13.786)
Cumulative survival
1-year	0.925	0.846	0.967	0.852	0.733	0.414 (13)*
2-year	0.779	0.530 (23)*	0.795 (25)*	0.531 (23)*	0.333 (27)*	0.059 (22)*
3-year	0.602	0.355	0.686	0.266 (35)*	0.133 (37)*	0 (26)*
4-year	0.437	0.163	0.388	0.100	0 (39)*	-
5-year	0.273 (57)*	0.114 (55)*	0.266 (55)*	-	0	-

^*No cumulative survival for the corresponding time and replaced by the survival rate of the month in brackets.

TDN, tumor deposits negative; TDP, tumor deposits positive.

Figure 2.

Survival curves of patients with different status and different number of TDs. A. Significant prognosis was found between group TDN and group TDP (P<0.001); B. Group TD = 1 had the best prognosis compared to group TD = 2 (P = 0.005), group TD = 3 (P<0.001) and group TD≥4 (P<0.001), and the worst prognosis was found in group TD≥4 compared to group TD = 2 (P<0.001) and group TD = 3 (P = 0.004). There was no significant prognostic difference between group TD=2 and group TD = 3 (P = 0.186); C. Group TD≥4 had a worse prognosis than group TD = 2/3 (P<0.001) or than group TD = 1 (P<0.001) when group TD = 2 was merged with group TD = 3. TDN, tumor deposits negative; TDP, tumor deposits positive.

Validating the rationality of TD staging by prognosis

We divided 91 cases of TDP into N stages which did not include TD and then performed a survival analysis. It showed that the prognosis of N stage from N0 to N2 was significantly lower, and it was consistent with N stage (Figure 3A). Kaplan-Meier curves of subgroups in N1 stage showed that N1c had a significantly better prognosis than N1b (P = 0.020); nevertheless, no significantly different prognosis was found between N1a and N1b (P = 0.420), as well as N1a and N1c (P = 0.379) in Figure 3B. As we had done previously in groups TD = 2 and TD = 3, we also merged groups N1a and N1b, and this showed that a significant difference on prognosis was found between N1c and N1a/1b (P = 0.042 in Figure 3C). Moreover, T3-4aN1c and T3-4aN1a/1b, both were classified as TNM IIIB stage, and demonstrated a significant difference in prognosis (P = 0.022 in Figure 3D).

Figure 3.

Comparison on prognosis of each N lesion and each N1 lesion, and prognosis analysis between N1a/1b and N1c, TNM IIIB (T3-4aN1a/1b) and TNM IIIB (T3-4a/N1c). A. When we excluded TD from N stage, prognosis significantly decreased along with an increase of N stage from N0 lesion to N2, and N0 had the superior prognosis compared to N1 (P = 0.034) or N2 (P<0.001). B. In three groups of N1, a significant prognostic difference was found between N1c and N1b (P = 0.020), but there was no significant difference on prognosis between N1a and N1b (P = 0.420), N1a and N1c (P = 0.379). C. Significant difference in prognosis recurred when N1a and N1b were merged, and N1c had a significantly better prognosis compared to N1a/1b (P = 0.042). D. There was a significant prognostic difference between TNM IIB (T3-4aN1a/1b) and TNM IIIB (T3-4aN1c), P = 0.022, although both T3-4aN1a/1b and T3-4aN1c were classified as TNM IIIB.

Discussions

Since TD were first accepted as part of the TNM 5^th edition, the definition and the staging of TD were revised in the every TNM classification. The latest edition of the TNM classification defined TD as tumor foci of CRC located in pericolic or perirectal fat discontinuous with primary tumor. Regardless of the number of TDs, in this edition, all patients with TD but without LNM were classified as N1c, and patients in T3-4aN1a/1b and T3-4aN1c were all counted as TNM IIIB, although they were different in status of TDs. In this study, we analyzed the correlations of TD with CRC with regard to clinicopathologic features and prognosis, and more importantly, we validated the rationale of TD staging in the TNM 7^th/8^th editions.

First, we identified the risk factors associated with TD by analyzing the difference in clinicopathologic features between patients with TD and without TD. We found that the occurrence of TD increased with the ascending levels of T stage, N stage and TNM stage when we examined the influence of the occurrence of TD and factors related to TD such as differentiation degree, T stage, N stage, and TNM stage. For example, the occurrences of TD in N0, N1, and N2 were 13.9%, 16.6% and 30.1% respectively. However, this rule was a little different in differentiation degree, in which moderately differentiated had a higher occurrence (22.5%) than well and poorly differentiated. On the whole, a positive correlation existed between the occurrence of TD and the malignancy of CRC, namely, the higher the malignancy, the more occurrence of TD. The definition in the TNM 8^th and most studies [14,15] considered TD to be the reason for tumor metastasis secondary to vascular and neural invasion, and a positive correlation was confirmed in other studies about gastric cancer [16] and CRC [17,18]. In this study, vascular and neural invasion were not independent risk factors although they associated with the occurrence of TDs. For prognosis, the TDP group had a significantly worse median survival time and 1- to 5-year cumulative survival rates than the TDN group, which illustrated that TDP was an indicator of worse prognosis for patients with CRC, same as the opinion of TNM 7^th/8^th and most studies at present.

Second, we grouped clinicopathologic features in terms of the number of TDs to elucidate the value of number of TDs on clinicopathologic features and prognosis. Similar to the occurrence of TDs mentioned before, we found that differentiation degree, T stage, N stage, TNM stage, CEA levels and vascular invasion were associated with the number of TDs, which disclosed that besides the occurrence of TDs, the malignancy of cancer was also related to the number of TDs, and advanced cancer might have more TDs. Our study indicated a negative correlation between the number of TDs and the prognosis of patients with TDs. Both the median survival time and survival curves decreased along with an increasing of the number TDs from 1 to ≥4, but no significant difference in median survival time and survival curve was found between groups TD = 2 and TD = 3, although group TD = 2 was higher than that in TD = 3. Therefore, we attempted to compare the prognosis again after merging the groups TD = 2 and TD = 3. Significantly lower of survival curves were found among groups TD = 1, TD = 2/3 and TD≥4. This showed that the prognostic influence of TD might not increase strictly with an increasing of the number of TDs, but it was a classified influence, and TD = 1, TD = 2/3, TD≥4 are three probably standards of the grade. We emphasize that the prognostic role of TD≥4 might not be adequately evaluated and other standards might exist as group TD≥4 was not able to be classified into more subgroups because of small size and censored data. The opinion that different number of TDs represent different prognoses had been confirmed in other studies, but disputes remain over the specific number of TD. Goldstein and Turner [8] reported that group TD = 1/2 was significantly different with group TD≥3 on 5-year survival (P<0.01), but a significant difference on prognosis was observed among groups TD = 1, TD = 2 and TD = 3, in which TD≥2 was an independent risk factor (P = 0.008, HR = 2.03, 95% CI: 1.20-3.44) according to Puppa et al. [19]. It has been confirmed that a different number of TDs indicated different prognoses, but the specific criteria for the number of TD still needs to be identified by large scale and multicenter studies.

Thirdly, the transition by which all patients with TD but without LNM were stratified as N1c in TNM 7^th/8^th upgraded the patients in N stage from N0 (with TD and without LNM) to N1c, behind N1a (with 1-2 LNM without TD) and N1b (with 3 LNM and without TD). Now the N0 in TNM 7^th/8^th should indicated patients without simultaneous TD and LNM. Owing to the upgrade of N0, meanwhile, TNM stage for patients with TD was reclassified from TNM I-II (any TN0) to TNM IIIA/B (T1-4aN1/1c). Hence N1c is now involved in two stages of TNM IIIA and TNM IIIB in TNM 7^th/8^th. TNM stage is the cardinal reference for prognostic evaluation clinically, but both T1-2N1 or T1-2N1c and T3-4aN1 or T3-4aN1c were classified as TNM IIIA and TNM IIIB respectively in the TNM 7^th/8^th, which meant that there was no difference between N1 and N1c on prognosis and ignored the influence of TD.

Next, we analyzed the survival curves of TDP patients while excluding TD from N stage, and the result showed that N0 had the best prognosis. This suggests there was no difference between prognosis and N stage in which TD was excluded. Would the same conclusion exist when we included TD as N1c according to the TNM 7^th/8^th? Therefore, we analyzed the prognosis among N1a, N1b and N1c, and a significant prognostic difference was found between N1c and N1b, but there was no significant difference between N1a and N1b, or N1a and N1c on prognosis. Actually, the subgroups of each T, and N stage did not exactly correspond to prognosis, and so TNM stage was used to evaluate prognosis. We next planned to compare the each TNM IIIA/IIIB stage which involved N1c on prognosis, but only TNM IIIB stage was performed because there were only 9 cases with TNM IIIA stage but 57 cases with TNM IIIB stage. The result showed that TNM IIIB (T3-4aN1a/1b) had a significantly worse prognosis than TNM IIIB (T3-4a/N1c), which meant it might be unreasonable to classify both T3-4a N1 and T3-4a N1c into the TNM IIIB stage.

Additionally, whether TD should be classified as anN stage is disputed. Researche [20,21] equated TD with LNM and a positive result was found when they counted TD as LNM, while an opposite result was mentioned by von Winterfeld et al. [22] and Goldstein and Turner [8] who considered that not only the prognostic value of TD was different from LNM, but also more obvious. In the meantime, TD has been also classified as a type of serosal invasion [23] or peritoneal metastasis [24] in some studies. Our study only investigated of the rationale for classifying all patients with different number of TDs into the N1c stage, but had not referred to the validity of including TD into N stage.

Our study has some limitations. We considered all cases of TD≥4 as one group due to the small size of cases, and we just confirmed the probable irrationality of TD staging in TNM 7^th/8^th editions. We did not try to propose a new possible TD staging. Moreover, our study is a one-center study, and the large size and multicenter research needs to be performed in the future. Meanwhile, all studies on TD aimed to analyze the relationships between the TDs and prognosis, but not one has involved the individualized treatment of patients with TD in CRC which we attempted.

Conclusion

For patients with CRC, advanced tumor will probably show more occurrence and numbers of TDs, and TD is an adverse prognostic indicator. Different numbers of TDs and patients with T3-4aN1/1c should not be all categorized as N1c and TNM IIIB stage respectively. TD staging in the TNM 7^th/8^th editions may not be reasonable on account of ignoring the number of TDs, which represent different and classified prognoses, and TD = 1, TD = 2/3, and TD≥4 are three likely subsets of the grade.

Disclosure of conflict of interest

None.