Histopathological Predictors of Recurrence in Stage III Colon Cancer: Reappraisal of Tumor Deposits and Tumor Budding Using AJCC8 Criteria

Abstract

Patients with stage III colonic adenocarcinoma have a spectrum of risk for recurrent disease, and histopathological variables that predict recurrence can help stratify patients into prognostic groups. To identify histopathological predictors of recurrence, we investigated the effect of implementation of the eighth edition of the American Joint Committee on Cancer (AJCC8) staging system definition of tumor deposits and International Tumor Budding Consensus Conference (ITBCC) criteria for tumor budding compared with other known prognostic variables in 256 resected colonic adenocarcinomas, including 150 stage III and 106 stage II tumors. In stage III colon cancer, tumor deposits and high tumor budding were the only independent histological variables that predicted disease recurrence. In a multivariable analysis in stage III colon cancer, tumor deposits and high tumor budding were associated with a 2.2- and 1.5-fold increased risk of developing disease recurrence, respectively (95% CI = 1.1-4,2, P = .02, and 95% CI = 1.1-2.1, P = .01, respectively). The negative prognostic effect of tumor deposits was most pronounced in patients with stage IIIB disease in which tumor deposits were associated with a 3.2-fold increased risk of disease recurrence (95% CI = 1.4-7.1; P = .005). Within the N1 cohort, patients with tumor deposits without concurrent positive lymph nodes (N1c) had a significantly decreased disease-free survival compared with patients with N0 tumors (P < .001) and patients with N1a/b tumors (P = .02). As independent risk factors for recurrence, tumor deposits and high tumor budding are important histopathological variables and should be included as a part of a routine comprehensive pathological risk assessment in stage III colon cancer.

Keywords

colon cancer tumor deposit tumor budding survival stage IIl MSI mismatch repair

Introduction

An estimated 140 000 individuals are diagnosed with colorectal carcinoma each year, and approximately 50 000 will die from this disease, making colorectal carcinoma the third leading cause of cancer deaths in the United States.¹ The treatment and prognosis of colorectal cancer is determined primarily by TNM staging, although significant prognostic heterogeneity remains within each stage category. Patients with stage III colonic adenocarcinoma are considered to have a clinically significant risk of distant metastasis after resection. However, patients with stage III colonic adenocarcinoma have a spectrum of risk for progressive disease. In addition to tumor stage, the National Comprehensive Cancer Network (NCCN) identifies lymphovascular invasion, tumor deposits, and perineural invasion as histopathological features associated with patient survival.² Of these histopathological features, only tumor deposits are incorporated into the AJCC staging scheme.³ The definition of tumor deposits in colorectal cancer has been revised with each successive American Joint Committee on Cancer (AJCC) staging system, resulting in therapeutically relevant changes to tumor staging.^4-6 Despite improvements in the AJCC definition, there remains considerable variability in the histopathological assessment of tumor deposits among pathologists.⁷ The eighth edition of the AJCC staging system (AJCC8) has appropriately revised the definition of tumor deposits compared with the seventh edition of the AJCC (AJCC7).^3,8 Both the AJCC7 and AJCC8 defined a tumor deposit as a discrete focus of tumor within the lymph node drainage area of the primary carcinoma with no identifiable lymph node. However, the AJCC7 tumor deposit definition allows for inclusion of tumor foci with “venous invasion with extravascular spread,” whereas the AJCC8 tumor deposit definition excludes tumor foci associated with an “identifiable vascular or neural structure.” Little data exist on the effect of this change in tumor deposit classification, and more study is needed to determine the effect of these changes on prognosis and therapy.

Incorporation of tumor budding into diagnostic pathology reporting has been problematic, given the disagreements in its definition and lack of standardized assessment. Recently the International Tumor Budding Consensus Conference (ITBCC)⁹ has devised a standardized approach to tumor budding definition and assessment. Based on consensus in the literature, the ITBCC has advocated for the assessment of tumor budding in colorectal cancer in 2 settings: (1) pT1 tumors, in which it is an independent predictor of lymph node metastases, and (2) stage II tumors, in which it is an independent predictor of survival. Although the AJCC8 provides no recommendations on tumor budding, the College of American Pathologists (CAP) colorectal carcinoma protocol¹⁰ has adopted the ITBCC recommendations and recommends tumor budding assessment in these 2 scenarios. However, the prognostic impact of tumor budding, if any, using the ITBCC scoring criteria has not been fully evaluated in stage III colon cancer.

Although prior studies have shown that tumor deposits and high tumor budding are prognostically significant in colon cancer, to our knowledge, the prognostic effect of these histopathological variables have not been investigated in stage III colon cancer following rigorous histological re-review using the revised AJCC8 and ITBCC criteria. In this study, we evaluated the implementation of both the AJCC8 criteria for tumor deposits and the ITBCC scoring scheme for tumor budding on predicting recurrence in stage III colon cancer. In so doing, our findings indicate that tumor deposits and tumor budding are the only histological variables that independently predict tumor recurrence in stage III colon cancer and should be included as part of a routine comprehensive pathological risk assessment.

Materials and Methods

Study Group

The clinicopathological records of 256 patients with colonic adenocarcinoma resected between 2011 and 2013 at the University of Pittsburgh Medical Center, Presbyterian Hospital, were reviewed, including 150 patients with stage III colonic adenocarcinoma and 106 patients with stage II colonic adenocarcinoma. Patients with colonic adenocarcinoma treated with neoadjuvant chemotherapy prior to surgical resection and patients with stage IV disease confirmed either by biopsy or review of radiographic imaging studies were specifically excluded. Rectal adenocarcinomas were specifically excluded given their different therapeutic management compared with colonic adenocarcinoma. The type of initial surgical procedure, extent of disease, and demographic information were obtained from medical records under the guidelines of the University of Pittsburgh Institutional Review Board (IRB No. PR016040136).

Pathological Evaluation

All slides from each tumor were reviewed, and the following histological features were recorded for each case based on review of the slides and digitally captured images of the surgical resection specimen: grade, tumor location in the colon, stage, mucinous component, signet ring cell component, tumor deposits, lymphatic invasion, extramural venous invasion, extramural perineural invasion, and tumor budding. Tumor location was stratified into 3 categories: cecum, right colon (which encompasses the ascending colon, hepatic flexure, and transverse colon), and left colon (which encompasses the splenic flexure, descending colon, and sigmoid colon).

The presence or absence of tumor deposits as documented in the original pathology report using AJCC7 criteria was recorded. All cases were reevaluated for tumor deposits using the AJCC8 definition of tumor deposits based on hematoxylin and eosin (H&E) stains and elastin stains (Verhoeff van Gieson, VVG), as needed. AJCC8 refined the definition of a tumor deposit to exclude tumor nodules associated with an identifiable vascular or neural structure. The CAP colorectal carcinoma protocol further detailed that a tumor focus in or around a “large” nerve or vessel wall should not be classified as a tumor deposit. The CAP colorectal carcinoma protocol is not entirely clear regarding how to classify tumor deposits associated with involvement of small nerves or vessels. Tumor foci with involvement of small nerves or vessels accounting for less than 10% of the overall surface area of the focus were classified as tumor deposits with limited vascular and/or neural invasion and were separately recorded to further evaluate their significance on survival. For each case, the number of tumor deposits and size of the largest tumor deposit measured microscopically were recorded. Nagtegaal et al¹¹ identified prognostic significance for tumor deposit size >12 mm and size ⩽3 mm; thus, tumor deposit size was further stratified into 3 categories (⩽3 mm, 3-12 mm, and >12 mm).

The criteria proposed by Rock et al⁷ were used to aid in the distinction between a tumor deposit and positive lymph node. Briefly, tumor foci were classified as positive lymph nodes if they displayed one or more of the following histological features: residual lymph node, completely round shape with thick capsule, peripheral lymphocyte rim, and peripheral lymphoid follicles.⁷

A 2-tier tumor grading scheme was used. Low grade was defined as >50% gland formation. High grade was defined as <50% gland formation and/or the presence of signet ring cell differentiation. Tumor budding was assessed using the method advocated by the ITBCC⁹ and adopted by the CAP colorectal carcinoma protocol. Tumor buds were defined as isolated cancer cells or a cluster of <5 neoplastic cells at the invasive front of the tumor. The tumor invasive front was assessed at a scanning (10× objective) magnification for the area with maximal tumor budding. In this area, the number of tumor buds was determined at the invasive front in a 0.785 mm² area. Tumors were classified using the CAP colorectal carcinoma protocol 3-tier scheme as follows: low tumor budding if 0 to 4 tumor buds were identified per 0.785 mm², intermediate tumor budding if 5 to 9 tumor buds were identified per 0.785 mm², and high tumor budding if ⩾10 tumor buds were identified per 0.785 mm².

Microsatellite Instability Polymerase Chain Reaction, DNA Mismatch Repair Protein Immunohistochemistry, and Mutational Analysis

Tumors were analyzed by either microsatellite instability, polymerase chain reaction (PCR), or DNA mismatch repair (MMR) protein immunohistochemistry, as previously described.¹² Detection of BRAF mutations was performed using real-time PCR and post-PCR fluorescence melting curve analysis on LightCycler (Roche Applied Science, Indianapolis, IN), as previously described.¹³ This assay detects BRAF mutations in codons 599, 600, and 601. Detection of mutations in exon 2 (codons 12 and 13) and exon 3 (codon 61) of the KRAS gene was performed as previously described.¹²

Statistical Analysis

The χ² or Fisher exact tests were used to characterize the relationship between categorical variables, as appropriate. Kruskal-Wallis tests were used to characterize the relationship between continuous variables. Disease-free survival was the primary end point. Death or recurrence occurring within 1 month of the initial operation was attributed to operative mortality or inadequate surgery and was not included in the survival analysis. Disease-free survival was defined as the time (measured in months) from the date of initial diagnosis to the date of distant disease recurrence (ie, tumor involving the peritoneum and/or other organs) and censored at the date of last clinical follow-up or death. Survival rates were determined by the Kaplan-Meier method, and differences between groups were evaluated by log-rank test. Hazard ratios (HRs) were calculated from a Cox proportional hazard model to identify individual predictors of disease-specific survival. Multivariate analysis of significant or borderline significant individual risk factors (P ⩽ .05) was performed using Cox proportional hazard regression to identify independent risk factors for survival. Data from univariate and multivariate analyses were reported as HRs with 95% CIs. All statistics were assessed using 2-sided tests, with P values <.05 considered statistically significant. Statistical analyses were performed using SPSS (for Windows 23, IBM, Armonk, NY).

Results

Reappraisal of Tumor Deposits and Tumor Budding and Association With Other Tumor Characteristics

All histological slides from all 256 cases were re-reviewed. Of the 256 resected colonic adenocarcinomas, tumor deposits were originally documented in the pathology report in 58 tumors using the criteria outlined by AJCC7. Of the 58 tumors originally reported as having tumor deposits, 6 (10%) had tumor foci associated with a large identifiable vascular structure identified on H&E and elastin stains and were reclassified as having no tumor deposits using the AJCC8 criteria (Figure 1). Of these 6 cases, 4 had no concurrent positive lymph nodes, resulting in a change from the N1c to N0 category. Thus, application of the AJCC8 definition for tumor deposits resulted in downstaging of 7% (4 of 58) of patients from stage III to stage II disease. Of the 198 tumors originally reported as lacking tumor deposits, histological re-review identified 1 case with a single 0.6-cm tumor deposit not documented in the original pathology report. Thus, overall, 53 of 256 cases (21%) had tumor deposits using the AJCC8 criteria. The vast majority had ⩽3 tumor deposits (48/53, 90%), with most cases having just 1 tumor deposit (32/53, 60%; Table 1). In most cases, the tumor deposits measured ⩽12 mm (39/53, 74%). Only 4 cases had tumor deposits ⩽3 mm, and in all 4 of these cases, the tumor deposits measured 3 mm. Concurrent positive lymph nodes were identified in 43/53 (81%) cases; 10 cases were classified as stage III (N1c) based solely on the presence of tumor deposits (Table 1). Of 53 cases, 18 (34%) had invasion in or around small nerves, accounting for less than 10% of the overall surface area of the tumor deposit (Figure 2); 8 of the 53 cases (15%) with tumor deposits had a focus of venous invasion accounting for less than 10% of the overall surface area of the tumor deposit (Figure 2).

Table 1.

Characteristics of Stage III Colonic Adenocarcinomas Associated With Tumor Deposits.

Histopathological Feature	Feature Present (%)
Number of cases	53
Number of tumor deposits
1	32 (60)
2-3	16 (30)
4-6	3 (6)
7 or more	2 (4)
Size of largest tumor deposit
⩽3 mm	4 (8)
>3 to 12 mm	35 (66)
>12 mm	14 (26)
Number of concurrent positive lymph nodes
0	10 (19)
1	6 (11)
2-3	14 (26)
4-6	10 (19)
7 or more	13 (25)
Small venous invasion seen within tumor deposit	8 (15)
Small perineural invasion seen within tumor deposit	18 (34)

Compared with stage III tumors without tumor deposits, stage III tumors with tumor deposits more often displayed extramural venous invasion (62% vs 29%, P < .001) and extramural perineural invasion (68% vs 33%, P < .001; Table 2). There were no differences in high tumor budding, lymphatic invasion, tumor grade, tumor location, mucinous or signet ring cell differentiation, KRAS mutation, BRAF mutation, or MMR protein deficiency for stage III tumors stratified by the presence or absence of tumor deposits. There were significant differences in the presence of tumor deposits and stage III subgroups (Table 2). Most stage III tumors were IIIB (63%), followed by fewer cases of IIIC (26%) and IIIA (11%) disease. Of the stage IIIB tumors, most were classified as T3-T4a N1 disease (84/95, 88%) and fewer cases classified as T2-T3 N2a disease (11/95, 12%). Tumor deposits were identified in 59% (23/39) of IIIC, 31% (29/95) of IIIB, and only 6% (1/16) of IIIA tumors (P < .001).

Table 2.

Clinicopathological Comparison of Colonic Adenocarcinomas Stratified by Stage and Tumor Deposits.

Clinicopathological and Molecular Features	Stage II (%)	Stage III			P Value, Stage II vs III	P Value, Stage III Stratified by Tumor Deposits
Clinicopathological and Molecular Features	Stage II (%)	Overall (%)	Tumor Deposits Absent (%)	Tumor Deposits Present (%)	P Value, Stage II vs III	P Value, Stage III Stratified by Tumor Deposits
Number of cases	106	150	97	53	NA	NA
Mean age in years (range)	70 (30-93)	65 (23-93)	65 (33-93)	66 (23-87)	.004	.5
Location^a
Cecum	25 (24)	36 (24)	25 (26)	11 (21)	.2	.8
Right colon	36 (34)	37 (25)	24 (25)	13 (25)
Left colon	44 (42)	77 (51)	48 (49)	29 (55)
Mean tumor size in cm (range)	5.1 (0.3-14.9)	4.4 (1.3-14.0)	3.9 (1.3-12)	5.0 (1.3-14)	.04	.08
Stage subgroups
A	81 (76)	16 (11)	15 (15)	1 (2)	NA	<.001
B	10 (9)	95 (63)	66 (68)	29 (55)
C	15 (14)	39 (26)	16 (16)	23 (43)
N stage
1a or 1b	NA	93 (62)	73 (75)	20 (38)	NA	NA
1c		10 (7)	0	10 (19)
2a or 2b		47 (31)	24 (25)	23 (43)
High tumor grade	10 (9)	25 (17)	13 (13)	12 (23)	.1	.1
High tumor budding	14 (13)	51 (34)	35 (36)	16 (30)	<.001	.6
Extramural venous invasion	18 (17)	61 (41)	28 (29)	33 (62)	<.001	<.001
Lymphatic invasion	42 (40)	123 (82)	74 (76)	49 (92)	<.001	.07
Extramural perineural invasion	11 (10)	68 (45)	32 (33)	36 (68)	<.001	<.001
Mucinous differentiation	30 (29)	42 (28)	28 (29)	14 (26)	.9	.6
Signet ring cell differentiation	4 (4)	11 (7)	7 (8)	4 (8)	.2	1.0
KRAS exon 2 or 3 mutation
Number tested	103	100	66	34
Present	37 (36)	37 (37)	27 (41)	10 (29)	.9	.3
BRAF mutation
Number tested	103	104	70	34	.7	.1
Present	16 (16)	14 (13)	12 (17)	2 (6)
MMR protein immunohistochemistry
Number tested	105	149	97	52	.1	.6
Loss of expression	24 (23)	23 (15)	14 (14)	9 (17)

Abbreviation: MMR, mismatch repair.

One tumor was located at a prior ileocolonic anastomosis and was not assigned a specific location in the colon.

Compared with tumors with low/intermediate tumor budding, high tumor budding was more often seen in tumors located in the cecum (27/61, 44%) compared with tumors of the right colon (11/73, 15%) and left colon (27/121, 22%); P < .001 (Table 3). High tumor budding was also more frequently identified in tumors with lymphatic invasion (82%), extramural venous invasion (49%), and extramural perineural invasion (49%), all with P <.001. However, high tumor budding was not significantly associated with tumor grade (Figure 3). Tumors with high tumor budding less frequently harbored MMR protein deficiency compared with those with low/intermediate tumor budding (8% vs 22%, P = .009). No differences in KRAS and BRAF mutation were identified with respect to tumor budding.

Table 3.

Clinicopathological Features of Stage II-III Colonic Adenocarcinomas Stratified by ITBCC Tumor Budding Score.

Clinicopathological and Molecular Feature	Low or Intermediate Tumor Budding	High Tumor Budding	P Value
Number of cases	191	65	NA
Mean age in years (range)	67 (23-93)	69 (36-92)	.2
Location^a
Cecum	34 (18)	27 (42)	.001
Right colon	62 (32)	11 (17)
Left colon	94 (49)	27 (42)
Mean tumor size in cm (range)	4.8 (0.3-14.9)	4.9 (1.7-14.0)	.7
N stage
0	92 (48)	14 (22)	<.001
1a or 1b	67 (35)	26 (40)
1c	5 (3)	5 (8)
2a or 2b	27 (14)	20 (31)
High tumor grade	24 (13)	11 (17)	.4
Extramural venous invasion	47 (25)	32 (49)	<.001
Lymphatic invasion	112 (59)	53 (82)	<.001
Extramural perineural invasion	47 (25)	32 (49)	<.001
KRAS exon 2 or 3 mutation
Number tested	158	45	.8
Present	57 (36)	17 (38)
BRAF mutation
Number tested	161	46	.8
Present	24 (15)	6 (13)
MMR protein immunohistochemistry
Number tested	189	65	.009
Loss of expression	42 (22)	5 (8)

Abbreviations: ITBCC, International Tumor Budding Consensus Conference; MMR, mismatch repair.

One tumor was located at a prior ileocolonic anastomosis and was not assigned a specific location in the colon.

Analysis of Factors Predicting Survival

A total of 233 of 256 (91%) patients had clinical follow-up data, including 97 of 106 patients with stage II tumors, 88 of 97 patients with stage III tumors without tumor deposits, and 48 of 53 patients with stage III tumors with tumor deposits. The median follow-up interval was 50 months (range 3 to 74 months) from the time of the initial diagnosis. In the cohort with available follow-up, there were 53 recurrences of disease occurring between 3 months and 48 months from the time of the diagnosis.

Tumor Budding and Tumor Deposits Predict Recurrence in Stage III Colon Cancer

Using Kaplan-Meier survival functions, patients with stage III tumors with high tumor budding (P < .001) and tumor deposits (P < .001) had significantly decreased disease-free survival compared with patients with stage III tumors without these histological features (Figure 4). Using Cox proportional hazards modeling for the cohort of stage III tumors with follow-up, features associated with decreased disease-free survival on both univariate and multivariable analysis (Table 4) were tumor deposits (multivariable HR = 2.2; 95% CI = 1.1-4.2; P = .02) and high tumor budding (multivariable HR = 1.5, 95% CI = 1.1-2.1; P = .01). There was no association between the molecular alterations evaluated (KRAS and BRAF), MMR protein status, or tumor location and disease-free survival (all with P > .1). The effect of tumor deposits on disease-free survival was most pronounced for patients with stage IIIB tumors, which represents the largest subgroup of patients with stage III disease (Figure 4). For patients with stage IIIB tumor, the presence of tumor deposits was the only independent predictor of decreased disease-free survival (multivariable HR = 3.2, 95% CI = 1.4-7.1, P = .005; Table 4). A trend to decreased disease-free survival was observed for patients with stage IIIB tumors and high tumor budding (HR = 2.0; 95% CI = 1.0-4.8), although this effect was of borderline significance (P = .05; Table 4). No difference in disease-free survival was detected for patients with stage IIIC tumors stratified by the presence of tumor deposits or high tumor budding (both with P > .05).

Table 4.

Univariate and Multivariate Disease-Free Survival Analysis in Stage III Colonic Adenocarcinomas.

Clinicopathological Feature	Stage III (All Patients), n = 136				Stage IIIB, n = 84
	Univariate Analysis		Multivariate Analysis		Univariate Analysis		Multivariate Analysis
	Hazard Ratio (95% CI)	P Value	Hazard Ratio (95% CI)	P Value	Hazard Ratio (95% CI)	P Value	Hazard Ratio (95% CI)	P Value
Patient age	1.0 (0.9-1.0)	.1	—	—	1.0 (0.9-1.1)	.4	—	—
Tumor deposits
Absent	Referent	.01	Referent	.02	Referent	.005	Referent	.005
Present	2.2 (1.2-4.0)		2.2 (1.1-4.2)		3.1 (1.4-7.0)		3.2 (1.4-7.1)
Extramural venous invasion
Absent	Referent	.04	Referent	.5	Referent	.3	—	—
Present	1.9 (1.1-3.6)		1.2 (0.6-2.5)		1.5 (0.7-3.6)
Extramural perineural invasion
Absent	Referent	.1	—	—	Referent	.3	—	—
Present	1.7 (0.9-3.1)				1.6 (0.7-3.5)
High tumor budding
Absent	Referent	.002	Referent	.01	Referent	.05	Referent	.05
Present	2.5 (1.4-4.6)		1.5 (1.1-2.1)		2.1 (1.0-4.7)		2.2 (1.0-4.8)

Tumor Deposits Are a Poor Prognostic Factor in N1 Disease

A total of 94 of 103 (91%) patients with N1 disease had clinical follow-up data, including 8 of 10 patients with N1c tumors and 86 of 93 patients with N1a/b tumors. All patients with N1c tumors had more than 12 lymph nodes examined (range 14 to 30 lymph nodes, mean 19 lymph nodes). Four of the 8 patients with N1c tumors developed tumor recurrence at 8, 9, 16, and 17 months after initial presentation. Four patients classified as N1c using AJCC7 were reclassified as N0 using AJCC8 criteria because the tumor foci were associated with a large identifiable vascular structure. All 4 of these patients had clinical follow-up, with 3 patients alive without recurrence at last follow-up of 53, 54, and 60 months and 1 patient developing recurrence at 18 months after initial presentation.

Although the limited number of patients with N1c tumors precludes multivariable analysis of survival, Kaplan-Meier survival functions were used to compare patients based on N1 subgroup status. Patients with N1c tumors had a significantly decreased disease-free survival compared with patients with N0 tumors (P < .001) and patients with N1a/b tumors (P = .02). The presence of tumor deposits in patients with N1a/b disease also affected survival (Figure 5). Patients with N1a/b disease without tumor deposits had improved disease-free survival compared with patients with N1a/b tumors with tumor deposits (P = .02). Patients with N1a/b disease without tumor deposits had improved disease-free survival (P = .004) compared with patients with N1c tumors. Patients with N1a/b tumors with tumor deposits and patients with N1c tumors had a similarly reduced disease-free survival (P = .5). These data indicate that within the N1 category, tumor deposits are associated with shorter time interval to disease recurrence even in the setting of concurrent positive lymph nodes.

Tumor

Deposit Size, Tumor Deposit Number, and the Presence of Limited Vascular or Perineural Invasion Within Tumor Deposits: Do They Matter?

The size of the largest tumor deposit was stratified into 3 groups based on the results of Nagtegaal et al¹¹: ⩽3 mm, 3 to 12 mm, and >12 mm. However, only 2 of 4 patients with tumor deposits ⩽3 mm had clinical follow-up. Thus, for the survival analysis, patients with tumor deposits ⩽12 mm were compared with those with tumor deposits >12 mm. Using Kaplan-Meier survival statistics, there was no significant difference in disease-free survival for patients with tumor deposits ⩽12 mm compared with those with tumor deposits >12 mm (P = .7).

Jin et al⁴ identified that a cutpoint of 4 tumor deposits resulted in significant differences in patient survival. Thus, in our data set, cases were stratified into 3 groups based on the number of tumor deposits: 1 tumor deposit, 2 or 3 tumor deposits, and 4 or more tumor deposits. Using Kaplan-Meier survival functions, there was no difference in disease-free survival for patients stratified by the number of tumor deposits (P > .05 comparing all 3 groups).

Limited vascular and/or perineural invasion in a tumor deposit was defined as small vascular and/or perineural invasion accounting for less than 10% of the overall surface area of the tumor focus and was identified in 22 of 53 cases with tumor deposits. Using Kaplan-Meier survival statistics, there was no significant difference in disease-free survival (P = 1.0) for patients with tumor deposits displaying limited vascular and/or perineural invasion and patients with tumor deposits lacking small vascular and/or perineural invasion. These data indicate that limited vascular and/or perineural invasion seen within a tumor focus should not preclude classification as a tumor deposit.

Discussion

The purpose of this study was to investigate histopathological features that predict recurrence in patients with stage III colon cancer incorporating AJCC8 criteria for tumor deposits and the ITBCC tumor budding scheme advocated by the CAP colorectal carcinoma protocol. We demonstrate that tumor deposits and high tumor budding are the only independent histological variables that predict tumor recurrence in patients with stage III colon cancer, supporting their inclusion as prognostically relevant histological variables in colon cancer. The negative independent prognostic effect of tumor deposits is seen even with concurrent positive lymph nodes and is most pronounced in patients with stage IIIB (pT3-T4a N1) tumors. The presence of tumor deposits and tumor budding appears to indicate an aggressive colorectal carcinoma phenotype in stage III disease, even in the presence of concurrent positive lymph nodes.

Tumor budding has been shown to be an independent predictor of recurrence and survival in stage II colorectal carcinoma.^14-19 However, to date, incorporation of tumor budding into diagnostic pathology reporting has been problematic given the disagreements in its definition and lack of standardized assessment. Recently, reporting of tumor budding has been recommended by the ITBCC⁹ in stage II colorectal carcinoma. However, assessment of tumor budding is not routinely performed in most institutions in the United States and has only recently been introduced in the CAP colorectal carcinoma protocol as an optional reporting element. Although the prognostic impact of tumor budding in stage II colon cancer has been established, the effect of tumor budding using the ITBCC scoring criteria has not been specifically evaluated in stage III colon cancer. A recent meta-analysis of 7821 patients with colorectal carcinoma of all stages demonstrated that tumor budding was significantly associated with lymph node metastasis, the development of disease recurrence, and cancer-related death.²⁰ However, of the 34 studies within the meta-analysis, none specifically evaluated stage III tumors, many included rectal carcinoma despite its different therapeutic management, and the definition and assessment of tumor budding varied widely between studies. Similarly, Graham et al²¹ evaluated 553 colon and rectal carcinomas of all stages and identified high tumor budding as an independent predictor of poor disease-specific survival. Our results indicate that patients with stage III colon cancer exhibiting high tumor budding have a more aggressive clinical course, with poor disease-free survival compared with patients with stage III tumors lacking high tumor budding. Tumor budding is thought to represent an epithelial to mesenchymal transition, wherein tumor cells at the invasive front acquire a mesenchymal phenotype with increased potential for vascular invasion and metastasis.²² Thus, it is not surprising that the presence of high tumor budding at the invasive front in stage III colon cancer is an independent predictor of tumor recurrence and metastasis. However, our data indicate that tumor budding is a prognostic feature that is independent of other commonly assessed histopathological variables, such as lymphatic, venous, and perineural invasion. Our findings suggest that histological evaluation of tumor budding is helpful in identifying patients with more aggressive disease in the setting of stage III colon cancer.

Tumor deposits have been extensively studied in colorectal carcinoma, with most studies demonstrating that they are an important prognostic variable.^4-6,11,23-30 However, many of the previous studies evaluating the prognostic impact of tumor deposits did not account for the changing definition of tumor deposits over the study periods analyzed and included rectal tumors and stage IV disease, which have a different therapeutic management compared with stage III colon cancer. To our knowledge, to date, no study has evaluated the impact of the revised AJCC8 definition for tumor deposits and its effect on survival. Our data indicate that tumor deposits are an independent histopathological predictor of recurrence in stage III colon cancer, even in the setting of concurrent positive lymph nodes. The negative prognostic effect of tumor deposits was most pronounced for the stage IIIB category, which represents the largest subgroup of patients with stage III colon cancers and includes T3-T4a N1 disease. For patients with stage IIIB tumors, the presence of tumor deposits was a significant independent predictor of recurrence, and such patients had a disease-free survival similar to that of patients with stage IIIC disease. Our findings also confirm the negative prognostic impact of tumor deposits in the setting of N1 disease. Patients with N1c tumors had significantly decreased survival compared with patients with stage II (N0) disease and compared with patients with N1a/b disease. In their analysis of 17 patients with N1c disease using AJCC7 criteria, Jin et al⁴ also demonstrated that patients with N1c disease had significantly worse outcomes compared with patients with N0 disease; however, they did not identify a significant difference in survival between patients with N1c disease and patients with N1a/b disease. Our limited data indicate that even in the presence of lymph node metastasis, patients with concurrent tumor deposits in the N1 category have decreased disease-free survival compared with patients without tumor deposits.

In recent multicenter and meta-analyses of colorectal carcinoma, tumor deposits were associated with poor disease-free survival, and these studies proposed that the N category should be determined by summing the number of tumor deposits and lymph node metastases.^11,27 However, the results of our study do not support summing the number of tumor deposits and lymph node metastases to arrive at the N category. The vast majority (90%) of the cases in our series had relatively few (ie, 3 or fewer) tumor deposits, with most (60%) having a single tumor deposit. Thus, the sum of the number of tumor deposits and lymph node metastases to arrive at a final N category would very infrequently result in a change in the overall stage and would not entirely capture the negative prognostic effect of tumor deposits in stage III colon cancer.

Previous changes to the AJCC definition of tumor deposits have resulted in stage reassignment in between 15% to 45% of patients with tumor deposits.^4,6 In our series, the AJCC8 revision to the definition of tumor deposits resulted in stage migration in only 7% of patients, with all changes resulting in downward migration from stage III to stage II. This downward stage migration is unlikely to affect patient management because all cases reassigned to stage II disease had tumors with large venous invasion, which is a histopathological variable recognized by the NCCN guidelines as a poor prognostic feature with a recommendation for adjuvant chemotherapy.2 In daily practice, distinguishing between tumor deposits and large vascular invasion within a focus of tumor within pericolic soft tissue can be challenging. In our analysis, elastin stains were used to confirm the presence of large vascular remnants identified by review of the H&E-stained sections. In their study, Goldstein and Turner²⁴ performed serial sectioning of tumor deposits and identified perineural growth in 77% and intravascular growth in 83% of cases. It is possible that evaluation of serial sections and the use of elastin stains would identify vascular remnants and/or invasion in or around nerves in our cohort of cases, potentially precluding the classification of a tumor focus as a tumor deposit. However, the exclusion of tumor foci within the pericolic soft tissues with limited involvement of vessels and nerves may be overly restrictive. Limited vascular and/or perineural invasion in a tumor deposit accounting for less than 10% of the overall surface area of the tumor focus was seen on routine H&E evaluation in 42% of cases in our series. We found no difference in survival for patients with tumor deposits displaying limited and small vascular and/or perineural invasion, and patients with tumor deposits lacking small vascular and/or perineural invasion, indicating that limited and small vascular and/or perineural invasion seen within a tumor focus should not preclude classification as a tumor deposit.

Our analysis has limitations, including the retrospective design, the size of the study cohort, and the inherent issue of lack of rigorously standardized treatment in retrospective analyses. Although the number of cases included in our study is relatively small by some standards, our analysis is one of the largest studies to date requiring rigorous re-review of tumor histology using the AJCC8 definition for tumor deposits and the ITBCC scoring scheme for tumor budding and with correlation with survival. Given the retrospective design, we were also not able to control for chemotherapy treatment. Our study also represents a single-institution study, with its inherent selection and referral bias. Finally, identification of vascular invasion within a potential tumor deposit can be challenging and may be aided by histochemical stains for elastin.^31,32 We only performed additional stains for elastin in select cases based on evaluation of the H&E-stained slides. This more likely reflects general practice patterns where elastin stains are not routinely performed on all tumor deposits to evaluate for potential remnants of vascular structures.

In conclusion, our analysis highlights significant prognostic heterogeneity for patients with stage III tumors and identifies tumor deposits and high tumor budding as the most important independent predictors of tumor recurrence in stage III colon cancer. The negative prognostic effect of tumor deposits is most pronounced in patients with stage IIIB disease, which represents the largest subgroup of patients with stage III disease. Our results further demonstrate that patients with N1c disease have poor disease-free survival compared with patients with N0 disease and N1a/b disease. As independent risk factors for recurrence in stage III colon cancer, tumor deposits as defined by AJCC8 and high tumor budding using ITBCC criteria are important histopathological variables that should be included as a part of a routine comprehensive pathological risk assessment of stage III colon cancer, even in the setting of concurrent positive lymph nodes.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Ethical Approval

Not applicable, because this article does not contain any studies with human or animal subjects.

Informed Consent

Not applicable, because this article does not contain any studies with human or animal subjects.

Trial Registration

Not applicable, because this article does not contain any clinical trials.

ORCID iD

Frances N. Akwuole, .

Reetesh K. Pai, .

References

Siegel

Miller

Jemal

Cancer statistics, 2018. CA Cancer J Clin. 2018;68:7-30.

Benson

Venook

Cederquist

et al . Colon cancer, version 1.2017, NCCN Clinical Practice Guidelines in Oncology. J Natl Comp Canc Netw. 2017;15:370-398.

Jessup

Goldberg

Asare

et al . Colon and rectum. In: Amin

, ed. AJCC Cancer Staging Manual. 8th ed. Chicago, IL: Springer; 2017:251-274.

Jin

Roth

Rock

Washington

Lehman

Frankel

WL.

The impact of tumor deposits on colonic adenocarcinoma AJCC TNM staging and outcome. Am J Surg Pathol. 2015;39:109-115.

von Winterfeld

Hoffmeister

Ingold-Heppner

et al . Frequency of therapy-relevant staging shifts in colorectal cancer through the introduction of pN1c in the 7th TNM edition. Eur J Cancer. 2014;50:2958-2965.

Nagtegaal

Tot

Jayne

et al . Lymph nodes, tumor deposits, and TNM: are we getting better? J Clin Oncol. 2011;29:2487-2492.

Rock

Washington

Adsay

et al . Debating deposits: an interobserver variability study of lymph nodes and pericolonic tumor deposits in colonic adenocarcinoma. Arch Pathol Lab Med. 2014;138:636-642.

Benedix

Meyer

Kube

et al . Influence of anatomical subsite on the incidence of microsatellite instability, and KRAS and BRAF mutation rates in patients with colon carcinoma. Pathol Res Pract. 2012;208:592-597.

Lugli

Kirsch

Ajioka

et al . Recommendations for reporting tumor budding in colorectal cancer based on the International Tumor Budding Consensus Conference (ITBCC) 2016. Mod Pathol. 2017;30:1299-1311.

10.

Kakar

Shi

Berho

et al . Protocol for the examination of specimens from patients with primary carcinoma of the colon and rectum. http://www.cap.org/ShowProperty?nodePath=/UCMCon/Contribution%20Folders/WebContent/pdf/cp-colon-17protocol-4000.pdf. Accessed January 19. 2018.

11.

Nagtegaal

Knijn

Hugen

et al . Tumor deposits in colorectal cancer: improving the value of modern staging—a systematic review and meta-analysis. J Clin Oncol. 2017;35:1119-1127.

12.

Landau

Zhu

Akwuole

Pai

RK.

Site-specific differences in colonic adenocarcinoma: KRAS mutations and high tumor budding are more frequent in cecal adenocarcinoma. Am J Surg Pathol. 2018;42:351-358.

13.

Yousem

Nikiforova

Nikiforov

The histopathology of BRAF-V600E-mutated lung adenocarcinoma. Am J Surg Pathol. 2008;32:1317-1321.

14.

Betge

Kornprat

Pollheimer

et al . Tumor budding is an independent predictor of outcome in AJCC/UICC stage II colorectal cancer. Ann Surg Oncol. 2012;19:3706-3712.

15.

Canney

Kevans

Wang

et al . Stage II colonic adenocarcinoma: a detailed study of pT4N0 with emphasis on peritoneal involvement and the role of tumour budding. Histopathology. 2012;61:488-496.

16.

Petrelli

Pezzica

Cabiddu

et al . Tumour budding and survival in stage II colorectal cancer: a systematic review and pooled analysis. J Gastrointest Cancer. 2015;46:212-218.

17.

Lai

et al . Tumour budding is a reproducible index for risk stratification of patients with stage II colon cancer. Colorectal Dis. 2014;16:259-264.

18.

Hayes

Maguire

Conlon

Gibbons

Wang

Sheahan

Reproducibility of the rapid bud count method for assessment of tumor budding in stage II colorectal cancer. Am J Surg Pathol. 2010;34:746-748.

19.

Horcic

Koelzer

Karamitopoulou

et al . Tumor budding score based on 10 high-power fields is a promising basis for a standardized prognostic scoring system in stage II colorectal cancer. Hum Pathol. 2013;44:697-705.

20.

Rogers

Winter

Heeney

et al . Systematic review and meta-analysis of the impact of tumour budding in colorectal cancer. Br J Cancer. 2016;115:831-840.

21.

Graham

Vierkant

Tillmans

et al . Tumor budding in colorectal carcinoma: confirmation of prognostic significance and histologic cutoff in a population-based cohort. Am J Surg Pathol. 2015;39:1340-1346.

22.

Koelzer

Zlobec

Lugli

Tumor budding in colorectal cancer—ready for diagnostic practice?

Hum Pathol. 2016;47:4-19.

23.

Al Sahaf

Myers

Jawad

Browne

Winter

Redmond

HP.

The prognostic significance of extramural deposits and extracapsular lymph node invasion in colon cancer. Dis Colon Rectum. 2011;54:982-988.

24.

Goldstein

Turner

JR.

Pericolonic tumor deposits in patients with T3N+MO colon adenocarcinomas: markers of reduced disease free survival and intra-abdominal metastases and their implications for TNM classification. Cancer. 2000;88:2228-2238.

25.

Puppa

Maisonneuve

Sonzogni

et al . Pathological assessment of pericolonic tumor deposits in advanced colonic carcinoma: relevance to prognosis and tumor staging. Mod Pathol. 2007;20:843-855.

26.

Tateishi

Arima

Futami

et al . A clinicopathological investigation of “tumor nodules” in colorectal cancer. Surg Today. 2005;35:377-384.

27.

Ueno

Mochizuki

Akagi

et al . Optimal colorectal cancer staging criteria in TNM classification. J Clin Oncol. 2012;30:1519-1526.

28.

Ueno

Mochizuki

Shirouzu

et al . Actual status of distribution and prognostic impact of extramural discontinuous cancer spread in colorectal cancer. J Clin Oncol. 2011;29:2550-2556.

29.

Yabata

Udagawa

Okamoto

Effect of tumor deposits on overall survival in colorectal cancer patients with regional lymph node metastases. J Rural Med. 2014;9:20-26.

30.

Nagtegaal

Quirke

Colorectal tumour deposits in the mesorectum and pericolon; a critical review. Histopathology. 2007;51:141-149.

31.

Kirsch

Messenger

Riddell

et al . Venous invasion in colorectal cancer: impact of an elastin stain on detection and interobserver agreement among gastrointestinal and nongastrointestinal pathologists. Am J Surg Pathol. 2013;37:200-210.

32.

Messenger

Driman

Kirsch

Developments in the assessment of venous invasion in colorectal cancer: implications for future practice and patient outcome. Hum Pathol. 2012;43:965-973.