Is Microsatellite Status Associated With Prognosis In Stage II Colon Cancer With High-Risk Features?

Abstract

BACKGROUND:

The influence of microsatellite instability on prognosis in high-risk stage II colon cancer is unknown.

OBJECTIVE:

To investigate the relationship between microsatellite instability and overall survival in high-risk stage II colon cancer.

DESIGN:

Retrospective review of the National Cancer Database from 2010-2016

SETTINGS:

This study included national cancer epidemiology data from the American College of Surgeons Commission on Cancer.

PATIENTS:

16,788 patients with stage II colon adenocarcinoma and known microsatellite status (1,709 microsatellite unstable)

MAIN OUTCOME MEASURES:

Overall survival

RESULTS:

Microsatellite unstable cancers with high-risk features had significantly better overall survival than microsatellite stable cancers with high-risk features (5-year survival 80% vs 72%, p=0.01), and had equivalent survival to microsatellite stable cancers with low risk features (5-year survival 80%). When stratifying by specific high-risk features, patients with lymphovascular invasion, perineural invasion, or high-grade histology had similar overall survival to patients without these features, only in microsatellite unstable cancers. However, patients with high-risk features of T4 stage, positive margins, and <12 lymph nodes saw no survival benefit based on microsatellite status. This was confirmed on multivariable cox regression modeling. A subgroup analysis of patients who did not receive chemotherapy similarly demonstrated that microsatellite unstable cancers with lymphovascular invasion, perineural invasion, or high-grade histology had similar overall survival to microsatellite unstable cancers without those features.

LIMITATIONS:

The study is limited by lack of specific clinical data and potential treatment bias.

CONCLUSIONS:

In microsatellite unstable cancers, lymphovascular invasion, perineural invasion, and high-grade histology are not associated with worse overall survival, even when deferring adjuvant chemotherapy. These data support National Comprehensive Cancer Network recommendations to forego chemotherapy in stage II cancers with microsatellite instability and these features. In contrast, some high-risk features were associated with worse survival despite microsatellite unstable biology, and therapies to improve survival need to be explored.

INTRODUCTION

The positive prognostic value of microsatellite instability (MSI) status in early-stage colon cancer has been widely reported in the literature. Specifically, in stage II colon cancer patients, MSI-high (MSI-H) cancers have been shown to have more favorable local recurrence rates and overall survival (OS) compared to microsatellite stable (MSS) or MSI-low (MSI-L) cancers.^1-4 In addition to MSI status, there are a variety of other clinicopathologic factors that are considered to be high-risk features that may potentially impact prognosis. These high-risk features are often cited as indications for adjuvant chemotherapy in patients with stage II colon cancer and include: T4 stage, high-grade histology, lymphovascular invasion (LVI), perineural invasion (PNI), bowel obstruction or perforation, inadequate lymph node sampling, and close/undetermined/positive margins.^5-7

Importantly, the interaction between MSI-H status and high-risk features in stage II colon cancer has not been studied and remains unclear. This relationship is of potential relevance for clinical decision making, as the most recent National Comprehensive Cancer Network (NCCN) guidelines recommend adjuvant chemotherapy for all stage II colon cancers with high-risk features, unless those cancers are also MSI-H.⁸ Meta-analyses have demonstrated the benefit of adjuvant chemotherapy in high-risk stage II colon cancer,^9,10 however there are little data to confirm that MSI-H cancers with high-risk features have similar prognosis to MSI-H or MSS cancers without high-risk features. The NCCN recommendation to forego adjuvant chemotherapy in MSI-H cancers is primarily based on literature demonstrating that MSI-H cancers do not respond to 5-fluorouracil based adjuvant chemotherapy.^11-13 Furthermore, population-level data have shown that not all high-risk features have similar adverse effects on OS, with T4 stage carrying significant risk and LVI, PNI, and high-grade histology potentially carrying negligible risk when adjusting for other variables.¹⁴ Ultimately, there is a lack of data to guide oncologic practice in early stage colon cancer patients with MSI-H cancers and high-risk features.

Therefore, the primary aim of this study was to describe the association between MSI status and prognosis in stage II colon cancer when considering high-risk features, and as a corollary, determine how specific high-risk features relate to survival in patients with MSI-H cancers. The secondary aim was to describe OS in the subgroup of patients with MSI-H cancers that did not receive adjuvant chemotherapy, when stratified by high-risk features.

METHODS

Data Source

We selected patients from the National Cancer Database (NCDB) 2016 Participant User File.¹⁵ The NCDB is a national hospital-based cancer registry which captures approximately 70% of adult cancer patients in the United States (US) from over 1500 Commission on Cancer-accredited hospitals. It contains information on patient demographics, disease stage, comorbidity, and cancer treatment course. Our study was considered exempt by the Institutional Review Board.

Patient Selection

We included patients with American Joint Committee on Cancer (AJCC) pathologic stage II primary colon adenocarcinoma (C18.0, C18.2-C18.9, C19.9) treated with a definitive surgical resection from January 1, 2010 to December 31, 2015 and with follow up until 12/2016. Adenocarcinoma was defined using International Statistical Classification of Diseases (ICD-O-3) histology codes (8140-8148 adenocarcinoma NOS, 8210-8223 adenocarcinoma in situ of adenomatous polyp, 8261-8263 papillary adenocarcinoma, 8480-8483 mucinous adenocarcinoma, and 8490-8493 signet ring cell carcinoma). We excluded patients who did not have their primary tumor resected and those with rectal cancers. Patients were additionally excluded if they had previous invasive cancers at any location, mortality within 90 days of diagnosis, resection over 180 days from diagnosis, radiation, chemotherapy beyond one year from diagnosis, missing lymph node evaluation data, missing surgical margins, and if all treatment was performed at a center other than the reporting facility. Detailed patient selection can be found in Figure 1.

Figure 1.

Patient selection CONSORT diagram.

Study design

MSI status was defined using the collaborative stage site-specific data provided by the NCDB, which classifies patients as MSS, MSI-low, MSI-high, or MSI-NOS. Patients were in the MSI-H group if they were specifically reported to be MSI-H. For the purpose of the study, patients were included in the comparison group if they were reported to be MSS or MSI-L (MSS/MSI-L). Patients reported as MSI-NOS or with missing MSI data were excluded. The primary outcome was OS from time of diagnosis and patients were censored if alive at the time of last follow up.

Demographic data and comorbid status were recorded for each patient. NCDB captures comorbidities using the Charlson-Deyo index, a weighted score derived from 15 comorbid conditions that correlates with survival.¹⁶ Oncologic variables were recorded including primary tumor side (right colon defined as cecum to distal transverse colon and left colon defined as splenic flexure to rectosigmoid) and receipt of adjuvant chemotherapy, which was recorded as a binary yes/no, regardless of single-agent or multi-agent therapy. High-risk features were defined as high-grade histology (poorly differentiated or undifferentiated), LVI, PNI, T4 stage, less than 12 lymph nodes (<12 LNs) examined in the surgical specimen, and positive surgical margins (either positive proximal/distal margin or positive circumferential resection margin).

Statistical Analysis

Baseline characteristics between groups were compared using 2-sample t tests for continuous variables and χ ² tests for categorical variables. We analyzed OS using Kaplan-Meier (KM) estimators, comparing groups stratified by MSI status and specific high-risk features with the log-rank test, as well as multivariable Cox proportional hazard models (separate models for MSS/MSI-L and MSI-H groups) to adjust for possible confounding variables. We additionally performed subgroup analyses on patients with MSI-H cancers who did not receive adjuvant chemotherapy to describe OS in this cohort. All statistical analyses were 2-sided and considered significant if p < .05. Variables in which data are reported as “unknown” had a separate indicator category created. All statistical analyses were performed using SPSS software, version 25.0 (IBM Corp, Armonk NY).

RESULTS

Baseline characteristics

We identified 16,788 patients with primary stage II colon adenocarcinoma and known microsatellite status, of which 1,709 were MSI-H (10.2%) and 15,079 were MSS/MSI-L (89.8%). Median follow up was 34.8 months (IQR 19.7-53.5 months). Patient characteristics are listed in Table 1. Patients who were female and white race were more likely to have MSI-H cancers. Furthermore, MSI-H cancers were more likely to be right sided, high-grade, and have 12 nodes examined in the surgical specimen, and less likely to receive chemotherapy.

Table 1.

Patient Characteristics

Variable	MSS/MSI-L	MSI-H
	(n = 15,079)	(n = 1,709)	p value
Demographics
Age (SD)	65.4 (13.6)	65.1 (15.4)	0.336
Female (%)	7,453 (49.4)	969 (56.7)	<0.001
Race
White	12,272 (81.4)	1,503 (87.9)	<0.001
Black	1,873 (12.4)	121 (7.1)
Other	934 (6.2)	85 (5.0)
Institution type
Community Cancer Program	1,243 (8.5)	133 (9.0)	<0.001
Comprehensive Community Program	6,158 (42.1)	674 (42.3)
Academic/Research	4,691 (32.1)	581 (36.5)
Integrated Network Cancer Program	2,528 (17.3)	193 (12.1)
Primary Payor
Not insured	581 (3.9)	50 (2.9)	0.269
Private insurance	5,825 (38.6)	648 (37.9)
Medicaid	887 (5.9)	90 (5.3)
Medicare	7,489 (49.7)	887 (51.9)
Other Government	141 9).9)	15 (0.9)
Unknown	156 (1.0)	19 (1.1)
Charlson-Deyo score
0	10,331 (68.5)	1,165 (68.2)	0.341
1	3,327 (22.1)	371 (21.7)
2	945 (6.3)	125 (7.3)
3+	476 (3.2)	48 (2.8)
Primary Tumor Site
Right colon	8,256 (54.8)	1,362 (79.7)	<0.001
Left colon/Rectosigmoid	6,823 (45.2)	347 (20.3)
Adjuvant chemotherapy	3,269 (21.7)	247 (14.5)	<0.001
T Stage
T3	12,906 (85.6)	1,466 (85.8)	0.830
T4	2,173 (14.4)	243 (14.2)
Lymphovascular Invasion
Absent	12,330 (81.8)	1,359 (79.5)	0.063
Present	1,980 (13.1)	247 (14.5)
Unknown	769 (5.1)	103 (6.0)
Perineural Invasion
Absent	13,505 (89.6)	1,542 (90.2)	0.108
Present	1,115 (7.4)	106 (6.2)
Unknown	459 (3.0)	61 (3.6)
Tumor Grade
Low Grade	13,032 (86.4)	1,088 (63.7)	<0.001
High Grade	1,774 (11.8)	585 (34.2)
Unknown	273 (1.8)	36 (2.1)
Positive Margins	874 (5.8)	102 (6.0)	0.773
< 12 Lymph Nodes	931 (6.2)	54 (3.2)	<0.001
High Risk Features	5,731/13,713 (41.8%)	818/1,537 (53.2%)	<0.001

MSI status and high-risk features

Overall, in patients with resected stage II colon cancers, MSI-H cancers were associated with improved OS compared to MSS/MSI-L cancers (5-year OS 83% vs 76%, p=0.01). Of the included patients, 7,202 patients had at least one high-risk feature (42.9%). The presence of at least one high-risk feature was associated with significantly worse OS when compared to patients with no high-risk features (5-year OS 73% vs 80%, p<0.001). In patients with MSI-H cancers, the presence of at least one high-risk feature was also associated with worse OS (5-year OS 80% vs 85%). Yet, based on MSI status, patients with cancers that were both MSI-H and high-risk had significantly better OS than patients with cancers that were MSS/MSI-L and high-risk (5-year OS 80% vs 72%, p=0.01) and equivalent survival to patients with MSS/MSI-L cancers and no high-risk features (5-year OS 80%). MSI-H cancers with no high-risk features had the most favorable OS (5-year OS 85%).

LVI, PNI, and high-grade histology

We next sought to determine if MSI status was associated with OS when considering specific high-risk features. On univariable KM analysis, we observed that patients with LVI, PNI, or high-grade histology present exhibited a survival benefit based on the presence of MSI-H compared to MSS/MSI-L. In these analyses, 5-year OS for patients with MSI-H compared to patients with MSS/MSI-L was 85% vs 69% for LVI-present cancers (p<0.01), 75% vs 59% for PNI-present cancers (p<0.01), and 82% vs 73% for high-grade vs. low grade histology (p<0.01). Interestingly, in patients with MSI-H cancers, LVI, PNI, and high-grade histology were not associated with worse survival; however, these high-risk features were associated with worse survival in patients with MSS/MSI-L cancers (Figure 2A-C).

Figure 2.

A-F. Univariable Kaplan Meier analysis stratified by MSI status and high-risk features. MSS = patients in MSS/MSI-L group; LVI = lymphovascular invasion; PNI = perineural invasion. RM = resection margin; LN = lymph nodes; NS= not significant (p > 0.05).

T4 stage, positive margins, and inadequate lymph nodes

In contrast to other high-risk features, patients with either T4 stage, positive margins, or <12 LNs saw no survival benefit based on MSI-H status compared to those with MSS/MSI-L cancers. Similarly, both subgroups of MSI-H and MSS/MSI-L cancers had reductions in OS when high-risk features of T4 stage, <12 LNs, or positive margins were present (Figure 2D-F).

Multivariable analyses

When adjusting for patient demographics, tumor side, receipt of adjuvant chemotherapy, and high-risk features, the association between MSI-H status and OS was again observed (HR 0.75, 95%CI 0.65-0.87, p<0.01) (Table 2). We next created separate models for MSS/MSI-L and MSI-H cancers to determine adjusted hazard ratios for individual high-risk features. In the regression model including only MSS/MSI-L patients, all high-risk features were associated with worse survival (Table 3). However, in the analysis of MSI-H patients, only T4 stage (HR 2.61, 95%C 1.83-3.72, p<0.01) and positive margins (HR 1.73, 95%CI 1.07-2.80, p=0.02) were associated with worse OS (Table 3). All other high-risk features (i.e. LVI, PNI, high tumor grade, and <12 LNs) were not associated with worse OS in patients with MSI-H cancers. Of note, unknown PNI status was associated with worse OS.

Table 2.

Adjusted hazard ratios for mortality in stage II colon cancer based on patient and tumor features.

Variable	HR (95% CI)	p-value
MSI-H (vs MSS/MSI-L)	0.75 (0.65-0.87)	<0.001
Age (SD)	1.06 (1.05-1.06)	<0.001
Female (%)	0.78 (0.72-0.85)	<0.001
Race
White	Ref	Ref
Black	1.29 (1.14-1.46)	<0.001
Other	0.69 (0.56-0.87)	<0.001
Institution type
Community Cancer Program	Ref	Ref
Comprehensive Community Program	0.88 (0.76-1.10)	0.066
Academic/Research	0.80 (0.69-0.93)	0.003
Integrated Network Cancer Program	0.89 (0.76-1.05)	0.170
Primary Payor
Not insured	Ref	Ref
Private insurance	0.72 (0.55-0.95)	0.022
Medicaid	1.21 (0.88-1.67)	0.247
Medicare	0.89 (0.67-1.17)	0.404
Other Government	0.99 (0.60-1.64)	0.961
Unknown	0.96 (0.59-1.57)	0.873
Charlson-Deyo score
0	Ref	Ref
1	1.22 (1.12-1.34)	<0.001
2	1.83 (1.62-2.08)	<0.001
3+	2.61 (2.23-3.05)	<0.001
Primary Tumor Site
Left colon/rectosigmoid	Ref	Ref
Right colon	0.89 (0.82-0.97)	0.005
Adjuvant chemotherapy	0.90 (0.80-1.02)	0.096
T4 Stage	1.77 (1.59-1.96)	<0.001
Lymphovascular Invasion
Absent	Ref	Ref
Present	1.19 (1.06-1.33)	0.003
Unknown	1.09 (0.92-1.30)	0.331
Perineural Invasion
Absent	Ref	Ref
Present	1.40 (1.22-1.60)	<0.001
Unknown	1.01 (0.81-1.27)	0.921
Tumor Grade
Low Grade	Ref	Ref
High Grade	1.12 (0.99-1.24)	0.052
Unknown	1.03 (0.78-1.36)	0.826
Positive Margins	1.36 (1.18-1.57)	<0.001
< 12 Lymph Nodes	1.27 (1.11-1.45)	<0.001

Table 3.

Adjusted hazard ratios for mortality from stage II colon cancer based on MSI status

	MSS/MSI-L		MSI-H
Variable	HR (95% CI)	p-value	HR (95% CI)	p-value
Adjuvant chemotherapy	0.87 (0.76-0.97)	0.027	1.42 (0.91-2.24)	0.124
T4 Stage	1.72 (1.04-1.32)	0.008	2.61 (1.83-3.72)	<0.001
Lymphovascular Invasion
Absent	Ref	Ref	Ref	Ref
Present	1.22 (1.09-1.38)	0.001	0.90 (0.59-1.37)	0.625
Unknown	1.15 (0.96-1.38)	0.123	0.67 (0.36-1.25)	0.212
Perineural Invasion
Absent	Ref	Ref	Ref	Ref
Present	1.44 (1.26-1.66)	<0.001	1.14 (0.63-2.05)	0.671
Unknown	0.92 (0.72-1.17)	0.499	2.06 (1.13-3.76)	0.018
Tumor Grade
Low Grade	Ref	Ref	Ref	Ref
High Grade	1.18 (1.04-1.32)	0.008	0.86 (0.63-1.17)	0.333
Unknown	0.95 (0.70-1.29)	0.733	1.82 (0.87-3.78)	0.111
Positive Margins	1.33 (1.14-1.54)	<0.001	1.73 (1.07-2.80)	0.025
< 12 Lymph Nodes	1.26 (1.09-1.44)	<0.001	1.56 (0.82-2.95)	0.173

^*Cox regression adjusting for age, sex, race, institution type, primary payor, Charlson-Deyo score, and tumor side

Subgroup analyses

Given the NCCN recommendation to observe all MSI-H stage II colon cancers, we performed a subgroup analysis of 1462 patients with MSI-H cancers who did not receive adjuvant chemotherapy. The aim was to describe the relationship between OS and high-risk features that potentially do not benefit from MSI-H status (T4 stage, positive margins, <12 LNs), in the absence of chemotherapy. Compared to patients with T3 stage tumors, patients with T4 stage tumors in this subgroup again had significantly worse 5-year OS (Figure 3A, 84% vs 68%, p<0.001). Due to smaller sample sizes, cases with positive margins or <12 LNs were combined for analysis. Compared to patients without positive margins or <12 LNs, patients with these features also had significantly worse 5-year OS (Figure 3B, 84% vs 67%, p<0.001). In a model adjusting for patient demographics, tumor side, and other high-risk features, T4 stage (HR 2.22 95%CI 1.47-3.34, p<0.01) and positive margins/<12 LNs (HR 1.66 95%CI 1.07-2.57, p=0.02) were again associated with worse OS.

Figure 3.

A-B. Univariable Kaplan Meier analysis of subgroup of MSI-H patients that did not receive chemotherapy, stratified by T stage and the presence of positive resection margins or <12 lymph nodes in the surgical specimen. LN = lymph nodes.

We next analyzed 481 patients with MSI-H cancers who did not receive chemotherapy and were only considered high-risk if they had pathologic features of LVI, PNI, or high-grade histology. Patients with T4 stage tumors, positive margins, or <12 LNs in their specimen were excluded due to their negative association with OS. On KM analysis, we observed a similar trend to the overall cohort, as LVI, PNI, and high-grade histology had similar 5-year OS to patients without these features (85% vs 85%, p=0.9).

DISCUSSION

While the association between MSI-H status and improved survival in stage II colon cancer has been reproducible in the literature, the prognostic importance of MSI status in stage II colon cancer with high-risk features has not been well described. Our data again show that for patients with stage II colon cancer, MSI-H cancers were associated with favorable OS compared to MSS/MSI-L cancers. Our data add to the body of literature by demonstrating that in patients with MSI-H cancers, certain high-risk features (LVI, PNI, and high-grade histology) were not associated with worse survival compared to cancers without these features, while other high-risk features (T4 stage, positive resection margins, and inadequate nodal sampling) were associated with worse survival. In other words, in MSI-H cancers, some conventional high-risk features were no longer high-risk for worse survival. In contrast, all high-risk features analyzed in this study were associated with worse survival in MSS/MSI-L cancers. It is unclear why only some high-risk features are still associated with worse survival in the presence of MSI-H biology, and others are not. One potential explanation is that features such as positive margins and inadequate nodal sampling represent risks associated with sub-standard surgery that are not outweighed by favorable biology. Similarly, T4 stage suggests potential involvement beyond the plane of surgical resection.

Given the NCCN guideline to observe all MSI-H stage II colon cancers based on lack of response to 5-fluorouracil-based chemotherapy,^11,12 we additionally performed subgroup analyses on patients with MSI-H cancers who did not receive adjuvant chemotherapy. We again showed that T4 stage, positive margins, and inadequate lymph node sampling were associated with worse OS. In contrast, the 5-year OS rate in patients with MSI-H and high-risk features of LVI, PNI, or high-grade histology was equivalent to patients without these features, and quite favorable at 85%. PNI-unknown status was associated with worse OS, possibly due to other negative risk factors with high prevalence in this group that were not accounted for. Notably, our data provide an important descriptive reference for oncologists, and support current NCCN recommendations to forego chemotherapy in MSI-H cancers with otherwise high-risk features, at least inclusive of LVI, PNI, and high-grade histology. Nevertheless, the data also highlight that patients with MSI-H cancers that are T4 stage, have positive margins, or <12 lymph nodes in the surgical specimen would greatly benefit from the identification of adjunctive therapies that may improve prognosis. While it would have been of interest to investigate the role of chemotherapy in this subgroup, the NCDB lacks detail on specific chemotherapeutic agents/regimens used, making it less than ideal for this purpose. Of note, some have suggested that oxaliplatin-based therapy may be beneficial in stage II MSI-H cancers,¹⁷ and the recent NICHE trial demonstrated early promising results in the treatment of mismatch repair-deficient early stage colon cancer with neoadjuvant immune checkpoint inhibitors.¹⁸

Current data describing the relationship between MSI status and high-risk features in stage II disease is limited, and the few studies that have attempted to investigate high-risk features in MSI-H patients are limited by small sample sizes. In a 2016 study of 329 stage II-IV colon cancer patients who underwent resection at a single center, Oh et al.¹⁹ reported that only PNI and disease stage were independently associated with disease free survival, and only disease stage was associated with OS. Interestingly, conventional high-risk features were all associated with worse survival in MSS cancers, but not in MSI-H cancers, which is consistent with our findings in stage II cancers. Baek et al.²⁰ similarly studied 237 patients, of which 76 were MSI-H. In their cohort, MSI status had no prognostic effect on OS or disease-free survival, however they were limited by a small sample size of patients with MSI-H cancers. A 2019 study using NCDB data from 2010-2013 included 843 MSI-H cancers from stage II colon cancer patients.²¹ The study was focused primarily on the effect of adjuvant chemotherapy on MSI-H cancers, however in the cox regression analysis displayed in their supplemental data, MSI-H status was protective, while T4, PNI, and <12 nodes were associated with worse survival. While this analysis is similar to our study, they did not consider the prognostic effect on OS for each feature separately, which proved to be informative in our study.

While previous studies have seldom examined the interaction between MSI status and high-risk features that warrant consideration for chemotherapy per NCCN guidelines, there are many studies that do investigate how MSI status impacts prognosis in early stage colon cancer in light of other important factors. For example, studies^22-24 analyzing the interaction between MSI status and BRAF mutation in colon cancer have observed a similar beneficial effect of MSI status as seen in the present study. That is, BRAF mutations do not appear to be associated with worse survival in the setting of MSI-H cancers, while they are frequently associated with worse prognosis in early-stage MSS colon cancer. These studies assessing the prognostic effect of combinations of markers on colon cancer survival further support our current findings of a complex interplay between MSI status and high-risk features. Furthermore, a 2019 study using a similar NCDB cohort of patients found that MSI-H cancers may not benefit from adjuvant chemotherapy when adjusting for high-risk features, and that MSS cancers benefitted from chemotherapy even in the absence of high-risk features.²¹ Taken together with our findings, the data support that MSI-H cancers seem to have a different natural history than MSS cancers, likely due to important differences in their underlying biology.

There are several limitations to consider when interpreting these data. First, considering our primary outcome of OS, the study cohort had a relatively short median follow-up of 35 months. Second, the NCDB does not provide clinical data for some variables that may impact OS. For example, NCDB does not record whether an operation is done urgent/emergently or if indications such as obstruction or perforation are present. Similarly, a number of important pathologic characteristics that influence survival are not recorded, and some of the variables that are recorded may not be standardized – most relevant to our study may be the lack of data on BRAF status and whether LVI represented extramural vs intramural vascular invasion. Additionally, data on comorbid conditions is limited to the Charlson-Deyo comorbidity index, and we are not able to adjust for specific disease. We attempted to limit any effect of surgical complications by excluding patients who died within 90 days of surgery. Lastly, the database is retrospective and is subject to selection bias for various oncologic treatments. However, the NCDB contains the largest number of incident cancer cases available for analysis, and we leveraged the size of the database to create the largest cohort of MSI-H stage II colon cancers to date.

CONCLUSION

In conclusion, our data demonstrate that MSI status is likely to be an important prognostic factor in stage II colon cancer, however its effect on OS may be dependent on the individual cancers’ high-risk features. Our data have several clinical implications. Importantly, by showing that patients with MSI-H cancers and LVI, PNI, or high-grade histology have equivalent survival to patients with MSI-H cancers without these high-risk features, even in the setting of deferring chemotherapy, the data provide some support for NCCN guidelines recommending no chemotherapy in this group of patients. However, further investigation is needed to determine if chemotherapy or other adjuvant therapies are beneficial in MSI-H stage II colon cancer with other high-risk features. Furthermore, in the research setting, MSI status needs to be a factor in selecting/randomizing patients for clinical trials and should universally be considered in risk adjustment models in retrospective studies.