Age Determines Adjuvant Chemotherapy Use in Resected Stage II Colon Cancer

Abstract

BACKGROUND:

The role of adjuvant chemotherapy in resected stage II colon cancer remains controversial. Recommendations to treat rely largely on the presence of certain high-risk features for recurrence.

OBJECTIVE:

We sought to define patient and clinicopathologic differences between early-onset and late-onset CRC and determine if these differences impact treatment. We hypothesized that high-risk features in stage II colorectal cancer differed between age groups and would most strongly influence administration of adjuvant chemotherapy.

DESIGN:

This was a retrospective cohort study.

SETTING:

The study was conducted at a Commission on Cancer designated hospital as well as the National Cancer Institute Intramural Research Program.

PATIENTS:

Patients with resected stage II colon cancer were identified in the National Cancer Database, and clinicopathologic characteristics were recorded. Patients were stratified into young (≤45), middle-aged (50–75), and older (>75) age groups.

MAIN OUTCOME MEASURES:

Incidence of high-risk clinicopathologic features and receipt of adjuvant chemotherapy were measured.

RESULTS:

A total of 14,966 patients met inclusion criteria. Young patients were found to have had at least one high-risk feature (n = 489, 44%) slightly more often than both middle-aged (n = 3734, 40%) and older patients (n = 1890, 42%). A total of 332 (7%) of older patients received adjuvant chemotherapy compared to 627 (56%) of young patients and 2854 (30%) of middle-aged patients. Age group was independently associated with receipt of adjuvant chemotherapy when controlling for relevant clinicopathologic factors.

LIMITATIONS:

This was a retrospective study without granular detail on treatment decisions.

CONCLUSIONS:

Young patients are frequently prescribed adjuvant chemotherapy for both high- and low-risk tumors despite questionable benefit in the latter. Older patients rarely receive adjuvant therapy. Both medical and surgical oncologists should be aware of disparities in cancer treatment and remain conscientious about making treatment decisions solely based on age. See Video Abstract at http://links.lww.com/DCR/B846.

INTRODUCTION

Colorectal cancer (CRC) is the second leading cause of cancer-related death in the United States.¹ Resection with regional lymphadenectomy is the mainstay of treatment for locoregional disease. Randomized trials have demonstrated a survival benefit with adjuvant chemotherapy (AC) for stage III disease.^2–5 The role of AC in stage II disease, however, remains controversial. One randomized trial has shown a statistically significant, albeit small, benefit from systemic therapy after resection of stage II disease.⁶ Furthermore, a recent study of the National Cancer Database (NCDB) found that adjuvant treatment was associated with prolonged overall survival in a propensity-score matched analysis.⁷ Meanwhile, other randomized trials and population-based analyses have failed to detect a difference.^8–11 Multiple clinicopathologic characteristics have been identified to categorize certain stage II patients as high-risk for recurrence.^12,13 Thus, AC is currently recommended for pathologic stage II tumors with high-risk features.¹⁴

In recent years, the incidence rates of CRC among those younger than age 50 have been rising, in contrast with the declining overall incidence of CRC.¹⁵ Patients with early-onset CRC tend to present at an advanced stage yet experience a more favorable stage-for-stage prognosis compared to those who develop CRC at an older age.^16,17 Lack of guideline-driven screening in the younger age group may provide some explanation for the higher frequency of advanced-stage tumors. However, there are recognized differences in molecular and histopathologic characteristics that may contribute to more rapid tumor progression.^18,19 Despite multiple studies examining difference in tumor biology across age groups, there is a knowledge gap in defining how this information is translated to guide treatment decisions. This is particularly true in stage II disease, where the benefit of AC is less clear.

In this study, we set out to determine who receives adjuvant chemotherapy for resected stage II colon cancer and why. To do this, we sought to define patient and clinicopathologic differences between early-onset CRC (EOCRC) and late-onset CRC (LOCRC) and determine if these differences impact treatment. We hypothesized that high-risk features in stage II colorectal cancer differed between age groups and would most strongly influence administration of adjuvant chemotherapy. The first aim of this study was to compare demographics and prevalence of high-risk tumor characteristics between young, middle-aged, and older patients. The second aim was to discover what patient features are associated with receipt of AC after resection of pathologic stage II CRC.

METHODS

Data Source

We performed a retrospective cohort study using the NCDB Participant User Files (PUFs). The NCDB is a joint program of the American College of Surgeons Commission on Cancer (CoC) and the American Cancer Society (ACS). Data is sourced from 1,500 Commission-accredited cancer programs in the United States. The NCDB captures about 70% of all newly diagnosed cases of cancer in the United States and it contains information on patient demographics, tumor characteristics, initial treatment and outcomes. The CoC and the ACS have not verified and are not responsible for the analytic or statistical methodology used or for the conclusions drawn from these data. This study was exempt from review by our respective Institutional Review Boards.

Selection of the Study Population

The evaluated PUFs were comprised of patients aged 18 and over who were diagnosed between 2010 and 2016. Cases were identified by searching PUFs of colon and rectosigmoid junction tumors for those patients who underwent resection and had pathologic stage II tumors, defined by the AJCC Cancer Staging Manual 8^th edition as T3 or T4 and N0. The cohort was further narrowed by selecting only those with using the International Classification of Diseases for Oncology, 3rd edition (ICD-O-3) morphology codes for adenocarcinoma, with delineation of mucinous and signet ring histologic subtypes. Those who either refused adjuvant chemotherapy, underwent neoadjuvant chemotherapy, or were not offered chemotherapy based on perceived risk were excluded. Those identified as having tumors that were microsatellite unstable (MSI) or had missing information on microsatellite stability were also excluded. To further eliminate the presence chronic systemic disease as a confounder, we included only those patients who had a Charlson-Deyo score of 1 or less. In order to test our hypothesis that early-onset colon cancer carries a different biology that late-onset disease, we sought to clearly demarcate the young group from the rest of the study population. To do this, and to eliminate ambiguity about those who fell within screening guidelines as they have evolved over time, patients aged 46–49 were not included in the analysis. Details of cohort selection are outlined in Figure 1.

Figure 1.

Diagram outlining selection of the study cohort. NCDB participant user files (PUFs) containing cases of colon cancer and rectosigmoid junction cancer were searched. The arrow box pointing towards the right denotes patients who were excluded from the analysis. High risk features were considered to be indeterminate/positive margin, T4, lymphovascular invasion, perineural invasion, and high or intermediate grade. NETs = neuroendocrine tumors, MSI = microsatellite instability, MS = microsatellite status

Patient and Tumor Characteristics

Patients were categorized into young (≤ 45), middle-aged (50–75), and older (>75) age groups. Age cutoffs were chosen based on screening recommendations of the ACS, NCCN, and USPSTF.^20–22 Specifically, the ACS recommends screening for average-risk individuals starting at age 45 compared to 50 for the NCCN and USPSTF. All three organizations recommend only selective screening for patients aged 76–85. Demographic information such as sex, race, ethnicity, income, education level, insurance status, tumor location, size, and histology were collected. Frequencies were compared among age groups. Tumor laterality was delineated by ICD-O-3 topography codes. High-risk pathologic features for recurrence were also compared, which included pT4 tumors, lymphovascular invasion, perineural invasion, high grade (Grade 3 or 4), positive or indeterminate margin, and fewer than 12 lymph nodes examined in the pathologic specimen.^11,12 Of note, the NCDB does not provide information on clinical risk factors, namely concomitant bowel obstruction or emergent resection.

Adjuvant Chemotherapy Utilization

In order to determine factors associated with the utilization of AC, frequencies of use were gathered for the entire unadjusted cohort and also after covariate stratification. Characteristics examined included age group, sex, socioeconomic factors, tumor location, histology, facility type, and previously defined high-risk characteristics. Bivariable comparisons were made. To identify those factors independently associated with chemotherapy use, multivariable analysis was then performed using variables that were statistically significant (p < 0.05) on bivariable analysis.

Statistical Analysis

Statistical analysis was done using SPSS® software version 25.0 (IBM Corporation, Armonk, NY, USA). Categorical variables were compared using a chi-squared or Fisher’s exact test where appropriate. Multivariable analysis was done using a binary logistic regression model. To avoid discovery bias, factors included in the regression model were limited to those with p < 0.05 on bivariable analysis. A two-tailed p value of < 0.05 was considered statistically significant.

RESULTS

Patient and Tumor Characteristics

A total of 14,966 patients met inclusion criteria. We identified 1,124 patients aged 45 and under, 9,352 between the ages of 50 and 75, and 4,490 patients over the age of 75. Among patients in the young age group, a marginally lower percentage were white when compared to the middle-aged and older age groups. When compared to young and middle-aged patients, older patients were less likely to be uninsured or to live in low-income or low-education level areas (Table 1).

Table 1.

Demographics and clinicopathologic characteristics of young, middle-aged, and older patients meeting inclusion criteria.

Stage II pT3/T4 N0	Age ≤45 (%) n = 1,124	Age 50–75 (%) n = 9,352	Age > 75 (%) n = 4,490	p
Female	567 (50)	4343 (46)	2420 (54)	<0.001
Race				<0.001
White	859 (76)	7610 (81)	3934 (88)
Black	169 (15)	1209 (13)	335 (8)
Asian	60 (5)	342 (4)	135 (3)
Other	36 (3)	191 (2)	86 (2)
Hispanic	29 (3)	150 (2)	43 (1)	<0.001
Uninsured	82 (7)	377 (4)	16 (<1)	<0.001
Charlson-Deyo Score				<0.001
0	1017 (91)	7070 (76)	3151 (70)
1	107 (9)	2282 (24)	1339 (30)
Area Med Income Quart				< 0.001
1	455 (41)	3493 (38)	1794 (40)
2	257 (23)	2125 (23)	1054 (24)
3	200 (18)	1963 (21)	923 (21)
4	203 (18)	1650 (18)	675 (15)
Area % w/o HS Degree Quart				<0.001
1	251 (22)	1921 (21)	770 (17)
2	267 (24)	2402 (26)	1037 (23)
3	312 (28)	2550 (28)	1331 (30)
4	287 (26)	2374 (26)	1318 (30)
Prior Malignancy	42 (4)	1307 (14)	1196 (27)	<0.001
Tumor Location				<0.001
Right Colon	455 (42)	4899 (54)	2878 (66)
Left Colon	642 (58)	4235 (46)	1482 (34)
Histology				0.034
Adenocarcinoma	1014 (90)	8564 (92)	4042 (90)
Mucinous	104 (9)	756 (8)	430 (10)
Signet Ring	6 (1)	32 (<1)	18 (<1)
High-Risk Features
Median Tumor Size (IQR)	5.2 (4.0,7.0)	4.7 (3.5,6.2)	4.7 (3.5,6.0)	<0.001
pT4	200 (18)	1416 (15)	628 (14)	0.005
Grade 3/4	158 (14)	980 (11)	616 (14)	<0.001
LVI+	174 (16)	1227 (13)	677 (15)	0.002
PNI+	102 (9)	730 (8)	341 (8)	0.251
Pos/Ind Margin	41 (4)	267 (3)	129 (3)	0.321
<12 LN examined	36 (3)	544 (6)	354 (8)	<0.001
Any High Risk	489 (44)	3734 (40)	1890 (42)	0.009
Adjuvant Chemo	627 (56)	2854 (30)	332 (7)	<0.001
Single Agent	206 (18)	1194 (13)	213 (5)
Multi Agent	390 (35)	1507 (16)	94 (2)
Unspecified	31 (3)	144 (2)	25 (1)

Area Med Income Quart = Indicates the median income of the patient’s zip code in quartiles.

Area % w/o HS Def Quart = Indicates the number of adults in the patient’s zip code who obtained a high school degree in quartiles.

IQR = Interquartile range

LVI = Lymphovascular invasion; PNI = Perineural Invasion

Pos/Ind = Indicates a positive or indeterminate margin

There were several histopathological characteristics that differed among age groups. Notably, prevalence of right-sided colon tumors increased with age (Table 1). Additionally, pT4 tumors were more frequently seen in young patients compared to middle-aged and older patients (Table 1). Middle-aged patients had the lowest proportion of high-grade tumors and of lymphovascular invasion, whereas these factors were equally prevalent in young and older patients (Table 1). Perineural invasion and positive/indeterminate margin status were similar across all age groups (Table 1). In total, young patients were found to have had at least one high-risk feature significantly more often than middle-aged and older patients (Fig. 2).

Figure 2.

Incidence of high-risk features broken down by age group. The “any high risk” column indicates the incidence of patients within each group possessing at least 1 high-risk feature. For context, frequencies of adjuvant chemotherapy use are displayed as well. LVI = lymphovascular invasion, PNI = perineural invasion, LN = lymph nodes

Adjuvant Chemotherapy Utilization

We identified large differences in frequencies of utilization of AC among age groups. A total of 489 patients in the young group received AC (44%) compared to just 3,734 middle-aged patients (30%) and 1,890 older patients (7%) (Table 1). Furthermore, older patients were significantly more likely to receive single-agent AC, whereas the majority of middle-aged and young patients received multi-agent regimens. Of note, a total of 2,039 patients who met all other inclusion criteria were offered AC but refused. When including these patients with the rest of the cohort, we found that just 12% of older patients refused treatment compared to 13% of middle-aged and 9% of young patients (p < 0.001).

Among those patients who demonstrated at least one high-risk feature for recurrence, there was a persistent disparity. A higher percentage of patients in the young age group received AC in the setting of high-risk disease (n = 347, 31%) compared to those in the middle-aged (n=1,832, 20%) and older age groups (n = 243, 5%). Only 13% of young patients with high-risk disease did not receive AC compared to 20% of middle-aged and 37% of older patients. Notably, a quarter of young patients received AC in the absence of high-risk features compared to 11% and 2% of middle-aged and older patients, respectively (Fig. 3). Furthermore, of those with high-risk features who received AC, multiagent therapy was given to young patients 75% of the time, compared to 61% and 31% of middle-aged and older patients respectively.

Figure 3.

Chemotherapy utilization stratified by risk status for each age group (≤45, 50–75, > 75).

A number of demographic, socioeconomic, and histopathologic factors were associated with use of AC after resection of stage II colon cancer on univariate analysis, namely age, race, insurance status, prior malignancy, tumor location, histology, tumor size, pT stage, grade, LVI, PNI, margin status, number of examined nodes, and facility type (Table 2). The multivariate analysis identified multiple patient and tumor characteristics that were independently associated with receipt of AC, the strongest of which was age group. Young and middle-aged patients were far more likely to receive AC than older patients (young OR 19.3, 95%CI 15.5–23.9; middle-age OR 7.09, 95%CI 6.2–8.2). Clinicopathologic factors independently associated with AC included Charlson-Deyo score of 0, pT4 stage, positive/indeterminate margin, fewer than 12 examined nodes, and presence of LVI (Table 2).

Table 2.

Bivariable and multivariable analyses of demographic and clinicopathologic factors associated with receipt of adjuvant chemotherapy for resected stage II colon cancer.

Stage II pT3/T4 N0	No Chemo n = 11,162	Adjuvant Chemo n = 3,804	P (bivariable)	Adjusted OR (95% CI)	p (multivariable)
Sex			0.854
Female	5462 (75)	1936 (25)
Male	5700 (75)	1868 (25)
Age			<0.001
≤45	497 (44)	627 (56)		19.3 (15.5–23.9)	<0.001
50–75	6507 (70)	2845 (30)		7.09 (6.2–8.2)	<0.001
> 75	4158 (93)	332 (7)		Ref
Race			0.037
Black	1242 (73)	471 (27)		0.98 (0.9–1.1)	0.778
Asian	386 (72)	151 (28)		1.10 (0.9–1.4)	0.409
Other	244 (78)	69 (22)		0.80 (0.6–1.1)	0.190
White	9290 (75)	3113 (25)		Ref
Hispanic			0.102
Yes	156 (70)	66 (30)
No	10322 (75)	3429 (25)
Charlson-Deyo Score			<0.001
0	8210 (73)	3028 (27)		1.19 (1.1–1.3)	0.001
1	2952 (79)	776 (21)		Ref
Area Med Income Quart			0.769
1	4303 (75)	1439 (25)
2	2548 (74)	888 (26)
3	2311 (75)	775 (25)
4	1874 (74)	654 (26)
Area % w/o HS Deg Quart			0.066
1	2148 (73)	794 (27)
2	2763 (75)	943 (25)
3	3179 (76)	1014 (24)
4	2968 (75)	1011 (25)
Insurance Status			< 0.001
Insured	10759 (75)	3581 (25)		0.86 (0.7–1.1)	0.189
Uninsured	292 (62)	183 (39)		Ref
Prior Malignancy			< 0.001
Yes	2093 (82)	452 (18)		0.81 (0.7–0.9)	0.002
No	9069 (73)	3352 (27)		Ref
Tumor Location			< 0.001
Right Colon	6525 (79)	1707 (21)		0.67 (0.6–0.7)	<0.001
Left Colon	4355 (69)	2004 (31)		Ref
Histology			0.003
Mucinous	922 (72)	368 (28)		1.18 (1.0–1.4)	0.036
Signet Ring	35 (63)	21 (37)		1.13 (0.6–2.2)	0.734
Adenocarcinoma	10205 (75)	3415 (25)		Ref
Median Tumor Size (IQR)	4.5 (3.5,6.0)	5.0 (3.7,7.0)	<0.001	1.02 (1.01–1.03)	0.001
T stage			<0.001
T4	910 (41)	2470 (19)		6.75 (6.0–7.6)	<0.001
T3	10252 (81)	1334 (59)		Ref
Grade			<0.001
High Grade	1157 (66)	597 (34)		1.56 (1.4–1.8)	<0.001
Low Grade	10005 (76)	3207 (24)		Ref
LVI			<0.001
Present	1318 (63)	760 (37)		1.83 (1.6–2.1)	<0.001
Absent	9844 (76)	3044 (24)		Ref
PNI			<0.001
Present	729 (62)	444 (38)		1.40 (1.2–1.6)	<0.001
Absent	10433 (76)	3360 (24)		Ref
Margin			<0.001
Pos/Ind	163 (37)	274 (63)		2.42 (1.9–3.1)	<0.001
Negative	10999 (76)	3530 (24)		Ref
Nodes examined			<0.001
< 12	637 (68)	297 (32)		1.89 (1.6–2.2)	<0.001
≥12	10525 (75)	3507 (25)		Ref
Facility Type			0.039
Academic	4596 (75)	1551 (25)		1.14 (1.0–1.2)	0.004
Non-Academic	6375 (76)	1985 (24)		Ref

Area Med Income Quart = Indicates the median income of the patient’s zip code in quartiles.

Area % w/o HS Def Quart = Indicates the number of adults in the patient’s zip code who obtained a high school degree in quartiles.

LVI = Lymphovascular invasion; PNI = Perineural Invasion; Pos/Ind = Indicates a positive or indeterminate margin

To better characterize factors associated with administration of AC within the specified age groups, multivariable analyses using age as a continuous variable were performed. Within the young age group, there was no per-year association of age with AC use (OR 1.00, 95%CI 0.91–1.1), whereas a decreased likelihood of AC administration was associated with each year increase in age within the middle and older age groups (middle-aged OR 0.95, 95%CI 0.94–0.95; older age OR 0.81, 95%CI 0.78–0.84). Within each group, the strongest determinant of AC use was the presence of a T4 lesion.

DISCUSSION

In this study, we have demonstrated that young patients with Stage II CRC were slightly more likely to harbor high-risk features when compared to their middle-aged and older counterparts, yet were disproportionately more likely to receive AC. We had originally hypothesized that the presence or absence of established high-risk pathologic features in Stage II disease, namely T4 stage, high grade, presence of LVI or PNI, positive margin(s), or fewer than 12 lymph node examined, would be the primary factors associated with AC utilization. In fact, we found that patient age carried the strongest association with AC administration. Additionally, while the presence of at least one high-risk feature was more likely among young and older patients, young patients were far more likely to be given AC even in the absence of such factors. Similarly, older patients were far less likely to receive AC even in the presence of established high-risk features. Furthermore, older patients were far more likely to receive single-agent chemotherapy compared to younger patients. While we recognize the importance of a nuanced approach to each patient’s care, we also expected rates of AC administration to more closely mirror rates of high-risk features. Although the underpinnings of our observations are certainly multifactorial, we believe that these data raise awareness of national practice patterns among both medical and surgical oncologists participating in the care of stage II CRC patients of all ages.

AC administration was exceedingly prevalent among patients with EOCRC, even among those whose lacked the indication for the use of adjuvant systemic therapy. Two prior studies analyzing data from the military healthcare system also found similarly high rates of AC utilization among younger patients across all stages.^16,23 Surprisingly, a substantial portion of EOCRC patients who underwent AC had no high-risk features for recurrence. This is of particular interest given the lack of data for use of AC outside of the high-risk population. Multiple studies have failed to identify a survival benefit from AC in stage II CRC.^3,8,9,24 However, a recent study focused on patients with high-risk features suggests that they may benefit.²⁵ Several potential explanations may help rationalize these findings. First, we believe that clinician bias may result in the assumption of higher physiologic tolerance or improved oncologic benefit for younger patients receiving AC, despite a lack of evidence for either concept. In fact, a study examining efficacy and tolerability of chemotherapy for advanced CRC across multiple trials found no differences between younger and older patients.²⁶ Second, some studies suggest that EOCRC is biologically distinct from LOCRC and requires a different treatment approach.¹⁹

Although we did find a slightly higher incidence of recognized high-risk pathologic features in the young population, this finding alone cannot entirely explain the discrepancy in management patterns. It is clear, based on our data, that histopathologic characterization is not the only factor determining management. Ultimately, oncologist perception of CRC among younger and older patients may result in dissimilar treatment approaches. Clearly, more tools are needed that serve the purpose of making data-driven decisions regarding AC for resected stage II CRC, regardless of age. A comprehensive geriatric assessment may serve to better select older patients who will derive the most benefit from AC and therefore guide treatment-related counseling.²⁷ Emerging technologies, such as ctDNA, may also fill this role to aid clinicians.²⁸

While it is challenging to quantify and isolate the causative relationship of age-related bias in cancer care, studies have attempted to describe its prevalence and resultant outcomes.^29,30 Our finding that older patients rarely underwent AC regardless of risk status underlines that there are non-biologic and non-data-driven factors governing treatment decisions. Furthermore, when older patients did receive AC, they were prescribed single-agent regimens much more frequently. Some of this underuse may be linked to patient preferences. However, when evaluating the patient cohort with patients who were documented as having declined chemotherapy, we observed similar rates of refusal across the age groups. There was a statistically significant association of age group with refusal, owing to the slightly smaller proportion of young patients who declined. Though we eliminated those patients who were documented to have refused AC when it was offered from our final analysis, this may not account for a shared decision-making process in which the patient expressed desire not to undergo further treatment and thus was not offered AC. Early-onset colon cancer certainly carries emotional weight for both patient and clinician that may not be present in older or even middle-aged patients. The risk-to-benefit ratio of adjuvant therapy may indeed appear different among patients at different ends of the age spectrum. This should not distract from the idea that an age bias on the part of the practitioner may still exist, despite data to suggest that elderly patients derive the same benefit from AC for CRC as do younger counterparts without additional adverse effects.^9,31,32 Thus, it stands to reason that utilization should not be so discordant between age groups. Our findings parallel those from a previous study of stage III CRC patients demonstrated that despite unquestionable benefit in patients with positive nodes, older patients received AC far less often than young patients.³³ It is important to realize that advanced chronologic age does not equate to diminished returns nor does it imply decreased ability to tolerate either single or multi-agent therapy.

Aside from chemotherapy use, several differences were found between age groups with respect to tumor characteristics in patients with CRC. For instance, right-sided tumors were more common in older and middle-aged patients, which is consistent with previous reports.^34,35 Surprisingly, despite its known poorer prognosis when compared to left-sided CRC, patients with right-sided CRC were less likely to undergo AC, even when controlling for age and the presence of high-risk features. Though right-sided tumors are more likely to be microsatellite unstable (MSI) and thus not receive the same benefit from AC, MSI tumors were excluded from this study. It should also be noted that despite having a higher prevalence of right-sided tumors, elderly patients had a similar or lower prevalence of multiple high-risk histopathologic features such as LVI, PNI and pT4 tumors.

This study is not without limitations. As with any retrospective analysis, there are unmeasured confounders that, when unaccounted for, may alter our results. For example, many patients have missing information on factors of interest such as chemotherapy administration and high-risk features which may alter conclusions. Additionally, tumor-level data such as mutational status, microsatellite stability, and tumor budding which may influence treatment decisions are frequently missing or not captured by the NCDB. We attempted to mitigate this by eliminating patients with missing or ambiguous data. Furthermore, our retrospective cohort may also be subject to selection bias. Although we used a highly selected study population in order to eliminate confounding factors, there may be patient-level information not captured by the NCDB that influenced decisions regarding use of AC. For instance, frailty assessments, performance status, and length of recovery from surgery are often considered when determining oncologic treatment. Unfortunately, these measures are not captured by the NCDB and are therefore unaccounted for in our study. Finally, while the NCDB does capture the number of systemic agents used, details on type and duration of chemotherapy are not present. A more detailed study including the agents, duration, and overall tolerance of AC may aid in further understanding patterns of its utilization.

In summary, we demonstrated that despite small differences in prevalence of high-risk features, elderly patients receive AC at an alarmingly low rate compared to young and middle-aged patients. Young patients frequently undergo AC for CRC with both high- and low-risk features despite questionable benefit in the latter. Both medical and surgical oncologists should be aware of conscious and unconscious biases that may lead to disparities in cancer treatment and remain conscientious about making treatment decisions solely based on chronologic age.