Risk Factors Associated With Early-onset Colorectal Cancer

Abstract

Background & Aims

The incidence of colorectal cancer (CRC) is increasing in individuals younger than 50 years, who do not usually undergo screening if they are of average risk. We sought to identify risk factors for CRC in this population.

Methods

We compared sociodemographic and medical characteristics of patients who received a diagnosis of CRC at an age of 18–49 years (early-onset) with patients who received a diagnosis of CRC at an age of 50 years or older (late-onset) and with age-matched, cancer-free individuals (controls) at a tertiary academic hospital. We collected data from all adult patients with a diagnosis of CRC from January 1, 2011 through April 3, 2017 from electronic health records. Associations with risk factors were assessed using univariable and multivariable logistic regression models.

Results

We identified 269 patients with early-onset CRC, 2802 with late-onset CRC, and 1122 controls. Compared with controls, patients with early-onset CRC were more likely to be male (odds ratio [OR], 1.87; 95% CI, 1.39–2.51), have inflammatory bowel disease (IBD) (3% vs 0.4% for controls; univariable P<.01), and have a family history of CRC (OR, 8.61; CI, 4.83–15.75). Prevalence values of well-established modifiable CRC risk factors, including obesity, smoking, and diabetes, were similar. Compared to patients with late-onset CRC, patients with early-onset CRC were more likely to be male (OR, 1.44; 95% CI, 1.11–1.87), black (OR, 1.73; 95% CI, 1.08–2.65) or Asian (OR, 2.60; 95% CI, 1.57–4.15), and have IBD (OR, 2.97; 95% CI, 1.16–6.63) or a family history of CRC (OR, 2.87; 95% CI, 1.89–4.25). Sensitivity analyses excluding IBD and family history of CRC showed comparable results. Early-onset CRC was more likely than late-onset disease to be detected in the left colon or rectum (75% vs 59%, P=.02) and at a late stage of tumor development (77% vs 62%, P=.01).

Conclusions

In a retrospective study of patients with early-onset CRC vs late-onset CRC or no cancer, we identified non-modifiable risk factors, including sex, race, IBD, and family history of CRC, to be associated with early-onset CRC.

INTRODUCTION

Colorectal cancer (CRC) is the third leading cause of cancer for women and men in the United States (US) [1]. CRC incidence has declined overall, and this has been attributed to population-level reductions in modifiable risk factors as well as increased participation in screening over the past three decades [2]. In contrast, CRC incidence among individuals younger than age 50 is on the rise, and at the current rate it is estimated to double by 2030 [3].

The reasons for this trend are unclear, but they likely include a combination of non-modifiable and modifiable risk factors. Current guidelines recommend screening at an earlier age for individuals with a family history of CRC and personal history of inflammatory bowel disease (IBD) [4, 5], whereas specific recommendations for individuals with obesity or history of smoking do not exist. The American Cancer Society (ACS) recently advocated lowering the age of screening to 45 years for the general population [6]. This approach has been shown to be cost-effective, but it would still add substantial cost to the healthcare system and there are concerns that it may divert resources away from older individuals who have a higher absolute risk of cancer [7, 8]. Alternatively, identifying specific risk factors in patients with early-onset CRC may permit risk stratification and targeted screening. We performed a single-institution study in a large, diverse metropolitan center to identify sociodemographic, medical, and histologic predictors of early-onset CRC.

MATERIALS AND METHODS

We conducted a retrospective chart review of patients who were seen at NYU Langone Health, a tertiary academic medical center in New York City. The protocol was approved by the NYU School of Medicine Institutional Review Board (Study #17-00077).

Patient selection

We identified all patients aged 18 and older with a diagnosis of CRC who received care at our institution between January 1, 2011 and April 3, 2017 using the electronic health record. Patients were identified as having a history of CRC based on ICD-9 (153.0–154.9) and ICD-10 (C18 to C20) codes. Based on recommendations from the US Multi-Society Task Force on Colorectal Cancer (USMSTF) and most other organizations to begin average-risk screening at 50 years, we defined early-onset CRC patients as those diagnosed between 18 and 49 years of age and created two comparison groups [4, 9]. The first group comprised late-onset CRC patients diagnosed at age 50 or older. The second group comprised controls without cancer who were age-matched to early-onset cases in a 4:1 ratio. Controls were randomly selected from patients who received care at our institution during the study period between 2011 and 2017, without matching for sex. Our initial automatic query identified 297 early-onset CRC cases, 2864 late-onset CRC cases and 1188 age-matched controls. After manual review, we re-classified 11 late-onset cases as early-onset and excluded 8 controls due to an incorrect birth year, leaving a total of 4341 patients (308 early-onset cases, 2853 late-onset cases, and 1180 age-matched controls). We excluded cases with no personal history of CRC (incorrect ICD coding), a personal history of a hereditary CRC syndrome, and substantial missing data. A hereditary CRC syndrome was defined as documentation of Lynch syndrome, familial adenomatous polyposis or MUTYH-associated polyposis, or pathogenic germline mutations in the mismatch repair or epithelial cellular adhesion molecule (EpCAM) genes. Controls with a prior history of cancer were also excluded. All patients received care at our institution, but some were diagnosed elsewhere.

Data collection

We collected data on demographics, clinical history, and CRC outcomes through a combination of automated extraction and manual chart review. Clinical variables of interest included personal history of known CRC risk factors such as IBD, obesity, smoking, and diabetes as well as a family history of CRC. These data were collected prior to diagnosis for most cases and within six months of diagnosis when prior assessment was unavailable. Area-level socioeconomic status including income and education were obtained using a crosswalk of residential ZIP codes and Primary Care Service Areas (PCSAs) from the Dartmouth Atlas [10]. Tumor data (location, stage, grade, and molecular testing) were available for a minority of cases (~17%) from our tumor registry and were supplemented by manual chart review. Missing data was excluded from frequency calculations. When data from the automated extraction conflicted with results of the manual chart review, the latter was considered more accurate and used for analysis.

Study aims

Our primary aim was to identify sociodemographic and clinical risk factors for early-onset CRC by comparing early-onset cases to age-matched controls. Our secondary aim was to compare the early- and late-onset groups to identify differences in sociodemographic factors and clinical risk factors, as well as tumor characteristics. Age-related medical conditions such as hypertension, hyperlipidemia, coronary artery disease, stroke, and diabetes were only compared between early-onset cases and controls. We also performed sensitivity analyses of our primary and secondary outcomes excluding early-onset CRC patients with any family history of CRC or personal history of IBD. To assess the generalizability of our results to the broader New York City population, we also performed an analysis comparing the demographics of our hospital-based controls to a similarly-aged community cohort from the 2013 New York City Health and Nutrition Examination Study (NYC HANES), a survey and physical exam study representative of the entire adult non-institutionalized NYC population [11].

Statistical analysis

For the univariable analysis, we used the chi-squared and Fisher’s exact tests for categorical variables and Student’s t-test and Mann-Whitney U test for continuous variables. A two-sided P < 0.05 was considered statistically significant. Variables with P < 0.20 were carried forward to the multivariable logistic regression model, where we obtained adjusted odds ratios (ORs) and 95% confidence intervals (CIs). Some variables had either insufficient sample size or too much missing data to carry forward to multivariable analysis. Thus, they were left as univariable comparisons for the sake of hypothesis generation. Analysis was performed using R Statistical Software (Foundation for Statistical Computing, Vienna, Austria). Descriptive statistics for NYC HANES were generated using sampling weights to account for the complex survey design.

RESULTS

A total of 4193 individuals met the inclusion and exclusion criteria (Figure 1), of whom 47% were male and 65% were non-Hispanic white. In addition, 0.9% had a personal history of IBD and 5% had a family history of CRC.

Figure 1: Study cohort flow chart.

Abbreviations: CRC, colorectal cancer

Sociodemographic and clinical risk factors of early-onset CRC

Compared to age-matched controls, patients with early-onset CRC were more likely to be male (OR 1.87, 95% CI 1.39–2.51) and have a family history of CRC (OR 8.61, 95% CI 4.83–15.75) or a personal history of IBD (Table 1). Of note, history of IBD was excluded from the multivariable logistic regression model because of the small sample size in the control group. Race as well as area-level mean household income and education level were not significantly different between these two groups. With respect to common age-related comorbidities, hyperlipidemia was more prevalent in the control group, and there was no difference in body mass index (BMI), smoking, coronary artery disease, hypertension, stroke, or diabetes between the two groups (Table 1).

Table 1:

Comparison of individuals with early-onset CRC vs. young controls without cancer

	Variable, n (%)	Early-onset (n=269)	Controls (n=1122)	Univariable Pa	Multivariable OR (95% CI)	Multivariable P
Sociodemographic	Ageb, mean (SD)	43 (6)	45 (6)
	Male sex	146 (54)	501 (45)	<.01	1.87 (1.39 – 2.51)	<.01
	Race/Ethnicity			.26
	Non-Hispanic White	154 (57)	617 (55)
	Non-Hispanic Black	26 (10)	108 (10)
	Non-Hispanic Asian	24 (9)	67 (6)
	Hispanic	10 (4)	45 (4)
	Other	55 (20)	285 (25)
	Income, mean (SD)	72325 (27127)	72703 (27543)	.84
	High school education, %, mean (SD)	85 (9)	85 (9)	.77
Medical	BMI, mean (SD)	27 (6)	28 (6)	.02	.98 (.95 – 1.00)	.06
	Family history of CRC	34 (13)	21 (2)	<.01	8.61 (4.83 – 15.75)	<.01
	Inflammatory bowel disease	7 (3)	5 (.4)	<.01
	Crohns disease	4 (57)	4 (80)
	Ulcerative colitis	3 (43)	1 (20)
	Smoker	70 (27)	312 (29)	.53
	Coronary artery disease	10 (4)	43 (4)	>.99
	Hypertension	50 (19)	221(20)	.85
	Hyperlipidemia	41 (16)	259 (23)	.01	.57 (.38 – .83)	<.01
	Stroke	1 (.4)	8 (.7)	>.99
	Diabetes	19 (7)	64 (6)	.48

Abbreviations: CRC, colorectal cancer; BMI, body mass index

^aChi-squared and fisher exact tests were used for categorical variables and student t-test for continuous variables.

^bAge at CRC diagnosis in early-onset cohort, at time of search in controls (cases and controls matched by birth year).

Compared to patients with late-onset CRC, those with early-onset CRC were more likely to be male (OR 1.44, 95% CI 1.11–1.87), black (OR 1.73, 95% CI 1.08–2.65) or Asian (OR 2.60, 95% CI 1.57–4.15), have a family history of CRC (OR 2.87, 95% CI 1.89–4.25), or have IBD (OR 2.97, 95% CI 1.16–6.63, Table 2). Additional sub-analyses on sex and race are shown in Supplementary Materials. Mean household income and education level were similar between the two groups. The prevalence of common comorbidities such as obesity and diabetes were not compared because these comparisons in older versus younger CRC patients are intractably confounded by age.

Table 2:

Comparison of individuals with early-onset CRC vs. late-onset CRC

	Variable, n (%)	Early-onset (n=269)	Late-onset (n=2802)	Univariable Pa	Multivariable OR (95% CI)	Multivariable P
Sociodemographic	Age at CRC diagnosis, mean (SD)	43 (6)	71 (11)
	Male sex	146 (54)	1335 (48)	.04	1.44 (1.11 – 1.87)	<.01
	Race/Ethnicity			<.01
	Non-Hispanic White	154 (57)	1970 (70)
	Non-Hispanic Black	26 (10)	219 (8)		1.73 (1.08 – 2.65)	.02
	Non-Hispanic Asian	24 (9)	126 (5)		2.60 (1.57 – 4.15)	<.01
	Hispanic	10 (4)	88 (3)		1.64 (.78 – 3.09)	.16
	Other	55 (20)	399 (14)		1.89 (1.34 –2.65)	<.01
	Income, mean (SD)	72325 (27127)	72307 (26135)	.99
	High school education, %, mean (SD)	85 (9)	86 (9)	.38
Medical	BMI, mean (SD)	27 (6)	28 (6)	.02	.98 (.95 – .99)	.04
	Family history of CRC	34 (13)	148 (5)	<.01	2.87 (1.89 – 4.25)	<.01
	Inflammatory bowel disease	7 (3)	27 (1)	.03	2.97 (1.16 – 6.63)	.01
	Crohns disease	4 (57)	21 (78)
	Ulcerative colitis	3 (43)	6 (22)
	Duration of IBD prior to CRC diagnosis, y, median (IQR)	22 (17–25)	13 (0–36)	.71

Abbreviations: CRC, colorectal cancer; BMI, body mass index

^aChi-squared and Fisher exact tests were used for categorical variables and Student’s t-test and Mann-Whitney U test for continuous variables.

Because patients with IBD or a family history of CRC are a high-risk population, we performed a sensitivity analysis that excluded early-onset CRC patients with any family history of CRC or personal history of IBD and compared them to both controls and late-onset CRC cases (Supplementary Tables 1 and 2). The sensitivity analysis did not substantively alter our outcomes and therefore we have reported the main findings including these individuals.

Comparison of control group to the 2013 NYC HANES cohort

To assess whether our hospital-based control group is representative of the local population, we compared demographics and medical co-morbidities with similarly aged participants (aged 27–56) from the 2013 NYC HANES. Comparing frequencies in the control group to the weighted frequencies in NYC HANES, our cohort was older, contained a higher proportion of non-Hispanic whites and persons with coronary artery disease, and contained a lower proportion of smokers (Table 3). The sex distribution, BMI, and prevalence of other common comorbidities such as congestive heart failure, hypertension, hyperlipidemia, and diabetes were similar in the two groups. These data suggest that, for the most part, our hospital-based control group is representative of the NYC population with respect to most key conditions and risk factors linked to CRC.

Table 3:

Comparison of NYU controls vs. NYC HANES 2013

	Variable	NYU, n	NYU, Raw %	NYC HANES, n	NYC HANES, Weighted % (95% CI)
Sociodemographic	Age at search/survey (by 10 years)
	27–36	131	12	397	38.9 (35.3, 42.7)
	37–46	435	39	247	30.5 (26.9, 34.4)
	47–56	556	50	264	30.5 (26.9, 34.5)
	Age at search/survey (by 5 years)
	27–31	41	4	216	21.2 (18.5, 24.1)
	32–36	90	8	181	17.7 (14.9, 21.1)
	37–41	117	10	114	13.6 (11.1, 16.6)
	42–46	318	28	133	16.9 (14.1, 20.0)
	47–51	376	34	126	14.7 (12.1, 17.7)
	52–56	180	16	138	15.9 (12.9, 19.5)
	Male sex	501	45	392	47.3 (44.0, 50.5)
	Race/Ethnicity
	Non-Hispanic White	617	55	294	33.2 (27.7, 39.1)
	Non-Hispanic Black	108	10	208	22.4 (17.3, 28.6)
	Non-Hispanic Asian	67	6	134	16.2 (12.1, 21.3)
	Hispanic	45	4	218	25.2 (20.8, 30.3)
	Other	285	25	54	3.0 (2.2, 4.2)
Medical	BMI (kg/m2), mean (SD)	28	6	28 (0.25)	(27.5, 28.5)
	Smoker	312	29	367	42.1 (38.1, 46.1)
	Coronary artery disease	43	4	8	0.95 (0.47, 1.9)
	Congestive heart failure	6	0.5	9	1.0 (0.5, 2.0)
	Hypertension	221	20	195	22.0 (18.8, 25.5)
	Hyperlipidemia	259	23	218	25.2 (22.0, 28.6)
	Diabetes	64	6	64	7.3 (5.6, 9.4)

Abbreviations: BMI, body mass index

Overall sample sizes: NYC HANES n=908, NYU n=1122

Tumor characteristics of early-onset vs. late-onset CRC

We also compared tumor characteristics of early vs. late-onset CRC in the subset of cases with available data (Table 4). Due to the limited available data, these analyses are exploratory and aimed at hypothesis generation. Early-onset CRC, compared to late-onset cases, was associated with distal tumors (left colon or rectum, 75% vs. 59%, P=.02) and late-stage disease at diagnosis (Stage III/IV, 77% vs. 62%, P=.01). Tumor grade was similar in both groups, with approximately 80% classified as low grade or well differentiated. Early-onset CRC had a lower prevalence of microsatellite instability than late-onset cases (6% vs. 18%, P=.03), but the prevalence of KRAS mutations was similar (48% vs. 41%, P=.52) in the two groups.

Table 4:

Comparison of tumor characteristics of early-onset CRC vs. late-onset CRC

Variable	Early-onset, known n	Early-onset, n (%)	Late-onset, known n	Late-onset, n (%)	Univariable Pa
Site	79		463		.02
Right colon		20 (25)		190 (41)
Left colon		32 (41)		162 (35)
Rectum		27 (34)		111 (24)
Stage	83		460		.01
Early stage		19 (23)		173 (38)
Late stage		64 (77)		287 (62)
Grade	73		412		.53
Low grade		58 (80)		343 (83)
High grade		15 (21)		69 (17)
KRAS mutation	98	47 (48)	49	20 (41)	.52
MSI	117	7 (6)	56	10 (18)	.03

Abbreviations: MSI: microsatellite instability

^aChi-squared and Fisher exact tests were used for analysis.

Age of diagnosis of early-onset CRC patients and family history of CRC

Among patients with early-onset CRC, 12 individuals (4%) were diagnosed at 20–29 years of age, 60 (22%) at 30–39 years, and 197 (73%) at 40–49 years. Thirty-four of the 269 patients (13%) in this group had a documented family history of CRC, of whom 17 had a first-degree relative with CRC and therefore should have received screening before age 50. A slight majority (19/34, 56%) of the early-onset CRC patients with a family history of CRC were diagnosed before current USMSTF guidelines would have recommended screening [4]. They were diagnosed a median 6 years (range 1–29) before the recommended screening age. The remaining 15 patients (44%), all of whom had indications for screening before age 50, were diagnosed a median 6 years (range 0–19) years after the recommended screening age.

The vast majority of early-onset CRC patients (252/269, 94%) were considered average-risk for screening. Of these 252 individuals, 101 (40%) were diagnosed between ages 46–49, 15 (6%) at age 45, and the other 136 (54%) between ages 21–44.

DISCUSSION

In this large retrospective study of early-onset CRC, we identified male sex, family history of CRC, and personal history of IBD as predictors of early-onset CRC compared to both age-matched controls and late-onset CRC cases. Additionally, early-onset CRC patients were more likely to be black or Asian compared to late-onset CRC patients. Common age-related comorbidities were not more prevalent in early-onset cases than age-matched controls, and socioeconomic factors were not significant risk factors. These data show certain non-modifiable risk factors contribute to early-onset CRC.

After excluding all known cases of hereditary cancer syndromes, we found early-onset CRC patients were more than eight times as likely to have a family history of CRC compared to controls and nearly three times more likely than late-onset CRC patients. CRC diagnosed before age 50 has been more strongly associated with family history of CRC or probable hereditary syndrome than CRC diagnosed later in life. Chen et al found an 8% higher absolute prevalence of family history in young-onset CRC cases compared to cases diagnosed at 50 years or older (25% vs. 17%, P=.03) [12]. Although our study had a lower overall prevalence of positive family history due to the exclusion of hereditary syndromes, we also found an 8% absolute difference in prevalence of family history in the early-onset and late-onset CRC groups (13% vs. 5%, P<.01). The prevalence of hereditary syndromes is significantly higher in early-onset CRC cases than in healthy controls, but to our knowledge the association with family history of CRC in the absence of hereditary syndromes has not previously been reported. This could be attributed to weaker genetic risk factors for early-onset CRC including intermediate penetrant genes, low-risk genetic variations with additive effect, and genetic variants that modify the expression of known CRC susceptibility genes [13]. Our results also confirm that IBD is a risk factor for early-onset CRC [14, 15]. The prevalence of IBD in early-onset CRC patients compared to controls has not been previously examined, but a recent Swedish study found that patients diagnosed with IBD as children were more likely to be diagnosed with early-onset CRC than individuals without IBD [16].

The finding that Asians and blacks have higher rates of early-onset CRC than non-Hispanic whites is supported by studies using the Surveillance, Epidemiology, and End Results (SEER) registry [17, 18]. A 1973–2009 SEER analysis found that the proportion of CRC patients who were diagnosed under age 50 was 1.8-fold higher in Asians/Pacific Islanders than in non-Hispanic whites, however overall CRC incidence in Asians was lower [17]. It has been hypothesized that the rising incidence of CRC in Asians may be attributed to the adoption of the Western diet [19]. Asians who are born in the US and exposed to a lifetime Western diet may carry a higher risk for early-onset and overall CRC. On the other hand, the majority of Asians in the US are immigrants who may have less exposure to the Western diet. Thus, lifestyle differences between Asians born in the US and those who immigrated may explain the discrepancy between overall and early-onset CRC risk. With respect to black patients, a study of rectal and rectosigmoid cancers in individuals under age 40 years found a higher absolute incidence in blacks compared to whites (0.67 per 100,000, 95% CI, 0.60–0.74 vs. 0.51 per 100,000, 95% CI, 0.48–0.53) [18].

In contrast to a large prospective cohort study that found higher risk for early-onset CRC in obese women [20], we did not observe an association between obesity and early-onset CRC. A large retrospective study comparing young CRC cases and controls found specific dietary components, but not obesity or diabetes, to be risk factors for early-onset CRC [21]. Therefore, the relationship between metabolic conditions such as obesity or diabetes and early-onset CRC may be confounded by unmeasured dietary or environmental factors. Additional prospective studies with dietary and environmental exposure data are needed to further investigate this relationship.

Socioeconomic disparities in CRC incidence are well-documented and have been attributed to a higher burden of predisposing comorbidities and lower rates of screening in groups with lower socioeconomic status [22]. That socioeconomic status was not a risk factor for early-onset CRC in our study may be explained by our patient demographics. The relative affluence of our New York City population (area-level mean household income greater than $72,000) and abundant local resources for CRC screening may have reduced our ability to detect disparities related to healthcare access. Moreover, it is not clear that socioeconomic disparities observed for CRC overall are applicable for early-onset disease, since biological mechanisms may be distinct and access to screening may be less relevant for early-onset cases.

The finding that early-onset CRC presents at a more advanced stage than late-onset disease is consistent with the literature [12, 23, 24]. Ample evidence suggests the existence of biological differences in early- vs. late-onset CRC. First, the rising incidence of early-onset CRC is predominantly due to left-sided disease, in contrast to a higher proportion of right-sided tumors in late-onset CRC [24–27]. Second, early-onset CRC exhibits a higher prevalence of mucinous or signet-ring histology with poor differentiation than that of older adults [27]. Lastly, as our study demonstrated, late-onset CRC also has a higher prevalence of microsatellite instability (MSI), which is associated with early-stage [23] and right-sided disease [28]. Prior studies have shown that approximately 12–17% of all CRC is positive for MSI, and the majority of these are sporadic [29].

Forty percent of the average-risk early-onset cases in our population were diagnosed between 46–49 years, which may support the ACS recommendation to start screening at age 45. However, it is unclear what proportion of these cases would have been prevented or detected at an earlier stage. Additionally, as a recent Markov model analysis demonstrated, targeted screening in high-risk individuals may be a more cost-effective approach [8]. Consequently, simple risk scores that can identify individuals at highest risk for early-onset CRC are needed. Our data highlight that efforts to construct these scores should include non-modifiable risk factors.

The strengths of our study include its large sample size, the availability of sociodemographic, medical, and tumor data, as well as a hospital-based control group that is representative of the local population. However, several limitations should be noted. First, certain established risk factors—such as dietary history, physical activity, and aspirin use—were not included in the analysis because they were either unavailable or could not be easily extracted from the medical record. Second, because data was collected retrospectively, it is possible that certain medical data (e.g., BMI) for cases may have been influenced by cancer itself. We tried to minimize this bias by including medical data obtained before or within 6 months of CRC diagnosis for cases diagnosed at our institution. Third, although we conducted a manual chart review to identify and exclude patients with hereditary syndromes, it is possible that some genetic testing was missing and a small number of patients with hereditary syndromes were included in the analysis. Fourth, because not all patients underwent surgery and some received resection outside of our institution, only a minority of patients had tumor data available. Therefore, the tumor analysis should be considered exploratory and the results interpreted with caution. Finally, there may be selection bias in that the hospital-based control cohort may not reflect a truly healthy population. However, our comparison to NYC HANES showed that cancer-free controls were comparable to the NYC population with respect to important predictors of CRC.

In summary, the majority of early-onset CRC cases in our single-center study were sporadic. Patients with early-onset CRC were more likely to be male and have a family history of CRC or personal history of IBD. In addition, they were more likely to be black or Asian compared to individuals with late-onset CRC. We did not observe associations with well-established modifiable risk factors such as obesity, smoking, and diabetes. These data suggest non-modifiable factors should be included in risk prediction models to facilitate targeted screening in individuals under age 50. Such efforts can be refined as more granular predictors of early-onset CRC, especially early-life exposures that are measurable and readily available in the clinical setting, are identified from future prospective studies.

What You Need to Know.

Background

The incidence of early-onset colorectal cancer (CRC, age younger than 50 years) is increasing, but associated risk factors are not well-understood.

Findings

In a retrospective study of patients with early-onset CRC vs late-onset CRC or no cancer, we identified non-modifiable risk factors, including male sex, black or Asian race, inflammatory bowel disease, and family history of CRC, to be associated with early-onset CRC. We did not identify an association with modifiable risk factors, including obesity, smoking, and diabetes.

Implications for patient care

Risk-stratification efforts for early-onset CRC should include these non-modifiable risk factors.

Abbreviations

ACS

American Cancer Society

BMI

Body mass index

CI

Confidence interval

CRC

Colorectal cancer

HANES

Health and Nutrition Examination Study

IBD

Inflammatory bowel disease

MSI

Microsatellite instability

NYC

New York City

OR

Odds ratio

PCSA

Primary care service area

SEER

Surveillance, Epidemiology, and End Results

US

United States

USMSTF

US Multi-Society Task Force