Early-onset Colorectal Cancer is Distinct From Traditional Colorectal Cancer

Accepted Manuscript Early-Onset Colorectal cancer is distinct from traditional colorectal cancer Heather Yeo, MD MHS, Doron Betel, PhD, Jonathan S. Abelson, MD, Xi E. Zheng, MD PhD, Rhonda Yantiss, MD, Manish A. Shah, MD PII:

S1533-0028(16)30216-X

DOI:

10.1016/j.clcc.2017.06.002

Reference:

CLCC 380

To appear in:

Clinical Colorectal Cancer

Received Date: 14 October 2016 Revised Date:

16 March 2017

Accepted Date: 16 June 2017

Please cite this article as: Yeo H, Betel D, Abelson JS, Zheng XE, Yantiss R, Shah MA, Early-Onset Colorectal cancer is distinct from traditional colorectal cancer, Clinical Colorectal Cancer (2017), doi: 10.1016/j.clcc.2017.06.002. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

Early-Onset Colorectal cancer is distinct from traditional colorectal cancer Heather Yeo, MD MHS1,2, Doron Betel PhD3,4, Jonathan S. Abelson, MD1;, Xi E. Zheng MD PhD2, Rhonda Yantiss MD5, Manish A. Shah, MD4 Department of Surgery, Center for Advanced Digestive Care, Weill Cornell Medicine, New

RI PT

1

York-Presbyterian Hospital, New York, New York 2

Department of Healthcare Policy and Research, Weill Cornell Medicine, New York

SC

Presbyterian Hospital, New York, New York

Institute for Computational Biomedicine, Weill Cornell Medicine, New York, New York

4

Department of Medicine, Division of Hematology and Medical Oncology, Center for Advanced

M AN U

3

Digestive Care, Weill Cornell Medicine, New York-Presbyterian Hospital, New York, New York. 5

Department of Pathology, Weill Cornell Medicine, New York-Presbyterian Hospital, New

Corresponding Author: Manish A. Shah, MD

TE D

York, New York.

EP

Bartlett Family Associate Professor of Gastrointestinal Oncology Director, Gastrointestinal Oncology Program

AC C

Co-Director, Center of Advanced Digestive Care 1305 York Avenue, 12th Floor New York, NY

[email protected] Phone: 646 – 962 – 2547 Fax: 646 – 962 – 1607

1

ACCEPTED MANUSCRIPT

Word Count: 2,976 Abstract: 250

RI PT

Total Pages: 23 Tables: 2 Figures: 3

SC

Supplemental Tables: 2

AC C

EP

TE D

M AN U

Supplemental Figures: 5

2

ACCEPTED MANUSCRIPT

Abstract Background: Early-onset colorectal cancer(E-CRC) is increasing in incidence, unlike traditional CRC(T-CRC). We sought to characterize differences between E-CRC and T-CRC.

RI PT

Materials and Methods: Data sources included the Surveillance Epidemiology and End

Results(SEER) database, the Behavioral Risk Factor Surveillance Survey(BRFSS), and The Cancer Genome Atlas(TCGA). We compared demographics, tumor characteristics, and

SC

incidence of CRC in subjects aged 20-49(E-CRC) with those aged > 50(T-CRC). We correlated

characteristics of CRC using TCGA.

M AN U

the incidence of E-CRC and T-CRC to CRC risk factors and age-dependent genomic

Results: A total of 369,796 CRCs were identified (2000-2011). E-CRC incidence has risen 1.4%/year while T-CRC has declined 3.1%/year(p<0.05). The incidence of E-CRC increases in a step-wise fashion from the ascending colon to rectum(p < 2.2e-16). E-CRC is more prevalent in

TE D

males(53.7% vs 46.4%,p<0.001), Black(14.6% vs 11.0%,p<0.001) and Hispanic(14.7% vs 8.3%,p<0.001) patients. E-CRC presents with aggressive histology, including high-grade(1.5% vs 1.3%,p<0.001), signet ring cell(1.9% vs 0.9%,p<0.001), and mucinous carcinomas(8.9% vs

EP

8.1%,p<0.001), and more often with distant disease(24.4% vs 18.8%,p<0.001). The geographical distribution of E-CRC mirrors US counties with higher Black population densities. Unlike T-

AC C

CRC, E-CRC prevalence is not correlated with known CRC risk factors. E-CRC is associated with a lower rate of mutations than traditional CRC. Limitations of this study include E-CRC sample size for the TCGA analysis, as well as lack of comorbidity information and family history.

3

ACCEPTED MANUSCRIPT

Conclusion: E-CRC tumors are clinically, pathologically, and molecularly distinct from T-CRC. Further evaluation of genetic and molecular differences is necessary to understand the

RI PT

pathophysiology of E-CRC and to help target treatment/surveillance strategies.

Keywords: colorectal neoplasms; epidemiology; risk factors; Behavioral Risk Factor

AC C

EP

TE D

M AN U

SC

Surveillance System; SEER Program

4

ACCEPTED MANUSCRIPT

Introduction Despite significant reductions in colorectal cancer (CRC) incidence and mortality over the last decade, CRC remains the fourth most commonly diagnosed cancer and the second

RI PT

leading cause of cancer death in the US.1 Not all age groups have experienced a decrease in CRC incidence. Annual incidence rates of CRC have increased by more than 1.5% per year for more than a decade among patients 20-49 years old, and now early-onset CRC (E-CRC) now

SC

comprises 10-18% of newly diagnosed cases2-5. The rise in cancer incidence in this population is startling given the lack of identifiable risk factors for early onset disease and decreasing rates of

M AN U

CRC in older age groups.

Most studies that focused on the rising incidence of sporadic E-CRC from the Surveillance Epidemiology and End Results (SEER) database were performed in the early 2000s1-3, 5-10. Studies that examined the pathologic characteristics of E-CRC were small and

TE D

heavily influenced by patients with heritable risk factors or inflammatory bowel disease9, 11-13. Early retrospective studies suggest that E-CRC presents at a later stage and with more aggressive disease characteristics8, 14, 15. There has been no recent in-depth characterization of these patients

EP

and their tumors, particularly with regard to CRC risk factors and demographics using national US registries. Additionally, despite the advances in molecular characterization of colorectal

AC C

cancer16, a specific molecular analysis of E-CRC has not been performed, and it remains unclear how this sporadic early-onset group differs from patients who develop CRC over the age of 50. We evaluated the SEER database to characterize the clinical and pathologic differences

between early-onset CRC (20-49 years, E-CRC) and the traditional group (> 50 years, T-CRC). The age cut-off of 50 years was chosen because it is the age that is recommended to begin CRC screening and because previous literature used this cut-off2-7, 10, 14, 15, 17. We also compared rates of age-adjusted CRC with known risk factors based on population data from the Behavioral Risk 5

ACCEPTED MANUSCRIPT

Factor Surveillance Survey (BRFSS). Based on differences in clinical presentation as well as differences in mapping distribution relative to the mapping distribution of T-CRC risk factors, we hypothesized that there are underlying molecular differences in E-CRCs that may explain

RI PT

their increased incidence and differences in their tumor biology. In addition to the clinical and epidemiological characterization of E-CRC patients and tumors, we therefore also evaluated molecular differences between E-CRC and T-CRC using colorectal cancer data from The Cancer

SC

Genome Atlas (TCGA) dataset. Materials and Methods:

M AN U

Analysis of SEER data

The most recent SEER database constitutes population-based data of cancer patients from 18 registries18. Using this database, we compared CRC incident and prevalent cases in subjects aged 20-49 and aged >50 from 2000-2011. We examined incidence trends using SEER*Stat

TE D

8.1.5 “rate session”. The incidence rates are per 100,000 and age-adjusted to the 2000 US standard population (19 age groups-Census P25-1130). Annual percent changes (APC) were calculated using the least squares method and were tested for whether it is statistically different

EP

from zero using the t distribution in the regression model. CRC incidence rates stratified by tumor stage were also assessed.

AC C

To evaluate patient and tumor characteristics, all prevalent CRC cases occurring in

patients >20 years of age were extracted using SEER*Stat 8.1.5 “case listing session”. We evaluated age, gender, race/ethnicity, tumor stage, histologic grade, tumor location, and histology (including adenocarcinoma, signet ring cell, and unknown). Patients with more than one primary CRC were excluded. Comparisons of E-CRC cases (20-49 years) and T-CRC cases (>50 years) were assessed by chi-square or ANOVA tests for each categorical variable. Logistic

6

ACCEPTED MANUSCRIPT

regression analyses were performed to estimate the risk of CRC presenting with each demographic and tumor characteristic in E-CRC relative to T-CRC. Statistical analyses were

RI PT

conducted using STATAv13 and p-values<0.05 were considered significant.

Risk Factor and CRC Incidence Mapping

For geographical mapping analysis the age-adjusted CRC incidence rates in 2011 were

SC

calculated using SEER*Stat software. SEER 18 data were used to obtain cancer rates by county. Rates were stratified by tumor location (cecum, ascending colon, hepatic flexure of colon,

M AN U

transverse colon, splenic flexure of colon, descending colon, sigmoid, rectosigmoid junction, or rectum), age (<50 years versus ≥50 years), and race/ethnicity (White, Black, Hispanic). Rates of self-reported risk factors for CRC, such as diabetes, obesity, smoking and excessive drinking, were collected using the 2011 BRFSS, a large population-based health survey conducted in the

TE D

US19. BRFSS is an annual, cross-sectional survey of 500,000 participants that has previously been described and validated20. These risk factors were matched to cancer rates using the Federal Information Processing Standard (FIPS) code. To distinguish differences in E-CRC from overall

EP

trends in CRC rates, when correlating with anatomical locations, we used the ratio of the E-CRC to T-CRC rate. To evaluate the distribution of E-CRC across different race/ethnic groups in the

AC C

US, we calculated the rate of E-CRC in each race/ethnic group relative to the at-risk population for that race/ethnicity.

Analysis of TCGA data

To investigate underlying genetic differences as a possible explanation for the observed differences between E- and T-CRC, we analyzed the most recent TCGA data21 that catalogues

7

ACCEPTED MANUSCRIPT

genome alterations in large cohorts of human tumors through validated integrated multidimensional analyses. For full methodology of the somatic mutation analysis, see online methods. This study was exempt by the Institutional Review Board at Weill Cornell Medical

RI PT

College (Protocol No. 1509016541) and conforms to the SEER data-use agreement.

Source code and data files

Results: Time Trends in CRC Incidence

M AN U

available at http://github.com/dbetel/EarlyOnsetCRC

SC

All source code and data files for the analysis of the TCGA and SEER mapping data are

A total of 369,796 cases of colon and rectal cancer were identified in SEER from 2000-

TE D

2011. E-CRC incidence has risen at an annual rate of 1.4%/year from 2000-2011, whereas TCRC incidence has declined by 3.1%/year during the same period (p < 0.05). Amongst T-CRC patients, all race/ethnic subgroups experienced a decline in CRC incidence from 2000-2011, with

EP

the greatest decline observed in White (5.6%/year) and Black (4.7%/year) populations, whereas Hispanic patients observed a modest annual decline of 1.9%/year (Supplemental Figure 1B). In

AC C

contrast, E-CRC incidence increased minimally in all race/ethnic groups, as follows: White (0.16%/year), Hispanic (0.15%/year) and Black patients (0.09%/year) over the same time period (Supplemental Figure 1A). Notably, the overall rates of E-CRC and T-CRC among Black populations are considerably higher in comparison to White or Hispanic. Amongst E-CRC, patients with distant disease experienced the greatest increase in incidence (3%/year), compared with 1%/year increase for regional and localized CRC (Figure 1a, p < 0.001). Conversely, the

8

ACCEPTED MANUSCRIPT

rate of decrease in T-CRC is significantly greater in the regional group (3.7%/year) compared with 2%/year in the distant group (p< 0.05). E-CRC incidence is increasing across all tumor anatomic locations, with the greatest increase observed in both the left colon (1.8%/year, p<0.05)

RI PT

and rectum (2.1%/year, p<0.05). In contrast T- CRC incidence has decreased across all locations, with the greatest decrease in tumors of the left colon (Figure 1b). The increase rate of E-CRC is not randomly allocated across the colon, but rises in a near linear fashion from the ascending

SC

colon to the distal colon and rectum (Figure 2, p-value < 2.2e-16 by Chi-squared test).

M AN U

Clinical and Pathologic Characteristics

E-CRC is more likely than T-CRC to occur in males (53.7% vs 46.4%, p<0.001), and Black (14.6% vs 11.0%, p<0.001) and Hispanic patients (14.7% vs 8.3%, p<0.001). E-CRC presents with higher pathologic grade (1.5% vs 1.3%, p<0.001) and more commonly with distant

TE D

disease (24.4% vs 18.8%, p<0.001) (Table 1). In addition E-CRC patients are more likely to have left-sided (27.7% vs 24.8%, p<0.001) or rectal (41.8% vs 28.3%, p<0.001) cancer compared with T-CRC. Histologically, the majority of both E-CRC and T-CRC are adenocarcinomas. However,

EP

signet ring carcinomas (1.9% vs 0.9%, p<0.001) and mucinous adenocarcinomas (8.9% vs 8.1%, p<0.001), the more aggressive subtypes, are more often diagnosed in E-CRC. Additionally, E-

AC C

CRC tumors are larger (>5 cm) and more likely to have positive lymph nodes, even when accounting for lymph node harvest, in addition to higher rates of perineural invasion and positive margins (p < 0.001 for all comparisons) (Supplemental Table 1).

Geographic Distribution of CRC and Relation to Race/Ethnicity and Risk Factors

9

ACCEPTED MANUSCRIPT

The risk of E-CRC as determined by the log-odds ratios is higher among Hispanic and Black populations relative to White population (Table 1). To investigate whether the increase in E-CRC risk in minority populations is due to changes in minority population densities, we

RI PT

correlated the Hispanic and Black E-CRC rates in each county (defined as the rate of E-CRC in the Hispanic or Black at-risk population), relative to the E-CRC rate in the White at-risk

population. The E-CRC rate of the Hispanic populations in each US county is closely correlated

SC

with the E-CRC rate in the White population (Figure 3a). In contrast, E-CRC rates among the Black populations are significantly greater than E-CRC rates in the White populations (p-value <

M AN U

8.27e-8, Wilcoxon signed rank test) (Figure 3b). This demonstrates that for Hispanics, the increase risk of E-CRC is related to increasing young Hispanic populations, whereas the increase risk of E-CRC in Black populations is distinct from that in White populations and not due to increased young Black population densities.

TE D

We also correlated geographical rates of E-CRC and T-CRC to rates of previously described CRC risk factors. In this analysis, the self-reported risk factor prevalence for each county in 2011 as reported from BRFSS, was graphed against the 2011 incidence of E-CRC and

EP

T-CRC. For T-CRC, diabetes, obesity and smoking were significantly associated with increased rates of CRC (p-values ≤ 0.05, F-test). In contrast, only smoking was marginally associated with

AC C

increased E-CRC rates, and diabetes and obesity were not associated with E-CRC incidence. Excessive drinking was not correlated with either E- or T-CRC (Table 2).

Genetic differences between early and traditional-onset CRC The time trend differences, demographic, pathological and clinical differences between E-CRC and T-CRC together suggest that E-CRC pathophysiology is distinct from T-CRC. 207

10

ACCEPTED MANUSCRIPT

tumors from patients with T-CRC and 16 tumors from patients with E-CRC were analyzed using TCGA (Supplemental Table 2). We found a significant, but weak correlation between mutation rates in CRC samples and age (Spearman rank=0.17, p-value≤ 0.009 by Wilcoxon rank-sum test)

RI PT

(Supplemental Figure 2), such that mutation rates increase with age. The same finding was

confirmed even when controlling for male and female populations (Supplemental Figure 3). This correlation was also observed when considering silent and non-synonymous mutations separately

SC

(p=0.028 and 0.0064, respectively) suggesting an increase in the genome wide mutation burden with increasing age22.

M AN U

We additionally identified a number of genes whose mutation status frequency correlated with age (p ≤ 0.1 by Wilcoxon rank-sum) (Supplemental Figure 4). For example, mutations in BRAF are more prevalent in tumors from older patients whereas NRAS and PTEN mutations are more common in tumors from younger individuals. A number of focal amplification and deletion

TE D

events had significant differences in their age distribution (Supplemental Figure 5). However, there were no clear relationships between patient age and global methylation levels, number of

Discussion:

EP

CNV events, or correlation with genomic clusters.

AC C

Our analysis shows a steady increase in the incidence of E-CRC from 2000-2011 in the

United States, only partially explained by changes in population densities. E-CRC tumors are not associated with typical CRC risk factors, such as obesity and diabetes. Sporadic E-CRC tumors more commonly develop in the distal colon and rectum with a continuous (linear) increased incidence from the proximal colon to the rectum. As suggested in previous studies8, 14, 15

, these tumors are usually of advanced stage, and display unfavorable features, including high-

11

ACCEPTED MANUSCRIPT

grade morphology, signet ring differentiation, and perineural invasion. Data from the TCGA indicate that somatic gene alterations increase with advancing age among CRC patients. These observations suggest that E-CRC cannot be entirely explained by known genetic etiologies. Our

RI PT

geographic mapping analysis suggests that the increased rate of E-CRC for Blacks cannot be explained by population increases alone. Additionally, the known risk factors for CRC in adults such as smoking and obesity are correlated with T-CRC only, and not with E-CRC. Taken

SC

together, considering the more aggressive presentation and histology, the association with Black patient populations, the lack of correlation with known CRC risk factors, as well as the

distinct disease from T-CRC.

M AN U

distribution of E-CRC in the distal colon and rectum, E-CRC appears to be a phenotypically

We identified several genes and local CNV events that correlated with age. Out of 46 genes commonly mutated in CRC, the mutation status of 14 of these genes significantly

TE D

correlated with different age distributions. It is of note that two of the genes, PLA2G4A and CRTC1, are implicated in COX-2 expression and CRC development in preclinical models23, 24. We also found that some mutations that are early in the Vogelstein pathway such as BRAF are

EP

less likely to be mutated in E-CRC, making it possible that these cancers may not follow the canonical pathway of progression from polyp to malignancy that has been described by

AC C

Vogelstein25, 26. Several genes identified as more commonly mutated in E-CRC are not identified in the top 20 genes mutated from adenoma to carcinoma in the Vogelstein model27. Collectively, these factors suggest that E-CRC tumors are genetically distinct from their older counterparts. Discerning the full spectrum and predictive genetic markers that differentiate between E-CRC and T-CRC will require directed studies with wider age distribution than profiled in the TCGA study.

12

ACCEPTED MANUSCRIPT

Other reports of the high prevalence of E-CRC describe similar disease characteristics, and note that the rate of simultaneous detection of cancer and polyps is very low (~10%)28, suggesting that the traditional pathway from polyp to malignancy in E-CRC may not hold. The

RI PT

increase in left sided colon and rectal cancers is also consistent with earlier reports and although underlying reasons are unknown5, 17, 29, they are suggestive of differences in tumorogenesis and mutational characteristics. It is therefore not certain that instituting CRC screening for younger

SC

patients would impact their incidence because their carcinogenesis may not match that of an older patient’s CRC, where it may take 10 years or longer for a cancer to develop30, 31.

M AN U

Earlier research suggested that part of the increase in younger patients might be attributed to increases in obesity and diabetes in younger patients3. However, while there has been an increase in obesity in the US, it alone does not seem to result in more cancers. Austin at al determined that changes in obesity rates over the past 10 years, do not account for the majority of

TE D

the changes in CRC incidence in E-CRC6. Particularly since obesity has increased at all age ranges, we would expect to see a concomitant increase in CRC in those over 50 also32. Our analysis of risk factor associations with CRC rates indicate that E- CRC is not associated with

EP

diabetes or obesity and that smoking is weakly associated with E- CRC suggesting a different mechanism of disease progression among the younger and older age groups. Possible

AC C

mechanisms involved in E-CRC pathogenesis include posttranslational modification of mRNA13 and genomic stability11,13. Other hypotheses to explain the predominance of E-CRC in the distal colon include differences in transit time, microbiome, and embryologic development. Nevertheless, a clearly understood explanation of the underlying biologic mechanism for this observation remains unknown and affirms the need for further research.

13

ACCEPTED MANUSCRIPT

As in any study using administrative data, there are limitations inherent in the data. Variations in coding at different sites may lead to errors and billing codes can be inaccurate. There are also limitations in measuring disease severity. However, these limitations should not

RI PT

be biased by patient age. There are potential sources of bias, such as ecological fallacy, when using FIPS codes by county to make comparisons between groups. Nevertheless, there are

presently no other methods by which to aggregate national data about risk factors. Furthermore,

SC

we compared risk factors from 2011 to incident cases of E-CRC and T-CRC in 2011, thereby allowing for direct assessment of risk factors in patients who were newly diagnosed with CRC.

M AN U

The BRFSS is a cross-sectional survey, which does not afford itself to trend analyses. In addition, TCGA does not allow for longitudinal analysis so our ability to assess correlations between mutations in CRC and trends in E-CRC and T-CRC using SEER over time is limited. The impact of significant CRC genetic factors, such as BRAF and NRAS mutational status, remains. The

TE D

TCGA patient cohort is significantly biased towards older patients and younger CRC patients are underrepresented, highlighting the need for more comprehensive evaluation of this distinct cohort. Another limitation of this study is the lack of data related to family history. We attempted

EP

to correct for this potential bias in the TCGA analysis by excluding the hypermutator phenotype that would be characteristic of Lynch syndrome. Strengths of this study include the large sample

AC C

size, use of the most updated SEER data available, combination of multiple national data sources, and focus on sporadic cases. For the first time to our knowledge, an analysis linking SEER and BRFSS to address questions of epidemiology of CRC has been performed. Additionally, no previous analysis of the E-CRC subtype using TCGA has been performed. By highlighting differences between E-CRC and T-CRC we provide compelling evidence that E-CRC represents

14

ACCEPTED MANUSCRIPT

a distinct subtype of CRC. We are working to understand the biologic mechanisms unique to ECRC and aim to developing more appropriate screening and treatment algorithms for E-CRC.

RI PT

Conclusions:

CRC incidence in patients younger than 50 years continues to rise. The demographic and pathologic characteristics of E-CRC indicate unique disease biology. The underlying mechanism

SC

of E-CRC is still largely unknown and needs to be studied further. Data from our study suggest that there are important differences in CRC in these two age groups. From a clinical perspective,

M AN U

an efficient screening method and adequate knowledge about risk factors for young patients are lacking. The impact on minority populations requires further evaluation as well. Future research and efforts are ongoing to better understand the biology of early-onset CRC and to improve a

TE D

CRC screening strategy for a younger patient population.

Funding

This study was funded in part by Michael’s Mission and the Center for Advanced Digestive Care

EP

(CADC) at New York-Presbyterian Hospital/Weill Cornell Medical Center. Neither Michael’s Mission nor CADC was directly involved in the conduct of the study, preparation, review, or

AC C

approval of the manuscript.

15

ACCEPTED MANUSCRIPT

REFERENCES 1. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2016. CA Cancer J Clin. 2016;66(1):7-30. 2. Dozois EJ, Boardman LA, Suwanthanma W, et al. Young-onset colorectal cancer in patients

RI PT

with no known genetic predisposition: can we increase early recognition and improve outcome? Medicine (Baltimore). 2008;87(5):259-263.

3. Siegel RL, Jemal A, Ward EM. Increase in incidence of colorectal cancer among young men

SC

and women in the United States. Cancer Epidemiol Biomarkers Prev. 2009;18(6):1695-1698. 4. You YN, Xing Y, Feig BW, Chang GJ, Cormier JN. Young-onset colorectal cancer: is it time

M AN U

to pay attention? Arch Intern Med. 2012;172(3):287-289.

5. Ahnen DJ, Wade SW, Jones WF, et al. The increasing incidence of young-onset colorectal cancer: a call to action. Mayo Clin Proc. 2014;89(2):216-224.

6. Austin H, Henley SJ, King J, Richardson LC, Eheman C. Changes in colorectal cancer

TE D

incidence rates in young and older adults in the United States: what does it tell us about screening. Cancer Causes Control. 2014;25(2):191-201. 7. Davis DM, Marcet JE, Frattini JC, Prather AD, Mateka JJ, Nfonsam VN. Is it time to lower

EP

the recommended screening age for colorectal cancer? J Am Coll Surg. 2011;213(3):352-361. 8. O'Connell JB, Maggard MA, Liu JH, Etzioni DA, Livingston EH, Ko CY. Rates of colon and

AC C

rectal cancers are increasing in young adults. Am Surg. 2003;69(10):866-872. 9. Chang DT, Pai RK, Rybicki LA, et al. Clinicopathologic and molecular features of sporadic early-onset colorectal adenocarcinoma: an adenocarcinoma with frequent signet ring cell differentiation, rectal and sigmoid involvement, and adverse morphologic features. Mod Pathol. 2012;25(8):1128-1139.

16

ACCEPTED MANUSCRIPT

10. Rahman R, Schmaltz C, Jackson CS, Simoes EJ, Jackson-Thompson J, Ibdah JA. Increased risk for colorectal cancer under age 50 in racial and ethnic minorities living in the United States. Cancer Med. 2015;4(12):1863-1870.

RI PT

11. Losi L, Di Gregorio C, Pedroni M, et al. Molecular genetic alterations and clinical features in early-onset colorectal carcinomas and their role for the recognition of hereditary cancer syndromes. Am J Gastroenterol. 2005;100(10):2280-2287.

SC

12. Pucciarelli S, Agostini M, Viel A, et al. Early-age-at-onset colorectal cancer and

Rectum. 2003;46(3):305-312.

M AN U

microsatellite instability as markers of hereditary nonpolyposis colorectal cancer. Dis Colon

13. Yantiss RK, Goodarzi M, Zhou XK, et al. Clinical, pathologic, and molecular features of early-onset colorectal carcinoma. Am J Surg Pathol. 2009;33(4):572-582. 14. Bailey CE, Hu CY, You YN, et al. Increasing disparities in the age-related incidences of

TE D

colon and rectal cancers in the United States, 1975-2010. JAMA Surg. 2015;150(1):17-22. 15. Fairley TL, Cardinez CJ, Martin J, et al. Colorectal cancer in U.S. adults younger than 50 years of age, 1998-2001. Cancer. 2006;107(5 Suppl):1153-1161.

EP

16. Muzny DM Bainbridge MN, Chang K, et al. Comprehensive molecular characterization of human colon and rectal cancer. Nature. 2012;487(7407):330-337.

AC C

17. Rozen P, Liphshitz I, Barchana M. The changing epidemiology of colorectal cancer and its relevance for adapting screening guidelines and methods. European Journal of Cancer Prevention. 2011;20(1):46-53.

18. Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov) SEER*Stat Database: Incidence - SEER 18 Regs Research Data + Hurricane Katrina Impacted

17

ACCEPTED MANUSCRIPT

Louisiana Cases, Nov 2012 Sub (1973-2010 varying) - Linked To County Attributes -Total U.S., 1969-2011 Counties., 2013. 19. County Health Rankings & Roadmaps. University of Wisconsin Population Health Institute,

RI PT

2016 update. Available from URL: http://www.countyhealthrankings.org [accessed August 2015].

20. Nelson DE, Holtzman D, Bolen J, Stanwyck CA, Mack KA. Reliability and validity of

SC

measures from the Behavioral Risk Factor Surveillance System (BRFSS). Soz Praventivmed.

M AN U

2001;46 Suppl 1:S3-42.

21. The Cancer Genome Atlas. National Institute of Health; National Cancer Institute; National Human Genome Research Institute, 2015 update. Available from URL: https://tcgadata.nci.nih.gov/tcga/tcgaHome2.jsp [accessed 2015].

TE D

22. Kennedy SR, Loeb LA, Herr AJ. Somatic mutations in aging, cancer and neurodegeneration. Mech Ageing Dev. 2012;133(4):118-126.

23. Dong M, Guda K, Nambiar PR, et al. Inverse association between phospholipase A2 and

EP

COX-2 expression during mouse colon tumorigenesis. Carcinogenesis. 2003;24(2):307-315. 24. Schumacher Y, Aparicio T, Ourabah S, et al. Dysregulated CRTC1 activity is a novel

AC C

component of PGE2 signaling that contributes to colon cancer growth. Oncogene. 2015;35(20):2602-2614.

25. Parsons DW, Wang TL, Samuels Y, et al. Colorectal cancer: mutations in a signalling pathway. Nature. 2005;436(7052):792. 26. Vogelstein B, Kinzler KW. Cancer genes and the pathways they control. Nat Med. 2004;10(8):789-799.

18

ACCEPTED MANUSCRIPT

27. Carvalho B, Sillars-Hardebol AH, Postma C, et al. Colorectal adenoma to carcinoma progression is accompanied by changes in gene expression associated with ageing, chromosomal instability, and fatty acid metabolism. Cell Oncol (Dordr). 2012;35(1):53-63.

RI PT

28. Zahir MN, Azhar EM, Rafiq S, Ghias K, Shabbir-Moosajee M. Clinical features and outcome of sporadic colorectal carcinoma in young patients: a cross-sectional analysis from a developing country. ISRN Oncol. 2014;2014:461570.

SC

29. Meyer JE, Narang T, Schnoll-Sussman FH, Pochapin MB, Christos PJ, Sherr DL. Increasing incidence of rectal cancer in patients aged younger than 40 years. Cancer. 2010;116(18):4354-

M AN U

4359.

30. Hofstad B, Vatn M. Growth rate of colon polyps and cancer. Gastrointest Endosc Clin N Am. 1997;7(3):345-363.

31. Winawer S, Fletcher R, Rex D, et al. Colorectal cancer screening and surveillance: clinical

TE D

guidelines and rationale-Update based on new evidence. Gastroenterology. 2003;124(2):544-560. 32. Ladabaum U, Mannalithara A, Myer PA, Singh G. Obesity, abdominal obesity, physical

AC C

EP

activity, and caloric intake in US adults: 1988 to 2010. Am J Med. 2014;127(8):717-727.e712.

19

ACCEPTED MANUSCRIPT

FIGURES Figure 1. Localization and staging trends of early- vs traditional-onset colorectal cancer from SEER 2000-2011 data. a) Average changes in E-CRC or T-CRC tumor staging from 2000-2011

RI PT

indicate an increase in distant staged tumors among E-CRC patients. b) Similarly, average

changes in E-CRC and T-CRC tumor localization from 2000-2011 indicate increased propensity of tumors in the distal colon among E-CRC patients. All trends are statistically significant (p-

SC

value< 0.05).

M AN U

Figure 2. The ratios of E-CRC rates relative to T-CRC rates by anatomical location from 2011 SEER data indicate a strong propensity for E-CRC tumors to localize towards distal colon and rectum. The near linear increase as one moves from the cecum to the rectum indicates a unique

TE D

underlying biology for E-CRC.

Figure 3. Comparison of E-CRC rates among race/ethnic populations by SEER counties. a) ECRC incidents (per 100K individuals in the same age and race/ethnic population) in White and

EP

Hispanic populations are well correlated as indicated by agreement along the diagonal red line (slope=1, p-value = 1). In contrast b) incidents of E-CRC among Black populations are generally

AC C

higher than expected relative to White population in the same county as indicated by large number of points above the red diagonal line (p-value <8.27x10-8, Wilcoxon signed rank test).

20

ACCEPTED MANUSCRIPT

Table 1. Demographics and Tumor Characteristics of Early-onset and Traditional-onset Colorectal Cancer, SEER, 2000-2011.

Female

39,787

Pvalue

330,009 <0.001

21,342 (53.7%) 18, 445 (46.4%)

166,300 (50.4%) 163,709 (49.6%)

23,701(59.6%) 5,822 (14.6%) 5,838 (14.7%) 3,664(9.2%) 762 (1.9%)

Other Stage Localized

13,143 (33.0%)

Regional

15,480 (38.9%)

II III IV

9,697 (24.4%) 1,467 (3.7%)

3,058 (7.7%)

AC C

Histograde I

P-trend

0.86

0.90

ref

130,871 (39.7%) 117,476 (35.6%) 61,981 (18.8%) 19,681 (6.0%)

1.62

1.57

1.67

2.14

2.07

2.21

1.59

1.53

1.65

1.97

1.82

2.13

1.31

1.28

1.34 <0.001

1.56

1.51

1.60

0.74

0.70

0.78

1.09

1.05

1.13 <0.001

1.26

1.20

1.31

1.36

1.24

1.50

ref

TE D

Unknown

239,082 (72.5%) 36,270 (11.0%) 27,530 (8.3%) 23,229 (7.0%) 3898 (1.2%)

95% CI

<0.001

EP

Distant

0.88

M AN U

Asian/PI

ref

<0.001

Ethnicity Non-Hispanic White Non-Hispanic Black Hispanic

Unadjusted OR (E-CRC vs T-CRC)

SC

Total N Gender Male

Age Group 50+ years

RI PT

20-49 years

22,526 (56.6%) 6,862 (17.3) 603 (1.5%)

Unknown

6,738 (16.9%)

Location Right colon

10,807 (27.2%)

Left colon

11,038 (27.7%)

<0.001 28,493 (8.6%) 192,523 (58.3%) 50,897 (15.4%) 4,124 (1.3%) 53,972 (16.4%)

ref

<0.001 139,078 (42.1%) 81855 (24.8%)

21

ref 1.74

<0.001 1.69

1.78

ACCEPTED MANUSCRIPT

Table 1. Demographics and Tumor Characteristics of Early-onset and Traditional-onset

Unspecified Histology Adenocarcinoma

16,646 (41.8%) 1,296 (3.3%)

Pvalue

93,378 (28.3%) 15,698 (4.8%) <0.001

25,168 (63.3%)

Adenocarcinoma in an adenoma Mucinous Adenocarcinoma Signet ring cell

5,953(15.0%)

Unknown

1,649 (4.1%)

3,544 (8.9%)

AC C

EP

TE D

738 (1.9%)

213,873 (64.8%) 58,194 (17.6%) 26,879 (8.1%) 2,881 (0.9%) 18,706 (5.7%)

Unadjusted OR (E-CRC vs T-CRC) 2.29

22

95% CI

2.24

P-trend

2.35

ref

M AN U

Rectum

Age Group 50+ years

SC

20-49 years

RI PT

Colorectal Cancer, SEER, 2000-2011, Continued

0.87

0.84

0.9

1.12

1.08

1.16

2.18

2.01

2.36

ACCEPTED MANUSCRIPT

Table 2. Age-adjusted rate of CRC stratified by early and traditional-CRC cases and known CRC risk factors 20-49 years

50+ years

p value

R2

p value

Smokinga

0.0208

0.0006

0.0803

<0.0001

Obesityb

-0.0011

0.5812

0.0846

<0.0001

Diabetesc

-0.0012

0.6465

0.0327

<0.0001

-0.0015

0.6649

-0.0019

a

SC

drinkingd

0.9953

M AN U

Excessive

RI PT

R2

Smoking: Adult Smoking is the percentage of the adult population that currently smokes every

day or most days and has smoked at least 100 cigarettes in their lifetime. b

TE D

Obesity: Adult Obesity is the percentage of the adult population (age 20 and older) that reports a

body mass index (BMI) greater than or equal to 30 kg/m2. c

Diabetes: Respondents were considered to have diagnosed diabetes if they responded "yes" to

EP

the question, "Has a doctor ever told you that you have diabetes?" Women who indicated that they only had diabetes during pregnancy were not considered to have diabetes. Excessive drinking: Excessive drinking, defined as the percentage of adults that report either

AC C

d

binge drinking (consuming more than 4 (women) or 5 (men) alcoholic beverages on a single occasion in the past 30 days) or heavy drinking (consuming more than one (women) or 2 (men) drinks per day on average)

23

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

18

C

at ic

ep

H

3−

18 .

fle xu re

in g

sc en d

AC C

on

co l

ec um

.0 −C

0.00

of co .4 lo −T n ra C ns 18 ve .5 −S rs e pl co en lo ic n fle xu re C 18 of .6 co −D lo n es ce nd in g C co 18 lo .7 n −S C ig 19 m .9 oi −R d co ec lo to n si gm oi d ju nc C 20 tio .9 n −R ec tu m ,N O S

C

.2 −A

18

C

18

C

EP TE D

0.05

M AN U

SC

E−CRC/T−CRC rates (age−adjusted) 0.10

RI PT

ACCEPTED MANUSCRIPT

Ratios of E−CRC rates over T−CRC rates by anatomical location

Anatomical Location

ACCEPTED MANUSCRIPT

●

●

M AN U

SC

150

Rates of Black E−CRC (per 100K)

Rates of Hispanic E−CRC (per 100K)

150

RI PT

Comparison of E−CRC Rates by Ethnicity

100

●

●

100

50 ●

●

●

●

●

● ● ●

0

● ●

●● ● ●● ● ●● ●● ● ●●● ● ● ●● ●● ● ● ● ● ● ● ● ● ●●● ● ● ● ●● ● ●●●● ●●● ●●● ● ●●● ● ● ● ● ● ●● ● ● ● ● ●● ●

10

[A]

● ●

● ● ● ●● ● ●

● ● ● ●

● ●

AC C

●

●

50

●

●

● ●

● ● ●

● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●● ●● ●● ● ● ● ● ●● ● ●● ● ● ● ● ● ●●●●● ● ● ● ●● ● ●● ● ●● ● ● ●● ●●● ● ●● ●● ● ● ● ● ● ●● ●● ● ●● ● ●● ●● ● ● ● ● ● ●● ● ● ●●●● ● ● ● ● ● ● ● ●●●●●● ●

●

20 Rates of White E−CRC (per 100K)

●

●

●

●

● ●

EP

●

TE D

●

●

0 30

0

[B]

30 60 Rates of White E−CRC (per 100K)

90