Early-onset colorectal cancer: what reported statistics can and can’t tell us, and their implications

Precis:

In their current article, Virostko and colleagues report on the distribution of adult colorectal cancer cases in the National Cancer Database from 2004 to 2015 based on age-eligibility for screening. The editorial discusses the findings, the features of the database, and uses population-based data to contextualize emerging trends in incidence and mortality and the need for research to improve outcomes in all people at risk for colorectal cancer irrespective of screening-eligibility.

The US Preventive Services Task Force (USPSTF) and other national groups agree on routine colorectal cancer screening in people who do not have symptoms or known familial risks and are between the ages of 50 and 75. Recent reports have drawn attention to an increase in the incidence rates for colorectal cancer in people younger than 50 years old.¹ While an increase in incidence in any population group is a concern, the magnitude of the absolute increase is small and colorectal cancer remains a very rare condition in younger individuals in whom heritable risks play a larger role than in older individuals. For instance, the incidence of colorectal cancer in 20–49-year-old people increased from 10.3 in 2004 to 13.2 per 100,000 people in 2016 (i.e., 0.010% versus 0.013%) and mortality rate was 2.53 in 2004 and 2.97 per 100,000 in 2016. In comparison over the same period in people ≥50 years, the incidence and mortality rates trended down from 167.0 to 115.3, and 61.5 to 44.8 (all per 100,000), respectively (see Figure 1).¹

Figure 1. US colorectal cancer incidence by race/ethnicity and age, SEER 2000–2015.

A: Incidence in people 50 years and older and 20–49 years old with fitted trend line

B: Incidence in people 50–64 years with fitted trend lines

In the current issue of the journal, Virostko and colleagues report on the number of colorectal cancer cases in the National Cancer Database (NCDB) in people between the ages of 18 and 49 as a proportion of the total number of cases in adults.² They identified 1,185,763 patients with colorectal cancer between 2004 and 2015, 11% of whom were 18–49 years old with heterogeneity according to diagnosis year and race, sex, socioeconomic status, and place of residence. The proportion of NCDB colorectal cancer cases in people aged 18–49 was higher in 2015 (12.2%) than in 2004 (10.0%). Over that period, the proportion of NCDB colorectal cancer cases in 18–49-year-old people increased in non-Hispanic white men and women and in Hispanic women, but not in African Americans or some geographic areas such as rural areas or the East South-Central Census Division (Alabama, Kentucky, Mississippi, and Tennessee). The proportion increased over time in people residing in areas in the highest zip-code income quartile and patients with private insurance, but was lower over time in those who had Medicaid or no insurance.

The results are provocative, but the absolute increase in incidence observed in population-based data in people younger than 50 is too small (see Figure 1)¹ to explain changes of the magnitude reported by Virostko and colleagues.² However, decreasing incidence in people >50 years old and population trends are plausible explanations and the strengths and limitations of the NCDB offer insights.³ The NCBD captures about 70% of all new cancers in the US, but unlike the National Cancer Institute’s Surveillance Epidemiology and End Results (SEER) program, the NCDB is not population-based and thus unable to provide incidence or mortality rates.³ The NCBD accrues people diagnosed and/or treated at centers accredited by the American College of Surgeons Commission on Cancer, and may not adequately capture patients from disadvantaged backgrounds – data from the Veterans Affairs and military health systems are not included in public use files.³ Also, the NCDB captures a greater proportion of younger patients than those 65 years and older. Because of these limitations, NCDB data are not suited for informing trends in cancer incidence or subgroups disproportionately affected.

Virostko and colleagues reported that more cancers in 18–49-year-olds were diagnosed at stage III or IV than in those ≥50 years old (51.6% versus 40.0%).² Increasing uptake of screening leads to a higher proportion of cancers detected at early stage. In a population-based study by Levin and colleagues, more colorectal cancers were detected at early stage and incidence and mortality rates decreased as screening uptake increased.⁴ Thus, a stable or decreasing proportion of cancer diagnoses in 18–49-year-old African American or uninsured patients, if true, may result from lower screening completion rates and historically high burden of colorectal cancer in those groups, but may also be due to selection bias in the NCDB. Notably, the East South-Central Census Division is in a colorectal cancer hotspot known for the highest mortality rates in the US (particularly in African American men) and low screening rates.⁵ Because current screening modalities are generally more effective at preventing cancer in the left colon/rectum, increasing uptake may shift the location of cancers to a higher proportion in the right colon.

Population trends can produce differences in the mean age at colorectal cancer diagnosis, unless appropriate age adjustment is performed. For instance, in 2005 and 2015, 26.7% and 33.6% of US non-Hispanic whites were aged 55 and older, but was 16.6% and 22.0% of African Americans, and 10.9% and 14.2% of Hispanics, respectively.⁶ Similarly, in 2005 and 2015, 14.6% and 18.8% of non-Hispanic whites versus 8.3% and 10.5% of African Americans, and 5.3% and 6.6% of Hispanics, respectively, were 65 years and older, when colorectal cancer risk is even higher. Because colorectal cancer risk increases with increasing age, the average age at diagnosis may be lower in African Americans and Hispanics as an epidemiological artifact from demographic trends (whether due to lower life expectancy or immigration) and not necessarily because of a shift to younger age at diagnosis in those groups.

The uptake of screening has increased over time in the US, and is considered the main contributor to decreasing incidence and mortality rates in the US.⁷ The screening initiation age is not an exact science, and screening occurs in people younger than 50 years.⁸ Because the number of colorectal cancer cases from inherited causes are much higher in younger individuals, it is unknown whether screening for sporadic cases in a group with such low disease rate can result in a favorable balance of harms and benefits. It is therefore imperative that the various hypotheses for increasing colorectal cancer incidence among people <50 be rigorously tested to determine if changing the current screening age in people who are not at increased familial risk represents the most appropriate public health response.

Other concerning epidemiological trends in colorectal cancer may be emerging in people >50 years old despite existing screening recommendations. In 50–64-year-old individuals, the colorectal cancer incidence rate in Asian/Pacific Islanders has been relatively unchanged in almost two decades, but in non-Hispanic whites, the rate declined and then leveled off or even rising after 2011, with similar patterns emerging in African Americans in recent years (see Figure 1). Among people 50–64, the rates were 64.9 and 68.9 in non-Hispanic whites and 65.2 and 64.9 per 100,000 in Asian/Pacific Islanders, in 2011 and 2016, respectively.¹ Thus, evidence is needed on the causes and interventions for such emerging trends, including the role of environmental factors or gene-environment interactions, and uptake and effectiveness of screening. Screening rates consistent with USPSTF recommendations are suboptimal in many subpopulations of Americans.⁹ A recent pragmatic modeling analysis showed that increasing the uptake of screening in current age-eligible groups may have a greater impact than lowering the age at which average-risk people begin screening.¹⁰ Similar studies focused on sporadic cases can help guide societal decisions on the age to begin routine colorectal cancer screening since the empirical evidence on the balance of harms and benefits is unlikely to be garnered.

Screening will not eliminate all colorectal cancer deaths and occurrence of cases in individuals outside current recommended age groups does not necessarily mean that expanding eligibility will optimize outcomes. Accurate documentation of family history, appropriate referral for genetic counseling and testing, and wider adoption of newer technologies such as next generation sequencing¹¹ may improve identification of familial colorectal cancer syndromes to guide preventive strategies for affected individuals and families. Although colorectal cancer symptoms are common in individuals without cancer, prompt evaluation of symptoms or signs that may indicate the presence of colorectal cancer may improve the chance for curative treatments in younger individuals, but requires further research.

A core principle of population-based screening is to do no more harm around the time of screening than the potential future health benefits. Evidence is thus needed on whether ‘routine screening’ with currently available strategies is as effective in preventing colorectal cancer death in individuals younger than is recommended by guidelines. Assumptions used in modeling studies to inform the age of screening initiation are unlikely to be realized in the real world, particularly given low incidence and higher prevalence of heritable causes in early-onset cases. Unfortunately, there is little possibility of adequately-designed empirical studies to determine the most appropriate age cutoffs to optimize the net benefits of screening. However, at a population level, a decrease in the average age at diagnosis or the proportion of colorectal cancers in people >50, and a shift to earlier stage at diagnosis may all suggest that current preventive interventions are effective. Because many groups do not realize potential benefits to same degree, efforts should continue on increasing access to screening in underserved populations in tandem with research to optimize the benefits of preventive interventions for colorectal cancer.