Background: In response to evidence that colorectal cancer (CRC) incidence is rising among individuals under age 50, consistent with a birth cohort effect, clinical guidelines were updated to recommend initiating average-risk CRC screening at age 45 instead of 50. Whether birth cohort trends in CRC risk also influence risk among individuals with a family history of CRC is unknown. To address this gap, we measured the association between family history of CRC and incident CRC using an age-period-cohort analysis.

Methods: We used data from the California Teachers Study (CTS), a cohort study with 133,477 female public-school teachers and administrators longitudinally followed from 1995-2020. After excluding those with missing family history data, prevalent and incident CRC were measured via linkage to the California Cancer Registry. We compared overall CRC incidence rates among individuals with vs. without family history of CRC, defined as having self-report of a first degree relative with CRC, including by 5-year age groups. Next, we examined birth cohort trends by decade among individuals with vs. without a family history. Lastly, we evaluated whether increased CRC risk for those with vs. without a family history were observed across the 25-year CTS. Follow up time began at age 18 and continued to the earliest of: CRC diagnosis, death, moving out of state, loss to follow up, or 12/31/20.

Results: 120,428 women were included; the sample was 91.1% White with median age 52.4 years at CTS entry. Over 6.7 million person-years of follow-up from age 18, there were a total of 2,337 CRC cases observed, with 426 occurring prior to, and 1,911 after CTS entry. Family history of CRC was associated with a 1.52-fold greater CRC risk (Incidence Rate Ratio 95% CI, 1.37-1.69, p-value <0.01) and numerically higher CRC age-specific incidence at nearly each 5-year age group (Figure 1). By birth cohort, CRC incidence rates were higher among those with vs. without a family history for women born before 1960, but not for women born after 1960 (Figure 2). When examining 5-year observation periods between 1995 and 2020, CRC incidence rates were significantly higher among those with vs. without a family history except from 2005-2009.

Conclusions: In the CTS cohort, consistent with prior studies, family history of CRC was associated with higher risk for incident CRC, including for nearly every recent time period between 1995 and 2020. When examining risk by birth cohort, family history was associated with higher incidence for pre- but not post-1960 cohorts. This could be due to the impact of more aggressive CRC screening for individuals with a family history, low study power, or a true lack of birth cohort effect. More studies are needed to evaluate whether the initiation age for family history-based screening should be lowered to account for trends in birth cohort CRC risk.

<b>Figure 1: Age-Specific CRC Incidence Rate Stratified by Family History Status.</b> Women with a family history of CRC were at 1.52-fold risk of CRC compared to those without (Incidence Rate Ratio 95% CI, 1.37-1.69, p-value <0.01). Across 5-year age groups, CRC incidence rate is numerically higher and significantly greater for Ages 50-54 (Incidence Rate Difference 1.69, 95% CI, 0.22-3.15, p-value = 0.02); for Ages 65-69 (Incidence Rate Difference 3.44, 95% CI, 0.78-6.09, p-value = 0.01); and for Ages >=75 (Incidence Rate Difference 5.32, 95% CI, 1.91-8.72, p-value < 0.01). *p-value < 0.05 and **p-value < 0.01 between those with vs. without a family history of CRC.

Figure 1: Age-Specific CRC Incidence Rate Stratified by Family History Status. Women with a family history of CRC were at 1.52-fold risk of CRC compared to those without (Incidence Rate Ratio 95% CI, 1.37-1.69, p-value <0.01). Across 5-year age groups, CRC incidence rate is numerically higher and significantly greater for Ages 50-54 (Incidence Rate Difference 1.69, 95% CI, 0.22-3.15, p-value = 0.02); for Ages 65-69 (Incidence Rate Difference 3.44, 95% CI, 0.78-6.09, p-value = 0.01); and for Ages >=75 (Incidence Rate Difference 5.32, 95% CI, 1.91-8.72, p-value < 0.01). *p-value < 0.05 and **p-value < 0.01 between those with vs. without a family history of CRC.

<b>Figure 2: Birth Cohort CRC Incidence Rate Stratified by Family History Status.</b> CRC incidence is highest among the oldest study cohort and trends down across younger birth cohorts. Within birth cohorts, CRC incidence rate is numerically higher and significantly greater for those born before 1920 (Incidence Rate Difference 2.29, 95% CI, 0.45-4.13, p-value < 0.02); 1920-1929 (Incidence Rate Difference 1.75, 95% CI, 0.44-3.07, p-value < 0.01); 1930-1939 (Incidence Rate Difference 1.70, 95% CI, 0.65-2.75, p-value < 0.01); and 1950-1959 (Incidence Rate Difference 1.05, 95% CI, 0.05-2.05, p-value = 0.04). *p-value < 0.05 and **p-value < 0.01 between those with vs. without a family history of CRC.

Figure 2: Birth Cohort CRC Incidence Rate Stratified by Family History Status. CRC incidence is highest among the oldest study cohort and trends down across younger birth cohorts. Within birth cohorts, CRC incidence rate is numerically higher and significantly greater for those born before 1920 (Incidence Rate Difference 2.29, 95% CI, 0.45-4.13, p-value < 0.02); 1920-1929 (Incidence Rate Difference 1.75, 95% CI, 0.44-3.07, p-value < 0.01); 1930-1939 (Incidence Rate Difference 1.70, 95% CI, 0.65-2.75, p-value < 0.01); and 1950-1959 (Incidence Rate Difference 1.05, 95% CI, 0.05-2.05, p-value = 0.04). *p-value < 0.05 and **p-value < 0.01 between those with vs. without a family history of CRC.

Maria Elena Martinez

Appearances