Change in Prevalence of Family History During Long-term Follow-up of Patients With Pediatric-onset Inflammatory Bowel Disease

Abstract

Objectives: 

The aim of the study was to prospectively study changes in prevalence of positive family history (FH+) in pediatric-onset inflammatory bowel disease (IBD) in contrast to previously published cross-sectional data.

Methods: 

An observational cohort study was performed using a prospective pediatric-onset IBD database including 485 patients with disease duration ≥10 years as of December 2016. Proband characteristics and FH+ were obtained at diagnosis and subsequently from the database, medical records, and follow-up telephone interviews in 2006 and 2016.

Results: 

Updated 2016 information was obtained from 322 (66%) patients and included in analysis with median follow-up of 18 years (interquartile range 14, 26). Prevalence of FH+ increased from 13.7% at diagnosis to 26.6% at 20 years for first-degree relatives and from 38.5% to 52.2% for all relatives. At 20-year follow-up, an additional 10.0% of probands had a sibling, 6.1% had a parent, 1.9% had a grandparent, and 4.5% had a cousin diagnosed with IBD. FH+ at diagnosis was associated with greater risk for additional FH+ at 20 years (43% vs 22%, P < 0.001). Non-Jewish Caucasians had significantly lower risk of a FH+ compared to Jewish Caucasians (P = 0.002), but similar risk to African Americans (P = 0.55). FH+ at diagnosis was not associated with disease type (P = 0.33) or age at diagnosis (P = 0.24).

Conclusions: 

This prospective study documents changes in family history of IBD in pediatric-onset IBD patients over time. Prevalence of FH+ increased for first-degree and all relatives at 20 years by 12.9% and 13.7%, respectively. FH+ at diagnosis was associated with a 2-fold greater likelihood of subsequent FH+ at 20 years.

What Is Known

• Prevalence of family history of inflammatory bowel disease at diagnosis has previously been ascertained through cross-sectional studies and varies among prior publications.
• Information regarding the observed change in prevalence of family history over time in pediatric-onset inflammatory bowel disease has not been previously reported.

What Is New

• The prevalence of family history of inflammatory bowel disease increases over time for all degree relatives.
• A positive family history at diagnosis is associated with a 2-fold greater risk of subsequent positive family history.
• Siblings demonstrate the greatest risk among relative groups of subsequently developing inflammatory bowel disease.
• The risk for any family history of inflammatory bowel disease in African Americans is similar to that of non-Jewish Caucasians.

Inflammatory bowel disease (IBD), consisting of Crohn disease (CD), ulcerative colitis (UC), and IBD unclassified (IBDU), affects 1 in 200 to 300 people in North America and Europe ^(1–3). The incidence of IBD peaks in the 15 to 29 year age group with a second smaller peak in the 5th to 7th decades ^(1,2,4). An estimated 3% to 25% of patients with IBD have a positive family history (FH+) at diagnosis ^(3,5–18). Of these, 3% to 17.6% are first-degree relatives ^{(13,14,16–18)}. This information is based on cross-sectional rather than longitudinal studies assessing changes in FH+ over time.

Both the incidence and prevalence of IBD are increasing worldwide with a significant increase in the pediatric population especially in those younger than 10 years ^(2,19). In a systematic review of international trends in pediatric-onset IBD, Benchimol et al ⁽²⁰⁾ found that 77.8% of the studies reported an increase in incidence over time with no studies finding a significant decrease. To our knowledge, however, there are no prospective reports documenting how family history of IBD changes over time for pediatric-onset IBD.

To further assess this question, we prospectively followed a population of pediatric-onset IBD patients for at least 10 years postdiagnosis to determine any changes in family history of IBD through a median of 18 years of follow-up.

METHODS

A longitudinal observational cohort study was performed using a prospective database comprised of pediatric patients with IBD treated at the University of Chicago Pediatric IBD Center. For the database, all pediatric patients with a diagnosis of IBD seen by the primary investigator at the IBD Center were approached for inclusion in the prospective database with the exception of those patients who were seen solely as a second opinion without any planned longitudinal follow-up. The database included data collection at time of enrollment and then prospectively at every follow-up visit while they continued under our care. For this study, only patients from the database with at least 10 years of follow-up since diagnosis as of December 2016 were included. Diagnosis was established through conventional criteria of clinical symptoms, laboratory findings, imaging, and confirmed with histopathology. Exclusion criteria included incomplete FH+ as of 2016, adopted probands, and change to a diagnosis other than IBD over time. Proband demographics (age, age at diagnosis, race/ethnicity), disease type (CD, UC, IBDU), disease location at diagnosis based on Paris Classification (CD: L1: distal 1/3 ileum ± limited cecal disease, L2: colonic, L3: ileocolonic, L4a: upper disease proximal to ligament of Treitz, L4b: upper disease distal to ligament of Treitz and proximal to distal 1/3 ileum; UC: E1: ulcerative proctitis, E2: left-sided, E3: extensive (hepatic flexure distally), E4: pancolitis) ⁽²¹⁾, FH+ at diagnosis, and subsequent changes in FH+ were recorded in the institutional review board–approved database or obtained from electronic hospital medical records. In addition to these prospectively collected records while patients were under our care, patients or family members were contacted via telephone interview in 2006 and 2016 for updated FH+ and changes in disease type/location. At least 2 attempts were made to contact the patient via telephone using the most up-to-date contact information that was available in our records and then Whitepages Premium was used to search for any additional updated contact information. If patients were unable to be contacted by telephone, a letter was sent to the most current address available in our records requesting a call to our office. All patients were asked the same computer-generated questions and all data were stored using the REDcap (Research Electronic Data Capture) tool hosted at the University of Chicago.

Race/ethnicity was based on patient self-report and was characterized as Caucasian Jewish, Caucasian non-Jewish, African American, Hispanic, or “all others.” Jewish descent was defined as having at least 2 out of 4 grandparents of Jewish descent ⁽²²⁾. Unless otherwise stated, a positive FH+ included any degree relative of the proband. First-degree relatives were defined as parents or siblings ⁽⁹⁾. Offspring were excluded in analysis as only 32% of probands had children at last follow-up and the majority of offspring were considerably younger at the last follow-up (mean age 8.3 ± 6.9 years) than the mean age of onset of pediatric IBD.

Statistical Analysis

Due to varying duration of follow-up since diagnosis, changes in FH+ were analyzed using the Kaplan-Meier method. An event was defined as the first occurrence of a family member being diagnosed with IBD after the proband. The time elapsed was calculated from the date of the proband's diagnosis to the date of the new diagnosis in a family member. Those who never reported having a family member with IBD were censored at the date of last contact. Univariable and multivariable Cox regression were used to examine risk factors for FH+ at diagnosis or subsequent development. Univariable and multivariable Cox regression and negative binomial regression were used to examine factors associated with a change from FH− to FH+. Continuous variables were expressed as mean and standard deviation or median and interquartile range, whereas categorical variables were expressed as number and percentage. The 2-sample t test or analysis of variance was used to compare continuous variables between groups. The chi-squared test or Fisher exact test was used to compare categorical variables. A P value of <0.05 was considered statistically significant. Analyses were performed using Stata (StataCorp, College Station, TX).

Ethical Considerations

This study was approved by the institutional review board of the University of Chicago Medicine and reported according to the Strengthening the Reporting of Observational Studies in Epidemiology guidelines ⁽²³⁾.

RESULTS

Study Population

Of the 485 patients in the database that met inclusion criteria of at least 10 years disease duration as of December 2016, 322 (66%) were successfully contacted for updated 2016 information and included in analysis. All patients were diagnosed between 1970 and 2006. One hundred sixty patients could not be contacted, 2 were excluded for incomplete data, and 1 family was contacted but refused continued participation in the study. The median follow-up after diagnosis was 18 years (interquartile range 14, 26). All 322 patients had at least 10 years of follow-up with 173 having 10 to 19 years, 109 having 20 to 29 years, 36 having 30 to 39 years, and 4 having 40 to 49 years of follow-up (Table 1).

TABLE 1: Patient demographics

For demographics, 49.1% of the study group was female. Mean age at IBD diagnosis for probands was 11.1 years (standard deviation [SD] = 3.7, range 0.7–19.0), and the mean age at 2016 follow-up was 31.0 years (SD = 8.0). The majority (88%) of patients were Caucasian, 9% were African American, 1% were Hispanic, and 2% did not fall into the above categories and included patients of Asian descent or those who self-described as “other." The Caucasian group was further divided into Jewish (30.4%) and non-Jewish descent (69.6%) with Jewish descent defined as having at least 2 grandparents of Jewish descent ⁽²²⁾. Had we used a stricter definition of having all 4 Jewish grandparents, 22.3% of the Caucasian group would be considered Jewish. The breakdown of IBD subtype included 72% with CD, 24% with UC, and 4% with IBDU at diagnosis (Table 1). At the 2016 follow-up, 299 patients (93%) had at least 1 sibling for a total of 581 siblings, and 103 patients (32%) had at least 1 offspring whose mean age was 8.3 ± 6.9 years with 43% being 5 years or younger. Disease location was defined using the Paris classification. For UC, 7.7% were E1, 18.7% were E2, 19.8% were E3, and 53.8% were E4. For CD, 15.6% were L1, 28.1% were L2, 53.7% were L3, 35.9% were L4a, 13.0% were L4b, and 21.2% had the perianal modifier.

The 163 patients who were excluded from analysis were similar in mean age at diagnosis (11.7 ± 3.6 vs 11.1 ± 3.7 years, P = 0.10), sex (49.7% vs 49.1% girls, P = 0.9) and diagnosis subtype (CD 64% vs 72%; UC 28% vs 24%; IBDU 8% vs 4%, P = 0.08) compared to the 322 patients included in the study. Patients who were excluded from the study, however, were less likely to have had a FH+ at diagnosis compared to those who were included (27% vs 39%, P = 0.009) (Table 1). A majority of patients had race/ethnicity documented only at the time of telephone follow-up; therefore, there was insufficient data on race/ethnicity of the excluded patients to compare them to the study patients.

Change in Family History Over Time

At diagnosis, 13.7% of probands had an affected first-degree relative and 38.5% had a positive FH+ of IBD in any-degree relative. Over a 20-year follow-up, the prevalence of FH+ of IBD increased to 26.6% for first-degree relatives and to 52.2% for all relatives (Fig. 1). Of the 99 probands with a new family diagnosis of IBD, 78% had a single new relation with IBD, whereas 16% had 2 and 6% had ≥3 new relations with IBD. The percentage of probands having a sibling with IBD increased from 1.0% at diagnosis to 11.0% at 20 years, whereas those having a parent with IBD increased from 13.0% to 19.1%, grandparent from 6.8% to 8.7%, and first cousin from 5.6% to 10.1% (Table 2).

FIGURE 1: Prevalence of new family history of inflammatory bowel disease over time based on relation to proband.

TABLE 2: Percentage of probands with positive family history of inflammatory bowel disease based on relation to proband over time

Of the 581 total siblings of the probands, 3 (0.5%) siblings had a diagnosis of IBD before the proband, and an additional 21 (3.6%) were diagnosed by 10 years after the proband's diagnosis. Of the 179 total offspring, only 3 (1.7%) had developed IBD by the conclusion of the study.

Stratifying the probands by FH+ at diagnosis demonstrated a statistically significant higher rate of new IBD diagnoses in family members of probands with a prior FH+ at diagnosis compared to those without (43% vs 22% at 20 years, P < 0.001 from Cox regression) (Fig. 2).

FIGURE 2: Prevalence of new family history of inflammatory bowel disease stratified by presence of family history at diagnosis.

Race/ethnicity was strongly associated with ever having a FH+ (P = 0.005). Jewish Caucasians had significantly greater risk of a FH+ compared to non-Jewish Caucasians (hazard ratio [HR] = 1.68, 95% confidence interval [CI] 1.21–2.33, P = 0.002); however, there was no statistically significant difference noted in the prevalence of a FH+ between non-Jewish Caucasians and the non-Caucasian population, of which the majority (76%) were African Americans (P = 0.55).

Disease type (P = 0.33), disease location (P = 0.44–0.98), and age at diagnosis (P = 0.24) were not associated with FH+ at diagnosis. The mean age at diagnosis for those probands with FH+ was 10.8 years (SD = 3.8) compared to 11.3 years (SD = 3.7) for those without FH+ at diagnosis. Of those patients diagnosed ≤5 years old, or very-early onset IBD (VEOIBD), 52.0% had FH+ at diagnosis compared to 37.4% of patients older than 5 years at diagnosis (P = 0.15). There was also no significant association between ever having a FH+, whether at diagnosis or subsequently, and disease type (UC vs CD; HR = 1.16; 95% CI 0.82–1.64; IBDU vs CD HR = 1.22; 95% CI 0.57–2.62; P = 0.64), age at diagnosis (HR = 0.98; 95% CI 0.94–1.02; P = 0.35), or number of first-degree relatives (HR = 1.05; 95% CI 0.94–1.17; P = 0.36).

When looking only at those probands with FH− at the time of diagnosis, there was no significant association with disease type, ethnicity, age at diagnosis, or number of first-degree relatives with likelihood of developing a FH+ by the time of follow-up (Supplemental Table 1, Supplemental Digital Content, https://links.lww.com/MPG/B553).

Concordance Rates

Concordance rates for disease type among probands with CD or UC and their affected family members were 56% (66/117) and 69% (31/45), respectively. For CD probands with a positive FH+, 56% had all affected relatives with CD, 17% had both CD and UC-affected relatives, and 27% had relatives only with UC. While for UC probands, 68.8% had all affected relatives with UC whereas 15.5% had both UC and CD-affected relatives, and 15.5% had only CD-affected relatives.

Disease subtype changed for 7% (23/322) of the probands at follow-up, most (11/23) of which were UC diagnoses reassigned to CD. Half of the probands initially diagnosed with IBDU were reassigned equally to CD or UC.

DISCUSSION

To our knowledge, this is the first cohort study examining changes in prevalence of FH+ of IBD over time in a large population of pediatric-onset IBD patients. Multiple cross-sectional studies have documented the prevalence of FH+ at diagnosis in both pediatric and adult-onset IBD patients ^(3,5–18), but change in family history following the initial diagnosis has not been well studied.

The change in prevalence of FH+ should be taken in the context of the worldwide increase in the incidence and prevalence of IBD perhaps secondary to environmental factors ^{(2,18–20,24,25)}. A significant rise in pediatric-onset IBD, especially in those younger than 10 years of age, consisted of a 9.7% annual increase in IBD incidence between 1999 and 2008 in Ontario, Canada and an overall increase in incidence of IBD in all age groups of 2.3% per year ⁽¹⁹⁾. It is unclear the degree to which the increase in prevalence of FH+ in our population of pediatric-onset IBD patients could have been influenced by the overall observed increase in IBD in the general population.

At diagnosis, 13.7% of our patients had an affected first-degree relative which is consistent with previous data ^{(13,14,16,17)}. The higher overall prevalence of a FH+ compared with prior publications ^(3,5–14) may be related to our inclusion of any degree relative. Furthermore, the long-term follow-up captured additional family members, previously undetected in cross-sectional studies.

We demonstrated a greater observed 10-year risk to siblings of 3.63% which is higher than the 2.09% estimated 10-year risk reported by Moller et al ⁽⁹⁾. The largest increase in prevalence of IBD, especially in siblings, occurred in the first 15 years after the proband's diagnosis which may reflect that siblings are likely entering the age of highest incidence of IBD during that time ^(1,2,4). Of interest, 2% of patients had a grandparent diagnosed after the pediatric proband reminding us that previous generations are still at risk.

Offspring of the probands were not considered in analysis since less than one third of the probands had children at last follow-up. In addition, most of these offspring would not yet be expected to have developed IBD as they were younger (mean age 8.3 ± 6.9 years) than the average reported age range for diagnosis of pediatric-onset IBD of 10 to 12.9 years ^(7,26–29).

If the proband had a FH+ at diagnosis, there was a 2-fold increased risk of having subsequent new familial IBD diagnoses. This is consistent with reports that FH+ of IBD is the strongest risk factor for IBD ^(3,13,18,30).

Race/ethnicity is known to affect risk of IBD likely owing to genetic predisposition. We determined race/ethnicity through patient self-report which has a 99.86% correspondence with genetic cluster membership of the self-identified race/ethnicity ⁽³¹⁾. We demonstrated a significant association between Jewish Caucasian descent and ever having a FH+ of IBD. Although the definition of Jewish descent has varied between studies, prior publications have identified genetic differences between non-Jewish persons and those with only one Jewish grandparent ⁽³²⁾. Yang et al ⁽³³⁾ reported that Jewish descent, defined as having at least 1 Jewish grandparent, was associated with a 2- to 4-fold increased frequency of IBD. Because of concerns of the accuracy of patient reporting, we did not further categorize Jewish descent as Ashkenazi versus non-Ashkenazi; however, it is unclear the extent of effect that the type of Jewish descent has on IBD risk as Zlotogora et al ⁽³⁴⁾ demonstrated similar rates of family history of IBD between Jewish patients with IBD in Israel of Ashkenazi verse non-Ashkenazi descent.

Historically, IBD was thought to predominantly affect Caucasians, although increasing incidence in minorities including African Americans has been noted ^(3,35). We found no statistically significant difference in rates of FH+ between our non-Caucasian group comprised mainly (76%) of African Americans and the non-Jewish Caucasians. Previously reported lower incidence/prevalence in African Americans may be related to underreporting and/or decreased access to care, or it may reflect a true increase in incidence over time.

Earlier age at diagnosis has been associated with FH+ ^{(10,11,36,37)} because genetic factors may contribute more to the development of IBD the younger the age at diagnosis ^(21,26,27). Heyman et al ⁽²⁶⁾ reported a significant increase in a FH+ in patients with UC 2 years or younger (44%) versus older than 2 years (28%). This may be attributed to genetic anticipation with future generations being affected by genetic diseases earlier in life or increased awareness of families which leads to earlier diagnosis from symptom onset (5,27,37). We found no significant difference in age at diagnosis between patients with and without FH+ at diagnosis. The VEOIBD patients were more likely to have FH+ at diagnosis compared to those diagnosed more than 5 years old; however, this did not reach significance which may be explained by the small number (N = 25) of VEOIBD patients in this study.

Concordance rates of IBD subtypes among relatives with IBD have been estimated at 69% to 88.8% ^{(3,6,8,13,15,17,38–40)}. Our concordance rates for CD were lower than previous reports at 56% but similar for UC at 69%.

Our study has many strengths including the large number and the longest reported duration of follow-up, median 18 years, of patients with pediatric-onset IBD ⁽²⁹⁾. Despite the length of follow-up, we reached 66% of patients at follow-up compared to another longitudinal pediatric study which reported contact rates closer to 30% in a similar population ⁽²⁹⁾. In addition, our study population was derived from a comprehensive prospective database and included patients with IBD from both the inpatient and outpatient settings. To our knowledge, this is the first study of its kind to assess changes in FH+ over time allowing us to report observed risks rather than risk estimates which provides more specific data for counseling families regarding the likelihood of disease in other family members.

Our study did have limitations. Our higher rates of FH+ may be related to selection bias since patients recruited from a tertiary pediatric IBD center may represent those with more severe disease which has been associated with familial IBD ^(6,41). We had a higher percentage of Jewish Caucasians than did other studies, although we analyzed the risk of each individual race/ethnicity independently. Furthermore, our small non-Caucasian population may limit our ability to make conclusions in this population. Our study and other IBD epidemiologic studies have used nonethnically diverse populations, which may not be accurately extrapolated to larger populations ^(4,42,43). Our higher female percentage compared to other pediatric-onset IBD studies may be related to the primary investigator being female ⁽⁴⁴⁾. Although we had a high response rate from patients based on the long duration of follow-up, those who were successfully contacted for follow-up were more likely to have a FH+ at diagnosis compared to those who were not reached. This may be secondary to participation bias as those with a FH+ at diagnosis may be more vested in research relating to changes in family history if they already had affected family members. There is also a concern for recall bias since the patients reported updates in FH+ which could have occurred years prior, although patients were contacted in both 2006 and 2016 to shorten the recall interval and ongoing prospective data were collected at each clinic visit while patients were still followed at our center. Information on relative's diagnosis were based on patient's or parent's report and not verified through medical records. Lastly, we likely underestimate lifetime risks as our follow-up lasted a median of 18 years, not the entire lifespan of those relatives.

This is the first study to assess changes in FH+ over time in which all subjects had pediatric-onset IBD and prolonged follow-up. Future studies which include larger multicenter studies and analysis of the proband's offspring should be considered to confirm the potential risk for the development of IBD in this group of first-degree family members. This information may be helpful in counseling pediatric-onset IBD patients and their families about the risks to other family members, particularly siblings, in the future.