Clinical Variables Associated With Statural Growth in Pediatric Crohn’s Disease Differ by Sex (The Growth Study) Free

Abstract

Background

Statural growth impairment is more common in male patients with Crohn’s disease (CD). We identified clinical variables associated with height z score differences by sex in children participating in the Growth Study, a prospective multicenter longitudinal study examining sex differences in growth impairment in pediatric CD.

Methods

Patients with CD (female patients with bone age [BA] ≥4 years 2 months and ≤12 years; male patients with BA ≥5 years and ≤14 years at screening) who had completed study visit 1 qualified. The height z score difference was computed as height z score based on chronological age minus height z score based on BA.

Results

One hundred thirteen patients with CD (36% female) qualified. The mean chronological age was 12.0 ± 1.8 (SD) years. The magnitude of the mean height z score difference was significantly greater in female patients (–0.9 ± 0.8) than in male patients (–0.5 ± 0.9; P = 0.021). An initial classification of inflammatory bowel disease as CD (P = 0.038) and perianal disease behavior at diagnosis (P = 0.009) were associated with higher standardized height gain with BA progression, and arthralgia at symptom onset (P = 0.016), azathioprine/6-merpcaptopurine (P = 0.041), and probiotics (P ≤ 0.021) were associated with lower standardized height gain with BA progression in female patients. Patient-reported poor growth at symptom onset (P = 0.001), infliximab (P ≤ 0.025), biologics (P ≤ 0.015), methotrexate (P = 0.042), and vitamin D (P ≤ 0.010) were associated with higher standardized height gain with BA progression, and initial classification as CD (P = 0.025) and anorexia (P = 0.005) or mouth sores (P = 0.004) at symptom onset were associated with lower standardized height gain with BA progression in male patients.

Conclusions

Different clinical variables were associated with statural growth in male patients vs female patients, suggesting that sex-specific molecular pathways lead to statural growth impairment in CD.

INTRODUCTION

Statural growth impairment, an important complication of pediatric Crohn’s disease (CD), occurs more frequently in male patients than in female patients for unknown reasons.^1-11 In pediatric CD, determination of bone age (BA) allows clinically meaningful interpretation of statural growth in the context of skeletal maturity.^{1, 12} Determining the height z score difference (Table 1)¹—height z scores based on chronological age (CA) (height CA-z scores) minus height z scores based on BA (height BA-z scores)—allows an assessment of standardized height gain in the context of skeletal maturation up to the date these measurements are obtained.¹

In a contemporary cohort of 171 children with CD screened to determine eligibility for participation in The Sex Differences in Statural Growth Impairment in Pediatric Crohn’s Disease Study (also known as the Growth Study),^{1, 13} our ongoing prospective multicenter longitudinal study, the absolute value of the mean height z score difference was lower in male patients than in female patients.¹ This observed sex difference in mean height z score differences indicates that standardized height gain was lower in male patients with BA progression (ie, skeletal maturation) up to the date of the screening visit, reflecting continued poorer statural growth in male patients.¹

This continued observed sex difference in statural growth impairment in a contemporary cohort of children with CD, despite the introduction of many new treatments to the CD armamentarium, suggests that clinicians’ approach to management should be sex-specific.¹ The need to consider sex-specific treatment approaches may reflect different underlying molecular pathways leading to statural growth impairment in male patients vs female patients.

In this report, our aims were to (1) assess sex differences in height z score differences, (2) identify clinical variables associated with height z score differences by sex, and (3) compare the strength of the association between these clinical variables and height z score differences by sex in children with CD at study visit 1 of the Growth Study. We hypothesized that the clinical variables associated with height z scores differences will be different in male patients and female patients, possibly reflecting sex-specific molecular pathways leading to statural growth impairment in CD. This research is critical to better understanding sex-specific differences in statural growth in CD. The novel findings in this report support the need for investigating, identifying, developing, and instituting effective sex-specific treatment strategies, an approach that is not currently part of the pediatric CD management algorithm.

METHODS

Eligibility and BA

Patients with CD satisfying the inclusion criteria for BA (ie, female patients with BA ≥4 years 2 months and ≤12 years; males with BA ≥5 years 0 months and ≤14 years) participated in study visit 1 at one of 8 medical centers between August 2015 and April 2020 after undergoing a screening visit to determine eligibility for enrollment in the Growth Study (1 study visit every 6 months for a total of 5 study visits)¹³ as previously described.¹ BA assessed by left-hand x-ray is regarded as the gold standard measure of skeletal maturity¹⁴ and growth potential; it allows for clinically meaningful interpretation of growth in the context of skeletal maturity in pediatric CD.^{1, 12} Puberty is of longer duration in male patients, explaining the broader BA range for eligibility criteria for male patients.

Left-hand x-rays were obtained at the time of the screening visit for this article and were blindly interpreted for BA by one of the investigators (Robert Lustig),^{1, 3, 12, 13, 15} using the standards of Greulich and Pyle.¹⁴ Because BA reference values vary by sex and CA, BA results were transformed into BA z scores using standard reference values.¹⁶

Description of Variables

Clinical information was collected, and self-Tanner staging was performed.^{1, 3, 12, 13, 15, 17} The first variable, CA, referred to the CA of the patient at the time of the study visit. Race was self-classified as White, Black or African American, Asian, American Indian or Alaskan Native, Native Hawaiian or Other Pacific Islander, Other, Unknown, or Declined. Asian was self-subclassified into South Asian or East Asian. Ethnicity was self-classified as Hispanic or Latino or Not Hispanic or Latino. Tanner stage refers to breast development in female patients and testes/scrotum/penis development in male patients.

Date of initial IBD diagnosis was defined as the date of initial inflammatory bowel disease (IBD) diagnosis based on the date of endoscopy with biopsy. Initial classification of IBD in patients with an ultimate diagnosis of CD was categorized as CD, ulcerative colitis, or IBD-unclassified. Date of diagnosis of CD was defined as the date of initial diagnosis of CD (vs date of initial IBD diagnosis).

Distal disease location at diagnosis was defined as distal 1/3 ileum ± limited cecal disease (L1), colonic disease (L2), ileocolonic disease (L3), or not applicable at the time of diagnosis of IBD.¹⁸Upper disease location at diagnosis was defined as upper disease proximal to the ligament of Treitz (L4a), upper disease distal to the ligament of Treitz and proximal to distal 1/3 ileum (L4b), both L4a and L4b, or not applicable at the time of diagnosis of IBD.¹⁸

Disease behavior at diagnosis was defined as nonstricturing/nonpenetrating disease (B1), stricturing disease (B2), penetrating disease (B3), or both penetrating and stricturing disease, at either the same or different times (B2B3) at the time of IBD diagnosis.¹⁸Perianal disease behavior at diagnosis referred to perianal fistula, anal canal ulcer(s), or perianal abscess at the time of IBD diagnosis.¹⁸

Symptoms and signs at onset, including extraintestinal manifestations, were patient- and/or parent-reported and included abdominal bloating, abdominal pain, abscess, anemia, anorexia/decreased appetite, anxiety, appendicitis, arthralgia, arthritis, body aches, Clostridioides difficile infection, clubbing, constipation, diarrhea, eczema, erythema nodosum, fatigue, fecal urgency, fevers, fistula, headaches, hematochezia, infection, lip swelling, mouth sores, nausea, nocturnal stooling, pallor, pancreatitis, perianal skin tag, poor growth/decreased height velocity, poor weight gain, pyoderma gangrenosum, rash, rectal bleeding, short stature, tenesmus, urinary symptoms, vomiting, and weight loss.

Reported use of medications of interest included 5-aminosalicylates (balsalazide, mesalamine, olsalazine, sulfasalazine), antibiotics (amoxicillin, azithromycin, ciprofloxacin, metronidazole, rifaximin, tobramycin, vancomycin), biologics (adalimumab, certolizumab, infliximab, natalizumab, ustekinumab, vedolizumab), corticosteroids (intravenous, oral, intranasal, inhaled, rectal, topical, eye drops, swish-and-spit/oral rinse, etc.), enteral nutrition, immunomodulators (azathioprine/6-mercaptopurine, cyclosporine, methotrexate, tacrolimus), probiotics, vitamins (folic acid, iron, multivitamin, vitamin B₁₂, vitamin D, zinc). Time of medication exposure was collected as <1 year ago, 1 to <5 years ago, and ≥5 years ago, with respect to time of evaluation. Medication ever referred to a report of medication exposure in any timeframe.

Weight and height were measured using a digital scale to the nearest 0.1 kg and a stadiometer to the nearest 0.1 cm, respectively; body mass index (BMI) was calculated as the weight in kg divided by the square of the height in meters. The z scores were calculated based on CA and BA for weight, height, and BMI (ie, weight, height, or BMI CA-z score and weight, height, or BMI BA-z score) using reference tables from the Centers for Disease Control and Prevention, National Center for Health Statistics.¹⁹ The height z score difference (Table 1),¹ the primary outcome in the analyses described herein, was calculated as height CA-z score minus height BA-z score.¹

Statistical Analyses

To describe the sample, we report the descriptive statistics for demographic characteristics and clinical variables of interest for our study participants. To compare female and male patients on clinical variables, we used the t test for continuous variables and the χ ² test for categorical variables. We employed linear regression to assess the association between clinical variables and height z score difference, adjusting for CA, race, and number of years since IBD diagnosis. Race was dichotomized (White compared with non-White) because of small numbers in each category when left as multilevel categories.

We constructed the above analysis for female and male patients and report the regression coefficient corresponding to the clinical variable of interest to represent its strength of association with the height z score differences. In addition, we further examined whether the strength of these associations differed by sex. The comparisons of sex difference were conducted via multiple linear regression analysis with a model that included the clinical variable of interest (1 in each model), potential confounding factors (ie, CA, race, and time since diagnosis of IBD), sex, and the clinical variable-by-sex interaction. Here we show the regression coefficient of the clinical variable-by-sex interaction term to represent the sex difference in strength of association between clinical variable and height z score difference. For each of the reported regression coefficients, we also present its corresponding 95% confidence interval and P value. P ≤ 0.05 was considered statistically significant. The SIR/XS version 19 relational database management system software was used to capture and store the data on a secure web-based Citrix platform. Data were analyzed using SPSS Version 26.

Ethical Considerations

Each participating site obtained institutional review board approval for the study protocol, and written informed consent and assent were obtained from parents and patients before participation in the Growth Study.

RESULTS

Baseline Information

As of April 2020, 113 patients (72 [64%] male patients) completed study visit 1. The mean time since screening visit was 3.4 ± 0.9 months (range, 1.9-5.8) for female patients and 3.4 ± 0.7 months (range, 2.5-5.6) for male patients (P = 0.997). Table 2 provides summary statistics for demographics and Tanner stage. Race, ethnicity, and Tanner stage did not differ by sex (P = 0.520, P = 0.666, and P = 0.183, respectively). Table 3 shows the distribution of baseline disease characteristics by sex. The mean CA at study visit 1 for female patients (11.2 ± 1.6 years; range, 7.5-13.8) was lower than for male patients (12.5 ± 1.7 years; range, 8.8-16.0; P < 0.001) based on our sex-specific CA eligibility criteria.¹

Disease Duration

Average disease duration since initial diagnosis of IBD was 1.9 ± 1.6 years (range, 0.3-6.1) in female patients and 2.7 ± 2.2 years (range, 0.2-9.0) in male patients (P = 0.026). Average disease duration since initial diagnosis of CD was 1.7 ± 1.5 years (range, 0.3-6.1) in female patients and 2.6 ± 2.1 years (range, 0.2-9.0) in male patients (P = 0.011).

BA and Anthropometrics

The mean BA at screening was 10.0 ± 1.6 years (6.8-12.0) for female patients and 11.8 ± 1.8 years (7.0-14.0) for male patients. The mean bone age z score at screening was –0.2 ± 0.5 (range, –0.6 to 0.5) for female patients and –0.2 ± 0.4 (range, –0.7 to 0.4) for male patients (P = 0.520). The sex difference remained nonsignificant when adjusted for CA, race, and number of years since IBD diagnosis. For females, the mean height CA-z score was –0.6 ± 0.9 (range, –2.3 to 1.0) and their mean height BA-z score was 0.4 ± 0.9 (range, –1.3 to 2.3). For male patients, the mean height CA-z score was –0.1 ± 1.0 (range, –2.3 to 2.3) and their mean height BA-z score was 0.4 ± 0.9 (range, –1.4 to 2.7). The mean height z score difference (Table 1)¹ was significantly greater in female patients (–0.9 ± 0.8; range, –2.6 to 0.5) than in male patients (–0.5 ± 0.9; range, –3.6 to 1.0; P = 0.021). The mean weight z score difference was significantly greater in female patients (–0.7 ± 0.6; range, –1.9 to 0.3) than in male patients (–0.4 ± 0.7; range, –2.4 to 0.8; P = 0.031). The mean BMI z score difference was greater in female patients (–0.3 ± 0.2; range, –0.8 to 0.1) than in male patients (–0.2 ± 0.3; range, –0.9 to 0.3) and trended toward significance (P = 0.057). A statistically significant sex difference was present for height z score difference, weight z score difference, and BMI z score difference when adjusted for CA, race, and number of years since IBD diagnosis.

Association of Clinical Variables With Height z Score Differences

Table 4 (clinical variables at diagnosis), Table 5 (symptoms/extraintestinal manifestations at onset), and Table 6 (medications) display the association of clinical variables with height z score differences by sex and the strength of association of the variables with height z score differences in male patients compared with female patients.

In female patients, initial classification of IBD as CD and perianal disease behavior at diagnosis were associated with higher standardized height gain with BA progression (Table 2), and arthralgia at symptom onset (Table 3), azathioprine/6-merpcaptopurine use ≥1 year to <5 years ago, probiotics use <1 year ago, probiotics use ≥1 to <5 years ago, and probiotics use ever (Table 4) were associated with lower standardized height gain with BA progression.

In male patients, patient-reported poor growth/decreased height velocity as a presenting symptom (Table 3), use of infliximab <1 year ago, use of infliximab ever, use of biologics <1 year ago, use of biologics ever, use of methotrexate ever, use of vitamin D <1 year ago, use of vitamin D ≥1 year to <5 years ago, and use of vitamin D ever (Table 4) were associated with higher standardized height gain with BA progression, and initial classification as CD (Table 2), and anorexia or mouth sores at symptom onset (Table 3) were associated with lower standardized height gain with BA progression.

There was a sex difference in the strength of the association between initial classification as CD, mouth sores at symptom onset, infliximab use ever, use of biologics <1 year ago, biologics use ever, vitamin D use <1 year ago, vitamin D use ≥1 year to <5 years ago, and vitamin D use ever and height z score differences.

DISCUSSION

In study visit 1 of our longitudinal Growth Study, we found that the absolute value of the mean height z score difference (Table 1)¹ was lower in male patients than in female patients. This finding reflects lower standardized height gain in male patients with BA progression up to the time of study visit 1. We identified several sex-specific clinical variables associated with height z score differences: initial classification of disease as CD, perianal disease behavior at diagnosis, specific symptoms/extraintestinal manifestations at presentation, and specific medication use. Each of the identified variables was either significantly associated with height z score difference in one sex and not the other sex or was more strongly associated with the height z score difference in one sex than in the other sex.

The association of initial classification of IBD as CD with higher standardized height gain with BA progression in female patients and lower standardized height gain with BA progression in male patients is consistent with sex differences in the risk for statural growth impairment in pediatric CD^1-11 and the basis for this current prospective longitudinal Growth Study.^{1, 13} Furthermore, a sex difference in the strength of the association between initial classification as CD and height z score differences was present.

Perianal disease behavior at diagnosis was associated with higher standardized height gain in female patients. Perianal disease has been linked to a greater inflammatory burden²⁰ and identified as a predictor for poor outcomes.^{21. 22} Assa, Amitai, et al²⁰ reported that children with perianal CD had lower height z scores (based on CA) than children without perianal involvement. In contrast, other studies found that height z scores (based on CA)^{23, 24} and height velocity z scores²⁴ (based on CA) did not differ based on the presence or absence of perianal disease in children with CD. Comparison of findings from these studies to ours is difficult because they did not examine height z scores based on BA or height z score differences. The association between perianal disease behavior at diagnosis (thought to reflect more severe disease/higher inflammatory burden) and higher standardized height gain with skeletal maturation in female patients supports the theory that female patients are more resistant to the negative effects of inflammation on statural growth.^{1, 3-5, 13}

Arthralgias at symptom onset in female patients and anorexia and mouth sores at symptom onset in male patients were associated with lower standardized height gain with skeletal maturation. In contrast, poor growth/decreased height velocity reported as a presenting symptom by patients/families was associated with higher standardized height gain with BA progression in male patients. Nir et al²⁵ reported that median height z scores (based on CA) did not differ in patients with pediatric CD based on joint involvement, but they did not examine height z scores based on BA or height z score differences for comparison to our study. Assa, Rinawi, et al²⁶ reported that growth delay (G1 phenotype) was associated with lower median height z scores (based on CA) but did not examine height z scores based on BA or height z score differences for comparison to our study. We will investigate the influence of study visit 1 growth status (based on actual height measurements rather than patient report) with statural growth (height velocity) by sex during the 2-year follow-up period of the Growth Study at the completion of data collection.

Arthritis and oral lesions are immunologically mediated extraintestinal manifestations and typically reflect intestinal disease activity related to active inflammation. A sex difference in the strength of the association between mouth sores at symptom onset and height z score differences was present. Anorexia is also attributed to active inflammation, with TNF-α as an important mediator.²⁷ Perhaps the pathophysiologic mechanisms that lead to joint manifestations and anorexia/mouth sores overlap with the underlying pathophysiologic mechanisms that lead to statural growth impairment in female and male patients, respectively. However, the etiology of these findings is currently unknown and requires thorough investigation.

Azathioprine/6-mercaptopurine and probiotics use were associated with lower standardized height gain with BA progression in female patients. In a different cohort of pediatric patients with CD, we found that thiopurine use was negatively associated with lean tissue mass (midarm circumference CA-z scores) and fat store (subscapular skinfold CA-z scores) measurements and with weight CA-z scores and BMI CA-z scores in pediatric CD.²⁸ In addition, we reported a persistent negative association with subscapular skinfold BA-z scores (although did not achieve statistical significance because of a smaller sample size available for BA analyses).²⁸ Furthermore, thiopurine treatment was associated with lower standardized bone age results.¹² These findings may reflect a lack of efficacy of thiopurine treatment in pediatric CD rather than a direct negative impact of the medication,²⁸ as thiopurines have not been shown to improve height velocity by other investigators.²⁹ Of the 5 female patients in this study reporting thiopurine use ≥1 to <5 years ago, 3 patients also reported exposure to infliximab during this timeframe. We do not know if these exposures represented combination therapy or sequential therapy. Furthermore, there was insufficient statistical power to assess the association of the overlap of these medications during this timeframe with height z score differences because of the small sample size of this subgroup.

Although studies have suggested that methotrexate,³⁰ infliximab,³¹ and adalimumab³¹ improve statural growth, no prospective randomized trials to test the effects of supplementation in targeted populations of patients with impaired growth and incorporating BA for optimal interpretation of growth have been conducted. In male patients, infliximab, biologics, and methotrexate were associated with higher standardized height gain with BA progression. Of the 47 male patients reporting use of infliximab <1 year ago, 3 patients reported use of thiopurines and 21 patients reported use of methotrexate in this same timeframe. We do not know if these exposures represented combination therapy or sequential therapy. Overlap of these medications during this timeframe was not associated with height z score differences.

A sex difference in the strength of the association between infliximab use ever, biologic use ever, and use of biologics <1 year ago and height z score differences was present. These identified associations between medication use and higher standardized height gain with BA progression in male patients only raise the possibility of sex-specific responses to treatment from the standpoint of statural growth, lending further support to our theory of sex-specific molecular pathways leading to statural growth impairment and the need for sex-specific treatment approaches.

Use of vitamin D was associated with higher standardized height gain with BA progression in male patients. Furthermore, a sex difference in the strength of the association for vitamin D use <1 year ago, vitamin D use ≥1 year to <5 years ago, and vitamin D use ever and height z score differences was present. In addition to its role in bone growth and mineralization,³² vitamin D is thought to interact with the growth hormone-IGF-1 axis.³² Vitamin D has been reported to increase circulating IGF-1 and IGFBP-3 and height.³³ The biochemical interaction of vitamin D with the GH/IGF-1 axis may depend on a patient’s vitamin D status.³² We are collecting serial vitamin D levels and serial information on vitamin D supplementation and thus will assess the impact of vitamin D on statural growth prospectively in our cohort of children with CD.^{1, 13} We aim to understand why the strength of association between vitamin D and height z score differences vary between male and female patients and understand the timing of vitamin D in relation to its impact on statural growth.^{1, 13}

Probiotics use was associated with lower standardized height gain with BA progression in female patients. Probiotic supplementation has improved height gain in undernourished preschool-age children,³⁴ whereas probiotics have not been shown to improve statural growth in healthy children.³⁵ In addition, probiotics have not been effective in the treatment of CD.³⁶ The etiology of our finding is unclear and highlights the need to understand the impact of the gut microbiota on statural growth in CD. Evidence suggests that the intestinal microbiota mediates growth through its effects on the growth hormone/IGF-1 axis.³⁷ In addition, the gut microbiota has been implicated in influencing sex hormones³⁸ and vice versa.³⁹ Furthermore, sex differences in immune responses may occur secondary to sex-specific microbiota.⁴⁰ Perhaps sex-specific microbiota differentially affect the hypothalamic-pituitary-gonadal axis and growth hormone/IGF-1 axis, contributing to the observed sex differences in statural growth impairment in pediatric CD.

We have described a significant sex difference in statural growth in a contemporary cohort of children with CD despite management with current therapies and have provided evidence that the association of multiple clinical variables with height z score difference differs between male and female patients with CD, strongly indicating a need for sex-specific treatment.¹ A sex-specific treatment approach for pediatric CD does not currently exist and will require development over time as we enhance our knowledge with continued scientific investigation. This suggested need for sex-specific treatment approaches may reflect different underlying molecular pathways leading to statural growth impairment in male vs female patients.

CONCLUSIONS

We found that different specific clinical variables were associated with statural growth in male patients compared with female patients and that a sex difference in the strength of the association between specific clinical variables and height z score differences was present, supporting our hypothesis that different variables will be important in risk stratification for statural growth impairment in male vs female patients.

Female patients appear to grow better independent of disease severity/inflammatory burden and medication interventions. As such, we suspect that the molecular pathways leading to statural growth impairment in female patients are different than those in male patients, and there may be a difference in the response of these sex-specific molecular pathways to current medications used to treat pediatric CD. Understanding the biological mechanisms that lead to statural growth impairment by sex will set the stage for identifying and developing management approaches that enable normal growth and development in both male and female patients.

Findings from the prospective multicenter longitudinal Growth Study will enable us to identify factors for each sex that predict which patients are at highest risk for statural growth impairment refractory to standard therapeutic approaches. We will conduct multivariate analyses incorporating disease phenotype, disease activity, and medication interventions simultaneously at the completion of data collection over the 2-year follow-up period of our study. Our findings will direct future studies focused on improving statural growth impairment in patients with high risk. Furthermore, we aim to define the optimal timing of and specific treatments for both male and female patients based on their individual risk for statural growth impairment.

Sex will likely be an important future determinant of treatment decisions, which will represent a major advancement in clinical decision-making for pediatric CD. Through the Growth Study, we aim to transform the care of pediatric patients with CD by providing an evidence-based approach for the appropriate early introduction of aggressive therapy in patients with high risk for each sex because there is only a narrow therapeutic window available for intervention to improve statural growth.

Author contributions: Neera Gupta obtained funding for this study and prepared the initial and final drafts of the manuscript. Neera Gupta, Robert Lustig, Howard Andrews and Cheng-Shiun Leu contributed to the concept, design, data management, and analyses. All authors contributed to data collection, data interpretation, editing and revising the manuscript, and final approval of the manuscript.

Supported by: This project was supported by NIH/NICHD R01HD075929 (NG), Crohn’s and Colitis Foundation Senior Research Award 545496 (NG), Weill Cornell Medicine Department of Pediatrics (NG); Weill Cornell Medicine Clinical and Translational Scientific Center NIH/NCATS UL1TR002384. The study sponsors had no role in the study design, collection, analysis, interpretation of data, or writing of the report.

ACKNOWLEDGMENTS

We thank the patients at all sites for participating in this study and thank their families for their support. We thank our colleagues for informing patients and families about the Growth Study. We thank Dr. Joel Rosh and his team for screening/enrolling patients and submitting data for the study. We thank the entire multicenter Growth Study team.

REFERENCES