Analysis of therapy monitoring in the International Congenital Adrenal Hyperplasia Registry

Abstract Objective Congenital adrenal hyperplasia (CAH) requires exogenous steroid replacement. Treatment is commonly monitored by measuring 17‐OH progesterone (17OHP) and androstenedione (D4). Design Retrospective cohort study using real‐world data to evaluate 17OHP and D4 in relation to hydrocortisone (HC) dose in CAH patients treated in 14 countries. Patients Pseudonymized data from children with 21‐hydroxylase deficiency (21OHD) recorded in the International CAH Registry. Measurements Assessments between January 2000 and October 2020 in patients prescribed HC were reviewed to summarise biomarkers 17OHP and D4 and HC dose. Longitudinal assessment of measures was carried out using linear mixed‐effects models (LMEM). Results Cohort of 345 patients, 52.2% female, median age 4.3 years (interquartile range: 3.1–9.2) were taking a median 11.3 mg/m2/day (8.6–14.4) of HC. Median 17OHP was 35.7 nmol/l (3.0–104.0). Median D4 under 12 years was 0 nmol/L (0–2.0) and above 12 years was 10.5 nmol/L (3.9–21.0). There were significant differences in biomarker values between centres (p < 0.05). Correlation between D4 and 17OHP was good in multiple regression with age (p < 0.001, R 2 = 0.29). In longitudinal assessment, 17OHP levels did not change with age, whereas D4 levels increased with age (p < 0.001, R 2 = 0.08). Neither biomarker varied directly with dose or weight (p > 0.05). Multivariate LMEM showed HC dose decreasing by 1.0 mg/m2/day for every 1 point increase in weight standard deviation score. Discussion Registry data show large variability in 17OHP and D4 between centres. 17OHP correlates with D4 well when accounting for age. Prescribed HC dose per body surface area decreased with weight gain.


| INTRODUCTION
Congenital adrenal hyperplasia (CAH) is an autosomal recessive condition leading to glucocorticoid deficiency, androgen excess, variable degrees of mineralocorticoid deficiency, salt wasting and a risk of life-threatening adrenal crisis. Poorly controlled CAH causes abnormal growth resulting in reduced adult height, reduced quality of life, increased comorbidities and shorter life expectancy. 1,2 Significant variation in treatment strategies has been noted in the United Kingdom and internationally, including using different formulations and dosing regimens. 3 An international consensus statement in 2002 4 was followed by a 2010 Endocrine Society guideline, 5 updated in 2018, that improved guidance for clinicians, 6 but there remain points of contention. The optimal balance of glucocorticoid, mineralocorticoid replacement, and need for salt replacement in infants is debated. It is acknowledged that treatment should be individualised, but precisely how to use the results from biochemical markers in the context of biometric measurements in children is unknown. 2,7 The recommended daily dose range of hydrocortisone (HC) is 10-15 mg/m 2 /day, with a recent review advocating doses up to 18 mg/m 2 /day. 7 However, others suggest doses over 17 mg/m 2 /day should only be used with care during puberty as adult height has been shown to correlate negatively with glucocorticoid dose. 1,2,8 Maintaining 17-OH progesterone (17OHP) concentrations in the upper end of the normal range is suggested, 6 with alternative targets including 17OHP of 10-20, 9 12-36 10 or 3-36 nmol/l across all ages and sexes. 2 Interpreting 17OHP and Androstenedione (D4) is challenging due to interindividual variability of their concentration profile in relation to glucocorticoid replacement, and variable practice in measurement in relation to timing of medication administration. 11,12 While alternative serum steroids 13 and urinary steroids 14 have been advocated for monitoring CAH, 17OHP and D4 are likely to be most frequently used in the medium term.
We analysed real-world data from the International Congenital Adrenal Hyperplasia Registry (I-CAH) (www.i-cah.org) to compare reported measurements of serum hormones in relation to prescribed doses of HC. We designed a longitudinal analysis, with repeated measures from patients managed in centres throughout different countries of the world, to gain insight into variations within patients as they age, between patients and to quantify the differences in results between different centres.

| Study design, setting and participants
This retrospective multi-centre cohort study, including 21 centres (14 countries), analysed information on patients from the I-CAH registry. The I-CAH Registry is an international database of pseudonymised information on patients with CAH and is approved by the National Research Ethics Service in the United Kingdom as a research database of information that is collected as part of routine clinical care (19/WS/0131). The data within this registry are deposited by clinicians following informed consent from patients or guardians. Participants were under 19 years of age with a diagnosis of 21-hydroxylase deficiency (21OHD) treated with oral HC as glucocorticoid replacement. All clinic visits that were recorded between January 2000 and October 2020 were analysed in this study. Data fields included in analysis are listed in Supporting Information: Table S1. Study design is limited by no overall quality assurance between the centres for laboratory assays, and variation in timing of sample collection and techniques of laboratory analysis and auxological assessments. This limitation is mitigated in part by advanced statistical analysis and separation of multilevel models into appropriate fixed and random effects.

| Data analysis
Serum 17OHP and D4 of patients within different centres was summarised alongside their height, weight, and most recent dose of HC, and serum biomarkers compared between centres. Recommended range thresholds for 17OHP were 12-36 nmol/L as recommended before morning medication by Merke et al., 10 although it should be noted that there is no international consensus on a precise target range for 17OHP in CAH and the timings of measurement around morning dose within this cohort varied (alternative 10-20 nmol/L analysis in Supporting Information: Table S2).
The same variables measured in patients over time were analysed using linear mixed-effects modelling (LMEM) to obtain insight into the within-patients and between-centres variability of HC dose and biomarkers. LMEMs are multilevel regression equations that allow stratification of different groups of data. Fixed effects are variables assumed to have a consistent effect across the whole cohort. Random effects are used to group aspects of the model that are interrelated, and thus have different coefficients that apply to each of the separate groups of data. 15 The model intercepts varied based on the random effects in our models of patient at level 1 and treatment centre at level 2. The fixed effects of each model are described within the results. This is not an appropriate technique to model nonlinear data, and thus height was not modelled.
Total daily dose of HC was expressed per BSA, calculated using the Mosteller formula. 16 Weight was converted to age and sex-  Biomarkers reported in the registry below the lower limit of detection of a centre's assay are rounded to zero. Median 17OHP was 35.7 nmol/l (IQR: 3.0 to 104), 15.9% within a target of 12-36 nmol/L and 50.0% above this range, and median D4 was 0 nmol/L (IQR 0 to 3.5).
Median 17OHP was inside the tighter control range of 10-20 nmol/l in just 0.6% of patients (Supporting Information: Table S2). Bayesian change-point analysis confirmed no suitable age categorisation for 17OHP, but a change point for D4 of 12 years, thus summary statistics were produced for those under and over 12 years However, differences in weight and dose are noted and thus interpreting causality from comparisons between these crosssectional cohorts is inappropriate, and instead warrants advanced statistical modelling.

| Comparison between sexes, centres and time periods
There was no significant difference in dose, weight SDS, BMI SDS, or biomarker readings between males and females (p > 0.05) ( Table 1)

| Correlation between biomarkers
There was good correlation between 17OHP and D4, strongest when controlling for patient age (

| Univariate LMEM analysis
We found an increase in D4 with age, an increase in weight and BMI standard deviation score (SDS) with age and a decrease in dose relative to BSA with age (  The biomarker 17OHP varies with age and sex in healthy children, 17 with guidance that values in CAH may be above the normal range in patients with adequate control. 4 Recent guidelines and reviews suggest that normalising 17OHP inappropriately is an indication of overtreatment in CAH, without specifying a precise target. 6,7 Alternative target ranges are based on expert opinion and include 3-36 or 12-36 nmol/l for 17OHP, with advice to normalise D4 into the sex and age-specific range. 10 The median 17OHP we report here is at the upper end of these target ranges with large interindividual variability. Painful phlebotomy can influence 17OHP, 18 thus variability will in part be due to heterogeneity in sampling techniques and timing of blood tests in relation to HC administration between different centres. These data highlight the difficulty in interpreting isolated measurements and therefore the importance of holistic patient assessment.
Both 17OHP and D4 have a short half-life, vary throughout the day and relative to treatment administration, 17OHP having greater variability than D4. 19,20 Some centres perform multiple measurements of serum steroids to accurately predict their 24 h profile. 12,19,21 Alternatives such as 21-deoxycortisol have been investigated for the diagnosis of CAH, but not routinely used for monitoring treatment. 13 Urinary steroid profiles can contribute to disease monitoring, although thresholds need validating in larger patient populations. 14 Serum 11-hydroxy-testosterone and 11-Ketotesterone have been shown to discriminate well between poor and good control in CAH.
However, it has been suggested they perform better in adults than children. 22 In our extraction of data for modelling, 66% of clinic assessments had either 17OHP or D4 measured, indicating the high prevalence of their use. Developing the evidence base of how best to interpret these hormones when taken as point measurements is important, and while interpretation is difficult, they will likely remain the most frequently used biomarkers of disease control for the foreseeable future.
While 17OHP varies with age in healthy children, 17,23 our crosssectional analysis shows higher levels in the cohort over 12 years.  Figure 2).
Values reported in this study are those that clinicians are faced with in routine practice and rely upon to make clinical decisions, which gives us valuable insight into these metrics outside of the confines of controlled settings. Limitations associated with this data from different centres include different laboratories using different assays, and blood drawn at different times. Differences in demographics and prevalence of different genotypes in different countries will contribute to differences in the metrics studied that will not exclusively be due to differences in clinical practice. The small number of large outlying values may indicate noncompliance with treatment or data entry errors.
In conclusion, children with increasing weight and BMI SDS are being prescribed less glucocorticoid dose per BSA. Assessment of biochemical markers within this relationship has not shown clear detriment to their disease control, although this warrants further investigation in relation to a more holistic assessment of control. Dose should be regularly reviewed taking into consideration their growth, pubertal development, biomarkers, side effects of treatment and compliance. Standard biomarker measurement practices are needed to evaluate biochemical evidence of disease control. Collection of realworld data within the established I-CAH platform is to be encouraged and will allow us to gain further insights into patients as they progress through puberty, helping to improve patient care and reducing unwarranted variation in practice.