Growth and Treatment in Congenital Adrenal Hyperplasia: An Observational Study from Diagnosis to Final Height

Congenital adrenal hyperplasia (CAH) due to 21-hydroxylase deficiency is an autosomal recessive disorder with varying degrees of compromised cortisol and aldosterone synthesis [1, 2]. Low levels of cortisol lead to increased pituitary secretion of ACTH, and excess adrenal production of androgen precursors and androgens [2]. The classical forms lead to a deficiency of cortisol and aldosterone that may cause hypoglycemia, hyponatremia, failure to thrive, and potentially lethal shock, especially in untreated neonates with the salt-wasting form (SW). Females with both SW and SV CAH have virilized external genitalia to varying degrees at birth due to the elevated levels of prenatal androgens [1]. Individuals with SV CAH have increased androgens, but typically not overt salt loss. The mildest form, nonclassical CAH (NC CAH), has no risk of salt loss, does not cause prenatal virilization, and is often cortisol-sufficient. It may be diagnosed in childhood due to signs of androgen excess such as accelerated growth velocity and premature adrenarche or due to infertility or hirsutism in adult women that benefit from treatment but can also remain undetected, especially in men [3]. Neonatal screening for CAH started in Sweden in 1986 [4].

There is a spectrum of CAH severity with good genotype-phenotype correlation. Clinical severity can thus be predicted from the CYP21A2 mutation [5, 6] and the patients can be classified into genotype groups depending on the severity of the mildest allele.

Treatment consists of replacement with glucocorticoid and, in most cases, with classic CAH, mineralocorticoid (fludrocortisone) [7]. There is a fine balance between avoiding overtreatment resulting in decelerated growth and weight gain and undertreatment leading to androgen excess with virilization, growth acceleration, advanced bone age, and compromised final height (FH) [2, 8, 9]. The recommended dose of glucocorticoid in growing children and adolescents with CAH is 10–15 mg of hydrocortisone equivalents/meter square body surface area per 24 h (HC-eq/m² and day), divided into three or four doses per day [1, 2, 7, 9]. Most studies suggest that hydrocortisone is preferable to long-acting, more potent synthetic glucocorticoids that may negatively affect growth [10]. Children with SW CAH require, in addition to the fludrocortisone, supplementation with sodium chloride during the first year of life or longer [1, 2, 9].

Meta-analyses have shown that patients with CAH lose about 1.4 SDS in FH, or −1.03 SDS (95% CI: −1.20 to −0.86) in FH corrected for target height [11]. Overall, no associations between the height outcome and sex, age at diagnosis, age at puberty onset, or glucocorticoid treatment have been observed. However, patients with mineralocorticoid treatment had a better height outcome in the more recent meta-analysis, despite the inclusion of patients with late diagnosed, non-classic CAH [11]. This is in accordance with other studies reporting mineralocorticoid treatment and sufficient salt supplementation to be important for optimizing growth in all classic forms of CAH [12‒14]. A low dose of fludrocortisone will also reduce the total glucocorticoid dose also in patients with SV CAH [15].

The FH in children with late diagnosed, non-classic CAH have been reported to be affected by the age at diagnosis and the severity of the phenotype, including the bone age advancement at diagnosis [13, 16]. Patients with early diagnosis have been reported to achieve a taller FH [14].

Growth in infancy and during the first 1.5–2 years of life appears to be relatively insensitive to androgens. Neither the growth velocity nor the bone age was affected in children below 1 year of age [17, 18]. Excessive glucocorticoid treatment to reduce androgen levels may therefore not be necessary during this first growth phase [17, 18]. Pubertal development has been reported to start earlier in CAH [19], with peak height velocity about 2 years earlier than in the normal population [20] and a reduced pubertal growth [19, 20]. The height gain during the pubertal growth was negatively correlated with the glucocorticoid dose [21, 22]. FH was negatively correlated with BMI during childhood in CAH [23] possibly through direct effects of the glucocorticoids on the growth plate as well as an earlier menarche in overweight girls with CAH [24].

The currently used hydrocortisone treatment regimen fails to mimic the circadian rhythm of cortisol secretion even when the highest dose is given in the morning [25]. The modified release hydrocortisone preparation with delayed and sustained release of hydrocortisone given late at night [26] is not yet available for children below the age of 12 years. New treatment strategies that aim to lower androgen synthesis by enzyme inhibition or CRF-1 receptor antagonists are being developed [27, 28].

The aim of the present study is to investigate differences in growth, FH, and BMI between patients with different severities of CAH, i.e., CYP21A2 genotype groups, and different glucocorticoid treatment strategies during the different phases of growth. We took advantage of the fact that treatment regimens differed between centers in Sweden and which allowed comparison between the outcome of treatment with hydrocortisone alone and hydrocortisone plus prednisolone. We were therefore able to observe the effect of adding a low dose of long-acting glucocorticoid in the evening.

All children with CAH detected by newborn screening or diagnosed at a later age in Sweden between January 1, 1989, and December 31, 1994, were prospectively included. A physician in the research group met all the patients for a clinical examination and to obtain informed consent from the parents. Laboratory investigations and genetic analyses were performed. Patients were followed until 18 years of age or until they had achieved FH. The local clinician reported data concerning auxological parameters and treatment.

The study included 86 cases detected between 1989 and 1994. Most cases (73/86) were diagnosed by neonatal screening using 17-hydroxyprogesterone (17OHP) or prenatal genotyping. A definite diagnosis within the first month of life was considered an early diagnosis. The mean age at diagnosis for the late-diagnosed patients was 6.6 ± 3.9 years (range: 1.8–16.3). Six patients were not followed until FH (two suffered from other serious diseases and four were lost to follow-up).

CYP21A2 genotyping was performed as previously described [29]. Based on the overall good correlation between genotype and phenotype [5, 6] the patients were divided into genotype groups based on the severity of the milder allele (Table 1). The patients with SW CAH comprised individuals with null genotype or compound heterozygous for the I2 splice mutation. The SV group included individuals in I172N and P30L genotype groups, and individuals with NC CAH most often carried V281L mutations (Table 1).

Auxological data sent by the local clinician were plotted onto the growth chart based on the Swedish reference population [30]. Height and weight were read from the growth charts at 0, 0.25, 0.5, 0.75, 1.0, 1.5, 2 years, and annually thereafter until FH or at 18 years of age. The height and weight SDS were calculated based on the Swedish reference population [30]. TH was calculated as the mid-parental height +/− 6.5 cm according to sex and body surface area using the DuBois formula [31].

All changes in treatment were recorded per the mean dose for the different time periods, 0–0.25, 0.25–0.5, 0.5–0.75, 0.75–1, 1–1.5, 1.5–2 years of age and thereafter between each full year were calculated. Glucocorticoid treatment was converted into HC equivalents/m² per day, assuming 1 mg of hydrocortisone to equal 1.25 mg of cortisone acetate [32] or 0.2 mg prednisolone [33]. The 9α-fludrocortisone was calculated as being 10 times more potent than hydrocortisone regarding the glucocorticoid effect [33] and included in the calculation of HC/m² and day. It was recorded whether the child was treated with hydrocortisone alone or with the addition of prednisolone. For late-diagnosed cases, auxological data were collected retrospectively until the age at diagnosis.

Depending on data distribution, the Student’s t test or the Mann-Whitney U test was used to compare continuous variables between the two groups. Repeated measure ANOVA with the Greenhouse-Geisser correction was used for comparisons between groups of repeatedly measured variables, and post hoc analyses were adjusted using Bonferroni corrections. The statistical difference in SDS between children with CAH and normal reference children was calculated using a one-sample t test. Spearman’s correlation test was used to analyze correlations. The χ² test or the Fischer’s exact test was used, when appropriate, for comparisons of proportions. A p value below 0.05 was considered significant.

The study population (n = 80) was followed from a median age of 7 days (range 0–16.3 years) to a median age of 18.0 ± 0.4 years. Patients were followed until they reached FH. Growth and treatment were assessed separately for the different growth periods; infancy, childhood, and adolescence. The mean glucocorticoid treatment doses, calculated as hydrocortisone equivalents, were within the international recommendations (Table 1). Dosing did not differ between the clinical severity groups or between males and females for any growth period.

The average glucocorticoid dose during the periods 0–2 years and 2–11 years of age correlated positively with height SDS from 2 to 9 years (p < 0.025, correlation 0.28–0.41) but not with height SDS at 10 and 11 years. The average glucocorticoid dose between 11 and 18 years of age was not associated with height SDS. The average glucocorticoid dose at any specific growth period and the total average glucocorticoid dose from 0 to 18 years were associated with the cFH SDS. The total average glucocorticoid dose was not related to BMI at 18 years of age.

Prednisolone treatment was used as an adjunct to hydrocortisone or cortisone acetate and was used only in individuals followed in one part of the country. Hence, it was dependent on the local protocol in the clinic rather than the treatment outcome for the individual patient. Prednisolone was used in combination with HC or cortisone acetate in patients from all of the genotype groups and was not overrepresented in any genotype group (p = 0.46, nor in males or females (p = 0.38). Combined treatment, typically administered in the evening, was used for a period of time, in a total of 26 patients (13 males, 13 females), including 4 patients (1 male, 3 females) with late diagnosis. The mean duration of prednisolone treatment was 4.4 years (range: 0.1–13.3 years) and the mean age at the start of treatment was 11.9 years (range: 4.6–17.8 years). The mean dose of prednisolone during treatment was equivalent to a hydrocortisone dose of 2.78 ± 2.5 mg/m² and day.

In the whole cohort, including early and late-diagnosed cases, the mean cFH SDS was −0.78 (95% CI: −1.03 to −0.54). Boys achieved a cFH SDS of −0.81 (95% CI: −1.21 to −0.40) and girls achieved a cFH SDS of −0.77 (95% CI: −1.08 to −0.45), meaning that both sexes differed from the reference data (p < 0.001). The cFH SDS did not differ between the sexes (p = 0.74). Girls reached their FH at an average age of 16.5 years and boys at an average of 17.2 years.

Children with early-diagnosed CAH exhibited impaired growth and weight development during infancy (Fig. 1a, b). In early childhood (Fig. 1c), boys displayed an increased growth rate, and reached a normal height SDS from 4 years of age to 10 years of age. At 11 years of age, boys with CAH were taller than the normal reference population (height SDS 0.57 [95% CI: 0.14–1.00, p = 0.01]). Girls were shorter than the normal population up to 5 years of age. From 6 years of age throughout the childhood growth phase, girls had a height SDS within the normal range. There was an overall reduction in gained height SDS during puberty (Fig. 1d) and FH was reduced compared to the normal population.

The mean height SDS from birth to 18 years of age differed between the clinical severity groups of CAH (p = 0.04). Post hoc testing revealed that children with SW CAH were taller than those with SV CAH from 3 to 7 years of age (p < 0.05). The growth acceleration before puberty seemed to be attributed to the boys with SW CAH.

In early-diagnosed children with classic CAH, SW, and SV forms, where data for calculation were available (n = 66), the mean cFH SDS was −0.86 (95% CI: −1.14 to −0.59), for boys it was −0.83 (95% CI: −1.28 to −0.38), and for girls, −0.89 (95% CI: −1.26 to −0.52) SDS. Girls achieved FH at 16.5 ± 1.3 years of age and boys at 17.2 ± 0.9 years. Age at the achieved FH correlated positively with the cFH SDS (p = 0.01, r² 0.32).

The mean height SDS at the time of diagnosis in late-diagnosed patients was 1.59 ± 1.57 (1.99 ± 1.44 when corrected for TH), i.e., they were taller than the reference population at the corresponding ages (p = 0.003). Late-diagnosed patients reached a mean cFH SDS of −0.49 (95% CI: −1.12 to 0.12), which was lower than the corrected height SDS at the time of diagnosis (p = 0.004). However, the height SDS at diagnosis in late-diagnosed patients did not correlate negatively with their achieved cFH (p = 0.30, r² −0.33). The mean hydrocortisone dose in these patients was slightly above 15 mg HC-equivalents/m² per day after the start of treatment.

It was not possible to obtain information on the exact onset of puberty in the majority of the study subjects, and peripubertal growth was thus defined as growth in girls from 8 years of age and in boys from 9 years of age until the achieved FH. This was used as a proxy for an estimation of pubertal growth. During this age period, boys grew 37.2 ± 7.2 cm and girls 34.1 ± 5.4 cm. Patients with SV CAH grew 39.3 ± 6.6 cm, which was more than patients with SW CAH, 34.3 ± 5.8 cm (p = 0.01). The magnitude of peripubertal growth correlated positively with the cFH for both boys (p = 0.01) and girls (p < 0.001). There were only 4 patients with NC CAH treated from infancy for whom there was adequately recorded peripubertal growth; therefore, this did not allow for statistical analysis.

Genotype group correlated with the cFH SDS and was significant for early-diagnosed patients (p = 0.04, r² 0.27), meaning that children with more severe forms achieved a shorter cFH than those with milder forms (Fig. 2). The correlation was also significant for early- and late-diagnosed patients combined (p = 0.01, r² 0.30).

BMI at 18 years of age was higher in males than in females, 26.1 ± 5.2 versus 23.1 ± 3.8 kg/m² (p = 0.01). In fact, 52% of the males and 25% of the females had a BMI >25 kg/m² at 18 years of age (p = 0.03). There were no significant differences in BMI between the genotype groups at 18 years of age but BMI was higher in patients with an early start of treatment (24.6 ± 4.7 kg/m²) than in patients with a late diagnosis and start of treatment (21.5 ± 2.9 kg/m²) (p = 0.03).

The patients who had been treated with an addition of prednisolone in the evening during some period had a shorter cFH SDS compared to those who had been treated with HC alone, with or without fludrocortisone, −1.1 ± 1.0 SDS compared to −0.60 ± 1.0 (p = 0.047) (Fig. 3a). BMI at 18 years of age was also higher in patients who had been treated with prednisolone (25.3 ± 4.7 kg/m²) than in patients not treated with prednisolone (23.4 ± 4.5 kg/m²) (p = 0.04) (Fig. 3b). Furthermore, the BMI correlated positively with the duration of prednisolone treatment (p = 0.02, r² 0.27), but not with the average HC-equivalent dose per m² per day (p = 0.13, r² 0.33). No other synthetic glucocorticoid was used.

In this population-based prospective observational cohort study, 80 children with CAH were followed from the time of diagnosis to FH. The FH corrected for target height was within 1 SDS below the mean for the whole cohort. The glucocorticoid doses were within the recommended range per m² body surface area. Corrected FH was significantly associated with the severity of the genotype group. The V281L genotype group was the tallest and included patients with late diagnosis (Fig. 2).

Both boys and girls with early-diagnosed CAH exhibited accelerated growth during childhood (Fig. 1c). However, this catch-up later ceased and resulted in an overall reduction in gained height SDS during puberty (Fig. 1d). Despite continued growth, a reduced FH compared to the normal population was ultimately reached. The accelerated early childhood growth, most pronounced in boys, may have contributed to the reduced FH. However, the whole cohort and both boys and girls separately achieved a mean cFH SDS within −1 SDS. Patients treated with an evening dose of prednisolone during some period had a significantly shorter cFH and a higher BMI at 18 years of age than patients who had not received additional treatment with prednisolone (Fig. 3a).

Growth and weight gain during the infancy period was impaired (Fig. 1a, b), which is in accordance with the observations presented by others [13, 14, 22, 34]. Both the reduction in height/length SDS and weight SDS appeared to affect children with severe forms more. This may be attributed to relative salt loss since mineralocorticoid treatment in combination with sodium chloride is essential in the SW form of CAH [9, 14]. Boys with SW form showed a prepubertal catch up growth after the first year and had more prepubertal growth acceleration than the girls which may indicate that the treating physician was less concerned about signs of androgen excess in the boys than in the girls resulting in the observed growth acceleration.

The 13 late-diagnosed patients obtained a mean cFH SDS of −0.49. They all had a period of growth acceleration before diagnosis, and their mean corrected height SDS at diagnosis was almost +2 SDS.

More than half of the patients in this study exhibited a pubertal growth spurt. Furthermore, both girls and boys reached their FH at an average comparable to the normal reference population. The cFH SDS correlated positively with the age of achieved FH. This may indicate that to optimize the FH in patients with CAH, it is important to normalize the longitudinal growth. Keeping glucocorticoid treatment at the lowest possible dose at which excess adrenal steroid production is inhibited during childhood seems to be fundamental. Interestingly, in this study, there was no correlation between the cFH SDS and the average glucocorticoid dose. This may possibly be explained by the fact that most children were treated according to the international recommendations [1, 2, 7] and that this corresponded to their biological need for replacement with infrequent under and overtreatment.

The magnitude of growth during the peripubertal period, defined as growth from 8 years in girls and from 9 years of age in boys, was important for FH. However, the study is limited by the fact that onset of puberty was not recorded uniformly by all clinicians and the time of the onset of puberty therefore cannot be presented properly.

Prednisolone, used in combination with hydrocortisone, was associated with a reduced FH (Fig. 3a), supporting an association between the use of more long acting glucocorticoids and reduced FH [10]. In this study, this appeared to be an independent contributing factor since the total dose of hydrocortisone equivalents did not differ sufficiently to explain the negative FH outcome. The growth-inhibiting effect of prednisolone may be greater than the one calculated from the conversion coefficients used in this study [32, 33]. In most cases, an evening dose of prednisolone was given to reduce the morning levels of 17-OHP, resulting in exposure to glucocorticoids during night-time when normal cortisol production is low. Furthermore, the use of prednisolone was associated with a higher mean BMI (Fig. 3b). In addition, prednisolone may also negatively affect bone health and glucose metabolism. Issues with compliance cannot be excluded as a contributing factor, although it is unlikely that compliance issues were more common in the one center using prednisolone.

Interestingly, the duration of treatment with prednisolone rather than the average dose correlated with BMI at 18 years of age. Our results indicate that the dose required to cause iatrogenic overweight is low and that prednisolone should be avoided in growing subjects with CAH. It is likely that the relatively higher glucocorticoid effect during the night in these cases, when cortisol production normally is at its lowest, is contributing to the effect on both growth and weight. The inhibition of the pituitary and the subsequently lower 17OHP in the morning may at first seem to be a positive effect; however, prednisolone per se and a low androgen production may have negative effects on growth, bone density, and metabolism such as increased insulin resistance in the longer perspective. This raises questions concerning the long-term effects of the newly developed and future medications aiming at inhibiting androgen production to enable lower glucocorticoid doses [26‒28]. Long-term follow-up is of great importance and there is a clear need for better biomarkers for follow up, both to improve treatment per se and to be able to evaluate the effects of newly developed treatments in the future.

Different treatment strategies should be used during the different growth phases to optimize growth and FH. Infancy and puberty seem to be the most vulnerable growth periods in CAH [13, 14, 22, 23, 34, 35]. Close monitoring to avoid over and undertreatment during childhood and adolescence is of utmost importance not only because both scenarios affect FH negatively. For all children with CAH, it is recommended to follow patients at 3–4 month intervals after 3 years of age and more frequently in the years before that [1, 7]. There is a lack of optimal laboratory parameters for follow up, for evaluation of treatment, and for tailoring the doses, while the observed growth velocity is the effect of the ongoing treatment. The 11 oxygenated androgens are promising but not clinically available [36]. Not all studies have reported an adverse effect from using more potent glucocorticoids in CAH [37, 38]. Optimal hormonal balance can be achieved with different treatment strategies but careful dosing and follow up is warranted. Not all centers have the possibility to perform genotyping for CYP21A2, but since there is a strong genotype and phenotype correlation, genotyping can contribute useful clinical information and guide management.

The strength of this study is that it is population-based and no subjects declined to participate. Thus, it is not biased in terms of patient selection. The patients were genotyped to determine the severity of CAH, and for comparison between the severity groups. The Swedish growth charts were used to control for normal growth. The data collection was detailed and every change in treatment was recorded. This study has weaknesses inherent to the study design. The study is purely observational, which makes conclusions concerning causality difficult to draw. It spans over an almost 30-year-long period and the patients included in this national study live in all parts of Sweden. They have been seen by the local physicians (in some cases, replaced due to retirement) and the treatment decisions have been made locally. Hormone assessments and bone age were not uniformly collected and therefore not included. Moreover, since a number of clinicians managed the children, subjective decisions concerning both diagnostics, such as the onset of puberty and decisions on changes in treatment, are inevitable, but nevertheless, weaknesses of the study. Data on compliance were not collected.

In conclusion, both boys and girls with CAH achieved a FH within 1 SDS of their TH. The study clearly showed that growth in CAH may continue for a longer period than could be expected from previous reports on pubertal growth [20] and that optimal peripubertal growth is an important determining factor for increasing the FH. In addition, we provide further evidence that prednisolone should not be used in growing subjects with CAH; even a relatively low dose as a supplement in the evening may have negative effects. Mineralocorticoid and sodium chloride treatment seem to be especially important during the early years. Close monitoring to normalize growth velocity through all growth phases is essential to optimize FH.

We are especially grateful to Astrid Thilén, who has passed away but initiated the study; she put energy and enthusiasm into the project. She was truly an inspiring and enthusiastic clinician and researcher. This study would not have been possible to conduct without the help of local pediatricians throughout Sweden who provided study data continuously. We also thank David Olsson who contributed to the data management and collection in a substantial way. We are especially grateful to all the patients and families that agreed to participate in the study over so many years. The authors wish to thank every contributing pediatrician and especially the following pediatricians (in alphabetical order of hospital): Dr. Ulla Iversen, Ängelholm Hospital; Dr. Anna Olivecrona and Dr. Anders Andersson, Falun Hospital; Dr. Åke Stenberg, Gällivare Hospital; Dr. Gunilla Kördel, Gävle Hospital; Dr. Stefan Aronsson and Dr. Nils-Östen Nilsson, Halmstad Hospital; Dr. Ulf Westgren, Helsingborg Hospital; Dr. Jonas Falås, Kalmar Hospital; Dr. Gudrun Jonsell and Dr. André Bachtiar, Karlstad Hospital; Dr. Jan Alm, Karolinska University Hospital and Sollefteå Hospital; Dr. Karin Larsson, Kristianstad Hospital; Dr. Björn Kornerup, Lidköping Hospital; Dr. Karin Segnestam, Mälarsjukhuset Hospital; Dr. Jan Åman, Örebro University Hospital; Dr. Anna-Lena Nilsson, Östersund Hospital; Dr. Annika Reims and Dr. Otto Westphal, Queen Silvia’s Children’s Hospital; Dr. Maria Elfving and Dr. Johan Svensson, Skåne University Hospital; Dr. Henrik Tollig and Dr. Kerstin Rex, Skövde Hospital; Dr. Fredrik Lindgren, Dr. Björn Rathsman, and Dr. Birger Winbladh, Stockholm Söder Hospital; Dr. Björn Stiernstedt, Sundsvall Hospital; Dr. Ragnar Hanås, Uddevalla Hospital; Dr. Berit Kriström, Umeå University Hospital; Dr. Jan Gustafsson and Dr. Maria Halldin Stenlid, Uppsala University Hospital; Dr. Hans Fors and Dr. Leif Inganäs, Vänersborg Hospital; Dr. Bengt Nylund, Västervik Hospital; Dr. Eva Karlsson, Växjö Hospital; and Dr. Bo Klintberg, Visby Hospital.

This study was approved by the Committee on Ethics in Research at Karolinska Institutet (dnr: 95:137) and Uppsala University (dnr: 89,136). According to the ethic approval by the two abovementioned ethic committees, written informed consent was not required, in accordance with the national guidelines and regulations at that time in Sweden and in accordance with the World Medical Association Declaration of Helsinki. A physician in the research group met all the patients and their parents to obtain informed consent from the parents; this was documented for each patient family.

The authors have no conflicts of interest.

This work was supported by the Swedish Research Council, the Center for Gender Medicine at Karolinska Institutet, the Stockholm County Council, the Novo Nordisk Foundation, the Sven Jerring Foundation, and Region Stockholm (clinical research appointment DNR RS2019-1140 to S.L.).

Sebastian Gidlöf: conception or design of the work; or the acquisition, analysis, or interpretation of data, drafting the work, and revising it critically for important intellectual content, final approval of the version to be published, agreement to be accountable for all aspects of the work. Daniel Eriksson Hogling: acquisition, analysis of data, revising the manuscript critically for important intellectual content, final approval of the version to be published, and agreeing to be accountable for all aspects of the work. Hanna Lönnberg: analysis of data, revising the manuscript critically, final approval of the version to be published, agree to be accountable for all aspects of the work. Martin Ritzén: conception and design of the work, interpretation of data, critically revising the manuscript for important intellectual content, final approval of the version to be published, and agreeing to be accountable for all aspects of the work. Svetlana Lajic: interpretation of data, revising the manuscript critically for important intellectual content, final approval of the version to be published, and agreeing to be accountable for all aspects of the work. Anna Nordenström: conception and design of the work, the acquisition, analysis, and interpretation of data, drafting the work and revising it critically for important intellectual content, final approval of the version to be published, and agreeing to be accountable for all aspects of the work.

Data are not publicly available due to ethical reasons. Further inquiries can be directed to the corresponding author.