Joint Modelling of Growth and Motor Function Centiles in Corticosteroids Treated Boys With Duchenne Muscular Dystrophy

ABSTRACT

Background

Corticosteroid (CS) treated boys with DMD display higher rates of height stunting, higher weight gain, improved motor function scores and delayed loss of ambulation compared to untreated patients. However, the relationship between growth and motor function has historically been understudied due to modelling complexities.

Methods

In this analysis, we use the newly developed motor function centiles for the NSAA, RFF and 10MWR. We consider each combination of growth (height and weight SD) and motor function using multivariate regression models controlling for differential CS treatment (prednisolone/deflazacort, daily/intermittent). This allows inference on the growth and motor function outcomes separately and on the relationship between the outcomes.

Results

We consider 559 steroid‐treated boys with DMD between the ages of 5 and 16 over 1643 assessments. Better motor function trajectories were observed in those treated with daily CS, with the deflazacort daily group displaying a positive NSAA centile trajectory (annual change of 0.07 SD). There was a mild, negative pairwise correlation between the annual changes in NSAA and 10MWR Z‐Scores, and height and weight Z‐Scores, ranging from −0.25 to −0.36. This indicated that patients with a milder weight gain or more severe height stunting trajectory with respect to their CS treatment were more likely to exhibit a more favourable NSAA or 10MWR trajectory over time.

Conclusions

This work describes the complex relationships between motor function, CS treatment and growth and provides insights for conversations about the relative benefits and negative effects of CS.

In this paper, we used data from 559 patients in the North Star Database to analyse the relationship between motor function, growth and corticosteroids (CS) groups (deflazacort vs prednisolone; daily vs intermittent) in Duchenne muscular dystrophy (DMD). We used the newly developed NSAA, 10MWR and RFF centiles to model motor function in this cohort and looked at differential trajectories by CS group. We also describe for the first time the correlation between these newly developed motor function centiles and the growth (height and weight) trajectories with respect to the different CS treatment groups.

Abbreviations

10MWR

10 m Walk Run

CI

Confidence Interval

CS

Corticosteroids

DMD

Duchenne Muscular Dystrophy

IQR

Interquartile Range

NSAA

North Star Ambulatory Assessment

RFF

Rise from Floor

SD

Standard Deviation

SoC

Standard of Care

1. Introduction

Duchenne muscular dystrophy (DMD) is an X‐linked recessive disease, characterised by muscle wasting, motor function decline and premature mortality. Typically, boys with DMD experience a peak in motor function abilities around the age of 7 years [1], after which they experience a continuous decline in motor function abilities, leading to a loss of ambulation in untreated boys by the age of 12 years [2, 3].

Corticosteroids (CS) form the primary treatment approach for boys with DMD and are the only recommended treatment for all patients in the standard of care (SoC) guidance [4, 5, 6], although novel therapies are at different stages of regulatory approval. Historically, two types of CS have been most prescribed, deflazacort and prednisone (or its active metabolite prednisolone), and in the UK, two treatment regimes are prevalent: daily and intermittent (10‐days‐on, 10‐days‐off), although the prescription of the CS type and regime has changed over the last 10 years and other regimes are popular in other geographical regions [7, 8, 9]. In the placebo arms of trials, CS have been shown to delay key motor function endpoints in both ambulatory and non‐ambulatory patients [9, 10], although differences in efficacy by CS regime were observed in the FOR‐DMD trial of CS in DMD [11].

Despite length and weight being in line with non‐DMD peers at birth, boys with DMD exhibit significantly different patterns of growth compared to their non‐DMD peers [12], as they are on average, shorter than their age‐matched peers within the first few years of life [13, 14, 15], even without CS treatment, and are on average heavier than their age‐matched peers by age 7–10 years [16]. After starting CS, documented growth‐related side effects like short stature and overweight are common in boys with DMD [17].

Several studies have reported specific patterns of growth changes related to the type and regime of CS used. Boys on deflazacort tend to be shorter than those on prednisolone [11, 13, 18, 19, 20], and, similarly, boys on daily regimes are consistently shorter than those on intermittent regimes [11, 13, 21]. Additionally, CS‐driven weight gain has been widely reported [10, 22], with some studies reporting a lower weight gain in boys treated with deflazacort compared to prednisolone [13, 18, 19] and, to some extent, a tendency to show higher weight gain with daily regimes [23].

The North Star Ambulatory Assessment (NSAA) is a 17‐item scale developed for ambulant boys with DMD [9]. NSAA trajectories typically follow a quadratic shape, with average peak NSAA scores observed in patients between the ages of 6 and 6.5 years [1]. The NSAA contains two timed items, the rise from floor (RFF) and the 10m walk/run (10MWR). These timed function tests capture differential trajectories, with the RFF in particular showing a decline at an earlier age, allowing trialists to capture functional change in patients who have reached the ceiling on the NSAA. To understand disease heterogeneity, we have recently developed centiles for the NSAA, RFF and 10MWR [1]. Similar to the centiles for weight and height, these centiles allow the NSAA, RFF and 10MWR to be modelled linearly with age.

The relationship between growth and motor function has been understudied in boys with DMD; however, a recent analysis of the interaction between growth and loss of ambulation in the NorthStar cohort of boys with DMD demonstrated that excess weight gain (i.e., drifting from one centile to the next one) had a significant, negative, impact on the risk of loss of ambulation. Conversely, excess height stunting had a positive impact on the risk of loss of ambulation [24].

1.1. Aim

With this work, we aim to analyse the relationship between growth for age and motor function for age (using the newly developed NSAA, RFF and 10MWR centiles) in CS‐treated boys with DMD.

2. Methods

The NorthStar database captures observational longitudinal data on boys with DMD seen at one of 23 paediatric neuromuscular centres in the UK. Boys attend clinics every 6 months on average, where medical and physiotherapy variables are collected, and care is in line with the standard of care guidance [4, 5, 6]. The NorthStar project is a natural history clinical audit of the National Neuromuscular Database, in which data from DMD boys have been collected since 2006. Caldicott guardian approval has been obtained at every site in the NorthStar network and the NorthStar network has Caldicott Guardian approval (protocol ID: 12DN14).

2.1. Participants

Participants recruited into the NorthStar database before July 2022 were considered for this analysis. Only boys with diagnosed DMD, who were not enrolled in trials or ataluren, and who had a known ambulatory status were included. Visits were only included after the patient‐initiated GC, and between the ages of 5 and 16 years. Only complete NSAA assessments that were non‐zero, and RFF/10MWR times greater than 0 were included. The cohort utilised in this analysis is derived from the same data cut used to develop the NSAA, RFF and 10MWR centiles for age [1], and the additional data cleaning steps are shown in Figure 1.

FIGURE 1.

Data cleaning process.

2.2. Variables

Boys were only included in the cohort if they had initiated CS. Boys were assigned to one of four CS groups: deflazacort daily, deflazacort intermittent, prednisolone daily and prednisolone intermittent based on the type and regime that they spent the most time on during the observation period, with boys with an unknown CS type or regime removed [21, 24].

Observations were only included if height, weight and the NSAA were all performed and recorded. Height was measured using standing height, as this cohort was all ambulant. Cleaning of the height and weight variables was performed: for heights over 500 cm, and under 20 cm, they were scaled by a factor of 0.1 and 10 respectively, as it was assumed they had been recorded in the wrong unit. Weights under 12 kg and over 100 kg were assumed to be errors and were removed. The Z‐Scores for both weight and height were calculated using the UK90 reference values [25]. We did not include BMI in this analysis due to difficulties in the interpretation of this metric in boys with DMD.

Observations were only included if all items of the NSAA were recorded, and there were valid RFF and 10MWR times, or the corresponding items were 0 [1]. The Z‐Scores for the NSAA, were calculated using the Stimpson et al., 2024 [1] reference values, whilst the Z‐Scores for the RFF and 10MWR were calculated for their respective velocities and then reciprocated to calculate the Z‐Scores for the RFF and 10MWR times.

2.3. Statistics

The analysis of the correlation between the motor function (NSAA, RFF and 10MWR) and growth (height and weight) Z‐Scores was performed using a bivariate longitudinal model with shared random effects structure, where the person‐specific effects for intercept and slope (denoted by $u_{\_}$) follow a multivariate Normal distribution. The linear form for growth is supported by previous modelling of the growth trajectories by CS group in this cohort [13]. Model parameters are presented with 95% confidence intervals (CI).

$$\begin{array}{cc}\mathit{\text{Growth}}Z-\mathit{\text{Score}}\left(\mathit{age}\right)\hfill & ={\mathit{\beta }}_1*\mathit{CS}\mathit{\text{status}}+{\mathit{\beta }}_2*\mathit{CS}\mathit{\text{status}}*\left(\mathit{age}-5\right)\hfill \\ & +u_{0i}+u_{1i}\mathit{age}+{ϵ}_{1i}\left(\mathit{age}\right)\hfill \end{array}$$

$$\begin{array}{cc}\mathit{\text{Motor Function}}Z-\mathit{\text{Score}}\left(\mathit{age}\right)\hfill & ={\mathit{\beta }}_3*\mathit{CS}\mathit{\text{status}}+{\mathit{\beta }}_4*\mathit{CS}\mathit{\text{status}}*\left(\mathit{age}-5\right)\hfill \\ & +u_{2i}+u_{3i}\mathit{age}+{ϵ}_{2i}\left(\mathit{age}\right)\hfill \end{array}$$

Here, ${\mathit{\beta }}_1$ reflects the mean growth Z‐Score at age 5 years for a given CS cohort, ${\mathit{\beta }}_2$ reflects the annual change in growth Z‐Score for a given CS cohort, reflects the mean motor function Z‐Score at age 5 years for a given CS cohort and ${\mathit{\beta }}_4$ reflects the annual change in motor function Z‐Score for a given CS cohort. The marginal correlation as a function of age was calculated for this model [26]. Where shown, patient trajectories were shown by selecting the patients on prednisolone daily CS (the largest subgroup) with the 10th, 30th, 50th, 70th and 90th quantile random slope for the NSAA Z score.

3. Results

In total, 559 participants with DMD were assessed across 1643 assessments (median 3 per patient). The largest CS subgroup was the prednisolone daily group (266 participants, 47.6%), with 27.4% in the prednisolone intermittent subgroup (N = 153), 17.4% in the deflazacort daily subgroup (N = 97) and 7.7% in the deflazacort intermittent subgroup (N = 43). An overview of the participant growth and motor function outcomes at the first visit is given in Table 1. Notably, the positive bias in the NSAA, RFF and 10MWR centiles (i.e., the median is greater than 50) is due to the exclusion of 0 scores in this analysis.

TABLE 1.

Summary of motor function and growth scores at first In‐study visit.

Function	NSAA	RFF	10MWR
M (number of assessments)	1643	1261	1602
N (number of patients)	559	339	526
Score [median (IQR)]	17 (10, 26)	5 (3.4, 7.2)	7.2 (5.2, 10.1)
Centile [median (IQR)]	54.4 (30.1, 79.7)	68 (45.8, 88.9)	58.5 (36.4, 81.9)
Z‐score [median (IQR)]	0.1 (−0.5, 0.8)	0.5 (−0.1, 1.2)	0.2 (−0.3, 0.9)

Growth	Weight	Height	BMI
M (number of assessments)	1643	1643	1643
N (number of patients)	559	559	559
Score [median (IQR)]	32.3 (25.4, 43)	126 (119, 134.5)	20.1 (17.2, 24.6)
Centile [median (IQR)]	72.2 (32, 92.6)	10.9 (1.6, 31.5)	95.6 (76, 99.5)
Z‐Score [median (IQR)]	0.6 (−0.5, 1.4)	−1.2 (−2.1, −0.5)	1.7 (0.7, 2.6)

3.1. Univariate Growth

Overall, the growth submodels were very similar to previously published work in this cohort [13], and as such, the growth results are reported fully in Supporting Section 6.1. Notably, the annual weight gain in the prednisolone‐treated cohorts was significantly greater than 0, whilst in the deflazacort‐treated cohort, this was not significantly different from zero. Significant height stunting over time was observed in the daily‐treated cohorts, but not in the intermittent‐treated cohort.

3.2. NSAA

The full analysis of the NSAA Z‐Scores with respect to the CS group is given in Supporting Section 6.2. Notably, the deflazacort daily cohort was the only CS subgroup that displayed a significant yearly increase in NSAA Z‐Score, with a value of 0.07 Standard Deviations (SD; 0.07 and 0.07 SD respectively), equivalent to moving from the 50th to the 53^rd centile in a year (Table 2A).

TABLE 2.

(A) Model coefficients for the multivariate modelling of growth Z‐Score and NSAA Z‐Score by CS group (Mean (95% CI)). Coefficients significantly different from 0 are bolded. (B): Pairwise correlations between the personal (random) intercept and slope of the NSAA and growth models.

Variable	CS group	Weight	Height
(A)
NSAA sZ‐Score at age 5 estimate (95% CI)	Prednisolone daily	0.204 (0.049, 0.36)	0.21 (0.052, 0.368)
	Prednisolone intermittent	0.038 (−0.154, 0.23)	0.026 (−0.169, 0.221)
	Deflazacort daily	−0.073 (−0.317, 0.17)	−0.077 (−0.324, 0.171)
	Deflazacort intermittent	−0.164 (−0.562, 0.234)	−0.143 (−0.546, 0.26)
Annual change in NSAA Z‐score estimate (95% CI)	Prednisolone daily	0.022 (−0.008, 0.052)	0.02 (−0.011, 0.05)
	Prednisolone intermittent	−0.026 (−0.068, 0.015)	−0.023 (−0.065, 0.02)
	Deflazacort daily	0.071 (0.022, 0.119)	0.074 (0.024, 0.124)
	Deflazacort intermittent	0.068 (−0.023, 0.159)	0.062 (−0.031, 0.155)
(B) Correlations
Growth Z‐Score and NSAA Z‐Score		0.043	0.08
Growth Z‐Score and Annual Change in Growth Z‐Score		−0.379	−0.377
NSAA Z‐Score and Annual Change in NSAA Z‐Score		−0.598	−0.623
Annual Change in Growth Z‐Score and Annual Change in NSAA Z‐Score		−0.255	−0.282

We visualise motor function trajectories for five patients on prednisolone daily CS (the largest subgroup) with the 10th, 30th, 50th, 70th and 90th percentile slope for the NSAA Z score in Figure 2A,D. The observed growth trajectories are shown for these patients in Figure 2B,C, with the corresponding observed Z‐Score trajectories for the NSAA and growth outcomes, and the fitted trajectories from the model are shown in Figure 2D–F. Here, Patient 5 has the worst annual change in NSAA Z‐Score, whilst Patient 2 displays the highest annual change in NSAA Z‐Score.

FIGURE 2.

Example growth and NSAA trajectories and comparative Z‐Score trajectories for 5 patients treated with prednisolone daily, who display varying NSAA trajectories (slopes of the 10th, 30th, 50th, 70th and 90th centiles).

We observed a negative correlation between the person‐specific change in the NSAA Z‐Score and the height Z‐Score of −0.282 (see Table 2B). When considered with age, the marginal correlation increased over time, rising to −0.45 at age 10 years (see Figure 3). Similarly, there was a negative correlation between the annual change in weight Z‐Score and the NSAA Z‐Score of −0.255, and the marginal correlation increases with age to −0.46 at age 10 years. Consequently, those who have increases in NSAA Z‐Score trajectory are more likely to have more severe height stunting and lower weight gain trajectories over time. This is shown by Patient 2, who has the steepest NSAA Z‐Score trajectory over time (i.e., a milder decline than expected on the NSAA), and the most severe height stunting trajectory in Figure 2D,E. Conversely, those with steeper declines in their NSAA trajectories (more decline than expected with age) are more likely to display increases in height Z‐Score and weight Z‐Score over time. This is shown in Figure 2D–F, where Patient 5 with the steepest annual decline in the NSAA Z‐Score has the steepest annual increase in both height and weight Z‐Score.

FIGURE 3.

Marginal correlations of motor function and growth with age.

3.3. Rise From Floor

The full analysis of the RFF Z‐Scores with respect to the CS group is given in Supporting Section 6.2. Most notably, the annual change in RFF Z‐score was significantly greater than 0 in all four CS cohorts, with the deflazacort daily subgroup displaying the largest slope of 0.14 (0.14, 0.14 respectively), indicative of a participant moving from the 50th to the 56th centile in 1 year (see Table 3A).

TABLE 3.

(A) Model coefficients for the multivariate modelling of growth Z‐Score and RFF Z‐Score by CS group (Mean (95% CI)). Coefficients significantly different from 0 are bolded. (B): Pairwise correlations between the personal (random) intercept and slope of the RFF and growth models.

Variable	CS group	Weight	Height
(A)
RFF Z‐Score at age 5 ESTIMATE (95% CI)	Prednisolone daily	0.156 (0.005, 0.307)	0.155 (0.003, 0.307)
	Prednisolone intermittent	−0.084 (−0.276, 0.107)	−0.092 (−0.285, 0.1)
	Deflazacort daily	−0.051 (−0.289, 0.187)	−0.053 (−0.292, 0.187)
	Deflazacort intermittent	−0.134 (−0.531, 0.264)	−0.118 (−0.517, 0.28)
Annual change in RFF Z‐Score estimate (95% CI)	Prednisolone daily	0.098 (0.065, 0.132)	0.097 (0.064, 0.131)
	Prednisolone intermittent	0.075 (0.025, 0.125)	0.075 (0.026, 0.125)
	Deflazacort daily	0.141 (0.084, 0.198)	0.143 (0.086, 0.2)
	Deflazacort intermittent	0.139 (0.025, 0.252)	0.133 (0.02, 0.246)
(B) Correlations
Growth Z‐Score and RFF Z‐Score		−0.033	−0.068
Growth Z‐Score and Annual Change in Growth Z‐Score		−0.218	−0.261
RFF Z‐Score and Annual Change in RFF Z‐Score		−0.732	−0.738
Annual Change in Growth Z‐Score and Annual Change in RFF Z‐Score		−0.049	−0.289

There was a negative correlation between the person‐specific change in the RFF Z‐Score and the change in the height Z‐Score of −0.289 (see Table 3B), although a similar correlation was not observed between the annual change in the RFF Z‐Score and the change in weight Z‐Score. The marginal correlation between the RFF Z‐Score trajectory and height Z‐Score increased initially with age and then stabilised. In contrast, the marginal correlation between the RFF Z‐Score trajectory and weight was much lower and generally trended towards zero over time (see Figure 3).

3.4. 10m Walk Run

The full analysis of the 10MWR Z‐Scores with respect to the CS group is given in Supporting Section 6.2. Notably, the annual change in the 10MWR Z‐Score in the daily‐treated cohorts was significantly greater than 0, but not in the intermittent‐treated cohorts (see Table 4A). We observed a negative correlation between the person‐specific change in the 10MWR Z‐Score and the change in both the weight Z‐Score (−0.267) and the height Z‐Score (−0.363, see Table 4B). When considered with age, the marginal correlation between the trajectories increased over time, rising to −0.45 between the 10MWR Z‐Score and height Z‐Score trajectory, although the marginal correlation between the 10MWR Z‐Score and weight Z‐Score was lower (see Figure 3).

TABLE 4.

(A) Model coefficients for the multivariate modelling of growth Z‐Score and 10MWR Z‐Score by CS group (Mean (95% CI)). Coefficients significantly different from 0 are bolded. (B) Pairwise correlations between the personal (random) intercept and slope of the 10MWR and growth submodels.

Variable	CS group	Weight	Height
(A)
10MWR Z‐Score at age 5 estimate (95% CI)	Prednisolone daily	−0.072 (−0.241, 0.096)	−0.064 (−0.234, 0.105)
	Prednisolone intermittent	−0.011 (−0.22, 0.199)	−0.027 (−0.237, 0.183)
	Deflazacort daily	−0.187 (−0.455, 0.081)	−0.186 (−0.455, 0.084)
	Deflazacort intermittent	0.02 (−0.409, 0.449)	0.04 (−0.391, 0.47)
Annual change in 10MWR Z‐Score estimate (95% CI)	Prednisolone daily	0.084 (0.05, 0.118)	0.081 (0.046, 0.115)
	Prednisolone intermittent	0.026 (−0.022, 0.075)	0.03 (−0.018, 0.079)
	Deflazacort daily	0.12 (0.063, 0.176)	0.12 (0.063, 0.178)
	Deflazacort intermittent	0.059 (−0.043, 0.16)	0.054 (−0.048, 0.156)
(B) Correlations
Growth Z‐Score and 10MWR Z‐Score		−0.111	0.012
Growth Z‐Score and annual change in growth Z‐Score		−0.367	−0.393
10MWR Z‐Score and annual change in 10MWR Z‐Score		−0.761	−0.761
Annual change in growth Z‐Score and annual change in 10MWR Z‐Score		−0.267	−0.363

4. Discussion

In this study, we describe the co‐occurring trajectories of growth and the NSAA score, including its timed function tests and how growth changes may impact the motor abilities of DMD boys on CS over time. This work builds upon previous modelling of growth with respect to the loss of ambulation risk [24], but uses the UK90 centiles [25] and the newly developed DMD‐specific motor function centiles [1] to describe the trajectories using a bivariate linear model.

Our results confirm previous studies that the prednisolone‐treated groups showed higher weight gain than the deflazacort‐treated groups, with the prednisolone intermittent subgroup having the highest weight gain overall [13, 18, 19]. When height was considered in the bivariate models, the daily‐treated groups showed more severe height stunting over time compared to the participants on the intermittent regime, which was also in line with previous results [9, 13, 21, 27].

In the NSAA submodels, a higher NSAA Z‐Score was observed in those on daily CS, with the deflazacort daily subgroup showing a significant annual improvement in the NSAA Z‐Score of 0.07. In the RFF submodels, a similar trend was observed, with the deflazacort daily group displaying the largest annual change of 0.14 SD, but all of the groups tracking above the 50th centile. This is likely caused by the restriction to only participants who were able to score on the motor function item (i.e., participants with a RFFV > 0), and this impacts the RFF model more as patients lose the ability to perform the RFF at younger ages than the 10MWR and the NSAA [1]. In the 10MWR submodels, significant annual improvements were observed in the daily‐treated subgroups. However, the validity of these findings is limited, as this analysis was performed in the same cohort for which the centiles were calculated. Analysis of the impact of CS on the motor function Z‐Scores should be validated in an external cohort.

Most interestingly, there was a mild, negative correlation between the yearly change in NSAA Z‐Score and the 10MWR Z‐Score and the yearly change in height Z‐Score and weight Z‐Score. These correlations, ranging from −0.25 to −0.36 indicated that patients with a more stable weight trajectory or more severe height stunting trajectory were more likely to exhibit a less severe decline in their NSAA and 10MWR trajectories, when the CS group was adjusted for. Additionally, there was a negative correlation between the annual change in RFF Z‐Score and the annual change in the height Z‐Score, suggesting that those with a higher rate of height stunting with age were more likely to have a milder decline on the RFF. This is in line with findings from the FOR‐DMD study, where negative correlations between change in height/weight and change in NSAA, RFF, 10MWR and 6 Minute Walking Distance (6MWD) were observed over the 3‐year trial period [28]. Notably, the functional scores were not adjusted for trend in that study, which is in contrast with the centile methodology used in our study. In the FOR‐DMD study, the younger cohort (4–5 years) displayed an increasing/stable trend in their motor function trajectories, while the older cohort (5–6 years) showed more of a decline. Consequently, the FOR‐DMD result could have been attributed to the trend of the functional scores. In our analysis, the use of the Z‐Scores allows us to identify trends despite the non‐linear trajectories.

When the marginal correlation between the growth Z‐Score trajectories and the motor function Z‐Scores, the height trajectories and the RFF Z‐Score trajectories, and the height trajectories and the 10MWR Z‐Score trajectories was considered with respect to age (as seen in Figure 3), there was a trend of increasing correlation magnitude with time, with relative stability of above −0.42 for both the weight and height models. The observed effect of growth on motor function may be a function of mechanics, with shorter DMD boys more able to stand and move around, and this is supported by the stronger marginal correlations with age between height and the 10MWR/RFF compared to weight and 10MWR/RFF.

One major limitation of this methodology is the lack of causality, and as such, it is not possible to quantify the extent to which excess growth causes motor function decline. In particular, in the multivariate modelling of weight and NSAA, it is possible that worsening function (indicated by lower NSAA Z‐Scores) could cause lower levels of activity and exercise, which in turn lead to an increase in weight Z‐Score, and as such, reduced NSAA could cause increased weight. It is also possible that, instead, a differential biological mechanism between prednisolone and deflazacort could be jointly responsible for reduced growth and increased motor function, such as a CS related anabolic effect. Additionally, this analysis generalised CS type and regime, so patients are assigned to the regime they spend the most time on between CS initiation and loss of ambulation. Consequently, as patients predominantly switch from prednisolone to deflazacort, and from intermittent to daily [29], these results may be biased by this switching. Additionally, due to missing data, it was not possible to model differences in CS dose (mg/kg) [29], which might also lead to differential growth‐related side effects.

Another crucial limitation of these analyses is that in each, it is only possible to account for one longitudinal outcome at a time, either height or weight. As there is a correlation between height and weight, adjusting for only one may omit a key part of the relationship. Whilst BMI theoretically accounts for both, it is much more difficult to interpret for DMD patients, as an increase in BMI Z‐Score could be due to a reduction in height Z‐Score or due to an increase in weight Z‐Score, both of which have been observed in the DMD population.

5. Conclusion

The findings of this study, alongside previous works understanding growth and function in DMD [13, 24, 27, 28], can support conversations with patients and their carers about the relative benefit of CS types and regimes, and the associated side effects. This is particularly pertinent given the recent approval of Vamorolone, a new generation of CS with differential side effect profiles [30, 31]. Although shorter height is correlated with slower motor function decline, it is also linked to low self‐esteem and social difficulties in DMD [32]. Consequently, individualised management of growth side effects remains a priority in CS‐treated boys with DMD and any benefit due to short stature must be weighed against quality‐of‐life considerations.

Author Contributions

G.S.; conceptualisation, data curation, formal analysis, methodology, visualisation, writing; original draft, writing; review and editing. D.R.; conceptualisation, formal analysis, methodology, supervision, writing; review and editing. A.W.; conceptualisation, data curation, writing; review and editing. E.M.; conceptualisation, writing; review and editing. EOR; conceptualisation, writing; review and editing. A.M.; funding acquisition, writing; review and editing. A.S.; writing; review and editing. F.M.; supervision, funding acquisition, writing; review and editing. G.B., conceptualistion, supervision, writing; review and editing.

Ethics Statement

The NorthStar project is a natural history clinical audit of the National Neuromuscular Database, in which data from DMD boys have been collected since 2006. Caldicott Guardian approval has been obtained at every site in the NorthStar network. Ethical approval, in order for the NorthStar network to be established as a research project, is ongoing (IRAS Project ID: 242567).

Conflicts of Interest

F.M. reports participation in Scientific Advisory boards and teaching initiatives for Novartis, Biogen, Roche; he is involved as an investigator in clinical trials from Novartis, Biogen, and Roche. Both institutions (UCL and GOSH) receive funding from Biogen and Roche for the SMAREACH SMA registry. G.B. is the PI of clinical trials by Pfizer, NS Pharma, and Reveragen, and has received speaker and/or consulting fees from Sarepta, PTC Therapeutics, Biogen, Novartis Gene Therapies Inc. (AveXis), and Roche and has worked as principal investigator of SMA studies sponsored by Novartis Gene Therapies Inc., and Roche. A.W. has received speaker or consulting fees from Biogen and Novartis Gene Therapies, and is involved as an investigator in clinical trials from Novartis, Biogen, and Roche. EOR reports participation in an advisory board for Roche. E.M. provides consultancy services through ATOM International Ltd. for the following companies: Amicus Therapeutics Pty Ltd., Genethon, Italfarmaco, NS Pharma, Pfizer, Edgewise Therapeutics, Sarepta, Dyne Therapeutics. G.S., D.R., A.M. and A.S. have no conflicts of interest to declare.

Supporting information

Appendix S1: Supporting Information.

Stimpson G., Ridout D., Wolfe A., et al., “Joint Modelling of Growth and Motor Function Centiles in Corticosteroids Treated Boys With Duchenne Muscular Dystrophy,” European Journal of Neurology 32, no. 11 (2025): e70414, 10.1111/ene.70414.

Data Availability Statement

The data used in this analysis are available on receipt of reasonable request to Francesco Muntoni and Adnan Manzur, and the subsequent approval from the NorthStar Steering Committee.