It is important to understand what variables influence change in predicted adult height (PAH) throughout GnRHa treatment for central precocious puberty (CPP) to individualize treatment decisions and optimize care.
Changes in PAH, chronological age (CA), bone age (BA), BA/CA, and height velocity (HV) were evaluated in girls with CPP throughout treatment with leuprolide acetate (n=77). A second analysis focused on changes in the 3 years preceding the first observed BA of ≥12 years. Relationships were characterized using plot inspection and linear mixed-effects analyses. Association between treatment duration and last assessed PAH was examined using multiple linear regression models.
BA/CA and HV showed a nonlinear change during treatment, with the largest changes and improvement in PAH observed in the first 6–18 months. Rate of BA advancement tended to decrease more slowly in girls initiating treatment at a younger BA. On-treatment change in PAH was predicted by concurrent BA/CA change, HV, and BA, as well as CA at treatment initiation. Last assessed PAH was positively associated with longer treatment durations (primary/exploratory models cut-offs of ≥33/≥55 months).
These findings support individualized monitoring during GnRHa treatment. Initial response should be interpreted with caution until 6–18 months after treatment initiation and failure should not be assumed based on continued bone maturation in girls starting therapy at a younger age. Treatment cessation should not be automatically based on a diminishing change in PAH or HV, as ongoing treatment may result in continued increase or maintenance of PAH.
Gonadotropin-releasing hormone agonist (GnRHa) therapy is the standard of care for treatment of central precocious puberty (CPP). One goal of GnRHa therapy is to slow the rate of bone maturation while maintaining an adequate height velocity (HV) in order to maximize the impact on adult height (AH) [
This study characterized changes in PAH during the course of GnRHa therapy and the relationship to concurrent changes in CA, BA, BA/CA ratio, and HV. The goal was to extend our previous work to inform individualized treatment decisions [
We evaluated GnRHa-naïve and non-naïve girls with CPP who received leuprolide acetate (LUPRON DEPOT-PED®, AbbVie, North Chicago, IL) treatment for at least 24 weeks in either of two Phase 3 open-label multicenter clinical trials and who had at least one post-baseline BA measurement (Study 1, NCT00660010 [
As previously reported, at enrollment patients were <9 years in CA and (if GnRHa-naïve) with Tanner stage for breasts ≥2 by CA of 8 years, BA advanced ≥1 year beyond CA, GnRH-stimulated luteinizing hormone (LH) peak of ≥8 [
BA, determined by a central reader using hand/wrist radiography and the FELS method, was obtained approximately every 6 months from treatment initiation until end of on-treatment follow-up. PAH was calculated by dividing actual height with the average percentage of AH associated with concurrent BA per the Bayley-Pinneau method [
This investigation evaluated 2 groups. Group 1 included GnRHa-naïve girls with data analyzed from the start of treatment until their last on-treatment assessment. Group 2 included GnRHa-naïve or non-naïve girls with data analyzed over the 3 treatment years preceding their first observed BA of ≥12 years. Plots were inspected to visualize changes in BA, BA/CA, HV, and PAH over time in Group 1 and changes in HV over time as well as changes in PAH in relation to BA/CA in Group 2.
Linear mixed-effects models examined average change in PAH between 2 consecutive time points during treatment (dependent variable) in Group 1. The base model included treatment duration (i.e., time) and its square term to account for non-linear (quadratic) change in PAH over time. Covariates were: change in BA/CA, HV, age at treatment initiation, BA, baseline height, baseline PAH, and MPH. Each of the covariates was first added separately to the base model to explore their relationship with change in PAH (while adjusting for time). Next, the final linear mixed-effects model used forward selection to add the covariate that maximized model prediction in each forward step. Beginning with time and square of time (i.e., base model), covariates were added in the order listed above until the stopping criterion of p>0.05 was reached. Individual patient plots were used to visualize key relationships in the final model.
To further evaluate the effect of treatment duration on PAH in Group 1, three multiple linear regression models were built, each using last assessed on-treatment PAH as the dependent variable and the same covariates as independent variables (i.e., treatment duration; MPH; race/ethnicity; and baseline BA, age, HV, height, and PAH). However, treatment duration was examined as a binary covariate with a cut-off of ≤33 vs. >33 months (the median treatment duration in Group 1) in Model 1 (primary analysis), and as a continuous covariate with a monthly increase in Model 2 (secondary analysis). Then, an exploratory classification and regression tree (CART) analysis was conducted to identify an “optimal” cut-off for treatment duration to separate girls with good PAH performance from others. Model 3 assessed treatment duration as a binary covariate with a CART-suggested cut-off.
SAS version 9.4 for UNIX (SAS Institute, Cary, NC) was used for these analyses, with a statistical significance threshold of p<0.05.
Group 1 included 77 girls with a mean CA of 7.0 ± 1.7 years (range 1–9 years), mean BA of 10.5 ± 1.9 years (range 2.5–12.7) at onset of treatment (
Patient characteristics at baseline.
Characteristic | Group 1 | Group 2 |
---|---|---|
n=77a | n=56a | |
Race or ethnicity, n (%) |
||
White/Caucasian | 47 (61.0) | 33 (58.9) |
Black/African descent | 18 (23.4) | 14 (25.0) |
Hispanic/Latinx | 8 (10.4) | 3 (5.4) |
Asian | 1 (1.3) | 2 (3.6) |
Multiracial or multiethnic | 3 (3.9) | 4 (7.1) |
|
||
CA, years | ||
|
||
Mean (SD) | 7.0 (1.7) | 7.6 (1.3) |
Range | 1–9 | 3–10 |
|
||
Tanner breast stage, n (%) | ||
|
||
1 | 1 (1.3)b | 3 (5.4)c |
2 | 13 (16.9) | 12 (21.4) |
3 | 46 (59.7) | 30 (53.6) |
4 | 16 (20.8) | 10 (17.9) |
5 | 1 (1.3) | 1 (1.8) |
|
||
Weight, kg | ||
|
||
Mean (SD) | 35.5 (10.4) | 37.0 (9.3) |
Range | 13.0–63.8 | 19.4–63.3 |
|
||
Height, cm | ||
|
||
Mean (SD) | 133.8 (13.9) | 136.2 (9.8) |
Range |
84.6–154.7 |
109.9–151.9 |
Height standardized score |
||
Mean (SD) | 1.6 (1.12) | 1.3 (1.2) |
Range | −1.3–3.4 | 1.3 (−1.2 − 3.4) |
|
||
PAH, cm | ||
|
||
Mean (SD) | 156.8 (7.4) | 156.4 (7.6) |
Range | 135.7–171.2 | 135.7–169.7 |
|
||
MPH, cm | ||
|
||
Mean (SD) | 163.4 (7.0) | 163.7 (5.8) |
Range | 138.3–175.6 | 144.6–172.6 |
|
||
BA, years | ||
|
||
Mean (SD) | 10.5 (1.9) | 10.9 (1.1) |
Range | 2.5–12.7 | 6.6–12.0 |
|
||
BA/CA | ||
|
||
Mean (SD, range) | 1.5 (0.3) | 1.4 (0.2) |
Range | 1.2–2.9 | 1.0–2.1 |
|
||
HV, cm/year | ||
|
||
Mean (SD) | 10.0 (3.8) | 9.0 (3.6) |
Range | 1.3–21.2 | 2.5–18.8 |
aGroup 1 included GnRHa-naïve girls with data analyzed from start of treatment; n=77 with the following exceptions: weight, BA, and BA/CA, n=76; PAH, n=67; MPH, n=70, and HV, n=73. Group 2 included girls with data analyzed over the 3 treatment years preceding the first observed BA of ≥12 years; n=56 with the following exceptions: weight, n=55; height, n=54; PAH=51; MPH, n=50; and HV, n=52. bA one-year-old was enrolled in the trial with breast at Tanner stage 1 based on qualifying peak stimulated luteinizing hormone (84.7 IU/L) and estradiol (90 pg/mL). cPatients with Tanner breast stage <2 were enrolled in Study 2, if they were not naïve to GnRHa therapy. BA, bone age; CA, chronological age; HV, height velocity; MPH, mid-parental height; PAH, predicted adult height; SD, standard deviation.
Group 1 plots revealed that changes in BA, BA/CA, and rate of change in PAH over time were nonlinear during treatment. Change in BA tended to decrease as treatment progressed, with the largest changes observed during the first year of therapy (
Individual patient plots illustrating changes in BA over time from start of therapy to BA measurements in Group 1*. *Group 1 included GnRHa-naïve girls with data analyzed from start of treatment. BA, bone age; y, year.
Mean BA/CA (A), mean HV (B) and mean annualized change in PAH (C) over time since start of therapy in Group 1*. Error bars reflect standard deviation. *Group 1 included GnRHa-naïve girls with data analyzed from start of treatment. BA, bone age; CA, chronological age; CI, confidence interval; HV, height velocity; PAH, predicted adult height.
Mean HV decreased rapidly in the first year of treatment, with the rate of change slowing thereafter (
Group 2 plots examining the 3 years of treatment preceding their first observed BA of ≥12 years showed that the change in BA/CA decreased and the change in PAH increased (
Individual patient plots illustrating changes in PAH in relation to changes in BA/CA (A) and HV over time (B) during the 3 years preceding BA of ≥12 years in Group 2*. Time (x-axis) of 0 in plot B indicates the first observed BA of ≥12 years. *Group 2 included girls with data analyzed over the 3 treatment years preceding the first observed BA of ≥12 years. BA, bone age; CA, chronological age; HV, height velocity; PAH, predicted adult height; y, year.
In Group 1, there was a significant association between change in BA/CA and change in PAH in the initial linear mixed-effects model adjusting for treatment duration (
Individual patient plots illustrating on-treatment changes in PAH in relation to change in BA/CA (A), HV (B), and BA (C) in Group 1*. *Group 1 included GnRHa-naïve girls with data analyzed from start of treatment. BA, bone age; CA, chronological age; HV, height velocity; PAH, predicted adult height; y, year.
Forward-selection (starting from a base model including only treatment duration and its square term) determined that incorporation of concurrent BA/CA change, HV and BA, as well as CA at treatment initiation significantly improved the model’s ability to predict change in PAH between consecutive time points (
Effect of the forward-selected independent variables on change in PAH between two consecutive time points during treatment in Group 1a.
Independent variableb, unit of increase | Dependent variable estimate (SE), cm | p-Value |
---|---|---|
Treatment duration at measurement, since start of treatment | 0.34 (0.168) | 0.0461 |
Treatment duration squared at measurement, since start of treatmentc | 0.06 (0.021) | 0.0076 |
Change in BA/CA at measurement, −0.1 y/y | 4.55 (0.167) | <0.0001 |
HV at measurement, 1 cm/y | 0.48 (0.051) | <0.0001 |
CA at treatment initiation, 1 y | 1.01 (0.132) | <0.0001 |
BA at measurement, 1 y | −0.53 (0.157) | 0.0008 |
aGroup 1 included GnRHa-naïve girls with data analyzed from start of treatment. bVariables are listed in the order in which they were included in the final model using forward selection, starting with the base model with treatment duration and its square term; addition of baseline height, baseline PAH, and MPH did not improve predictive ability and were excluded from the final model. cSquare term of treatment duration was used to account for nonlinear (quadratic) change in PAH over time. BA, bone age; CA, chronological age; HV, height velocity; MPH, mid-parental height; PAH, predicted adult height; SE, standard error; y, year.
In the three multiple regression models in Group 1, the last assessed PAH near the end of on-treatment follow-up was found to be positively associated with longer duration of treatment in the primary (Model 1), secondary (Model 2), and exploratory (Model 3) analyses (
As previously reported, relying on CA, BA, or HV as a single variable is not sufficient to predict AH, consistent with the understanding that the impact of treatment on AH is multifactorial [
Based on our measurements, an initial modest treatment response is not necessarily of concern, as the BA/CA does not decrease and PAH does not increase significantly until at least 6 months of treatment. This is consistent with a lag of impact of sex steroids on BA advancement [
The rate of change in PAH during treatment was observed to be greater on average in girls who initiated therapy at an older CA in the multivariate model. This suggests that older girls with CPP who (1) are at a CA nearing normal onset of puberty and (2) have a BA in the pubertal range, experience more rapid decreases in bone maturation with treatment. In contrast, bone maturation continued in younger girls at a year-to-year rate. This aligns with typical growth patterns observed prior to and during puberty and the physiological changes occurring in the growth plate [
PAH was observed to continue to improve after the first year, even up to 5 years of treatment, albeit to a lesser degree. With more prolonged treatment, the rate of change of the variables slows. However, PAH continues to increase, as long as BA/CA continues to decrease (or stabilize) and an adequate HV is maintained, indicating maintenance of PAH with ongoing treatment. Therefore, it can be expected that if treatment is stopped upon stabilization of BA/CA, unless BA is very advanced and approaching complete growth plate fusion [
There is also a tendency to discontinue treatment based on stabilization in PAH alone [
Strengths of this study include modelling an average on-treatment change in PAH as treatment starts and progresses (as opposed to reporting only on total PAH change). From the data obtained with the models utilized, better-informed decisions can be made, based on the patterns of change of the variables measured. Additional strengths included examining a long-term treatment course (median 2.75 years [33 months]; range, 6–118 months) and drawing from two trials, thus ensuring larger sample size and increased patient heterogeneity. A potential limitation of this study was that only PAH during treatment is reported. However, we previously published an analysis of a large subset of patients included in Group 1 of this study cohort, with data through AH [
In conclusion, these data support individualized monitoring of variables during GnRHa treatment. Changes in PAH vary depending on CA and time since initiation of treatment [
Medical writing support was provided by The Curry Rockefeller Group, LLC (Tarrytown, NY) and was funded by AbbVie (North Chicago, IL).
The online version of this article offers supplementary material (