Many short children and their families ask for an adult height (AH) prediction at the beginning of puberty. If this prognosis falls below their expectations, they often inquire about methods to increase AH by maximizing their pubertal height gain. Unfortunately, the remaining height gain at the beginning of puberty represents only 15-20% of total height. Closure of growth plates limits the intervention time (1, 2). This process is driven by gonadal steroids, mainly estrogens (3, 4). Gonadotropin-releasing hormone analogues (GnRHa) efficiently suppress the progression of puberty and delay the closure of growth plates. In girls with precocious puberty, GnRHa treatment results in a gain in AH of 3 to 10 cm (5). In children with normal pubertal timing but with poor AH prognosis the results of GnRHa treatment are rather disappointing, with a gain in AH ranging from 1 to 4.2 cm (6–8). These modest results can be explained by the fact that GnRHa not only slow down skeletal maturation but also reduce growth hormone secretion and growth velocity (9).

To prevent this slowdown of growth velocity during GnRHa therapy, a combined treatment of GnRHa with growth hormone (GH) has been explored in different clinical settings. In girls with precocious puberty, AH minus predicted adult height (PAH) at start was higher in girls treated with combined GH/GnRHa therapy (4.7 to 11.4 cm) compared with girls treated with GH alone (10–13). In children with GH deficiency (GHD) who entered puberty early, Tauber et al. reported no difference in AH minus PAH between the combined GH/GnRHa treatment and the GH-only treatment, although the duration and dosage of GH therapy was variable (14). Mericq et al. however, found a significant increase in near-adult height (NAH) after three years of combined GH/GnRHa treatment compared with GH alone in GHD (15). The first reports on this combined GH/GnRHa treatment in children with idiopathic short stature (ISS) did not show any improvement of AH compared with controls (16, 17). The National Cooperative Growth Study (USA) found that NAH exceeded initial PAH with 3 ± 6.1 cm in a pooled group of patients with GHD or ISS treated with combined GH/GnRHa treatment (18).

More recent studies with higher GH doses and longer treatment durations in children born small-for-gestational age (SGA) or with ISS reported a difference in AH vs PAH of 4.9 to 10.8 cm (19–26). Bennabad et al. found no difference in NAH-SDS between the combined GH/GnRHa treatment and the GH alone treatment in ISS children after 2.4 years, but the study was prematurely discontinued (27). With this longer duration of treatment however, by postponing puberty development for such a long time, psychosocial functioning can be impaired (28). Significant better height gains of combined GH/GnRHa treatment have also been reported in short adopted girls (29, 30), and patients with SHOX deficiency (31) ( Supplementary Table 1 ).

The level of evidence of many of these studies, however, was limited due to either small numbers of patients, the lack of a control group or a short follow-up time (32). Furthermore, most studies did not consider the limited accuracy of height prediction in this group of patients. Therefore, we conducted a prospective, case-controlled clinical trial of four-year combination treatment with GH and GnRHa in a cohort of girls with ISS at the beginning of puberty and with follow-up to AH.

In this multicenter, open-label, case-control study we evaluated a 4-year GH/GnRHa combination therapy in early pubertal girls with a PAH below -2.5 SDS and compared the results to untreated historical controls.

Mean AH in girls after 4-year treatment with GH/GnRHa was 6.4 cm higher than in matched controls despite the fact that PAH was higher in the control group. We also found that AH surpassed initial PAH by 12.0 ± 2.6 cm in treated girls and by 4.2 ± 3.6 cm in controls. Only a few studies reported the effect of GH/GnRHa combination treatment on AH in girls with ISS. Van Gool et al. reported a gain in AH vs PAH of 7.4 cm (20) and Maniati Christidi et al. of 5.7 cm (21). In a randomized controlled trial with GH versus GH/GnRHa, Pasquino et al. reported a gain of AH vs PAH of 6.1 cm with GH alone and of 10.0 cm with the GH/GnRHa combination treatment (19). In a retrospective study, Lazar et al. saw a gain of AH vs PAH of 7.2 cm with GH and 9.5 cm with GH/GnRHa in pubertal ISS girls (23). Benabbad et al. found no difference between GH alone and GH/GnRHa combination treatment in ISS patients when initiated at puberty onset (27). In our study we did not have a GH-only arm, so we are unable to measure the contribution of each treatment component separately. A meta-analysis and systematic review on the effect of GH-only treatment in ISS revealed a treatment effect of 4-6 cm versus untreated controls and about 4 cm compared with PAH (39, 40). The large treatment effect in our study supports the notion that concomitant GnRHa treatment adds to the effect of GH on height gain.

An important finding in the control group is that AH surpassed PAH by 4.2 ± 3.6 cm. The comparison between AH and PAH, heavily relies on the accuracy of the AH prediction. To establish the prediction error, we compared AH to the PAH in a matched control group and found that the Greulich and Pyle-Bayley and Pineau method underestimated AH. In most studies, AH prediction in untreated ISS patients was reported to be very accurate albeit with a wide prediction error (41–45). However, other authors found an average under-prediction of 2.8 cm and 2.3 cm in cohorts of girls comparable to ours (46, 47). This may be explained by the observation that the accuracy of AH prediction in ISS patients depends on the degree of bone age delay at the time of the prediction. If the bone age delay is close to 2 years, the AH prediction matches the measured AH, but if there is minimal bone age delay, as in the present study, under-prediction is expected (46).

Another important finding is the fact that PAH at the end of treatment overestimated AH. PAH continued to rise during the treatment phase because the bone age progression was kept at a low pace during the entire treatment period. When the GnRHa treatment stopped, bone age accelerated significantly. As a result, PAH at the end of the treatment overestimated AH by a mean of 2.6 cm. This was also demonstrated in the Dutch ISS study, which showed a gain in PAH of 7.8 cm between treated and untreated girls after the 3-year treatment period (25), however at AH the gain was only 5.5 cm (20).

In our study the PAH after 4 years of treatment was the same as the PAH after 3 years of treatment, suggesting that a fourth year of GH/GnRHa treatment might not have a major contribution to AH. On the other hand, the fourth treatment year added more cm in actual height and reduced the post treatment period that normally results in a loss of height SDS due to a fast bone maturation. The comparison of results between a shorter and longer treatment duration published in the literature is in favor of a longer treatment period (19) but only a comparative trial can resolve this question.

Most treated girls reached a normal height > -2 SDS (87.5%) and reached a height equal or more than TH (68.7%), while the minority of the controls did (37.5% and 6.2%, respectively). Despite these promising results, one must ask whether the number of girls attaining a normal height justifies the cost of the treatment protocol. Further research with larger study populations might be required to establish the cost-effectiveness of this treatment.

In our study we did not find an effect of birth weight and length, gestational age, maternal height, paternal height, TH, height deficit in relation to TH, nor bone age at onset on the outcome parameters AH, AH-PAH, and AH-H₀. However, the negative correlation between H₀ SDS and height gain SDS, and the positive correlation between bone age delay at start and height gain SDS, suggests that girls with a smaller height at start and a more delayed bone age could benefit more from this treatment. With a small sample size, however, caution must be applied, and these findings must be confirmed in larger studies.

The safety profile of the GH/GnRHa combination treatment in this study was acceptable. Most adverse events were typical for this age group of active adolescents. One adverse event possibly related to the intervention was reported: a fracture of metatarsals in a ballet dancer. Benabbad et al. also reported that bone fractures occurred more frequently in the combined GH/GnRHa group than the GH-alone group, and that they were associated with an abnormally low BMD (27). It has been reported that GnRHa treatment (with or without GH) decreases bone mineral density (BMD) in children with short stature (7, 48, 49), but that BMD returns to normal after cessation of therapy (20, 50). Future research is needed to better understand the risk of decreased BMD associated with this GH/GnRHa treatment. Another concern is the reduction in insulin sensitivity due to GH therapy. Therefore, fasting insulin and glucose levels and HbA1c levels were monitored during the treatment period. The observed increase in fasting insulin levels was not associated with abnormal fasting glucose nor HbA1c levels; fasting insulin levels normalized after cessation of treatment. This agrees with the findings of the Dutch study group that GH/GnRHa combination therapy did not adversely affect glucose metabolism in SGA children (51, 52).

The key strengths of this trial are its prospective and rigorous experimental design, with a uniform treatment schedule, a long GH/GnRHa treatment duration, a homogeneous study population and a follow-up period until AH. In contrast to most previous studies, treatment with GnRHa and GH was started simultaneously. The patients in our study were treated with a GH dose of 50µg/kg/day, since Wit et al. showed a dose effect for GH in ISS patients (46) and in a dose response GH/GnRHa trial in SGA children, Lem et al. demonstrated a significantly better outcome with the higher GH dose in combination with GnRHa (24). In some studies, GH treatment was continued until near AH. This could have further increased AH, but the effect would probably be limited by the acceleration of bone age and fast closure of the growth plates after stopping GnRHa. To maximize the effect of GH and to obtain a uniform treatment schedule, we choose a 4-year combination treatment duration. After 4 years, most of the pubertal height gain was obtained and we did not see a decrease in height SDS after stopping treatment, as seen in studies with a shorter treatment duration. In addition, the PAH after 3 and 4 years of treatment was similar, suggesting that a three-year treatment period might result in equal height gain, and reduce the number of dropouts. This hypothesis should be explored in a future randomized comparative trial. Moreover, with such a long treatment duration, treatment adherence sometimes became a problem due to the large number of injections for a long period of time, and the psychological effect of postponing puberty for 4 years. This was apparently less tolerated by boys than by girls, as was demonstrated by the failure to include enough boys and the high drop-out rate.

To obtain a homogeneous study population, only girls with an adult height prediction of less than -2.5 SDS for the Flemish population were included. This study group predominantly included girls with ISS, although familial short stature and short stature following SGA was not excluded. Most girls were early pubertal, and this may contribute to the positive outcome of this study. Some studies included girls with a more advanced pubertal development. A more advanced puberty stage, or a small size at birth may reduce the treatment effect and therefore underestimate the efficacy of this combination treatment in puberty stage B2 girls without SGA. Our post hoc analysis removing and replacing these patients did not substantially change the results. However since there were only 2 patients with a B3 stage and only 3 SGA patients in the control group, this study was not powered enough to answer this question.

Furthermore, one of the major strengths of this study is that we followed the treated girls until AH. Some studies report only on PAH at end of treatment. This may lead to overestimation of treatment results as again demonstrated in our study. Our results based on measured AH better reflect the true treatment effect.

This study has certain limitations. A first limitation is the nonrandomized design, the inclusion of a selection of historical controls and the relatively small study population. One third of the participants dropped out due to low compliance. Importantly, as demonstrated in Table 1 , our analysis shows that the excluded patients did not respond differently from the PP group in terms of response, excluding a selection bias. Our small study should encourage the conduction of a large, randomized, placebo controlled study to firmly establish the value of this treatment in early pubertal girls with a poor adult height prediction.

An additional limitation of our study is the lack of a centralized bone age reading, which may have led to inter–observer variances. This problem could be avoided by using an automated bone age estimation, which is unfortunately not available in all hospitals. A last limitation is the lack of quality-of-life assessment. Due to the long treatment protocol some girls were relatively old before puberty started and a future study should evaluate the effects on quality of life of postponing puberty for 4 years, to weigh the effectiveness of treatment against the psychological burden that it may entail.

Taken together, our findings suggest that a 4-year combination treatment of GH and GnRHa results in a clinically relevant and statistically significant increase in AH in early pubertal GH-naïve girls with a poor adult height prognosis. The impact of increasing AH must be weighed against the financial and the psychological burden of this intensive treatment. Moreover, the long-term effect on bone health remains to be further explored.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Etisch Comité UZA University Hospital Antwerp Edegem Belgium. Written informed consent to participate in this study was provided by the participants’ legal guardian/next of kin.

HD: First authorship. AF, CH, ST, CB, M-CL, KD, GM, ML, IG, JV, CD, MZ, JD, VB: co-authorship. SV: senior authorship. RR and MD: Equal contribution and last authorship. All authors contributed to the article and approved the submitted version.