Main

Ach is the most common genetic form of disproportionately short stature. It is inherited as an autosomal dominant trait and is usually the result of a spontaneous missense mutation in the transmembrane domain of the FGFR3 gene, resulting in a glycine to arginine substitution at codon 380 (1, 2). This mutation constitutively activates the receptor in the absence of ligand (3, 4) and results in the consistent phenotype of Ach, observed worldwide with similar frequency (5). The characteristic phenotype with disproportionately short stature (short limbs), macrocephaly, and the later development of lumbar lordosis and bow legs (6) results in an average adult Ht of approximately 130 (118–145) cm in untreated males and 120 (112–136) cm in females (7).

Linear skeletal growth relies on enchondral ossification of the growth plate cartilage within which chondrocytes undergo a well-defined process of proliferation and maturation (8). The pathogenesis in Ach is a defect in enchondral ossification of bone with normal membranous ossification (9).

GH is an important regulator of linear skeletal growth (10). It directly increases linear growth by stimulating proliferation of epiphyseal growth plate precursor cells, and it also enhances local production of IGF, the latter stimulating clonal expansion of differentiating chondrocytes (11–13). Given the skeletal abnormality, a reduced sensitivity to the action of GH and IGF would be expected (14).

The increased availability of GH with the introduction of r-hGH in 1986 made it possible to test its use in a wider range of conditions than simple replacement therapy for GH deficiency. Its short-term use (1–2 y) in small numbers (n= 6–15) of Ach patients has shown potential benefits of GH in Ach (15–20). Recently, Tanaka et al. (21) also confirmed this in a larger Japanese group (16 to 42 patients treated for a period of 1–3 y).

We describe the effect r-hGH therapy on stature and body proportion, using a wide range of GH doses in a larger number of Ach patients (n= 35) treated over a longer period (1–6 y, median 3 y).

METHODS

We undertook an open study of 35 consecutive children with Ach. The diagnosis was suspected on clinical grounds (22) and confirmed by standard radiologic criteria (6). All patients had cranial and spinal imaging (either computerized tomography or magnetic resonance imaging) after diagnosis with subsequent follow-up scans only if there were clinical concerns (e.g. increased intracranial pressure).

Blood samples for DNA isolation were available from 21 of the 35 patients. DNA was isolated from lymphocytes as previously described (23). The 164-bp product of the transmembrane domain of FGFR3 was amplified from genomic DNA with published oligonucleotide primer sequences (24). Samples were digested with restriction enzyme Sfc 1, Msp I, and Nla III to test for the G1138A, G1138C, and G1138T mutations, respectively.

Anthropometric measurements were taken by accredited measurers during outpatient visits to endocrine clinics at The London Centre for Paediatric Endocrinology. Standard measurements of Ht, SH, and weight were made before treatment and every 6 mo thereafter. Measurements of SILL were derived by calculating the difference between Ht and SH, whereas annual HV was calculated as annualized Ht increments. These measures were converted to age- and sex-appropriate SDS by comparison with normal reference standards (not Ach) as described by Tanner et al. (25). Because Ht is normally distributed, the 50th, 3rd, and 95th percentiles represent the mean and ±2 SD of the normal population. SDS are calculated according to the following formula:EQUATION 1 In this way, statistical comparisons can be accurately made between individuals of different ages and sexes. SILL SDS was not calculated for those <2 y of age, as normal standards are unavailable. Growth was also assessed by plotting serial Ht measurements on disease-specific Ach growth charts (7).

The onset of puberty was defined as testicular volume of at least 4 mL in boys and the onset of breast development in girls and was staged according to Tanner (26). Bone age and Ht predictions were not performed routinely because the abnormal bones prevent interpretation of x-rays in Ach as in other skeletal dysplasias (27).

The dose of GH used in each individual was variable, ranging from physiologic replacement doses of 15–20 U/m2 per week to supraphysiologic doses of >25 U/m2 per week and up to 40 U/m2 per week. The present study is effectively an audit of the long-term results of previous patients randomized to 20 or 40 U/m2 per week for 2 y together with other individuals arbitrarily treated with 30 U/m2 per week and largely reflects personal practice. Those who received <25 U/m2 per week were similar to those receiving >25 U/m2 per week with respect to pretreatment Ht and HV SDS (p> 0.05), and so we were able to correlate response to dose range.

r-hGH (Genotropin, Pharmacia-Upjohn, Stockholm, Sweden or Norditropin, Novo-Nordisk, Denmark) was administered by daily subcutaneous injections at a median dose of 30 (15.8–40) U/m2 per week [i.e. 0.06 (0.04–0.08) mg·kg−1·24 h−1]. Patients were treated for a median (range) of 3 (1–6) y. The number of patients treated year by year were the following: y 1 = 35, y 2 = 27, y 3 = 21, y 4 = 18, y 5 = 9, and y 6 = 6.

Ethical approval was obtained for the present study from the Joint University College London/University College London Hospitals Committee on the Ethics of Human Research. Informed consent was always obtained from parents and assent from the children when appropriate. All patients attended the endocrine clinics at the Middlesex Hospital or Great Ormond Street Hospital for Children, London.

Statistics.

Data are expressed as SDS with respect to Tanner et al. (25) normal values. Plotting raw data of Ht, SH, and SILL on percentiles is an acceptable and easy way of explaining an individual's growth problems to parents. However, in extremely short stature as in Ach in which data lie well below the lowest percentiles, this is not an accurate representation. Further, the statistical analysis of group data requires that the data are presented as a summary statistic with respect to age and sex standardized means, and, in this way, SDS are a well-recognized and robust measure. The SPSS v7.0 statistical package was used for analysis. Examination of data was undertaken to confirm its symmetrical distribution. Our population before treatment was also confirmed to be normally distributed across the Ach-specific growth charts described by Horton et al. (7). Changes in group mean data were compared by 1-way ANOVA with the patients as the blocking variable and Scheffe's post hoc tests for multiple contrasts. Stepwise multiple regression was used to ascertain the effects of age at therapy and dose of GH on the auxologic response, this being expressed as first-year changes in both Ht and HV SDS. Lowess locally weighted regression scatter plot smoothing procedure was used to show the yearly change in HV SDS curves in treated patients and those before treatment (28). This is a robust statistical method that fits a line or curve to include a minimum of 50% of the data points. In this case, it was used to show the inexorable decline in HV seen in Ach patients and as compared with its attenuation in treated Ach (Figure 4). As the data were symmetrically distributed, the mean and median were very similar. We have presented mean and SEM in graphics because this is what is most common, but we thought it important to include median and range to show the wide individual variability of response.

Figure 4
figure 4

HV SDS changes before and during therapy. Lowess locally weighted regression scatter plot smoothing procedure comparing treated HV SDS data with similar data before treatment and according to age and time. Although the pretreatment value at 11.3 y is shown, it was not included in the analysis of the data, as this case was pubertal at the start of therapy. For three patients, pretreatment HV SDS are not given, as the pretreatment HV was calculated at other centers.

Full size image

RESULTS

Mutation analysis.

The G1138A mutation in the transmembrane domain of FGFR3 was detected in all 21 patients with Ach from whom DNA was available.

Baseline auxology and puberty status.

The median age at start of therapy was 2.25 (1.2 to 9.3) y. The study included 23 males and 12 females. The median Ht SDS at the start of therapy in the whole group was −4.6 (−6.5 to −3.24), median SH SDS −0.74 (−2.98 to 1.8), and SILL SDS −7.52 (−9.27 to −6.02), confirming the known disproportion.

All were prepubertal at the start of therapy. During the treatment years, the number of patients in puberty were as follows: 2/35 in y 1, 2/27 in y 2, 2/21 in y 3, 3/18 in y 4, 2/9 in y 5, and 2/6 in y 6. Therefore, the majority of our patients remained prepubertal during the period of assessment. These patients will be followed up through puberty until final Ht.

Auxologic response to r-hGH.

The first-year response to r-hGH, expressed as a change in Ht SDS over 1 y of therapy, was greatest in the younger patients (p< 0.05) (stepwise linear regression analysis, R2 = 0.41;p< 0.01).

There was a progressive increase in Ht SDS from baseline to y 4 (F= 46.94, p< 0.001) as shown in Figure 1. Years 5 and 6, although different from baseline, were not significantly different from y 4 (p> 0.05, Fig. 1), but the number of cases are small by then. Although it seems that there is no further catch-up in Ht SDS after y 4, it must be remembered that maintaining the SDS position with respect to normal standards is a significant improvement on the expected inexorable decline in Ht SDS seen in Ach and, as such, represents a confirmed benefit. When the raw data are plotted on disease-specific Ach charts, the continued individual improvement becomes evident (Fig. 2).

Figure 1
figure 1

Change in Ht SDS over time. Serial Ht SDS in 35 Ach children during 6 y of treatment with r-hGH. Data shown as mean ± SEM.

Full size image
Figure 2
figure 2

Ht increments over time. Individual linear Ht growth responses of 35 Ach children treated with r-hGH plotted on disease-specific Ach growth percentiles (solid) and shown together with Tanner's normal reference percentiles (dotted) chart (left panel, males;right panel, females). Longitudinal Ht measurements of these patients, plotted on disease-specific growth curves, were continually maintained above the pretreatment percentiles.

Full size image

We did not observe an obvious dose effect after the first year of treatment with r-hGH. This is shown in Figure 3 and shows the first-year change in Ht SDS with respect to baseline values. All except three patients responded to therapy, but no obvious dose effect was seen. In a small minority, the Ht SDS position was maintained from pretreatment values with no catch-up. However, given the natural history of Ach in which you would expect a decline in Ht position with time, this would be consistent with a positive response.

Figure 3
figure 3

Dose response. The response to r-hGH therapy in the first treatment year expressed as a change in the first-year Ht SDS in 35 Ach patients does not show a significant dose effect.

Full size image

The HV of the treated patients before starting r-hGH followed an expected progressive decline. By contrast, treatment significantly and positively changed HV SDS in the first year, and the benefit was sustained in subsequent years (F= 4.28, p< 0.001). GH maintained the growth rates of Ach children close to the mean for normal children, and the slope of the HV SDS decline was significantly less steep than what would be expected from the history of Ach (Fig. 4).

Significant stepwise year-to-year increments were noted in SH SDS from baseline through 3 y (F= 26.25, p< 0.01), but again there were no significant increments thereafter, as shown in the upper panel of Figure 5. The increment in SILL SDS was significant at all years (F= 9.04, p< 0.01), but there were no further increments between treatment years, as shown in the lower panel of Figure 5. In comparison with SH gains, the increase in SILL was minimal, thereby accentuating the existing disproportion (y-4 change in SH SDS versus SILL SDS change, p> 0.05).

Figure 5
figure 5

Effect on body proportions. Stepwise year-to- year increments in SH SDS (upper panel) and SILL SDS (lower panel) from baseline through 6 y:F= 26.25, p< 0.01;F= 9.04, p< 0.01, respectively. Data shown as mean ± SEM.

Full size image

DISCUSSION

Since the introduction of r-hGH, the wider indications for GH therapy in short stature conditions not associated with GH insufficiency have attracted increasing interest.

Ach is one of the most severe forms of skeletal dysplasia with individuals on average being 50 cm shorter than the normal adult population. The nature of the growth problem in such conditions is most likely due to reduced sensitivity to the action of GH and IGF (14). Hence, supraphysiologic GH therapy has been used in an attempt to overcome skeletal resistance. Undoubted initial short-term benefits in both Ach and other skeletal dysplasias including Turner's syndrome (15, 16) have been described, although not all have been translated into significant increments in predicted adult Ht (29, 30). Differences in age at onset of therapy, duration and dose of therapy, and accuracy of the method of Ht prediction as well as a secular trend to increased adult Ht (31–33) may have contributed to these discrepancies.

We also report a significant benefit on longitudinal measurements of Ht SDS in an uncontrolled intervention study, but our data were strengthened by a larger number of a mostly prepubertal cohort of much younger age (2.25 y) and longer treatment duration (up to 6 y). In addition, we have been able to assess the effects of a wide range of r-hGH doses on the response. We were not able to show a significant dose response relationship even in the first year of treatment, and this was surprising given the experience in other related conditions (15, 16). However, beyond the first year of life, higher doses have been shown to advance skeletal maturity and to shorten the duration of puberty (34), thereby limiting the eventual growth potential. Thus, it will be important to compare outcomes of low-versus high-dose therapy at adult Ht before making recommendations on the dosage at therapy.

In the natural history of the disease, there is a progressive accumulation of Ht deficit. After the first year of life, the growth rate in untreated children approximates only the third velocity percentile of normal children and remains at this slow rate for the rest of childhood (7). In comparison with the growth rates of normal children, which oscillate about the 50th percentile, this is clearly abnormal. Early r-hGH therapy in children with Ach has prevented the accumulating Ht deficit by maintaining growth velocity near the normal range and maximizing the potential of normal growth in the spine. Although SILL remained significantly compromised, no further deficit was incurred over the treatment period. The accentuation of the existing disproportion in Ach due to the variable SH and SILL responses to r-hGH therapy that was shown in this study has not been reported and, in most cases, has not been examined in previous studies of r-hGH in Ach (15–21). The greater number of patients who had a longer duration of therapy in our study may account for this accentuation of disproportion being shown.

The abnormal bones in Ach make estimates of skeletal maturity unreliable. The inter- and intraobserver error even in the normal situation is well documented (31) and is increased in children with skeletal dysplasias. We have, therefore, not used this as a method of predicting final Ht. Instead, we have used growth charts available from historical Ach patients for comparison (7). Such a method has been used to assess the treatment effects in other conditions (32) and carries with it the potential of misrepresenting final outcomes if there has been a significant secular trend. We showed that our subjects were comparable to historical controls at the time of onset of treatment, whereas the magnitude of the secular trend is unlikely to significantly affect the reported outcome. The magnitude of growth improvement documented was an increase of one Ht SDS over 6 y, an increment of 8 cm in terms of “Ht gained.” Nevertheless, the individual responses reported were variable.

The effects of r-hGH on final Ht are not known, but GH therapy coupled with the opportunity of leg lengthening will alleviate the disproportion and carry the possibility of adult stature for patients within the lower end of the normal range. The dose of r-hGH and the timing(s) of surgical intervention(s) need to be established after adult Ht data are confirmed.