In children with idiopathic short stature, what advantage does administering recombinant growth hormone have over observation in final adult height?

Abstract

The research literature suggests that recombinant growth hormone, when compared with no therapeutic agents, increases the final adult height in short normal children. However, it seems that the increase in adult height may be of marginal magnitude (Grade of recommendation: A, based on systematic reviews of Controlled Clinical Trials). Two separate groups, Bryant et al (1) and Finkelstein et al (2), completed systematic reviews on the effect of growth hormone on children with idiopathic short stature (ISS). Finkelstein et al conducted a meta-analysis of 10 controlled and 28 uncontrolled trials of growth hormone. Of the 10 controlled clinical trials, only four observed final adult height or near final adult height, with only one of the trials being randomized. In Finkelstein et al’s analysis of the four trials, the pooled adult heights in the treatment group were greater than that of the controls. By translating the height standard deviation scores (SDS), Finkelstein et al estimated the average height in the treatment group to be 166.3 cm and 153.5 cm for American men and women, respectively. In comparison, the height SDS for the control groups translated to a significantly lower mean height of 160.7 cm and 148.3 cm for men and women, respectively. When Finkelstein et al reanalyzed the data of the three studies that included both baseline and adult measurements, the treatment group’s estimated gain in height over the control group was 5 cm to 6 cm (range 2.3 cm to 8.7 cm, height extrapolation from American population data). These values were in keeping with data analyzed from the uncontrolled trials that compared the final adult height against the predicted adult height in treated subjects. In a systematic review for the United Kingdom Health Technology Assessment Programme, Bryant et al (1) examined two nonrandomized trials of growth hormone that measured final height as well as one randomized trial that observed near final height; all three trials were also used in Finkelstein et al’s analysis. While none of these studies was statistically pooled, the individual outcomes were reported. In one study observing boys with ISS and a subnormal integrated concentration of growth hormone, the treatment group reached a mean final height of 164.5±3.9 cm (3). This value was significantly greater than the height of the control group, which had a mean height of 157.6±4.5 cm. In the other study that measured final height, the average final adult height increase over predicted adult height was found to be 2.8 cm in men and 2.5 cm in women (4). However, the difference in final adult height between the treatment and control group was not significant. Finally, in the last study comparing near final height in girls (5), the treatment group was significantly taller than both the control group and those that refused to be randomized (155.3±6.4 cm, 147.8±2.6 cm and 149.3±3.3 cm, respectively). When looking at this data, it seems reasonable to conclude that growth hormone does increase the final adult height of children with ISS. However, there are several complicating matters that must be noted. There was a large amount of heterogeneity between the four studies that examined final height or near final height (3–7). As stated above, only one of the studies was a randomized controlled trial (5), while the other studies used patients who refused growth hormone or had reached final height before recombinant growth hormone was readily available to serve as controls. One paper reported only near final height (as opposed to final height) in short normal girls (5), while another article, generated from the same experimental population, only reported height SDS after five years (which is not explicitly stated as final height) (7). Another study examined boys with ISS and concurrent subnormal integrated concentrations of growth hormone (this could be seen as an identity separate from ISS) (3). Additionally, half the studies examined children with an approximate baseline age of eight years (4,5), while the average baseline age of the other two studies was approximately 12 years (3,6). Finally, while the two systematic reviews agreed that growth hormone may have a beneficial effect, both reviews included roughly the same studies to reach their respective conclusions. The study by Bryant et al (1) was of strong methodological quality but did not statistically pool the results of the individual studies due to the large amount of heterogeneity. Meanwhile, the study by Finkelstein et al (2) pooled the individual study outcomes but failed to examine the methodological quality of the studies. While these trials and reviews suggest that growth hormone increases final height, the magnitude of the increase must also be considered. Those treated with growth hormone still appear to be about one standard deviation below average height. The estimated gain in height provided by growth hormone ranges from 2 cm to 7 cm. Thus, while it is likely that growth hormone will increase final adult height in short normal children, the gain in stature appears relatively minimal. These modest gains must be weighed against the possible adverse effects and the cost of growth hormone. While serious adverse effects have not been observed, increases in serum insulin have been noted, raising the possibility of insulin resistance (1,2). Additionally, growth hormone is administered through frequent injections (several times per week) that must be continued for a minimum of several years. Finally, the cost of growth hormone is estimated at several thousands of dollars per year per child (1,2). Thus, when considering this treatment option, the patient, family and physician must weigh the cost, discomfort and adverse effects against the potentially modest gains in height. On a more global level, it must be determined whether public funds should be used to support such a treatment at the expense of other interventions of potentially greater benefit.