Anastrozole Increases Predicted Adult Height of Short Adolescent Males Treated with Growth Hormone: A Randomized, Placebo-Controlled, Multicenter Trial for One to Three Years

Abstract

Context: The process of epiphyseal fusion during puberty is regulated by estrogen, even in males.

Objective: Our objective was to investigate whether anastrozole, a potent aromatase inhibitor, could delay bone age acceleration and increase predicted adult height in adolescent boys with GH deficiency.

Methods: Fifty-two adolescent males with GH deficiency treated with GH were randomized to cotreatment with anastrozole or placebo daily for up to 36 months.

Results: Fifty subjects completed 12 months, 41 completed 24 months, and 28 completed 36 months. Linear growth was comparable between groups; however, there was a significantly slower increase in bone age advancement from baseline in the anastrozole group vs. placebo group after 2 yr (+1.8 ± 0.1 vs. +2.7 ± 0.1 yr, P < 0.0001) and after 3 yr (+2.5 ± 0.2 vs. +4.1 ± 0.1 yr, P < 0.0001). This resulted in a net increase in predicted adult height of +4.5 ± 1.2 cm in the anastrozole group at 24 months and +6.7 ± 1.4 cm at 36 months as compared with a 1-cm gain at both time points in the placebo group. Estradiol and estrone concentrations increased less in the anastrozole group compared with placebo group. All boys on the aromatase inhibitor had normal tempo of virilization. Safety data, including glucose, and plasma lipid concentrations were comparable between groups.

Conclusions: Anastrozole increases adult height potential of adolescent boys on GH therapy while maintaining normal pubertal progression after 2–3 yr. This treatment offers an alternative in promoting growth in GH-deficient boys in puberty. Long-term follow up is needed to elucidate fully the safety and efficacy of this approach.

Treating short children in puberty can be a challenge due to the limited time available to increase linear growth before the completion of epiphyseal fusion. This study finds that co-treatment of adolescent boys with GH and the aromatase blocker, anastrozole, for 2–3 years significantly increased predicted adult height. These results were accompanied by maintenance of normal virilization, pubertal hormones, IGF-I and lipid concentrations, and a strong safety profile.

A number of strategies have evolved to increase height potential in GH-deficient children who are in puberty, such as using high-dose GH therapy (1) or GnRH analogs in addition to GH (2,3,4,5). The latter strategy has also been used in non-GH-deficient children, with mixed results (6,7,8). The consequences of gonadal suppression regarding bone accretion/bone density and the psychological impact of suppressing physiological puberty in an already short child have not been fully studied to date.

Studies of male patients with mutations in the estrogen receptor gene (9) or in the aromatase enzyme gene (10,11) as well as animal data (12) have shown that estrogen, in both females and males, is a principal regulator of epiphyseal fusion. Hence, a third strategy has evolved with more selective suppression of either estrogen production or estrogen action in puberty in those children who are very short. Administering 10 wk of the aromatase inhibitor anastrozole, which blocks conversion of Δ4-androstenedione to estrone and testosterone to estradiol, in young males, we observed no negative effects of estrogen suppression on a host of metabolic measures despite a 50% reduction in circulating estradiol concentrations (13). This is in sharp contrast to the deleterious effects of GnRH analog therapy described by us in males (14,15).

Data in boys treated with letrozole (another aromatase inhibitor) and testosterone (16) and results from a 2-yr trial using letrozole as monotherapy in boys with idiopathic short stature (17) also suggest that aromatase blockers may be a suitable alternative to promote growth in short boys in puberty. We designed this double-blind, randomized, placebo-controlled clinical trial to investigate whether treatment with a selective and potent aromatase inhibitor (anastrozole) delays the rate of bone age maturation and whether it increases adult height potential in GH-deficient adolescent boys also treated with GH. Secondary aims included measuring changes in growth velocity and height sd score adjusted for bone age, as well as changes in pubertal hormones, bone mineral density (BMD), and body composition. A thorough assessment of safety was conducted as well.

Subjects and Methods

Studies were approved by the Nemours Clinical Research Review Committee and the Institutional Review Boards of all seven participating sites. This was an original, investigator-initiated trial registered at www.clinicaltrials.gov, identifier NCT00133354.

Study subjects

Fifty-two boys with GH deficiency were recruited after informed written consent from their parents and them.

Inclusion criteria

GH deficiency was defined as evidence of either 1) short stature (>2 sd below average) or 2) significant growth deceleration (growth velocity ≤ 25% of corresponding chronological age population) and 3) peak GH responses to two pharmacological stimuli of no more than 10 ng/ml. They had to be on stable daily doses of GH for at least 6 months before, using average doses of about 0.3 mg/kg·wk. Subjects had to be in puberty [genital Tanner stage > II (>4 ml testicular volume)] and had to have residual height potential (bone age > 11.5 yr and <15 yr) at study entry.

Exclusion criteria

Subjects were excluded if they participated in other trials involving hormone therapy within 6 months, had chronic illnesses requiring long-term medication that impair growth, or had any hereditary disease or scoliosis at study entry.

Study protocol

A physical examination, including pubertal Tanner staging, accurate height measurement, blood tests, and a bone age x-ray were obtained at screening. Once determined eligible, treatment with anastrozole or placebo tablets was determined by a randomization schedule prepared by AstraZeneca, who provided the study drug and placebo tablets for the trial. Investigators, staff, and patients were unaware of the randomization codes until completion of the trial and data analysis. Dual-energy x-ray absorptiometry (DXA) was performed before initiation of study drug whenever possible but was not required. After the baseline visit, patients took 1 mg anastrozole (Arimidex) or placebo daily in addition to daily GH. Patients were followed at 3-month intervals during routine clinic visits, and blood samples and accurate anthropometric measurements were repeated each time. GH doses were titrated based on growth and body weight as needed. Left hand and wrist x-rays were obtained at 0, 12, 24, and 36 months for bone age determinations, and DXA was repeated after 24 and 36 months whenever possible. The study was continued until 36 months or until completion of linear growth, whichever happened sooner, the latter defined as the achievement of a bone age of at least 16 yr and a growth velocity of less than 2 cm/yr. Patients who voluntarily dropped out of the study were studied at termination with repeat height and weight measurements, blood tests, bone age, and DXA.

Assays

Blood concentrations of IGF-I, testosterone, lipids, glucose, estradiol, and osteocalcin were measured at 6-month intervals and complete blood count and chemistry profile with liver enzymes first every 3–6 months. Hormone concentrations were measured by RIA using commercially available Diagnostic Systems Laboratories kits (Webster, TX) at the study’s core laboratory at Nemours Children’s Clinic-Jacksonville. The intraassay coefficients of variation were 1.5% for IGF-I, 8.1% for testosterone, and 5.3% for osteocalcin. Estradiol and estrone assays were run at Dr. Ravinder Singh’s laboratory at the Mayo Clinic (Rochester, MN), using samples solvent extracted by liquid chromatography (LC) and subjected to tandem mass spectrometry (MS/MS) as previously described (18). The lower limit of sensitivity was 2 pg/ml with an intra assay coefficient of variation of about 4–20% in the lowest ranges.

Bone age and DXA readings

Bone age determinations were performed at Fels Institute (Yellow Springs, OH) by one single experienced reader blinded to the treatment arm. Predicted adult heights were estimated based on the Bayley-Pinneau table (19). DXA scans of the lumbar spine were performed using either Hologic (Waltham, MA) or Lunar instruments (Minster, OH).

Body composition

Whenever possible, body composition was assessed using the DXA software; however, not all participating centers had access to this, and it was considered an optional criteria for study participation. Because body composition was not a principal aim of these studies, in patients without DXA, those data were obtained through the sum of skinfold thicknesses (20). In nonobese individuals, the assessment of body composition correlates well with those obtained by DXA (21). This aim was considered exploratory.

Drugs

Anastrozole (Arimidex) (1-mg tablet) and placebo tablets were provided by AstraZeneca (Wilmington, DE) and dispensed for daily dosing under protocol using the principal investigator’s FDA IND number. All unused and empty vials were returned to the clinical centers for drug accountability. Recombinant human GH (somatropin) was provided to the study subjects as Nutropin, Nutropin AQ, or Saizen by generous drug-supply grants from Genentech (South San Francisco, CA) and EMD Serono (Rockland, MA), respectively. Those subjects that did not wish to switch brands were continued on their commercially prescribed somatropin products. The daily dose used throughout the study was about 50 μg/kg·d (or 0.35 mg/kg·wk).

Safety monitoring

A data safety management board composed of two pediatric endocrinologists and another pediatric subspecialist, not associated with the conduct of the studies, was established since the protocol’s inception, and all data, results and safety, were reported to them quarterly. Side effects were monitored, and all illnesses reported (gastroenteritis, respiratory, allergies, etc.) were recorded as adverse events. Any hospitalizations, surgery, or other significant events were recorded as serious adverse events and reported to the institutional review boards and data safety management board.

Statistical analysis

The primary endpoint of this trial was the change in predicted adult height between baseline and the end of therapy. The primary endpoint was compared between the treatment groups using two-sample t test. The analyses were not adjusted for the effect of trial center or for any interaction between covariates and the treatment group. Secondary endpoints included changes in bone age, growth velocity, height SDS adjusted for bone age, and changes in hormone concentrations. Tolerability and safety endpoints included adverse events, withdrawals, and laboratory data including glucose, lipid panel, and liver profile concentrations. By study design, data were unblinded to one single individual and analyzed after all available subjects completed 24 months of treatment. Neither the investigators nor participating subjects had any knowledge of the treatment arms. The protocol was continued in all available subjects through 36 months. Afterward, whenever possible, subjects are being followed and measured at each site until completion of near adult height. Those measurements are ongoing.

Subjects were analyzed on the basis of the treatment received and those without post-baseline measurements were excluded from the analyses of that variable. The analyses of endpoints at specific time points during the study included only patients with measurements available at that time point (and baseline, for change from baseline variables). No adjustments were made to any of the analyses for multiple testing, because some of the endpoints may be relatively well correlated with one another. The results of the analyses are presented using the simple means for the anastrozole and placebo groups along with the corresponding sem and the two-sided t test P value for treatment comparison. The analysis was carried out using SAS software. Significance was established as P < 0.05.

Results

Fifty-two subjects were recruited for participation, with clinical characteristics shown in Table 1. All but one of the subjects had isolated idiopathic GH deficiency; one had GH and TSH deficiency, the latter well replaced. Of the 52 subjects, 50 completed 12 months, 41 completed 24 months, and 28 completed 36 months.

Table 1.

Clinical characteristics of study subjects at baseline

	Anastrozole	Placebo
n	26	26
Chronological age (yr)	13.8 ± 0.3	14.2 ± 0.2
Bone age (yr)	13.7 ± 0.2	13.4 ± 0.2
Height (cm)	149.7 ± 1.6	151.6 ± 1.3
Height SDS	−1.4 ± 0.2	−1.5 ± 0.2
Body mass index (kg/m2)	20.0 ± 0.6	20.4 ± 0.8
Genital stage	II–IV	II–IV
Max stimulated GH (ng/ml)	6.4 ± 0.5	5.5 ± 0.5
Previous GH (yr)	3.6 ± 0.6	2.6 ± 0.5
Midparental height (cm)	169.8 ± 1.6	173.1 ± 1.2

P values were not significant between the groups. 

Pubertal and bone hormones (Fig. 1)

Figure 1.

Changes in testosterone, estradiol, estrone, IGF-I, and osteocalcin concentrations vs. baseline in the anastrozole (AN, solid lines) vs. placebo (PL, broken lines) groups. *, P = 0.01 for testosterone; *, P < 0.001 for all estradiol and estrone comparisons. To convert to SI units, multiply by 0.03467 for testosterone (ng/dl to nmol/liter), by 3.671 for estradiol (pg/ml to pmol/liter), by 0.131 for IGF-I (ng/ml to nmol/liter), and by 1.171 for osteocalcin (ng/ml to nmol/liter).

Serum testosterone concentrations increased during the 3 yr in both groups with the analyte concentrations for the whole groups at 0, 12, 24, and 36 months: anastrozole, 215 ± 35, 667 ± 71, 628 ± 71, and 510 ± 63 ng/dl; placebo, 162 ± 30, 416 ± 51, 570 ± 57, and 338 ± 31 ng/dl, respectively (P = 0.01 between groups at 12 months). When changes from baseline were analyzed, there was a greater increase in testosterone at 12 months in the anastrozole than the placebo group (+452 ± 67 vs. 254 ± 39 ng/dl, P = 0.01); however, this difference was not significant at 24 months (anastrozole, +445 ± 73 ng/dl; placebo, +450 ± 57 ng/dl, P = 0.74) or at 36 months (anastrozole,+325 ± 54 ng/dl; placebo, 257 ± 24 ng/dl, P = 0.25) (Fig. 1).

Estradiol concentrations were stable in the anastrozole group, whereas they increased in the placebo group at 0, 12, 24, and 36 months: anastrozole, 9.6 ± 1.1, 6.7 ± 0.9, 6.6 ± 1.0, and 8.1 ± 2.0 pg/ml; placebo, 7.5 ± 1.2, 11.0 ± 1.3, 16.2 ± 1.1, and 16.8 ± 1.7 pg/ml (P < 0.01 between groups at all time points except baseline, not significant). Estrone concentrations followed a similar pattern at 0, 12, 24, and 36 months: anastrozole, 14.0 ± 1.5, 4.7 ± 1.1, 6.0 ± 1.2, and 8.1 ± 2.7 pg/ml; placebo, 12.2 ± 1.4, 14.2 ± 1.7, 18.1 ± 1.5, and 18.5 ± 1.3 pg/ml (P < 0.01 between groups at all time points except baseline, not significant). Analysis of changes in plasma estradiol concentrations vs. baseline showed decreased levels in the anastrozole vs. placebo group at 12 months (anastrozole, −2.8 ± 1.1 pg/ml vs. placebo, +4.1 ± 0.8 pg/ml, P = 0.0001), at 24 months (anastrozole, −1.9 ± 1.5 pg/ml vs. placebo, +9.5 ± 1.4 pg/ml, P < 10⁻⁵), and at 36 months (anastrozole. −0.2 ± 2.2 pg/ml vs. placebo, +12.5 ± 1.5 pg/ml, P = 0.0002. Differences in estrone concentrations vs. baseline followed the same trends of the estradiol concentrations at 12 months (anastrozole, −9.1 ± 1.8 pg/ml vs. placebo, +3.7 ± 1.4 pg/ml, P = 1.2 x10⁻⁵), at 24 months (anastrozole, −5.3 ± 2.3 pg/ml vs. placebo, +6.6 ± 1.6 pg/ml, P = 1.7 x10⁻⁵), and at 36 months (anastrozole, −4.3 ± 2.6 pg/ml vs. placebo, +8.3 ± 1.9 pg/ml, P = 0.0001) (Fig. 1).

IGF-I concentrations were similar at baseline and increased comparably in both groups throughout the study at 0, 12, 24, and 36 months: anastrozole, 737 ± 64, 935 ± 79, 986 ± 107, and 879 ± 112 ng/dl; placebo, 812 ± 56, 1020 ± 80, 1059 ± 93 933 ± 76 ng/dl (P value not significant). There was no difference in the increase from baseline between the groups (Fig. 1). The total group concentrations of osteocalcin were remarkably stable during the entire study period at 0, 12, 24, and 36 months: anastrozole, 54 ± 2, 55 ± 2, 53 ± 2, and 52 ± 4 ng/ml; placebo, 58 ± 3, 51 ± 2, 51 ± 3, 46 ± 3 ng/ml (P value not significant). There was no difference in changes in osteocalcin concentrations vs. baseline between the groups (Fig. 1).

Lipid and glucose concentrations (Table 2)

Table 2.

Changes in fasting lipids and glucose concentrations vs. baseline

	Anastrozole					Placebo
	12 months	18 months	24 months	30 months	36 months	12 months	18 months	24 months	30 months	36 months
Cholesterol (mg/dl)	−18 ± 4a	−5 ± 13	−16 ± 4	−20 ± 5	−14 ± 6	−1 ± 4	−10 ± 4	−6 ± 5	−8 ± 6	−11 ± 4
LDL-cholesterol (mg/dl)	−13 ± 5	−12 ± 5	−11 ± 5	−12 ± 6	−9 ± 7	−4 ± 4	−10 ± 5	−5 ± 5	−8 ± 6	−8 ± 5
HDL-cholesterol (mg/dl)	−6 ± 2b	−6 ± 2	−6 ± 2	−7 ± 3	−4 ± 3	−2 ± 2	−4 ± 1	−3 ± 2	−1 ± 2	−4 ± 2
Triglycerides (mg/dl)	7 ± 13	26 ± 23	5 ± 17	−4 ± 13	0 ± 9	20 ± 12	14 ± 16	0 ± 17	9 ± 11	13 ± 18
Glucose (mg/dl)	2 ± 2	1 ± 4	5 ± 4	3 ± 3	−10 ± 8	−1 ± 4	−2 ± 2	−3 ± 4	−3 ± 5	−10 ± 5

To convert to SI units, multiply by 0.02586 for cholesterol, high-density lipoprotein (HDL)-cholesterol, and low-density lipoprotein (LDL)-cholesterol (mg/dl to mmol/liter); by 0.0113 for triglycerides (mg/dl to mmol/liter); and by 0.05551 for glucose (mg/dl to mmol/liter). 

^aP < 0.002. 

^bP < 0.05. 

Plasma lipids and fasting glucose concentrations were normal with no significant differences in either the anastrozole or the placebo group over time in analyte concentrations in the total cohort (data not shown). When comparing the changes at each time point vs. baseline, there were only differences between the two groups at 12 months, with a statistically greater drop in high-density lipoprotein cholesterol in the anastrozole group at 12 months compared with the placebo group (−18 ± 4 vs.−1 ± 4 mg/dl, P < 0.002) but a greater drop in total cholesterol also in the anastrozole group compared with the placebo group (−6 ± 2 vs.−2 ± 2 mg/dl, P < 0.05). There were no differences in the changes in these concentrations compared with baseline between the two groups at any other time point.

Anthropometry and body composition (Table 3)

Table 3.

Anthropometric, body composition, and bone mineral density changes vs. baseline

	Anastrozole			Placebo
	12 months	24 months	36 months	12 months	24 months	36 months
Growth velocity (cm/yr)	−0.8 ± 0.8	−3.0 ± 1.0	−4.1 ± 1.3a	−2.3 ± 0.6	−4.8 ± 1.0	−7.6 ± 1.0
Body mass index (kg/cm2)	1.1 ± 0.3	2.2 ± 0.2	3.1 ± 0.3	0.6 ± 0.2	1.9 ± 0.3	3.7 ± 0.4
Fat-free mass (kg)		12.4 ± 0.6	17.7 ± 1.8		13.2 ± 1.1	18.9 ± 1.9
% fat mass		−2.9 ± 0.8	−3.0 ± 1.5		−1.7 ± 1.1	−0.4 ± 1.5
BMD lumbar Z score		0.28 ± 0.12b	0.65 ± 0.28		0.80 ± 0.17	1.09 ± 0.23

A subset of subjects had DXA data available: at baseline, n = 39; at 24 months, n = 28; and at 36 months, n = 14. 

^aP = 0.04 vs. placebo. 

^bP = 0.02 vs. placebo. 

There were no significant differences in height between the two groups as a whole throughout the study period at 0, 12, 24, and 36 months: anastrozole, 149.7 ± 1.6, 157.7 ± 1.5, 162.9 ± 1.4, and 165.8 ± 1.3 cm; placebo, 151.9 ± 1.2, 160.6 ± 1.3, 166.6 ± 1.4, and 167.8 ± 1.8 cm (P value not significant). Analysis of the changes in growth parameters vs. baseline showed that growth velocity, as expected, decreased in both groups over time; however, the decrease vs. baseline was more significant in the placebo group at 36 months (P = 0.04). Changes in fat-free mass and percent fat mass from baseline were not different between the groups (Table 3). There were no differences in BMI vs. baseline between groups.

Bone age and predicted adult height

The tempo of bone age acceleration was significantly different between the groups (anastrozole, 13.7 ± 0.2 (baseline), 14.7 ± 0.2 (12 months), 15.4 ± 0.2 (24 months), and 15.9 ± 0.3 yr (36 months); placebo, 13.4 ± 0.2 (baseline), 14.8 ± 0.2 (12 months), 16.0 ± 0.2 (24 months), and 17.2 ± 0.3 yr (36 months). When the same subjects were compared at the same time points, there were significant differences between groups apparent after 24 months: anastrozole vs. placebo at 12 months, +1.1 ± 0.1 vs. +1.4 ± 0.1 yr, P = 0.08; at 24 months, + 1.8 ± 0.1 vs. +2.7 ± 0.1 yr, P < 0.0001; and at 36 months, +2.5 ± 0.2 vs. +4.1 ± 0.1 yr, P < 0.0001 (Fig. 2). There was a corresponding improvement in the gain in height sd score (SDS) adjusted for bone age at 24 and 36 months in the anastrozole group compared with the placebo group with net gains at 12 months of +0.22 ± 0.07 vs. +0.03 ± 0.13 SDS (P = 0.2); at 24 months, +0.52 ± 0.12 vs. +0.04 ± 0.13 SDS (P = 0.009); and at 36 months, +0.77 ± 0.16 vs.−0.10 ± 0.16 SDS (P = 0.0009) in the anastrozole and placebo groups, respectively (Fig. 3). These differences translated into greater gains in predicted adult height from baseline in the anastrozole group compared with the placebo group in paired analysis: anastrozole vs. placebo: at 12 months, +1.3 ± 0.7 vs. +0.3 ± 1.0 cm (P = 0.4); at 24 months, +4.5 ± 1.2 vs. +1.1 ± 1.1 cm (P = 0.04); and at 36 months, +6.7 ± 1.4 vs. +1.0 ± 1.1 cm (P = 0.004) (Fig. 4). For the primary, intent-to-treat analysis of change in predicted adult height at the end of therapy (i.e. including the last measurements for a subject withdrawing early), the P value for treatment comparison was 0.03.

Figure 2.

Changes in bone age compared with baseline at 12, 24, and 36 months in the anastrozole and placebo groups. Insets indicate mean change from baseline. *, P value is significant vs. placebo group at 24 and 36 months (P < 0.001 both times).

Figure 3.

Change from baseline in SDS adjusted for bone age in the anastroloze and placebo groups at 12, 24, and 36 months. *, P value is significant vs. placebo group at 24 (P = 0.009) and 36 months (P = 0.0009).

Figure 4.

Changes in predicted adult height based on Bayley Pinneau tables compared with baseline in the anastrozole and placebo groups. *, P value is significant vs. placebo group at 24 (P = 0.04) and 36 months (P = 0.004).

Pubertal changes

The pace of pubertal progression was similar between the groups as measured by changes in testosterone concentrations and testicular volumes. On average, right and left testes increased in volume from about 10–25 ml bilaterally in both groups. The anastrozole and placebo groups both were reported at Tanner stage V genital and pubic hair at the study completion.

BMD (Table 3)

DXA analysis was not a principal aim of the study, and it was performed mostly as a safety measure; hence, subjects were not denied entry into the study because of lack of available DXA. We had DXAs on 39 of 52 subjects available at baseline, 28 of 41 at 24 months, and 14 of 28 at 36 months. There were no significant differences between the groups in lumbar spine BMD (data not shown). Paired analysis of the available subjects showed there was a statistical difference in lumbar spine BMD Z score between 24 months and baseline (P = 0.02), whereas there was no significant differences between the groups in gain in Z score compared with baseline at 36 months (P = 0.2).

Safety

Anastrozole was well tolerated and largely free of side effects. There were no differences between the groups on complete blood counts, urinalysis, or liver profiles. The frequency for adverse events (head and neck, respiratory, gastrointestinal, genitourinary, and musculoskeletal) was similar in both groups. There were seven serious adverse events; four were in the same subject who had repeated episodes of cyclical vomiting, diagnosed after extensive workup in gastroenterology as psychogenic emesis. Another fractured an arm after severe sports trauma, one had an elective forearm bone surgery from previous trauma before the study, and the last one had a septic arthritis from a presumed focus of sinus infection. All reported serious adverse events were in the anastrozole group, none were considered related to the study medication by the investigators, and in all, anastrozole was quickly resumed after hospital discharge.

Long-term follow up

Nine subjects withdrew from the study because of either satisfaction with their achieved height or psychosocial issues at home or school, or they got tired of taking medications. Five subjects were discontinued by the investigators because of poor drug compliance or were lost to follow-up. Whenever possible, subjects are being followed to measure their final adult height; these measurements are ongoing.

Discussion

One of the challenging issues in dealing with short children in puberty is the limited time available to increase linear growth during concomitant and rapid epiphyseal fusion. The use of a potent aromatase inhibitor, anastrozole, resulted in a significant delay in the tempo of bone age acceleration compared with placebo in adolescent boys with GH deficiency who were also treated with GH. This slowing of epiphyseal fusion caused by estrogen blockade resulted in a significant net gain in predicted adult height, as calculated based on bone age, of +4.5 ± 1.2 cm and +6.7 ± 1.4 cm after 2 and 3 yr, respectively, compared with +1 cm at both time points in the placebo group. This translated into a net gain from baseline in height SDS adjusted for bone age in the anastrozole vs. the placebo group. We intentionally targeted an age range in which a natural deceleration of growth occurs after peak growth velocity. The drop in growth velocity during the course of these studies was also greater in the placebo group than in the anastrozole group at 36 months.

The effects of estrogen blockade are remarkable and cannot be fully appreciated after only 1 yr of treatment but are particularly evident at 24 months and in those that continued through 36 months. This observed benefit increasing height potential takes at least 2 yr. Comparable results were observed in boys with idiopathic short stature treated for 2 yr with letrozole, another aromatase inhibitor (17). The findings are also similar to the increase in predicted adult height observed after coadministration of the GnRH analog and GH (2) or GH in high doses in adolescents reported previously (1) in which positive results were apparent only after the second year of treatment. This observation that gains in adult height potential take up to 3 yr of treatment to be observed was also shown in low-birth-weight girls treated with the insulin sensitizer metformin reported by Ibáñez et al. (22). The changes in predicted adult height reported here were accompanied by normal progression of puberty, with normal virilization and testicular enlargement, not different from placebo. Testosterone concentrations increased initially more in the anastrozole group than in the placebo group at 12 months but not after continued treatment at 24 and 36 months. The ability of these adolescent boys to continue to virilize while slowing down epiphyseal fusion offers a potentially better alternative to the complete biochemical suppression of the gonadal axis caused by treatment with GnRH analogs.

Bayley Pinneau was the height prediction method used, but all current methods of height prediction can either under- or overestimate ultimate height because the accuracy depends on height gained during the pubertal growth spurt, which is highly variable. It seems clear that there is no best method and that each has its advantages and disadvantages under specific circumstances (23,24,25,26,27,28). In a metaanalysis of 10 studies in children with delayed growth followed to final adult height, the prediction error for the Bayley Pinneau method was approximately +2.4 cm (28), lower than other methods (24). The method used throughout this study was the same for all patients in both groups, using one single observer at the Fels Institute, making comparisons still robust. A recent mathematical model has shown that epiphyseal maturation and sex steroids may be more important determinants of pubertal growth spurt than GH dose (29). Hence, we believe that the effect of aromatase blockade was clinically significant, independent of GH, because the GH dose was controlled in the placebo group.

Plasma IGF-I concentrations were similar between the two groups throughout the study, in contrast to the decrease we and others observed after aromatase blockade in males (13,16,17). This is likely related to coadministration of GH in our experiments. Combining GH and anastrozole treatment might allow for normal pubertal height gain without having to increase the GH dose.

There is at present no consensus on the best estradiol assay to use, and often commercially available assays are not sensitive enough to pick up differences after given interventions. However, using a sensitive LC-MS/MS assay, we observed a significant decline in estradiol and estrone concentrations in the adolescents treated with anastrozole, whereas the placebo group continued to increase their estradiol and estrone levels as they progressed through puberty. These changes are not as marked as the 50% decline in estradiol concentrations after 1 yr of treatment in boys reported by us previously (30) using a highly sensitive recombinant cell bioassay (31). However, the results reported here show marked differences in the concentration changes over time between the two groups, reflecting the significant aromatase blockade caused by the anastrozole. Regardless of plasma estradiol concentrations, suppressing tissue aromatase at the epiphyseal growth plate is most important, and anastrozole can achieve more than 98% tissue aromatase blockade (32,33). When following these patients, measuring estradiol concentrations is only useful if the assay has a very low sensitivity and is properly validated, which most commercial assays are not. The clinician needs to know the sensitivity and limits of the assay used.

Our results on body composition cannot be considered definitive due to the combination of methods used, however they show interesting trends, with comparable increases in fat-free mass accrual and reciprocal decreases in % adiposity. This differs to the observed decrease in lean body mass and increase in % fat mass in healthy young males studied with a GnRH analog (14). It is possible that the lack of difference in body composition between the 2 groups studied here is due in part to the continued use of GH, a potent enhancer of lean body mass accrual in youth (34).

The increase in lumbar BMD was less in the anastrozole group at 24 months but not at 36 months. We acknowledge that correction for body size or height might offer a better estimate of the status of bone health (35,36) and that typically short subjects show even fewer bone abnormalities when areal BMD is corrected for size (37). However, there is presently no adequate normative data for volumetric BMD to compare against in children and no consensus on standards for adjusting BMD for size (38). Osteocalcin concentrations, a measure of bone formation, were similar at all time points between the groups. This contrasts with the significant decrease in measures of bone formation observed in GnRH analog-treated young men (15). Although human estrogen receptor and aromatase knockouts had osteopenia (9,11), when estrogen suppression was short-term, BMD was normal in letrozole-treated boys followed for 2 yr, reported previously (17). Our limited DXA data suggest that for up to 3 yr, timed aromatase blockade given with GH therapy is not detrimental to bone health. However, longitudinal follow-up is still needed.

Most of the boys studied had been on GH an average of about 3 yr before study entry and represent a typical group of GH-treated GH-deficient adolescents in the United States. We intentionally had no cutoff criteria for IGF-I concentrations at the original diagnosis because the variability in IGF-I assays is significant (39,40,41,42). It also depends on nutritional status, making it less useful for diagnosis unless values are severely low. We chose instead to include GH-deficient patients with peak GH responses to two pharmacological stimuli less than 10 ng/ml, a common (although arbitrary) cutoff used to identify patients with GH deficiency. Whether use of aromatase inhibitors in children with even more profound GH deficiency would be beneficial in increasing height potential in puberty awaits further study.

Extensive 3-yr safety data in these boys showed no differences in lipid or glucose concentrations, liver function tests, or adverse event rates between the groups. Anastrozole was well tolerated and safe for up to 3 yr. However, long-term data are still needed and continued surveillance a must as well as careful monitoring of BMD accrual.

Aromatase inhibitors are a class of compounds presently approved for treatment of postmenopausal women with early or metastatic hormone receptor-positive breast cancer. However, this class of drugs has been clinically tested in a variety of other experimental situations to selectively decrease estrogen (43,44,45,46,47,48). Previous studies in boys with constitutional growth delay treated with letrozole and followed to adult height have shown that the reported increase in adult height potential was sustained after drug discontinuation (16,49). Whether or not the observed increase in predicted adult height in our present study in GH-deficient adolescent boys will also translate into taller adult height for the subjects requires long-term follow-up.

In summary, 2–3 yr of aromatase blockade with anastrozole in GH-treated boys caused a significant slowing of the tempo of bone age maturation, resulting in a significant gain in predicted adult height compared with placebo. This was accompanied by normal tempo of virilization, pubertal hormones, IGF-I concentrations, and lipid concentrations, with a strong safety profile. Anastrozole treatment in GH-treated males offers promise and may be a useful choice in the approach to the growth-retarded male in puberty. Long-term follow-up to final adult height is still needed to fully characterize the safety and efficacy of this approach.