Prospective Study Confirms Oxandrolone-Associated Improvement in Height in Growth Hormone-Treated Adolescent Girls with Turner Syndrome

Abstract

Background/Aims

Untreated girls with Turner syndrome (TS) have growth failure, and adult height is, on average, 20 cm less than predicted height. Treatment with growth hormone (GH) is now standard of care. The objective of this study was to investigate the benefit of adding oxandrolone (Ox) to GH in a long-term, randomized, placebo (Pl)-controlled prospective trial to near adult height in TS.

Methods

Prospective, randomized, Pl-controlled study: 76 girls with TS (ages 10–14.9 years) were randomized to receive Ox (0.06 mg/kg/day) or Pl in combination with GH (0.35 mg/kg/week, daily) over 2 years. Auxologic data, breast and pubic hair Tanner stages, and hormone and lipid levels were measured. Subjects who chose to continue were followed in a 2-year double-blind extension, also received estrogen therapy (years 3, 4), and had dual-energy X-ray absorptiometry evaluation of bone density (years 3, 4).

Results

At year 4, the change in absolute height and height SDS was greater in the GH/Ox versus GH/Pl group [26.2 ± 6.7 vs. 22.2 ± 5.1 cm, analysis of covariance (ANCOVA) p < 0.001; 1.8 ± 0.9 vs. 1.2 ± 0.7 standard deviation scores, ANCOVA p < 0.001]. Bone mineral density (BMD) of the wrist (0.51 ± 0.17 vs. 0.54 ± 0.05 g/cm²) and spine (0.91 ± 0.34 vs. 0.96 ± 0.13 g/cm²) in the GH/Ox versus GH/Pl groups was similar after 4 years. Breast development was slower in the GH/Ox versus GH/Pl group [year 4: Tanner stage 2.9 ±1.3 (Ox) vs. 4.1 ± 1.3 (Pl), p = 0.003], and menarche was approximately 1 year later.

Conclusions

The addition of Ox to GH at mean age 12.0 ± 1.7 year augmented height gain after 4 years of treatment, slowed breast development and did not affect BMD in girls with TS. Whether initiation of Ox prior to initiation of pubertal development would optimize height gain without impeding breast development will require further study.

Introduction

Turner syndrome (TS), which occurs in 1/2,500 live-born females, is due to complete or partial absence of an X chromosome. The phenotype, described by Henry Turner in 1938, commonly includes short stature, ovarian failure and cardiac and renal abnormalities. Untreated women with TS, on average, attain adult height 20 cm less than their predicted target height (mean final height of 143 cm, or 4 ft 8 in) [1]. Growth failure occurs throughout childhood and is related both to haploinsufficiency of the X chromosome homeobox-containing SHOX gene and to the absence of a pubertal growth spurt.

Both growth hormone (GH) and oxandrolone (Ox) treatment have been used separately to stimulate growth in girls with TS [2,3,4,5,6,7,8]. The combination of GH and Ox improved height more than GH alone in several large, long-term prospective TS studies, most notably those of Menke et al. [9] (2010), Nilsson et al. [10] (1996), Rosenfeld et al. [11] (1998) and Stahnke et al. [12] (2002). This additive effect on height, however, was not found in several other studies [13,14,15]. GH was approved for treatment of short stature in TS in 1997. While recent clinical practice guidelines recommend that addition of Ox to GH therapy be considered in TS if the child is extremely short [4], a strong evidence-based recommendation had not yet been possible until the recent 2010 publication of Menke et al. [9].

Whether Ox treatment increases bone mineral density (BMD) is also relevant to girls with TS. During adolescence, girls with TS have decreased BMD [16] and an increased fracture rate, compared to peers (9:1,000 vs. 3.5:1,000) [16,17], which is likely a consequence of estrogen deficiency and ovarian failure. Androgen treatment improved BMD in postmenopausal women [18]. Therefore, we wished to determine whether treatment with Ox would improve BMD in GH-treated girls with TS.

Our hypothesis was that addition of Ox to GH would increase height velocity, height gain, and BMD in girls with TS. This is the first placebo (Pl)-controlled, double-blind study to investigate the benefit of 4 years of combined Ox and GH treatment compared to Pl and GH (plus estrogen in both groups in the final 2 years), in girls with TS.

Subjects and Methods

Subjects

Inclusion criteria included karyotype diagnosis of TS, no prior treatment with estrogen or androgen, no prior treatment with GH exceeding 12 months or in the preceding 3 months, chronological age of 10.0–14.9 years, and bone age ≤12 years. Exclusion criteria included a Y chromosome component in the peripheral karyotype.

Methods

This 4-year study (2 years with a 2-year extension) enrolled patients from 1989 to 1999 at Thomas Jefferson University and at the Developmental Endocrinology Branch of the National Institute of Child Health and Human Development. The study was approved by Human Studies Committees at both institutions and informed consent and assent were obtained in all cases. The clinical trial was registered with the NIH (NCT00029159). A Data and Safety Monitoring Board reviewed the study yearly from 1995 to 2003 and recommended study continuation until all subjects had completed the trial. Subjects were randomized (by the NIH Clinical Center Research Pharmacy) at baseline to receive daily treatment for 2 years with either (1) the oral nonaromatizable androgen, Ox (0.06 mg/kg p.o., daily) (maximum dose of 3.75 mg) or (2) Pl, both supplied by Bio-Technology General Corporation, Iselin, N.J., USA (fig. 1). Subjects were asked to record daily treatment in a patient diary. All study participants also received GH (Humatrope, donated by Eli Lilly & Co., Indianapolis, Ind., USA, 0.05 mg/kg, s.q., daily). After the initial 2 years of the clinical trial, the investigators received Institutional Review Board approval to continue the trial for an additional 2 years (extension) in order to gain longer-term height data and, additionally, to study the impact of 4 years of Ox on BMD. Therefore, after completing the first 2 years of the study, participants were given the option of continuing the study in the same treatment arm for another 2 years. Subjects and investigators remained blinded to treatment group. During the 2-year extension as part of the protocol, subjects received estrogen replacement as ethinyl estradiol, 50 ng/kg/day during year 3 and 100 ng/kg/day, during year 4. The estradiol dosage was approximately 2 µg/day during year 3 (for a 40-kg subject) and 4 µg/day during year 4.

Fig. 1.

Enrollment, randomization and drop-outs from baseline to 2 years and from the beginning of the extension (in gray box) to 4 years.

Study participants were monitored at baseline and every 6 months by obtaining a history and performing physical examination, measurement of height, weight, waist and hip circumference, breast and pubic hair Tanner staging (determined by a physician blinded to treatment group), and hormone measurements [thyroxine, thyroid-stimulating hormone, insulin, luteinizing hormone, follicle-stimulating hormone (FSH), estradiol, free testosterone, insulin-like-growth factor 1 (IGF-1) and insulin-like growth factor binding protein 3 (IGFBP-3)]. Bone ages (by X-ray) were assessed by the method of Greulich and Pyle [19] and were read by a single-blinded investigator (J.L.R.). Mid-parental height was determined using the formula [maternal height (cm) + paternal height (cm) − 13]/2. Predicted adult height was determined using the Bayley-Pinneau method [19]. Near-adult height was defined as height obtained when bone age was ≥13.5 years.

Bone mineral content (g/cm²) of the wrist and spine was measured annually by dual-energy X-ray absorptiometry (using both Lunar and Hologic systems) at the beginning of the extension (year 3) and at the end of years 3 and 4. For the spine, standard deviation scores (SDSs) were calculated based on the particular technology used and the available normative data for Caucasian females [20]. The technology was the same for individuals over the 2 years, and the individual difference from the end of year 2 to the end of year 4 was the outcome that was measured.

Patients were evaluated for adverse effects at each visit. Criteria for Ox/Pl dose reduction were any sign of excess androgen exposure (clitoral enlargement, hirsutism, or moderate acne), bone age advancement >12 months/year, abnormal liver function test (alanine aminotransferase (ALT), also known as [serum glutamic pyruvic transaminase (alanine aminotransferase) >100 IU/l], or total cholesterol >250 mg/dl. If these adverse effects were observed, the dosage of Ox or Pl was reduced by 50% and the patient was re-evaluated at the next study visit, 6 months later. If there was resolution of the adverse effect at the next visit, the reduced dose was maintained. Criteria for estradiol dose reduction were vaginal bleeding, bone age advancement (>12 months/year) or rapid breast development. If there was resolution of the adverse estrogen effect at the next visit, the reduced dose was maintained.

Compliance with Ox/Pl was estimated at each patient visit from counts of returned capsules, and confirmed by examination of patient diaries. The percent compliance with the prescribed Ox/Pl doses was calculated as: (number of capsules dispensed minus number of capsules returned)/(number of days from last visit to current visit).

Assays

Liver function (serum glutamic pyruvic transaminase), alkaline phosphatase, thyroxine, thyroid-stimulating hormone, insulin, total cholesterol, HDL cholesterol, triglycerides and glucose were measured using commercial assays. Luteinizing hormone, FSH, estradiol, free testosterone IGF-1, and IGFBP-3 were determined by Esoterix using RIA.

Statistical Analysis

The analyses compared GH/Ox versus GH/Pl-treated subjects at baseline and at years 1, 2, 3, and 4 by t tests. Height standard deviation scores were calculated based on US National Center for Health Statistics standards [21] and according to published TS-specific standards [1]. Results are presented as mean ± standard deviation. Analysis of covariance (ANCOVA) for repeated measures was performed for the change in height from baseline to year 2, and from baseline to year 4 of the extension, adjusting for baseline Tanner stage, karyotype, and chronologic age. ANCOVA for repeated measures was also performed for progression of breast and pubic hair Tanner stage in both groups, with covariates of age, karyotype, body mass index (BMI), estradiol, and testosterone. Two-tailed p values ≤0.05 were considered significant. Sample size was estimated for height velocity based on the data of Rosenfeld et al. [11]: 60 patients (30 per group) yielded a >95% probability to detect a statistically significant difference in height velocity between the GH/Ox and GH/Pl groups at year 2.

Results

Subjects (fig. 1)

Seventy-six TS subjects (ages 10.0–14.9 years at entry) were enrolled in the study and randomized (40 GH/Ox and 36 GH/Pl). The distribution of karyotypes was: 45,X, 50%; mosaic (45,X/46,XX, 45,X/46,X isochromosome X, or 45,X/46,X ring X), 41%; and nonmosaic partial deletion of one X chromosome (Xp or Xq deletion), 9%. None had previously received GH or estrogen. Of the 76 girls randomized, 70 completed the initial 2 years of the study (fig. 1). After the 2-year initial study, 35 Ox and 31 Pl subjects (66) chose to continue in the 2-year double-blind (Ox or Pl and GH plus estrogen) extension study (2 GH/Ox and 2 GH/Pl subjects did not continue). Forty-seven subjects completed the study and were evaluated at the end of year 4. Most of the study discontinuations occurred between years 3 and 4 of the study. Of the 18 patients who dropped out during this period, the only significant difference found at year 3 between those who later dropped out and those who continued was the study site (67% were from the NIH site, p < 0.01). There was no difference in age (p = 0.5), bone age (p = 0.08), height (p = 0.3), change in height from baseline (p = 0.2), or height SDS. Patients generally traveled from greater distances to the NIH site versus the Thomas Jefferson University site.

Compliance with Ox/Pl at 2 and 4 years (the end of the extension) was similar in the Ox and Pl groups: 92 ± 11 versus 88 ± 16% (year 2) and 85 ± 21 versus 85 ± 18% (year 4).

Evidence of spontaneous puberty (greater than Tanner 1 breast development) was noted in 8/40 girls in the GH/Ox group (6 Tanner 2, 1 Tanner 3, 1 Tanner 4) and 15/36 in the GH/Pl group (13 Tanner 2, 2 Tanner 3) at visit 1 (baseline). Estrogen replacement was administered per the protocol in 27/35 GH/Ox and 25/31 GH/Pl at year 3, the beginning of the extension period. Of the 8 in the GH/Ox group and 6 in the GH/Pl group who did not start ethinyl estradiol at the beginning of the extension, 6 in the GH/Ox group and 3 in the GH/Pl group started estrogen per this protocol within 6 months to 1 year. One girl in the GH/Ox group and 2 girls in the GH/Pl group received no supplemental estrogen because of ongoing spontaneous puberty. Reasons for not receiving estrogen per protocol were continued spontaneous puberty in 3 (normal gonadotropin levels: 1 Ox, 2 Pl), personal preference in 9 (7 Ox, 2 Pl), and advanced bone age in 2 girls in the Pl group. Estrogen dose reductions occurred at year 3 in 2 GH/Ox and 1 GH/Pl subjects, and at year 4 in 1 GH/Ox and 1 GH/Pl patient because of rapid breast development or bone age advancement.

Baseline Physical Findings

Baseline height, height SDS, waist/hip ratio, breast Tanner stage and bone age were statistically similar in the two groups. BMI SDS at baseline was slightly higher in the GH/Pl group.

Height Gain during Study (fig. 2a, b, online suppl. table 1)

Fig. 2.

a Mean height SDS in GH/Ox and GH/Pl annually from baseline to year 4. b Delta height SDS in GH/Ox and GH/Pl from baseline to years 1–4. c Breast Tanner stage in GH/Ox and GH/Pl from baseline to year 4.

There was a small but sustained increase in absolute height and change in height from baseline in the GH/Ox versus the GH/Pl group. After 2 years, the changes in height and height SDS from baseline were trending toward an increase in the GH/Ox versus GH/Pl groups (17.0 ± 3.5 vs. 15.5 ± 2.8 cm, p = 0.06; 1.1 ± 0.8 vs. 0.9 ± 0.7 SDS, p = 0.06). For the 47 subjects who were evaluated at year 4, the changes in height and height SDS from baseline were significantly greater by ANCOVA in the GH/Ox versus GH/Pl groups (26.2 ± 6.7 vs. 22.2 ± 5.1 cm, p < 0.0001; 1.8 ± 0.9 vs. 1.2 ± 0.7 SDS, p < 0.0001). The change in height SDS from baseline was also significantly greater when compared to normative data for TS (2.3 ± 0.9 vs. 1.6 ± 0.6, p < 0.001) [1]. At year 4, 21/24 in the GH/Ox group and 20/23 in the GH/Pl group had reached near adult height (bone age of ≥13.5 years). For those who had reached near adult height, the change in height from baseline was nearly significant, with those having received Ox growing on average 4 cm more [25.4 ± 6.7 (Ox) vs. 21.8 ± 5.3 cm (Pl), p = 0.07 (t test)]. Mean bone age did not statistically differ between Ox and Pl groups at any point in the study. Mean final bone age was 14.5 ± 1.3 years in the GH/Ox group and 14.1 ± 0.9 years in the GH/Pl group.

Significant covariates in the analysis of height gain at years 2 and 4 for the two treatment groups were baseline bone age and chronologic age (both p < 0.01, younger bone age and chronologic age were associated with greater height gain), but not karyotype. Baseline breast Tanner stage was a significant covariate associated with more height gain at year 2 but not at year 4.

The difference between mean patient height and mid-parental height at year 4 was 7.3 ± 6.7 versus 13.9 ± 5.3 cm for the GH/Ox and GH/Pl groups (p < 0.01), respectively, indicating that girls who received GH/Ox achieved a height closer to their genetic potential. Mean predicted adult height by the Bayley-Pinneau method was similar between the groups at baseline [145.8 ± 6.1 (GH/Ox) vs. 146.8 ± 5.6 cm (GH/Pl), p = 0.5], but the predicted height from baseline to visit 9 increased more in the GH/Ox group [9.9 ± 5.8 (GH/Ox) vs. 6.9 ± 3.3 cm (GH/Pl), p = 0.03].

Weight, BMI, Body Proportions

Weight SDS and change in weight SDS from baseline did not differ between groups throughout the study (online suppl. table 1; for all online suppl. material see www.karger.com?doi=10.1159/000317529). The GH/Pl group had a slightly greater BMI SDS only at baseline, but there was no effect of Ox on BMI thereafter. Waist:hip ratio was similar in the two groups throughout the study.

Breast Tanner Stage (fig. 2c, 3a–c)

Fig. 3.

a Breast development (Tanner stage) by visit in GH/Ox group. Box-and-whiskers plot: top and bottom of box represent 25% and 75%, dark line is median. b Breast development (Tanner stage) by visit in GH/Pl group. c Box-and-whisker plot of breast Tanner at final visit (visit 9) by karyotype and treatment group.

At baseline, Tanner stage breast development was similar in the GH/Ox and GH/Pl groups (1.3 ± 0.6 vs. 1.5 ± 0.6; fig. 2c). At year 2 (prior to estrogen replacement), fewer girls in the GH/Ox than GH/Pl group had greater than Tanner 1 breast development [11/37 (30%) vs. 19/33 (58%)] (p = 0.02). There was no difference in gonadotropin or estradiol levels between the groups. By year 4, mean breast Tanner stage in the GH/Ox group was approximately one Tanner stage behind that of the GH/Pl group (2.9 ± 1.3 vs. 4.1 ± 1.3, p = 0.003) (fig. 3a, b). Progression of breast Tanner stage over 4 years in the GH/Ox versus GH/Pl groups revealed a significant difference (Ox less developed than Pl, p = 0.01). There was no significant effect of karyotype (fig. 3c), BMI, BMI SDS, or estradiol on breast development, and luteinizing hormone, FSH and estradiol levels were similar between groups except at year 1. Additional follow-up of patients after study end revealed that 11/20 patients who had received Ox still had not achieved Tanner 5 breast development (age at final follow-up visit ranged from 14.5 to 18.5, mean age of 16.1 years), versus 3/13 patients who had received Pl (age at final follow-up visit 13.9–19.7, mean 15.7 years).

Menarche

By the end of year 4, 17/24 girls (71%) in the GH/Ox group and 15/23 girls (65%) in the GH/Pl group had attained menarche. Mean age at menarche was slightly higher in the Ox than the Pl group (16.3 ± 2.0 years Ox vs. 15.0 ± 1.4 years Pl, p = 0.04).

Pubic Hair Tanner Stage

Mean pubic hair Tanner stage was similar between groups throughout the study. Most patients had Tanner 2 pubic hair throughout the study. ANCOVA for pubic hair Tanner stage revealed no significant difference between treatment groups and no significant effect of karyotype, BMI, free testosterone, or estradiol.

Bone Mineral Density

Mean BMD of the wrist and spine was similar between the groups at baseline (beginning of year 3) and at the end of year 4 for the raw values (see online suppl. table 1). Based on normative age-adjusted data only available for the spine and Lunar systems, the mean BMD SDS at years 2 and 4 was slightly below average, but was similar in the GH/Ox and GH/Pl groups (year 4: −0.6 ± 1.5 Ox vs. −0.9 ± 0.9 Pl, p = 0.3) [20].

Hormone Values (online suppl. table 2)

Estradiol and TSH levels did not differ between the GH/Ox and GH/Pl groups at any visit. Sex hormone-binding globulin, thyroxine, and thyroxine-binding globulin levels were lower in the GH/Ox group at all post-baseline visits. IGF-I levels were similar at baseline, lower in the GH/Ox group from 6 months to 2 years, and similar thereafter between the groups. IGFBP-3 levels were similar in the GH/Ox and GH/Pl groups except for the baseline and 6-month visits. Free testosterone levels were similar for all study visits.

Safety Data (online suppl. table 2)

There were no deaths or treatment-related serious adverse events. Eighteen non-treatment-related serious adverse events associated with hospitalizations were reported in 16 subjects during the study (10 GH/Ox subjects and 6 GH/Pl subjects) among the 76 randomized subjects. The hospitalizations were related to childhood illness (n = 9), infections (n = 2), kidney disease associated with TS (n = 3), accidents (n = 2), congenital heart disease associated with TS (n = 1), and hip surgery (n = 1). No doses were reduced because of these hospitalizations.

Doses were reduced in response to potential androgen-associated side effects, such as alteration in liver function, hypertension, and virilization. The dose of Ox versus Pl was similar during the first year, but thereafter the GH/Ox group received lower mean doses than the GH/Pl group (expressed as percent protocol-specified weight-based dose and reflecting more dose reductions in the Ox group). By year 4, mean doses were 81 ± 24% Ox versus 97 ± 10% Pl, p < 0.01. Once dose reductions occurred, no further dose increases were made. Dose reductions for SGPT elevations (>100 U/l) occurred in 3 GH/Ox (137, 105, 118 U/l) and 1 GH/Pl (117 U/l) subject. SGPT decreased to normal on subsequent visits (data not shown). Hypertension was infrequent (1 GH/Ox, no GH/Pl). Virilizing signs leading to dose reduction included subjective clitoral changes in 2 GH/Ox and 2 GH/Pl patients, acne in 1 GH/Ox patient, and mild hirsutism in 3 GH/Ox patients, and resolved after dose reduction.

Hemoglobin A1C, glucose, and insulin levels were similar between groups throughout the study. Mean alkaline phosphatase levels were increased in the Ox versus Pl group at years 2.5 and 3. Total cholesterol levels were lower in the GH/Ox versus GH/Pl group at years 2, 2.5 and 3.5. HDL cholesterol levels were similar between groups at baseline, were lower in the Ox group at study visits 6 months through 2 years, and then increased in the Ox group once estrogen replacement was begun so that there was less of a difference by year 4. There was no significant difference between groups in LDL measurements for any visit. Triglycerides were lower in the GH/Ox versus GH/Pl group at years 3 and 4. SHBG levels were decreased in the Ox group, although free testosterone values remained similar throughout the study.

Discussion

The results of this randomized, double-blind, Pl-controlled study in girls with TS, treated with Ox or Pl for 4 years, plus GH for 4 years and pubertal estrogen replacement therapy for the 3rd and 4th years, are consistent with the hypothesis that Ox treatment has a positive effect on height velocity and height gain. Mean additional height gained in the Ox versus Pl group was 1.5 cm after 2 years of treatment and was 4 cm after 4 years. This effect was more pronounced in girls with younger chronologic and bone age at study entry. There was no apparent effect of Ox on BMD of the wrist or spine. Breast development and menarche were somewhat delayed in the GH/Ox treatment group, but there was no difference in pubic hair development or bone age progression after 4 years of Ox treatment.

This study adds to the current literature in several ways: this protocol is consistent with current GH dosage guidelines [2], the Ox dose protocol did not change throughout the study duration, and most importantly, there was a blinded Pl-control group. Recently, another long-term prospective double-blind study was published that had a consistent Ox dosing protocol, a Pl group, and stable GH dosing [9]. Unlike most other studies, there was a dose individualization protocol in this study that allowed for reducing the Ox dose if there was any sign of virilization or safety concerns. The weekly GH dose was 0.35 mg/kg/week in this study. The GH was given at a lower weekly dose or less frequently in the three other major prospective studies [10,11,12]. The Ox dose in this study was 0.05 mg/kg/day for all 4 years. In the studies by Rosenfeld et al. [11], Stahnke et al. [12], and Nilsson et al. [10], the Ox dosage was initially higher (0.1 mg/kg/day), with later reduction in all three studies to a similar dosage used in this study (0.05 mg/kg/day), due to virilization. The dosage changes at various time points in the other major studies complicate the interpretation and the applicability of the results. For example, it is possible that the initial higher dose of Ox or lower dose of GH in the previous studies impacted height velocity. The GH/Ox treatment group in this study was compared to a Pl control group (GH/Pl). The girls given GH with Ox in the study by Rosenfeld et al. [11] were compared to a small cohort of girls who received only GH (n = 17) and to historical control subjects. The GH/Ox-treated girls in the study by Stahnke et al. [12] were compared to those treated with GH alone and to a second group, GH-transient Ox, which was created during the study from those who stopped Ox due to virilizing adverse effects. The girls treated with GH and Ox in the study by Nilsson et al. [10] were compared to a group treated with Ox alone and to a third group treated with GH/Ox and estradiol (100 ng/kg/day). Thus, the previous studies by Rosenfeld et al. [11], Stahnke et al. [12] and Nilsson et al. [10] were limited by variability of study groups and dosages of both GH and Ox and lack of Pl-controlled comparison groups.

Despite the differences in protocol, the results of Ox treatment in this study are consistent with the previous prospective studies of Menke et al. [9], Nilsson et al. [10], Stahnke et al. [12], and Rosenfeld et al. [10], underlining the conclusion that Ox causes a modest but significant height gain when added to GH. Results are reported differently in the previous studies; however, a comparison can still be made. In Menke et al. [9], the addition of low-dose Ox (0.03 mg/kg/day) increased the mean height gain in both the intention-to-treat and per protocol analysis when compared to GH/Pl by 2.3 and 3 cm, respectively. In Nilsson et al. [10], the addition of Ox to GH for 4 years increased the height gain above projected adult height by 8.5 cm, compared to a net gain of 3 cm in the GH/Ox/estradiol group, 7.4 cm in the GH alone, and 4.3 in the GH/estradiol groups. In Stahnke et al. [12], the change in TS height SDS from baseline (mean age of 11 years) after 5 years of treatment was 1.77 ± 0.4 SDS in the GH/Ox group (height gain of approx. 10 cm) and 1.26 ± 0.4 SDS in the GH alone group (height gain of approx. 7 cm) versus our finding, after 4 years of treatment (mean baseline age of 12 years), of 1.8 ± 0.9 SDS GH/Ox (height gain of 11 cm) versus 1.2 ± 0.7 SDS GH/Pl (height gain of 7 cm). In the study of Rosenfeld et al. [11] (mean baseline age of 9 years), after 6–7 years of treatment, adult height minus mean projected height was 10.3 ± 4.7 cm in the GH/Ox group versus 8.4 ± 4.5 cm in the GH group. In the current study, the change in height from baseline after 4 years of treatment was 26.2 ± 6.7 cm in the GH/Ox group versus 22.2 ± 5.1 cm in the GH/Pl group.

The estrogen replacement protocol differed when compared to prior GH/Ox studies. Current recommendations are to start estrogen replacement as early as 12 years because of the detrimental psychosocial effects of late puberty induction and the lack of evidence for deleterious effects on height of age-appropriate estrogen initiation and dosage [2,21]. In this study, estrogen replacement therapy was initiated at a younger age (13.9 ± 1.6 Ox and 13.8 ± 1.6 Pl years) than in the study by Rosenfeld et al. [11] (age 15.2 ± 0.9 GH, 14.9 ± 0.9 GH/Ox) and that of Stahnke et al. [12] (14.9 years), but at an older mean age than in the study of Nilsson et al. [10] (12.4 years). Estrogen was initiated at 12 years in the study by Menke et al. [9]. The initial dose of estrogen in this study was 50 ng/kg/day ethinyl estradiol during year 3 and 100 ng/kg/day during year 4, while the initial dose was higher in the study of Nilsson et al. [10] study (100 ng/kg/day ethinyl estradiol). Rosenfeld et al. [11] used conjugated estrogen starting at 0.3 mg/day of conjugated estrogen, then increased to 0.625 mg/day after 6 months and Stahnke et al. [12] used estradiol valerate 0.2 mg daily for 6 months, followed by 0.5 mg daily for 6 months, followed by 1 mg daily for 1 year and 1.5 mg daily for the subsequent year. In sum, in this study, the estrogen replacement was started at lower doses and using ethinyl estradiol. The goal of initiating pubertal changes was therefore achieved along with increasing stature.

The finding in this study of significant slowing of breast development in the GH/Ox group may stimulate reconsideration of the optimal age of initiation and duration of Ox treatment. Carel et al. [22] recently found that late onset of puberty had a negative impact on quality of life in young women with TS. Therefore, both age of pubertal induction and optimization of height should be considered in developing optimal treatment regimens. Although delayed ‘feminization’ was noted in one of the first trials of a nonaromatizable androgen in TS [23], this issue has not, to our knowledge, been studied systematically. In the recent study by Menke et al. [9], the authors also found that breast development was slower in the GH/Ox-treated girls. Breast development is a complex issue in TS. According to a study by Doerr et al. [24], nearly 20% of girls with 45,X did not achieve breast development past Tanner 3 despite long-term estrogen therapy, whereas those with mosaicism were more likely to have complete (Tanner 5) development. Our results did not show a statistically significant difference by karyotype. It does appear that the girls with 45,X were somewhat less likely to develop than the girls with mosaicism (fig. 3c), despite estrogen replacement. The results also indicate later spontaneous onset of thelarche in the girls receiving Ox. Because the observed slowing of breast development is likely a direct androgen effect [25], we recommend consideration of this factor when deciding the age of onset and duration of Ox treatment. It is possible that Ox treatment starting as early as age 8 or 9, for 3 or 4 years, will provide the optimum height benefit, allow subsequent estrogen initiation as early as age 12, and thereby minimize Ox-mediated interference with breast development. This issue will require further study.

Fourth, this study investigated the impact of 4 years of treatment with Ox on BMD in girls with TS. Androgen treatment of osteoporosis has been proposed because osteoblasts possess androgen receptors, and because androgens stimulate bone cell proliferation and differentiation. Additionally, short-term treatment with androgen in combination with estrogen reverses the inhibitory effect of estrogen on bone formation in postmenopausal women [26], who share a similar lack of estrogen effect on bone. However, studies of postmenopausal women given testosterone replacement have shown either increased bone mass or no effect [17]. Moreover, Ox treatment of adult women with TS did not alter bone density [27] and, similarly, Ox treatment in this study did not affect bone density over the 4-year treatment period.

In terms of safety and side effects, Ox, at the low dosage of 0.06 mg/kg/day, did not affect weight, BMI, waist:hip ratio, or bone age progression. The virilizing effects reported at higher doses of Ox were rarely seen in this study using only the lower dose and resolved with further dose reduction. Despite the reduced dose, the growth was better in the Ox-treated group. The lower SHBG is an effect seen with Ox, as well as other androgens [23]. The lower HDL is likely to be a temporary androgen effect that is modulated by estrogen replacement and later reversed with cessation of Ox. Furthermore, positive effects of Ox on cognitive function were recently published from this study [28].

There are limitations of this study, including the following: there were missing data from those patients who did drop out although after analysis there was no apparent bias in terms of adverse effects or differences in height SDS in the drop-outs. There was no difference between those who dropped out and those who remained in the study in age, bone age, height, change in height from baseline, or height SDS. This study included girls with TS older than 10 years and therefore these results may not be generalizable to younger girls who have received GH. Despite the trend toward more advanced pubertal development at baseline in the GH/Pl group, the change in Tanner stage from baseline to 4 years was a full Tanner stage ahead in the Pl group and the difference was highly statistically significant. The analysis of BMD was dependent on available technology at the time, and there were no ideal normative data for comparison. Two different systems were used to analyze the data (Lunar and Hologic). Thus, an Ox effect on bone density may have been missed. Ox may influence BMD if used for a longer period of time or at different dosages. While most girls achieved near final height, they were not followed to final height in this study. Given the similar bone ages in the two groups throughout the study and at year 4, it is unlikely that additional major differences would have emerged.

Conclusion

This trial tested whether the addition of Ox to a GH treatment regimen increased growth rate and near-final height, and whether the Ox affected the progress of puberty and BMD. Supplemental Ox modestly increased growth velocity and near-final height compared to those receiving GH only, without significant change in bone age or bone density, and with a slower rate of breast development. This study is valuable because it is a carefully conducted, long-term study of girls with TS with many relevant measures collected over time. The design is noteworthy as well, being randomized, double-blind, and Pl controlled. The use of Ox might be considered in girls with spontaneous and satisfactory breast development who remain very short despite the use of GH or for prepubertal girls.

Disclosure Statement

M.Z., K.S., K.K., and H.K. have nothing to disclose. G.B.C is a former employee, retiree (with pension) and current consultant of Eli Lilly & Company. J.L.R. has received grant support from Eli Lilly & Company, Pfizer, and Novo Nordisk and has served as a consultant for Eli Lilly & Company, Pfizer, and Novo Nordisk.