Vertebral morphology in aromatase inhibitor–treated males with idiopathic short stature or constitutional delay of puberty

Abstract

Aromatase inhibitors (AIs), blockers of estrogen biosynthesis, delay bone maturation and therefore are used increasingly to promote growth in children and adolescents with growth disorders. The effects of treatment on skeletal health are largely unknown. Since estrogen deficiency is associated with various detrimental skeletal effects, we evaluated in this cross-sectional posttreatment study vertebral body morphology, dimensions and endplates, and intervertebral disks by the use of magnetic resonance imaging (MRI) in two cohorts of males previously treated with the AI letrozole or placebo. Males with idiopathic short stature received treatment with letrozole or placebo for 2 years during prepuberty or early puberty; males with constitutional delay of puberty received letrozole or placebo in combination with low-dose testosterone for 1 year during early or midpuberty. In males with idiopathic short stature, mild vertebral body deformities were found in 5 of 11 (45%) letrozole-treated subjects, whereas in the placebo group no deformities were detected (p = .01). In the cohort of males with constitutional delay of puberty, a high prevalence of endplate and intervertebral disk abnormalities was observed in both the letrozole- and the placebo-treated groups. We conclude that AI therapy during prepuberty or early puberty may predispose to vertebral deformities, which probably reflect impaired vertebral body growth rather than impaired bone quality and compression fractures. If AIs are used in growth indications, follow-up of vertebral morphology is indicated. © 2010 American Society for Bone and Mineral Research

Introduction

Loss of estrogen action owing to defective aromatase enzyme or inactive estrogen receptor α leads to delayed bone maturation, postponed growth plate fusion, and exceptionally tall adult stature.1–5 These findings have prompted clinical researchers to investigate whether in children or adolescents with growth disorders a delay in bone maturation and ultimately an increase in adult height can be achieved by blocking of estrogen biosynthesis with aromatase inhibitor (AI) treatment. The results of recently published clinical trials indeed suggest that in some growth disorders, predicted adult height increases with AI therapy.6–9 However, treatment effect on final adult height is still unclear.

The effects of AI therapy on bone health in children and adolescents have been inadequately characterized. In boys, AI treatment does not influence bone mineral density (BMD) significantly, as evaluated by dual-energy X-ray absorptiometry (DXA).6, 7, 9 However, based on serum markers of bone formation and resorption, the treatment appears to suppress bone turnover.10 In pubertal boys, the treatment stimulates cortical bone growth,10 but little is known about the treatment effects on trabecular bone mass and bone quality.

In our recent randomized, placebo-controlled study on AI treatment in boys with idiopathic short stature (ISS), abnormal vertebral body shape was detected with surprisingly high frequency in both placebo and AI groups, as evaluated by posttreatment DXA-derived vertebral imaging.10 This finding prompted us to thoroughly investigate the impact of AI therapy on vertebral morphology in two different populations previously treated with the AI letrozole or placebo during childhood or adolescence by using magnetic resonance imaging (MRI). The boys with ISS received therapy with letrozole for 2 years during prepuberty or early puberty7 and the boys with constitutional delay of puberty (CDP) for 1 year during puberty in combination with low-dose testosterone.6, 7 In this study, we characterize the treatment effects on vertebral morphometry and morphology, as well as on vertebral endplates and intervertebral disks.

Materials and Methods

Subjects and study design

The study population included males from two prospective, randomized, placebo-controlled studies. The details of these studies have been reported previously6, 7 and are briefly summarized here. The Ethics Committee for the Hospital for Children and Adolescents and the National Agency for Medicines approved the study protocols. Before initiation of treatment, written informed consent was obtained from each subject and his parents or guardian(s).

Males with idiopathic short stature (cohort 1)

Cohort 1 included boys with ISS followed up at the Pediatric Endocrine Clinic, Hospital for Children and Adolescents, University of Helsinki, Finland. The inclusion criteria were (1) calendar age 9.0 to 14.5 years and (2) height at least 2 SD units below the mean for age or at least 2 SD units below the midparental target height. Subjects with a chronic or endocrine illness and those with bone age above 14 years were excluded. We enrolled 31 subjects; 30 of them (16 on letrozole, 14 on placebo) completed the 2-year treatment. The study was conducted as a randomized, double-blind, placebo-controlled clinical study; the subjects received either letrozole 2.5 mg orally once daily (Femar, Novartis AG, Stein, Switzerland) or placebo orally once daily for 2 years.

Males with constitutional delay of puberty (cohort 2)

Cohort 2 included boys with CDP, defined as (1) a Tanner genital (G) or pubic hair (P) stage observed at an age exceeding +2 SD of the mean for healthy Finnish boys or (2) a testis volume of less than 4 mL after 13.5 years of age. However, all males with CDP included in this study showed physical signs of onset of puberty (Table 1), but not the pubertal growth spurt, at the start of the study. The participants (23 males) were randomised to receive testosterone 1 mg/kg intramuscularly every 4 weeks for 6 months in combination with either placebo or letrozole 2.5 mg orally for 12 months. The participants were carefully followed during treatment and assessed again at near-final height.8

Table 1. Physical Characteristics of the Study Subjects at the Onset of Treatment and at the Time of the MRI

	Males with ISS		Males with CDP
	Letrozole (n = 11)	Placebo (n = 12)	Testosterone + letrozole (n = 6)	Testosterone + placebo (n = 6)
a In males with ISS, sitting heights were available only for 4 and 6 participants in the letrozole and placebo groups at the time of MRI, respectively.
Age (years)
At treatment onset	10.7 (1.7)	10.9 (1.5)	15.5 (0.7)	14.6 (0.8)
At the MRI	16.9 (1.8)	17.3 (1.4)	23.7 (0.7)	22.7 (0.8)
Height (cm)
At treatment onset	127.1 (6.9)	127.6 (6.8)	156.2 (5.9)	149 (8.9)
At the MRI	159.1 (5.2)	161.1 (5.9)	177.6 (7.7)	170.1 (5.5)
Height (SDS)
At treatment onset	−2.3 (0.2)	−2.4 0.4)	−1.9 (0.6)	−2.0 (0.8)
At the MRI	−2.0 (0.5)	−2.1 (0.6)	−0.2 (1.3)	−1.4 (0.9)
Sitting height (%)
At treatment onset	53.2 (1.4)	53.2 (1.5)	50.9 (1.0)	52.2 (1.7)
At the MRI	52.1 (0.3)a	53.5 (1.6)a	52.5 (1.0)	53.3 (1.3)
Sitting height (SDS)
At treatment onset	−0.03 (1.2)	−0.1 (1.4)	−0.02 (0.7)	0.7 (1.3)
At the MRI	0.6 (0.2)a	1.1 (1.3)a	−0.2 (0.7)	0.3 (0.8)
BMI (kg/m^2)
At treatment onset	17.0 (2.3)	15.8 (1.3)	18.5 (1.4)	20.9 (2.2)
At the MRI	20.2 (2.4)	20.8 (2.0)	22.3 (2.8)	24.9 (2.8)
Bone age (years)
At treatment onset	8.8 (2.4)	8.8 (1.8)	13.2 (0.5)	12.3 (1.7)
At the MRI	15.8 (1.9)	16.6 (1.6)	19.0 (0)	19.0 (0)
Testis volume (mL)
At treatment onset	1.2 (0.9)	0.9 (0.5)	5.6 (2.3)	6.9 (5.5)
At the MRI	NA	NA	NA	NA
Tanner genital stage
At treatment onset	1 (1–2)	1 (1–2)	2 (2)	2 (2–3)
At treatment cessation	1 (1–5)	2 (1–4)	4 (3–5)	4 (3–5)
Tanner pubic hair stage
At treatment onset	1 (1)	1 (1)	1 (1–2)	1 (1–2)
At treatment cessation	1 (1–4)	1 (1–4)	3 (2–4)	3 (2–4)

Study protocol

For this study, the subjects were invited during early 2008 (cohort 1) and mid-2006 (cohort 2) for a visit including anthropometry (ie, height, sitting height, and weight), bone age X-ray, and spinal MRI; these findings were correlated with previously obtained blood biochemistry.7, 10 In cohort 1, 23 subjects (11 from the letrozole group and 12 from the placebo group) consented to participate, and in cohort 2, 12 subjects (6 from the testosterone plus letrozole group and 6 from the testosterone plus placebo group) consented to participate. The study visit was, on average, 4.2 and 7.0 years after completion of the initial placebo-controlled intervention in cohorts 1 and 2, respectively. In cohort 1, 4 participants (2 in letrozole group, 2 in placebo group) received either seasonal or continuous treatment with inhaled corticosteroid for asthma during the study. In cohort 2, 1 participant received inhaled corticosteroid treatment. No other medications known to influence bone metabolism were used at the time of the study in either population.

Imaging studies

Radiographs of the left hand were obtained, and bone age was determined by MH and LD (cohort 1) and SW and LD (cohort 2) according to Greulich and Pyle.11 Predicted adult heights were calculated by the Bailey-Pinneau method according to data for males with an average tempo of maturation.12

Spinal MRIs were performed with a 1.5-T imager (Intera Achieva, Philips Medical Systems, Best, Netherlands) using a total-spine imaging protocol; all thoracic and lumbar vertebrae were imaged. The protocol consisted of 17 sagittal slices with T₁-weighted (TR 500 ms, TE 5.5 ms, FOV 310, flip angle 90 degrees, slice thickness 3 mm, gap 1 mm) and T₂-weighted (TR 4500 ms, TE120 ms, FOV 310, flip angle 90 degrees, slice thickness 3 mm, gap 1 mm) turbo spin-echo sequences for assessment of vertebral anatomy, endplate abnormalities, and intervertebral disks. Sagittal slices with short-inversion-time inversion-recovery (STIR; TR 3000 ms, TE 70 ms, FOV 310, IR delay 170 ms, slice thickness 4 mm) sequence were obtained for the assessment of bone marrow edema. In addition, a coronal T₁-weighted (TR 400 ms, TE 7.4 ms; FOV 420, flip angle 90 degrees, slice thickness 4 mm) turbo spin-echo sequence was obtained for ISS patients.

All MRI images were visually analyzed for vertebral anatomy and shape, endplate changes, disk height and water content, and muscle status of the back by a radiologist (ST-S) blinded for the participants' treatment. Measurements of vertebral body dimensions and assessments of the shape of each vertebral body were performed with a digital calliper in the radiologic workstation. The shapes of all vertebral bodies were classified as normal, wedged, or compressed using a newly developed classification for pediatric vertebral body morphology.13 In this classification, anterior wedge deformities are classified as mild (grade 2a) when anterior vertebral height reduction is 20% or greater but less than 50% and as severe (grade 2b) when the reduction is greater than 50% compared with posterior vertebral height. In mild and severe compression deformities, the anterior, middle, and posterior vertebral heights are decreased by 20% to 30% (grade 3a) and by more than 30% (grade 3b), respectively, compared with adjacent normal vertebrae. In an effort to distinguish osteoporotic vertebral fracture from anterior wedging associated with anterior vertebral body growth disturbance, an index of vertebral body dimensions (AM:MP) was calculated as follows: (anterior height/middle height)/(middle height/posterior height). In adults with anterior wedging, an AM:MP ratio greater than 1.0 suggests osteoporotic compression fracture, whereas a value of less than 1.0 suggests Scheuermann disease.15 Thus in Scheuermann disease the anterior height of the vertebral body in particular is reduced, whereas in osteoporotic compression fracture the vertebral body midportion height is also reduced.

In order to assess vertical growth of the vertebrae, the anterior, middle, and posterior heights and the anteroposterior diameter of vertebral bodies Th₆ through Th₉, and L₁ and L₂ were measured from the midsagittal slice at the level of the central vein in each subject. In addition, an index of vertebral body height was calculated by dividing posterior vertebral body height by vertebral body anteroposterior diameter. Intervertebral disks at the midthoracic (Th₆ through Th₁₀) and upper lumbar (L₁ through L₃) spine were classified with the use of a previously described grading system (grade 0 = normal disk; grade 1 = slight decrease in signal intensity; grade 2 = normal-height hypointense disk; grade 3 = hypointense disk with disk space narrowing).14 All endplate abnormalities of the thoracic and lumbar spine were recorded; they were defined by endplate irregularity or the presence of Schmorl's node (a protrusion of intervertebral disk material through the vertebral body endplate into the adjacent vertebra).

The muscle status of the back was assessed both visually and by measuring the thickness of the psoas and back muscles. In the visual evaluation, apparent size loss compared with the adjacent vertebral body and high-signal-intensity streaks on T₁- and T₂-weighted images representing fat deposits in the muscle were used as criteria for muscle atrophy. The thickness of the back muscles (ie, erector spinae and multifidus spinae together) was measured from T₁- and T₂-weighted sagittal images at the level of L₄–L₅ disk; the maximal anteroposterior muscle thickness on each side was recorded. The psoas muscle measurements were obtained from coronal T₁-weighted images (available for ISS boys) and from coronal scout images (from CDP patients); the maximal thicknesses of the right and left muscles were recorded. In addition, musculature was evaluated using the ratio of total-body bone mineral content (BMC) and lean body mass, which were measured 12 months after cessation of treatment by DXA, as described previously.10

Statistical analysis

We used two-sided tests of hypotheses, and we considered a p value of less than .05 significant. Values are expressed as means ± SD unless otherwise stated. Vertebral body measurements are expressed as medians. We conducted the analyses with SPSS statistical software for Windows, Version 16.0 (SPSS, Inc., Chicago, IL, USA). In both cohorts we analyzed the between-group differences in physical characteristics and vertebral body dimensions by the Mann-Whitney test. Fisher´s exact test was used to compare the frequency of abnormal vertebral findings in the treatment groups. In cohort 1, the differences in predicted adult height between the treatment groups were analyzed by the Mann-Whitney test.

Results

Physical characteristics

Males with idiopathic short stature (cohort 1)

At the start of the letrozole intervention, there were no significant differences between the letrozole- and placebo-treated participants in age, height, weight, bone age, sitting height, or testis volume (Table 1); 2 (18%) of 11 participants in the letrozole group and 1 (8%) of 12 participants in the placebo group were pubertal (testis volume ≥ 2 mL). After 24 months of treatment, puberty had progressed in 5 (45%) participants receiving letrozole and in 7 (58%) participants receiving placebo (Table 1). Twelve months after cessation of treatment, BMC:lean-body-mass ratios, as analyzed by DXA, showed no significant difference between the letrozole and placebo groups (median values 0.048 versus 0.045, respectively, p = .60). At the time of the study, no significant differences between the treatment groups were evident in age, height, height standard deviation score (SDS), sitting height SDS, weight, body mass index (BMI), and bone age (Table 1). In addition, the median predicted adult heights were not significantly different at the time of MRI examination in the letrozole and placebo groups (166.5 and 162.4 cm, respectively, p = .57).

Males with constitutional delay of puberty (cohort 2)

Before treatment onset, no significant differences in baseline characteristics were found between the letrozole and placebo groups (Table 1). During treatment, participants in both groups progressed in puberty (Table 1). At the time of the present evaluation, no significant differences in age, height, sitting height, weight, BMI, or bone age were observed between participants treated with testosterone plus letrozole and testosterone plus placebo (Table 1). Subjects treated with testosterone plus letrozole were taller than those treated with testosterone plus placebo, but the difference failed to reach statistical significance (p = .06).

Spinal MRI findings

Males with idiopathic short stature

In cohort 1, mild vertebral body deformities were found in 5 participants, all of whom had received letrozole (Table 2 and Fig. 1). All observed deformities were located in the thoracic spine between vertebrae Th₆ and Th₉. Altogether, four grade 2a deformities were observed in 4 participants, with anterior wedging ranging from 20% to 22%. In addition, four grade 3a compression deformities were detected in 2 participants, with vertebral body height reductions ranging from 21% to 23%; both subjects had received letrozole. One or more vertebral bodies with milder anterior wedging ranging from 12% to 19% were found in 6 subjects, of whom 5 had received placebo; these changes were regarded as normal. No bone marrow edema indicating recent bone trauma was noted in association with vertebral deformities.

Table 2. Vertebral Morphology in Participants With Idiopathic Short Stature and Constitutional Delay of Puberty, as Evaluated by MRI

	Males with ISS			Males with CDP
	Letrozole (n = 11)	Placebo (n = 12)	p Value	Testosterone + letrozole (n = 6)	Testosterone + placebo (n = 6)	p Value
Note: Wedged and compressed vertebrae were graded as 2a, 2b, 3a, or 3b using a classification for pediatric vertebral body morphology. In this classification, anterior wedge deformities are classified as mild (grade 2a) when anterior vertebral height reduction is greater than 20 percent but less than 50% and as severe (grade 2b) when the reduction is greater than 50%. In mild and severe compression deformities, the anterior, middle, and posterior vertebral heights are decreased by 20% to 30% (grade 3a) and by greater than 30 percent (grade 3b), respectively. a Region Th6 through L3 was evaluated for intervertebral disk abnormalities.
Patients with vertebral deformity (n)	5	0	.01	1	1	NS
Deformed vertebrae (n)	8	0		2	1
Grade 2a	4	0		1	1
Grade 2b	0	0		0	0
Grade 3a	4	0		1	0
Grade 3b	0	0		0	0
Patients with endplate abnormalities (n)	2	5	NS	3	4	NS
Endplate abnormalities (n)	15	28		31	25
Patients with intervertebral disk abnormalities (n)a	4	3	NS	4	4	NS
Intervertebral disk abnormalities (n)a	10	10		15	12
Grade 1	0	0		6	6
Grade 2	4	3		4	3
Grade 3	6	7		5	3

Figure 1. (A) Normal vertebrae and intervertebral disks on MRI in a placebo-treated participant with ISS (T₂-weighted midsagittal slice, TR 4500 ms, TE120 ms). (B) Grade 3a vertebral body deformities in Th₇ through Th₈ (arrows) in a letrozole-treated participant with ISS. (C) Grade 2a and 3a vertebral body deformities (Th₇ through Th_9;arrows) with endplate irregularities and intervertebral disk changes in a letrozole-treated participant with ISS. (D) Normal vertebrae with disk and endplate abnormalities (arrowheads) including Schmorl's node (arrow) in a letrozole-treated participant with ISS.

Endplate deformities were found in 2 subjects treated with letrozole and in 5 subjects treated with placebo (Table 2). Two patients, one treated with letrozole and one with placebo, had extensive endplate abnormalities, including several Schmorl's nodes (Fig. 1). Intervertebral disk abnormalities in the midthoracic and upper lumbar spine showed equal distribution between the treatment groups (Table 2). Four patients (two treated with letrozole and two with placebo) had central Schmorl's nodes and intervertebral disk abnormalities in three adjacent vertebrae.

There was a trend toward smaller vertebral height measurements in the letrozole-treated group, but no statistically significant differences in anterior, middle, or posterior heights of vertebral bodies Th₆ through Th₉ and L₁ and L₂ were observed between the letrozole and placebo groups, and their vertebral body height indices (ie, posterior vertebral body height/diameter) were similar (detailed data not shown). We next calculated the mean values for anterior (16.8 versus 18.0 mm, p = .21), middle (16.8 versus 18.0 mm, p = .19), and posterior (17.0 versus 18.3 mm, p = .09) heights in thoracic (Th₆–Th₉) vertebral bodies and found no statistically significant differences between the letrozole and placebo groups. In a similar fashion, the mean values for anterior (24.0 versus 25.0 mm, p = .15), middle (24.0 versus 25.0 mm, p = .17), and posterior (24.0 versus 25.0 mm, p = .17) heights in lumbar (L₁–L₂) vertebral bodies showed no statistically significant differences. In order to further evaluate vertebral body anterior wedging, the mean anterior height for thoracic (Th₆–Th₉) and for lumbar (L₁–L₂) vertebral bodies was divided by the mean anteroposterior vertebral body diameter in the same region; no significant differences in these ratios were seen between the treatment groups (p = .32 to .35).

Spinal muscle size, as analyzed by anteroposterior diameters of erector spinae and multifidus muscles, did not differ significantly between the treatment groups (median values 34.8 versus 36.4 mm in the letrozole and placebo groups, respectively; p = .7). Similarly, the maximal width of the psoas muscle showed no significant difference (36.5 versus 38.9 mm, p = .65). Spinal muscle size did not differ between those with vertebral deformity and those without (median anteroposterior diameter of erector spinae and multifidus muscles 37.6 versus 36.1 mm, respectively, p = .54; median psoas muscle maximal width 39.4 versus 37.0 mm, p = .40).

When comparing letrozole-treated subjects with and without vertebral deformities, no statistically significant differences were detected in pretreatment height SDS, weight, age, bone age, serum testosterone, serum estradiol, testis volume, BMD, or markers of bone formation [eg, serum amino-terminal propeptide of type I procollagen (S-PINP)] and resorption [urinary excretion of amino-terminal telopeptide of type I collagen (U-NTX)] (data not shown). Similarly, changes in growth velocity, Tanner G and P stage of puberty, testis volume, bone age, lumbar spine BMD, predicted adult height, serum testosterone, serum estradiol, and bone turnover markers (ie, S-PINP and U-NTX) during the 2-year treatment did not differ (data not shown). The only near-significant difference between the deformity-positive and deformity-negative letrozole-treated males in cohort 1 was found in pretreatment predicted adult height (173.4 versus 162.1 cm, respectively, p = .05).

In all vertebral bodies with deformity (n = 8), the AM:MP ratio was 1.0 or less (median 0.94, range 0.87–1.00). This suggests that anterior rather than midportion height was reduced in wedged vertebral bodies.

Males with constitutional delay of puberty

In cohort 2, vertebral deformities were detected in two participants (Table 2). One male in the testosterone plus letrozole group had two vertebral deformities (Th₁₀ and Th₁₁, grades 2a and 3a). However, he had injured his back in a high-energy motocross accident 10 years previously, and therefore, these findings may be posttraumatic. In the testosterone plus placebo group, one male had a grade 2a deformity in Th₇.

Endplate deformities were observed in 3 of 6 men previously treated with testosterone and letrozole (50%) and in 4 of 6 men treated with testosterone plus placebo (67%; Table 2). No significant differences in the number of patients with endplate deformity or in the number endplate deformities were found between the treatment groups. Further, there were no significant differences in anterior, middle, or posterior heights in vertebrae Th₆ through Th₉ and L₁ and L₂ between the groups (detailed data not shown). The mean thoracic (Th₆ through Th₉) vertebral body anteroposterior diameter was slightly larger in the testosterone plus letrozole group compared with the testosterone plus placebo group (28.1 versus 26.4 mm, respectively, p = .06). Resulting from a greater diameter, the index of vertebral body height (mean Th₆ through Th₉ posterior height/diameter) was slightly lower in the testosterone and letrozole group (0.76 versus 0.82, p = .06). The mean lumbar (L₁ and L₂) vertebral body anteroposterior diameter (33.3 versus 31.3 mm, p = .24) and the index of vertebral body height (0.82 versus 0.83, p = .70) showed no significant differences.

Comparison between cohorts 1 and 2

When comparing the baseline characteristics in the two study populations, we observed several differences that may have contributed to the differences in treatment efficacy and impact on vertebral morphology (Table 1). At treatment onset, males in cohort 2 were older and had more advanced bone age and pubertal status and greater testis volume. In addition, the males in cohort 2 received letrozole for a shorter duration, in combination with adjuvant testosterone therapy. They also were evaluated by MRI at an older age and after a significantly longer interval since letrozole treatment cessation than the males in cohort 1 (mean 7.0 versus 4.2 years).

Discussion

This is the first study to describe the effects of aromatase inhibitor (AI) therapy on the growing human spine. We used MRI to assess vertebral body morphology and dimensions, endplate changes, and intervertebral disk characteristics in adolescents and young adults who had received letrozole for 1 to 2 years during prepuberty or puberty; the findings were compared with those in subjects treated with placebo. Our key finding was that in boys with ISS, letrozole treatment for 2 years during prepuberty or early puberty was associated with a significantly increased prevalence of vertebral body deformities compared with placebo-treated subjects. Males treated only for 1 year during puberty did not show an increased prevalence of vertebral changes. Furthermore, the prevalence of endplate or intervertebral disk changes was not increased in either group.

The mechanism by which letrozole may predispose to vertebral deformities is unclear. All observed vertebral deformities were mild, with anterior wedging or compressions ranging from 20% to 23%. There were no MRI signs suggesting recent traumatic fractures. Without prospective imaging data, it remains unclear whether the observed changes reflect disturbed or delayed growth and maturation of vertebral bodies or true compression fractures caused by increased bone fragility. The AM:MP ratios, which were of 1.0 of less in all deformed vertebral bodies, suggest that disturbed growth is more likely, although the index has been validated only in adults.15 Estrogen is a well-known regulator of endochondral bone growth and is critical for bone maturation and epiphyseal fusion,1–5, 16 but it is not known whether estrogen (or estrogen deprivation) influences endochondral bone growth and maturation similarly in the axial and appendicular skeleton. In juvenile rabbits and mice, estrogen treatment reduces chondrocyte proliferation and growth plate height, whereas letrozole may decrease the differentiation rate of growth plate chondrocytes and increase growth plate height.17, 18 It remains unclear whether human growth plates respond similarly to estrogen deprivation. In addition, it is unclear if such growth plate changes could predispose to vertebral deformities. The deformities also could represent subchondral aseptic necrosis under the cartilaginous endplate. Such changes are typical for Scheuermann disease, which is a condition of unknown etiology affecting predominantly males with disease onset in preadolescence.19

As regards bone fragility, AI treatment in boys has not been associated with impaired bone mass in studies employing DXA,6, 7, 9, 10 and in the short term, the AI anastrozole had no harmful effect on calcium kinetics in adolescent boys.20 Additionally, in our cohort of pubertal males with ISS, letrozole treatment stimulated cortical bone growth.10 Our DXA-derived findings obtained after treatment also suggest that letrozole treatment was not associated with an imbalance between bone mass and muscle strength in ISS males. Spinal muscle measurements obtained by MRI further support the view that differences in spinal muscle mass (and mechanical stress on vertebral bodies) were not the cause for the different prevalence of vertebral changes between letrozole and placebo groups in ISS males. However, the influence of letrozole on trabecular bone quantity and quality in adolescent boys remains obscure.

The prevalence of vertebral body deformities in spinal MRI scans was high in letrozole-treated boys with ISS. Even though all deformities were mild, the proportion of subjects with deformities was even greater than what has been observed previously with similar or comparable classification in children with chronic disease,13 rheumatoid arthritis,21, 22 or solid-organ transplantation.23, 24 To our knowledge, no data on the prevalence of intervertebral disk abnormalities or endplate irregularities in children or adolescents have been reported previously. Compared with MRI data in males in the general population, the prevalence of such findings was surprisingly high, especially in the CDP population.25

Interestingly, letrozole was associated with an increased number of vertebral deformities only in males with ISS and not in the subjects with CDP. Several factors may explain this difference. Participants with CDP were older and had more advanced bone age and puberty at the start of treatment. They also received treatment for shorter durations, combined with low-dose testosterone. Additionally, males with CDP were evaluated after a longer interval since treatment cessation, which may have allowed recovery of vertebral changes.

AI-treated boys with CDP showed increased predicted adult height after the cessation of treatment.6 Our current findings suggest that this increase in growth potential achieved by blocking estrogen synthesis also translated into taller final adult height. At near-final height, the mean difference in height in favor of participants treated with testosterone plus letrozole was 8.1 cm, and at final adult height it was 7.5 cm. In contrast, the increase in predicted adult height achieved by AI treatment in boys with ISS appears to have mostly disappeared during later posttreatment growth and maturation, although definitive final adult heights in this cohort are not yet available. This difference in treatment efficacy between the two groups is potentially explained by the difference in skeletal maturation at cessation of treatment. To prevent this potential loss of treatment effect after discontinuation of AI therapy, it is advisable to continue treatment until late skeletal maturity and near termination of linear growth. Long-term results of AI treatment are needed to confirm the optimal skeletal maturity at treatment onset and the optimal treatment duration and skeletal maturity at cessation.

Several methodologic issues need to be mentioned. This study was cross-sectional in design, with a significant time interval from treatment cessation to MRI. Clearly, a prospective design would be more appropriate for establishment of a causal relationship between letrozole and susceptibility to vertebral deformities. In addition, the number of patients was low, particularly in the subgroups.

In conclusion, this study suggests that despite the potentially beneficial effects on final adult height in males, the use of AI therapy, especially in prepuberty or early puberty, may predispose to vertebral changes. These changes may reflect delayed vertebral growth and maturation or vertebral body growth disturbance owing to subchondral aseptic necrosis under the cartilaginous endplate, as in Scheuermann disease. Less likely, the changes represent compression fractures owing to impaired bone quality. We recommend the use of AIs in growth indications only in clinical research settings, which should include serial evaluation of vertebral morphology by radiography. In addition, it is our view that patients with disorders or systemic medications associated with increased bone fragility should not receive AI therapy. For sustained gain in growth potential, treatment at least through the early and middle stages of puberty appears necessary, whereas treatment during prepuberty may be less effective and associated with an increased risk of vertebral deformities.

Disclosures

All the authors state that they have no conflicts of interest.

Acknowledgements

This work was supported by the Foundation for Pediatric Research and the Academy of Finland, both in Helsinki, Finland.