Skeletal Microstructure in Addison's Disease

Abstract

Context

Addison's disease (AD) is characterized by deficient adrenal glucocorticoid (GC) production. Treatment involves just GC replacement, but patients often receive high doses, leading to side effects. Bone mineral density (BMD) data in AD are conflicting. High-resolution peripheral quantitative computed tomography (HRpQCT) evaluates volumetric BMD, microarchitecture, and mechanical properties of the tibia and radius. No studies have assessed bone quality by HRpQCT on subjects with AD.

Objective

Evaluate the bone health of patients with AD using HRpQCT.

Design

Cross-sectional study.

Setting

Ambulatory of a tertiary medical center.

Participants

Nineteen patients with AD on GC were compared to 38 matched controls.

Main Outcome Measures

Dual-energy x-ray absorptiometry and HRpQCT measurements.

Results

Patients with AD had lower lean mass and BMD. At the radius, the AD group had 11% lower trabecular (Tb) number (P = .03). At the tibia, patients had lower Tb number and Tb volumetric BMD, greater Tb separation, and lower cortical area and thickness (17-26% difference in these parameters, P < .03 for all). Tibial stiffness was 21% lower in the AD group (P < .03). In this group, there was a positive correlation between lean mass and stiffness (radius r = 0.53, tibia r = 0.51; both P < .04) and a negative correlation between cumulative GC dose and spine BMD (r = −0.67, P < .01).

Conclusion

This is the first study to assess bone using HRpQCT in patients with AD on GC. Our findings suggest that AD patients have a loss of lean mass and skeletal fragility, mainly of the trabecular compartment and at the tibia. Bone loss may be related to loss of lean mass and GC.

Primary adrenal insufficiency or Addison´s disease (AD) is a rare life-threatening condition characterized by adrenocortical glucocorticoid (GC) and mineralocorticoid (MC) deficiency. Autoimmune adrenalitis is the most common cause of acquired AD in high-income countries, with a prevalence of approximately 100 cases per million. Age onset is around 30 to 50 years, and women are more affected than men [1-3]. In developing countries, tuberculosis and paracoccidioidomycosis are also important causes of AD [4, 5].

Glucocorticoid and MC replacement therapy for patients with AD is often a challenging task [6]. Available GC formulations fail to replicate the physiological circadian and ultradian rhythmicity of cortisol secretion [7], and the threshold between physiologic and supraphysiologic GC doses is difficult to achieve [6]. Not infrequently, patients are exposed to higher than physiologic GC doses and to long-term side effects, including poor quality of life and high cardiovascular risk [8, 9].

Regarding bone health, there are several mechanisms by which chronic exposure to GC excess may result in skeletal deterioration [10, 11]. Long-term effects include direct inhibition of osteoblasts and, consequently, reduced bone formation. In addition, increased renal calcium excretion and decreased intestinal calcium absorption (due to inhibition of vitamin D's physiological actions to promote calcium absorption) induce a secondary hyperparathyroidism, in which bone resorption may be enhanced at the outset of therapy. Moreover, GC can have a myopathic effect, in which muscle weakness leads to a higher risk of falls, which may result in fractures. Also, in AD, adrenal androgens are also lacking. They may also have reduced hepatic IGF-1 secretion, leading to additional risks [12-14]. There is limited data regarding bone involvement in patients with primary adrenal insufficiency chronically receiving GC replacement.

The recommended method for evaluating bone mineral density (BMD) and for diagnosing osteoporosis is dual-energy x-ray absorptiometry (DXA) [15, 16]. However, DXA depicts an area-based BMD (aBMD) rather than a volumetric BMD (vBMD). Thus, either larger or smaller bones can induce a measurement artifact. Furthermore, DXA provides information about the amount of mineralized bone (density) that does not necessarily reflect bone quality.

High-resolution peripheral quantitative computed tomography (HRpQCT) is a noninvasive in vivo method by which skeletal features not accessible by DXA can be determined, including vBMD, cortical and trabecular microarchitecture, and mechanical properties of the distal tibia and radius [17-19]. Bone strength is estimated by applying the finite element analysis (FEA) technique to HRpQCT images.

While bone histomorphometry was not assessed in AD patients, data on BMD are controversial. Moreover, there is essentially no available information about bone structure in these patients. This is due, in part, to the fact that HRpQCT has not heretofore been applied to subjects with AD. This study aims to evaluate the bone health of patients with AD on chronic GC therapy using an HRpQCT technique. We hypothesized that such patients would have deteriorated bone microstructure and strength compared to controls.

Materials and Methods

Subjects

This cross-sectional study included postpubertal, nonpregnant individuals aged 16 years or older with an established diagnosis of primary adrenal insufficiency, defined by low basal serum cortisol levels (<5 mcg/dL) on at least 2 occasions and/or an impaired response to ACTH stimulation testing [6], in addition to classic clinical signs and symptoms. Participants were receiving GC replacement therapy for at least 1 year and were under routine follow-up at the Adrenal Clinic of the Federal University of Sao Paulo, Brazil.

Individuals with conditions or medications known to affect bone metabolism other than AD/GC were excluded. These included untreated hyperthyroidism, prolactinoma, Cushing's disease, primary hyperparathyroidism, hypoparathyroidism, osteomalacia, osteogenesis imperfecta, malabsorption syndromes, malignancies, chronic kidney or liver disease, organ transplantation, and conditions characterized by androgen excess (eg, congenital adrenal hyperplasia). Individuals with a history of bisphosphonate use (within the past 3 years), anticoagulants, methotrexate, aromatase inhibitors, thiazolidinediones, teriparatide, raloxifene, or denosumab were additionally excluded.

An age-, sex- and race-matched control group was included for comparison. The group was recruited through hospital flyers or by direct invitation. Controls were healthy individuals with no reported pathologies or use of medications that could significantly affect bone health as assessed by history. Race was defined by self-designation.

The study was approved by the Committee on Ethical Research of the Federal University of Sao Paulo. After being informed of the study's purposes and procedures, the participants signed a formal written consent form. Data were analyzed exclusively by the study researchers.

Study Protocol

Data on previous fractures, use of medications (including GC, MC, and vitamin D supplements); calcium intake; and the duration, etiology, and treatment of AD were obtained from medical records and patient interviews. Different GCs were converted to hydrocortisone equivalents (mg/m²/day). The GC equivalence ratios were based on the doses that lead to suppression of the corticotrophic axis and growth, as follows: 80 mg of hydrocortisone = 16 mg of prednisone = 13.3 mg of prednisolone = 1 mg of dexamethasone [20, 21]. The cumulative GC dose was calculated since 2012, when electronic medical records were initiated at the institution. Anthropometric data, including weight and height, were recorded from all participants.

Biochemical Analyses

Fasting morning blood samples were collected to measure biochemical parameters related to the adrenal and bone metabolism of AD patients. The laboratory evaluation was performed using standard methods for sodium, potassium, and total calcium. Plasma renin activity (PRA) was measured by ELISA, with intra- and interassay coefficients of variation of 4.1% and 11.0%, respectively. PTH and 25-hydroxyvitamin D were measured by electrochemiluminescence, as well as carboxyterminal telopeptide of type 1 collagen (CTX) and osteocalcin. Intra- and interassay coefficients of variation were, respectively, 3.0% and 3.5% for PTH, 8.5% and 9.2% for 25-hydroxyvitamin D, 1.1% and 1.2% for CTX, and 1.5% and 2.2% for osteocalcin (Elecsys—Cobas, Roche, Mannheim, Germany). Blood sample analyses were carried out in the same laboratory for all patients. All samples were stored at −20 °C.

DXA Analyses

aBMD was measured at the lumbar spine (LS; L1-L4), total hip, femoral neck, and distal radius using DXA (Hologic, Waltham, MA, USA). Vertebral fracture assessment (VFA) and whole-body composition, including total body fat percentage (BF%), android/gynoid ratio, fat mass index (fat mass/height²), and appendicular skeletal muscle index (Baumgartner index = appendicular skeletal mass/height²) were also evaluated. While BF%, android/gynoid ratio, and fat mass index are fat indexes, the Baumgartner is a lean mass index (higher levels mean proportionally higher muscle mass) [22].

HRpQCT Analyses

HRpQCT scans of the nondominant side were acquired using a second-generation system (XCT2, Scanco Medical, Brüttisellen, Switzerland). Limbs were secured in a manufacturer-provided fiber-carbon cast and positioned within the scanner's central cavity.

For each scan, a 2D x-ray image (scout view) was obtained to identify the region to be analyzed and to position the reference line (on the inflection point on the endplate of the distal radius and tibial plafond). Subsequently, using the standard manufacturer protocol, the XCT2 system acquired 168 slices of a 10.24 mm section with a 61 µm isotropic resolution, beginning 9.0 and 22.0 mm proximal to the reference line, respectively, at the radius and tibia [23]. The entire acquisition process took approximately 2 minutes per site.

Sectional and 3D images were generated. Cortical and trabecular bone compartments were segmented using a semiautomated technique. Subsequently, microarchitectural and volumetric density parameters were quantified for both compartments. Finite element analysis, through the parameter Stiffness, was used to assess mechanical properties.

Statistical Analysis

The statistical analysis was performed using Microsoft Excel Version 16.87 (Microsoft, Redmond, WA, USA) and SAS Version 9.4 (SAS Institute, Cary, NC, USA). Absolute (n) and relative (percent) frequencies were measured for qualitative variables and mean ± SD or median (minimum-maximum) and 95% confidence intervals for quantitative variables.

Between-group comparisons using the chi-square test for categorical variables and a 2-sided t-test for continuous variables and standardized mean differences between groups were calculated. The significance level adopted was P ≤ .05. Adjustment for multiple comparisons for highly correlated trabecular HRpQCT parameters using Bonferroni correction was tested, as suggested in a recent publication [24]. Pearson correlations between clinical, biochemical, and bone parameters were performed in the AD group.

Results

Demographics, Clinical, Biochemical, and DXA Analyses

The descriptive and biochemical characteristics of the 19 AD patients are shown in Table 1. The median age was 52 years (24-82), 37% were women (71% postmenopausal), and 63% were White. The most common etiologies were autoimmune (63%) and infectious (16%). The median time since diagnosis was 6 years (2-33), and the initial age at GC treatment was 33 years (0-78). Six patients had a history of previous fractures, all in the extremities. No vertebral fractures were identified on imaging evaluation by VFA.

Table 1.

Demographic, clinical, and biochemical characteristics of Addison’s disease patients, mean ± SD (n = 19)

		Reference values
Age (years)	50.8 ± 17.7
Female, n (%)	7 (37)
Race, n (%)
White	12 (63)
Black	3 (16)
Brown	4 (21)
Years since diagnosis	13.2 ± 11.2
Addison's disease etiology, n (%)
Autoimmune	12 (63)
Infectious	3 (16)
Othera	4 (21)
Regular exercise, n (%)	10 (53)
History of smoking cigarettes, n (%)	6 (32)
History of alcohol abuse, n (%)	2 (11)
Diabetes therapy with insulin, n (%)	3 (16)
Previous fractures, nb (%)	6 (32)
Prevalent vertebral fractures by vertebral fracture assessment, n	0
Calcium intake (diet plus supplementation), mg/day	903 ± 440
Vitamin D supplementation, n (%)	16 (84)
Vitamin D supplementation dose IU/day	1675 ± 755
Fludrocortisone use, n (%)	18 (95)
Fludrocortisone dose, mcg/day	108 ± 65
Age glucocorticoid was begun, years old	37.7 ± 19.6
Current glucocorticoid, n (%)
Prednisone	14 (74)
Prednisolone	5 (26)
Current glucocorticoid dose (hydrocortisone equivalence/body surface area), mg/m2/day	17.4 ± 6.0
Cumulative glucocorticoid dose (hydrocortisone equivalence), grams	80.7 ± 50.3
Sodium, mmol/L	138.6 ± 3.8	137-148
Potassium, mmol/L	5.0 ± 0.6	3.5-5.0
Plasma renin activity, ng/mL/h	6.1 ± 8.1	0.30-4.37
PTH, pg/mL	36.6 ± 11.3	15-65
Calcium, mg/dL	9.7 ± 0.5	8.6-10.2
Osteocalcin, ng/mL	18.0 ± 6.4	10-37
25OH-vitamin D, ng/mL	34.1 ± 11.3	20-60
s-CTX, ng/mL	0.305 ± 0.160	0.161-0.737

Abbreviations: CTX, carboxyterminal telopeptide of type 1 collagen.

^a Other causes: 2 adrenoleukodystrophy, 1 idiopathic, 1 adrenal hypoplasia.

^b Fracture locations: 1 wrist, 3 arm, 2 hand, 1 forearm.

Prednisone was used by 74% of patients, whereas the remaining took prednisolone. The mean daily GC dose was 29.6 ± 8.3 mg (17.4 ± 6.0 mg/m²), and the cumulative GC dose since 2012 was 80.7 ± 50.3 grams, expressed in hydrocortisone equivalents. All but 1 patient received fludrocortisone, with a mean daily dose of 108 ± 65 mcg. Sixteen patients (84%) took vitamin D supplements, and the estimated mean calcium intake (diet plus supplementation) was 903 ± 440 mg/day.

Despite MC (and GC) replacement, PRA was above the normal range, with potassium in the upper limit and sodium near the lower limit. All other blood parameters were within the normal range. The mean bone turnover markers (BTM) were within the reference range, with 4 patients presenting lower osteocalcin levels and 1 with lower CTX levels. No patient was above the reference range for BTM (Table 1).

The control group consisted of 38 individuals with no significant differences in anthropometric measurements compared to the AD group. Table 2 presents the group comparisons for DXA parameters. Compared to controls, patients with AD had lower areal BMD at the LS, femoral neck, and total hip by 10.3%, 10.8%, and 9.5%, respectively, with no difference at the radius.

Table 2.

Anthropometric and regional BMD and body composition by DXA in Addison’s patients and healthy controls, mean ± SD (95% CI)

	Addison’s disease, n = 19	Control, n = 38	P-value (SMD)
Age, years	50.8 ± 17.7	48.3 ± 14.7	.57
Female, n (%)	7 (37)	14 (37)	1.00
Race, n (%)
White	12 (63)	24 (63)	.83
Black	3 (16)	8 (21)
Brown	4 (21)	6 (16)
Height, m	1.63 ± 0.11	1.65 ± 0.10	.42
Weight, kg	68.9 ± 13.2	71.9 ± 12.9	.42
BMI, kg/m2	26.0 ± 4.5	26.3 ± 4.4	.80
Lumbar spine, g/cm2	0.942 ± 0.131 (0.879, 1.005)	1.051 ± 0.17 (0.996, 1.107)	.02 (−0.69)
T-score	−1.1 ± 1.2 (−1.7, −0.6)	−0.2 ± 1.5 (−0.7, 0.3)	.02 (−0.66)
Z-score	−0.3 ± 1.3 (−0.9, 0.4)	0.3 ± 1.7 (−0.2, 0.9)	.19 (−0.37)
Femoral neck, g/cm2	0.753 ± 0.132 (0.690, 0.817)	0.845 ± 0.157 (0.794, 0.897)	.03 (−0.62)
T-score	−1.1 ± 1.0 (−1.6, −0.7)	−0.4 ± 1.2 (−0.8, −0.1)	.03 (−0.63)
Z-score	−0.3 ± 1.0 (−0.8, 0.2)	0.3 ± 1.2 (−0.1, 0.6)	.10 (−0.47)
Total hip, g/cm2	0.857 ± 0.144 (0.787, 0.927)	0.947 ± 0.145 (0.899, 0.995)	.03 (−0.62)
T-score	−1.0 ± 0.9 (−1.5, −0.6)	−0.4 ± 1.0 (−0.7, 0.0)	.02 (−0.66)
Z-score	−0.4 ± 0.9 (−0.9, 0.0)	0.1 ± 1.0 (−0.3, 0.4)	.07 (−0.51)
Distal radius, g/cm2	0.709 ± 0.180 (0.622, 0.796)	0.716 ± 0.082 (0.689, 0.743)	.88 (−0.06)
T-score	−1.4 ± 1.6 (−2.2, −0.6)	−1.0 ± 1.1 (−1.4, −0.6)	.32 (−0.28)
Z-score	−0.5 ± 1.7 (−1.3, 0.3)	−0.4 ± 1.1 (−0.7, 0.0)	.76 (−0.10)
Body fat, %	35.6 ± 12.7 (29.5, 41.7)	30.0 ± 9.0 (27.0, 32.9)	.07 (0.54)
Android to gynoid ratio	1.00 ± 0.14 (0.93, 1.07)	1.09 ± 0.28 (1.00, 1.18)	.13 (−0.36)
Fat mass index, kg/m2	9.4 ± 4.7 (7.1, 11.7)	7.9 ± 3.3 (6.8, 9.0)	.19 (0.38)
Baumgartner, kg/m2	6.39 ± 1.09 (5.86, 6.91)	7.39 ± 1.32 (6.95, 7.82)	<.01 (−0.80)

Bold numbers indicate P ≤ .05.

Abbreviations: BMD, bone mineral density; BMI, body mass index; CI, confidence interval; DXA, dual-energy x-ray absorptiometry; SMD, standardized mean difference.

The Baumgartner index was 13.5% lower in the AD patients (P < .01) with no differences in the fat indexes (Table 2), although there was a trend to higher BF% in the AD group (35.6 ± 12.7 vs 30.0 ± 9.0, P = .07).

HRpQCT Comparisons

As shown in Table 3, there were several between-group differences in bone microarchitectural parameters, more significant at the tibia. At this site, AD patients demonstrated lower total volumetric BMD, trabecular vBMD (TbvBMD), trabecular bone volume fraction (TbBV/TV), and trabecular number (TbN), with higher trabecular separation (TbSp) (18-26% difference in these parameters, P < .02 for all). There was also a trend toward decreased trabecular thickness (TbTh). Cortical differences at the tibia were also observed, with 22% and 14% lower cortical thickness and cortical area, respectively, and a trend toward decreased cortical vBMD. Correcting for multiple comparisons between highly correlated trabecular parameters (TbBV/TV, TbN, TbTh, TbSp) using the Bonferroni correction with a P-value threshold of .0125, the differences remained significant.

Table 3.

HRpQCT-derived bone parameters at the distal radius and tibia in Addison’s disease patients and controls, mean ± SD (95% CI)

Bone parameter	Addison’s disease, n = 19	Control, n = 38	P-value (SMD)
Distal radius parameters
Cortical porosity, %	0.67 ± 0.38 (0.49, 0.85)	0.77 ± 0.66 (0.55, 0.99)	.47 (−0.17)
Cortical volumetric BMD, mgHA/cm3	897.0 ± 57.2 (869.4, 924.6)	916.8 ± 59.1 (897.3, 936.2)	.23 (−0.34)
Cortical thickness, mm	1.07 ± 0.30 (0.93, 1.22)	1.16 ± 0.23 (1.08, 1.24)	.24 (−0.33)
Cortical area, mm2	63.2 ± 20.8 (53.2, 73.3)	69.0 ± 14.6 (64.3, 73.8)	.23 (−0.34)
Trabecular bone volume fraction	0.221 ± 0.075 (0.185, 0.257)	0.260 ± 0.073 (0.236, 0.284)	.07 (−0.53)
Trabecular number, 1/mm	1.35 ± 0.25 (1.23, 1.47)	1.50 ± 0.23 (1.42, 1.57)	.03 (−0.62)
Trabecular thickness, mm	0.23 ± 0.03 (0.22, 0.25)	0.24 ± 0.02 (0.23, 0.24)	.53 (−0.18)
Trabecular separation, mm	0.73 ± 0.21 (0.63, 0.83)	0.62 ± 0.12 (0.58, 0.67)	.053 (0.67)
Trabecular volumetric BMD, mgHA/cm3	152.5 ± 51.9 (127.5, 177.5)	180.3 ± 48.9 (164.3, 196.4)	.052 (−0.56)
Trabecular area, mm2	215.1 ± 71.6 (180.6, 249.6)	218.6 ± 61.0 (198.5, 238.6)	.85 (−0.05)
Total area, mm2	274.6 ± 76.2 (237.9, 311.3)	283.8 ± 64.6 (262.6, 305.1)	.63 (−0.13)
Stiffness, kN/mm	70.2 ± 30.0 (55.7, 84.6)	81.0 ± 25.8 (72.5, 89.4)	.16 (−0.40)
Distal tibia parameters
Cortical porosity, %	2.71 ± 1.29 (2.08, 3.33)	2.40 ± 1.66 (1.85, 2.95)	.49 (0.20)
Cortical volumetric BMD, mgHA/cm3	879.5 ± 60.9 (850.1, 908.8)	911.7 ± 60.1 (891.9, 931.4)	.06 (−0.53)
Cortical thickness, mm	1.35 ± 0.32 (1.19, 1.50)	1.64 ± 0.34 (1.53, 1.76)	.003 (−0.89)
Cortical area, mm2	121.7 ± 32.2 (106.2, 137.3)	141.9 ± 29.5 (132.2, 151.6)	.02 (−0.66)
Trabecular bone volume fraction	0.217 ± 0.047 (0.194, 0.240)	0.265 ± 0.063 (0.244, 0.286)	.005 (−0.82)
Trabecular number, 1/mm	1.11 ± 0.20 (1.01, 1.20)	1.30 ± 0.24 (1.22, 1.38)	.004 (−0.85)
Trabecular thickness, mm	0.25 ± 0.02 (0.24, 0.26)	0.27 ± 0.02 (0.26, 0.27)	.055 (−0.55)
Trabecular separation, mm	0.91 ± 0.24 (0.80, 1.03)	0.75 ± 0.15 (0.70, 0.80)	.01 (0.88)
Trabecular volumetric BMD, mgHA/cm3	142.6 ± 35.1 (125.7, 159.4)	179.3 ± 46.1 (164.1, 194.4)	.003 (−0.86)
Trabecular area, mm2	617.6 ± 130.2 (554.8, 680.3)	554.3 ± 139.5 (508.4, 600.1)	.10 (0.46)
Total area, mm2	733.8 ± 137.6 (667.5, 800.1)	690.9 ± 149.2 (641.9, 740.0)	.30 (0.29)
Stiffness, kN/mm	177.5 ± 57.1 (150.0, 205.0)	225.1 ± 103.6 (191.0, 259.1)	.03 (−0.52)

Bold numbers indicate P ≤ .05.

Abbreviations: CI, confidence interval; HRpQCT, high-resolution peripheral quantitative computed tomography; SMD, standardized mean difference.

No significant differences were observed between groups in cortical parameters at the radius, although the trabecular compartment appeared slightly more affected in the AD group, with 11% lower TbN and a trend toward reduced TbvBMD and TbBV/TV, together with increased TbSp.

FEA revealed 21.1% lower tibia stiffness in the AD group (P < .03), with no significant difference at the radius. Figure 1 illustrates HRpQCT scans of 1 AD patient and 1 healthy control, both males of similar age.

Figure 1.

HRpQCT images of the radius (A) and tibia (C) of a 24-year-old Addison patient, and the radius (B) and tibia (D) of a 21-year-old healthy control, both males.

Correlations Concerning Bone Health

In the AD group, we found a moderate positive correlation between the Baumgartner index and stiffness at both the radius and tibia, indicating that greater muscle mass was associated with better mechanical competence (Fig. 2). Several microarchitectural parameters, including both trabecular and cortical, showed moderate negative correlations with sodium levels and positive correlations with potassium and PRA levels (Table 4). There was no significant correlation between fludrocortisone or BTM and any HRpQCT parameter.

Figure 2.

Scatter plots showing positive correlations between the Baumgartner Index and stiffness at both the radius (A) and tibia (B) in Addison’s disease patients.

Table 4.

Correlations between Na, K, PRA, and HRpQCT parameters in the Addison’s disease group

	Radius TotvBMD		Radius TbvBMD		Radius TbTh		Radius CtAr		Radius CtTh		Tibia TotvBMD		Tibia CtTh
	r	P	r	P	r	P	r	P	r	P	r	P	r	P
Na	−0.66	.005	−0.51	.04	−0.71	.002	−0.53	.04	−0.66	.006	−0.64	.007	−0.68	.004
K	0.56	.03	0.54	.04	0.60	.02	0.55	.03	0.58	.02	0.41	.13	0.49	.06
PRA	0.72	.001	0.55	.03	0.62	.01	0.51	.04	0.67	.004	0.62	.01	0.52	.04

Bold numbers indicate P ≤ .05.

Abbreviations: CtAr, cortical area; CtTh, cortical thickness; HRpQCT, high-resolution peripheral quantitative computed tomography; K, potassium; Na, sodium; PRA, plasma renin activity; TbTh, trabecular thickness; TbvBMD, trabecular volumetric bone mineral density; TotvBMD, total volumetric bone mineral density.

There were no significant correlations between bone microarchitecture parameters and disease duration, although we observed some trends in tibia total vBMD (r = −0.45, P = .054) and tibia cortical thickness (r = −0.43, P = .067). However, cumulative GC dose was moderately negatively correlated with spine aBMD (r = −0.67, P < .01), as illustrated in Fig. 3. This correlation remained statistically significant after excluding an outlier patient (r = −0.45, P < .03).

Figure 3.

Scatter plot showing a negative correlation between cumulative glucocorticoid and lumbar spine bone mineral density in Addison’s disease patients.

Discussion

To our knowledge, this is the first study to assess bone microarchitecture using HRpQCT in patients with AD. Our results indicate that AD patients on GC have reduced aBMD, impaired bone microarchitecture, and lower muscle mass, which was more pronounced in the trabecular compartment and at the tibia. This was associated with reduced strength at the tibia. These results imply that patients with AD on GC replacement are at risk for the consequence of more fragile bone.

Previous studies are conflicting with regard to aBMD in patients with AD. Some show no difference [25-27], while others indicate lower aBMD, especially in postmenopausal women [28, 29], hypogonadal men [30, 31], and those on more potent GC (prednisolone compared to hydrocortisone) [32] or higher GC doses [33]. In our study, AD patients had lower aBMD at the LS and hip sites compared to controls. This could be anticipated, as none of our patients were on low-potency steroids (hydrocortisone), and most women were postmenopausal. In addition, AD patients used a mean hydrocortisone equivalent GC dose of 17.4 mg/m²/day, which is above physiologic levels, estimated as 9 to 11 mg/m²/day [34-36]. Some studies suggest that aBMD impairment occurs when the GC dose is above 12 to 13 mg/m²/day in AD patients on long-term treatment [33, 37].

Bone mineral density at the LS was inversely correlated with cumulative GC dose, a finding previously observed [27, 37, 38] that corroborates the relationship between GC and bone loss in AD patients.

We also found that AD patients have impaired microarchitectural parameters. Bone microarchitecture in AD patients has only been evaluated previously with trabecular bone score (TBS) [39]. A study comparing AD patients with matched controls showed no differences in TBS (1.32 ± 0.1 vs 1.30 ± 0.1, P = .63). Unlike ours, the study showed no difference in aBMD between the groups [27]. Compared to our data, the patients in the study had similar age and disease duration. However, the mean GC dose (25.8 ± 6.2 mg/day) was lower than that of our patients (29.6 ± 8.3 mg/day), which may explain the neutral aBMD results. Nevertheless, the analysis showed an inverse correlation between TBS and disease duration in AD patients (r = −0.48, P = .009), suggesting that prolonged GC use may be related to microarchitecture deterioration [27]. TBS, as an indirect method, has lower sensitivity to detect microarchitectural changes than HRpQCT and is just weakly to moderately associated with microarchitectural and FEA indices by HRpQCT [40-42].

Data from other conditions that require chronic GC use have shown impaired microarchitecture by TBS [43, 44] and HRpQCT [45-47], although it's important to highlight that these patients use GC at anti-inflammatory, not replacement, doses.

Notwithstanding, patients with endogenous hypercortisolism, even subclinical, also seem to have microarchitectural skeletal impairment. In a study with 75 patients with adrenal incidentalomas, those with subclinical hypercortisolism, compared to those with nonfunctioning nodules, had lower aBMD and worse trabecular architecture (lower TbvBMD, TbBV/TV, and TbTh) at the radius but not at the tibia [48], as assessed by HRpQCT. These data suggest that even minimal increases in cortisol levels may be associated with microstructural deterioration, endorsing our results.

Microarchitectural abnormalities and decreased BMD were accompanied by impaired bone strength by FEA at the tibia in our patients. Data with chronic GC users on anti-inflammatory doses have also demonstrated decreased stiffness [45, 46]. The reduction of bone strength observed in our study suggests that GC-treated AD patients are at increased risk for fracture [49, 50]. Studies that have assessed fracture risk in AD are consistent with this finding. Bjornsdottir et al found a higher risk of hip fractures in AD patients compared to controls (hazard ratio = 1.8, 95% confidence interval 1.6-2.1, P < .001), especially in females diagnosed longer [51]. Camozzi et al demonstrated that AD patients have more than a 3-fold risk of vertebral fractures as compared to healthy individuals, and the risk was also associated with a longer disease duration [52]. Although our patients had lower bone stiffness than controls, none had a vertebral fracture by VFA or a history of hip fractures. Nonetheless, 32% had a history of limb fractures.

Consistent with prior studies, mean BTM levels were within the reference range in our patients [25, 29, 37, 38]. In contrast, in individuals using supraphysiologic anti-inflammatory doses of GC, a decrease in formation markers is observed [53-55], consistent with the long-term effect of GCs to reduce bone formation. These differences may be due, at least in part, to differences in the mean GC dose. Most AD patients use replacement GC doses, which are often above the physiologic level but not high enough to suppress formation markers. Yet, a few patients in our study had suppressed BTM and mean osteocalcin levels that were in the lower normal range. The normal levels of PTH and CTX suggest no remarkable increases in bone resorption or urinary calcium loss.

Changes in body composition in patients using supraphysiologic GC doses are well recognized [56-58]. However, it is notable that even replacement doses of GC can lead to body composition changes, as our patients had remarkably lower muscle mass and a trend toward higher BF% compared to healthy individuals.

A relationship between muscle mass/strength and aBMD has been previously observed in anti-inflammatory doses GC users [56, 58] and non-GC users [59]. In older [60] and obese [61] adults, the lean mass explained 28% to 48% of variance in bone strength. However, the relationship between muscle mass and bone microstructure or strength in individuals on GC has not been investigated so far. We demonstrated a positive correlation between muscle mass and bone stiffness in AD patients on replacement GC.

Decreased muscle mass in patients with AD may be due, in part, to GC use, a condition described as “steroid myopathy” [62]. Our data suggest that patients with AD, on “replacement doses“of GC that do not affect BTM, usually use supraphysiological doses and are not exempt from musculoskeletal impairment caused by the GC. In addition, adrenal insufficiency may be associated with higher levels of myostatin, a protein that negatively regulates muscle mass, as suggested in a preclinical study [63]. The fact that the mean BTMs are normal, with significant muscle loss (and a correlation between lean mass and bone stiffness), may suggest that these body composition changes play an important role in the pathophysiology of bone impairment of AD patients on replacement GC.

The degradation of the microarchitecture was surprisingly more prominent at the tibia. One might expect that bone damage could be mitigated at this site, since it is a weight-bearing area, as observed in studies with endogenous hypercortisolism and general GC users [45, 48]. Further studies are needed to confirm and possibly explain this difference. We hypothesize that the loss of lean mass would affect predominantly the lower limb skeleton, as it usually receives more muscle stimulus and mechanical stress than the upper limbs.

One possible confounding effect for our results is the gonadal status, especially in men, as they comprised most of our patients. In addition to directly affecting bone health, chronic GC use may decrease testosterone levels, which can potentiate bone degradation [12, 64]. Studies assessing aBMD in AD patients show that hypogonadal men are among the populations at risk for bone loss [30, 31]. Furthermore, the drop in testosterone levels intensifies muscle loss, which also contributes to bone impairment [12, 64]. As we did not measure testosterone levels, we could not determine the gonadal status of the male participants.

Microarchitectural parameters correlated negatively with sodium and positively with potassium and PRA. We postulate that these correlations are due to the GC action on mineralocorticoid receptors, so the higher the GC dose, the higher the sodium and the lower the potassium and PRA, and the worse the bone parameters. Also, mineralocorticoids have a potential impact on bone, leading to urinary calcium loss (calcium/sodium transport is coupled in the nephron), as observed in patients with primary aldosteronism [65, 66]. Although 95% of our patients were on fludrocortisone, they were not on excessive doses (as verified by normal potassium and not suppressed PRA); it may, however, have contributed to bone impairment.

Our study has several strengths. It is the first study to assess bone microstructure using HRpQCT in patients with AD. We used the second-generation HRpQCT instrument, which has superior image resolution compared to first-generation scanners and provides a direct measurement of trabecular thickness and separation (these are measured indirectly on XCT1). Moreover, the XCT2 scan is acquired faster, resulting in fewer movement artifacts [17, 18]. In addition to volumetric BMD and microarchitecture, we assessed bone strength using FEA. Although we used the standard acquisition protocol and not a relative offset, our cases and controls were height-matched to ensure the region of interest was the same.

Our study also has some limitations. It is a cross-sectional study with a relatively small, convenience-based sample size. As this is the first study using HRpQCT in AD, we did not perform a power calculation for sample size. However, the current sample size is similar to many other HRpQCT studies, and we could detect several differences in bone parameters. Due to the relatively small sample size, we did not analyze the data separately for men and women. However, patients and controls were sex matched. Another limitation is the lack of control group biochemical data for comparison, although this group was healthy, and there were no remarkable changes in the biochemical parameters of the patients. Finally, the cumulative GC dose may have been underestimated since the information was only available from 2012. Even so, we found an inverse correlation with aBMD.

In conclusion, this is the first study that presents data on bone microarchitecture, as measured by HRpQCT, in patients with AD. Beyond the bone and lean mass loss by DXA, our findings suggest that AD patients on chronic GC replacement therapy have bone microarchitectural and strength abnormalities, more pronounced in the trabecular bone and at the distal tibia. Bone fragility may be related to muscle loss and cumulative GC dose. Our data highlight the importance of assessing bone health in patients with AD and suggest clinicians should address and prevent muscle loss, avoiding high GC doses as much as possible.

Funding

This work was supported by the Conselho Nacional de Desenvolvimento Científico e Tecnológico (Grant number 142470/2020-1) and the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - Finance Code 001, which provided fellowships. The HRpQCT machine was funded by the Financiadora de Estudos e Projetos (Grant number 0208/12-RPMI).

Disclosures

The authors have nothing to disclose.

Data Availability

Some or all datasets generated during and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.