PREVALENCE AND PREDICTORS OF BONE DENSITY LOSS IN MEN WITH LOW TESTOSTERONE

Abstract

Background:

Low Testosterone (T) is associated with several sequelae, including a high prevalence of bone density loss (BDL).

Aim:

We aimed to identify predictors of BDL in men with low T.

Methods:

The sample included (i) men ≥50 years old, (ii) with low total T levels (<300 ng/dL using LCMS on 2 early morning blood draws), (iii) who had dual-energy X-ray absorptiometry (DEXA) within 6 months of T measurement. Men with prior T therapy were excluded. On DEXA, osteopenia was defined as a T score between −1.0 and −2.5, and osteoporosis as a T score below −2.5. Additionally, demographics and comorbidity data were collected. We report BDL rates and identify predictors of any BDL using logistic regression.

Outcomes:

BDL, which was defined as the presence of osteopenia or osteoporosis

Results:

997 men were analyzed, with a median age of 64 (IQR 59, 70) years. Median total T was 202 (IQR: 107, 256) ng/dL, median free T 5.8 (IQR: 4.3, 7.5) ng/dL. 24% had a total T ≤ 100 ng/dL. 33% had ≥ 3 comorbidities. 74% of our cohort had an oncological history; most of them had a history of prostate cancer. BDL was present in 36% of the patients (89% osteopenia, 11% osteoporosis). On MVA, significant predictors of BDL were lower total T levels (OR = 1.09), older age (OR = 1.59), and the presence of ≥3 comorbidities (OR = 1.60).

Clinical Implications:

It is critical to be aware of the significant prevalence of BDL in men with low T. Early identification may enable targeted counseling, the initiation of preventive or therapeutic strategies to improve bone density outcomes, and the reduction of the risk of fractures and other serious skeletal complications.

Strengths & Limitations:

Limitations related to retrospective design, including selection bias and unmeasured confounding factors. Another limitation is that most cases had an oncological history. However, several strengths include the use of a large, systematically audited database; T labs were assessed using gold-standard methodology; and bone mineral density was assessed using a standardized scanning protocol.

Conclusion:

In our cohort of older men, most with oncological history and low T, around one-third had BDL, and around 10% of those with BDL had osteoporosis. Lower T levels, older age, and a greater number of comorbidities were significant predictors of BDL.

INTRODUCTION

Low testosterone (T) is defined by the American Urological Association (AUA) guidelines as two early morning serum total T levels below 300 ng/dL.¹ T deficiency is defined as low T levels combined with symptoms and/or signs. Such symptoms/signs include variable degrees and combinations of diminished sexual function, difficulties in reaching orgasm, low sex drive, mood changes, feeling depressed, irritability, subjective low energy levels, reduced strength, reduced endurance, reduced exercise response, reduced work productivity, muscle loss, fat gain, and cognitive problems. Long-term sequelae of low T include increased risk of cardiovascular events, bone density loss (BDL), including osteopenia and osteoporosis, perturbations in glycemic control and overall mortality.^2–5

The prevalence of low T in the general population is reported to range from 2-50%.^6–8 The Baltimore Longitudinal Study of Aging reported the prevalence of low T at 12% among men in their 50s, 20% among men in their 60s, 30% among men in their 70s, and 50% among men in their 80s.⁹ BDL is associated to low T.¹⁰ In men with low T, bone density decreases due to reduced bone formation and increased bone resorption.¹⁰

It is crucial to recognize the potential consequences of missing a BDL diagnosis and to take the necessary actions to protect and improve bone density, including vitamin D and calcium supplementation, as well as medications such as bisphosphonates, denosumab, and parathyroid hormone-related protein analogs (teriparatide and abaloparatide) that stimulate new bone formation.

This analysis was performed to assess the prevalence of BDL and its predictors in men with low T.

METHODS

Study Population:

Data were collected from a departmental database approved by the institutional review board (IRB 16-1526). This retrospective study included patients referred to an Andrology clinic from 2002-2025. We evaluated (i) men aged ≥50 years (ii) with symptomatic low total T levels defined as less <300 ng/dL, using liquid chromatography tandem mass spectrometry (LC-MS) based on two early morning blood draws, and (iii) who underwent dual-energy X-ray absorptiometry (DEXA) within 6 months of the T measurements. DEXA was ordered as part of the low T assessment before starting potential T therapy. All patients were diagnosed with low T according to the AUA guidelines, which require two early-morning total T levels ≤300 ng/dL. Men who had received prior T therapy were excluded from the study. Patient demographics, health history, clinical characteristics, and laboratory assessments were prospectively entered into the database, which undergoes regular audits to ensure the accuracy of the collected data. Information regarding demographics and comorbidities (such as diabetes, sleep apnea, coronary artery disease, hypertension, dyslipidemia, and smoking status) were recorded. All total T measurements were obtained early in the morning and analyzed by a single institutional laboratory using LC-MS to minimize inter- and intra-test variability. Free T was assessed by equilibrium dialysis. Finally, the use of medications associated with bone density loss were also recorded, including proton pump inhibitors and corticosteroids.

Dual-energy X-ray absorptiometry (DEXA):

DEXA is a non-invasive imaging test that utilizes low-dose X-rays to measure bone mineral density. It is considered to be the gold standard for diagnosing and monitoring BDL and the data generated permits the calculation of future fracture risk. Bone density measurements are reported at three key locations: the lumbar spine, hip, and femoral neck. The T-score is used for these assessments, which compares an individual’s bone density to that of a healthy young adult of the same gender. The T-score is used to assess bone density in men aged 50 and older. T-scores are reported for each of the three points measured. Osteopenia is defined as a T-score between −1.0 and −2.5, while osteoporosis is indicated by a T-score below −2.5. A normal bone density result is reported when the T-score is above −1.0. For this purpose of this analysis, the assignment of bone density abnormality was based on the lowest T-score among the three sites assessed. Abnormal BDL was diagnosed when any of the three reported T-scores indicated osteopenia or osteoporosis.

Statistical Analysis:

We report BDL rates for this population based on DEXA T-scores, and we tested the association between the predictors of interest and BDL to identify significant predictors of BDL. Univariate and multivariable Logistic regression analyses were used to evaluate predictors. A multivariable logistic regression model was employed to assess factors associated with BDL. The dependent variable was defined as the presence of BDL, following established clinical criteria. The independent variables included total T level (modeled as a decrease of 50 ng/dL), age (measured as a decade increase), number of comorbidities (specifically, having three or more comorbidities), and exposure to ADT (yes). These variables were selected in advance based on our clinical experience as relevant predictors of BDL. All covariates were entered into the model simultaneously to estimate adjusted odds ratios along with 95% confidence intervals. Statistical significance was defined a priori as a two-sided p-value of less than 0.05. All analysis was conducted using SPSS v29.

RESULTS

Patient Population:

The final analysis included 997 patients, with a median age of 64 years (IQR: 59, 70 years). Among them, 39% had obstructive sleep apnea, 24% had diabetes, and 9% had coronary artery disease. The median number of comorbidities was 2 (IQR: 1, 3), and 1/3 of the cohort had ≥3 comorbidities. In terms of hormone assessment, the median total T level was 202 ng/dL (IQR: 107, 256), and median free T level 5.8 ng/dL (IQR: 4.3, 7.5). The median estradiol level was 16 pg/mL (IQR: 11, 20), the median LH was 4.4 mU/mL (IQR: 2.8, 8.1), and the median PSA was 0.02 ng/mL (IQR: 0.02, 0.77). Additionally, 49% of patients had total T levels ≤200 ng/dL, and 24% levels ≤100 ng/dL. Most of the cohort had a history of prostate cancer, 46% undergoing radical prostatectomy, 39% received prostate radiation therapy, and 28% were treated with androgen deprivation therapy (ADT). The median duration of ADT exposure was 7 months (IQR: 3, 26), 52% ≥6 months, 39% ≥12 months and 26% ≥24 months or more. Additionally, 17% reported exposure to proton pump inhibitors, and 8% reported exposure to corticosteroids. Table 1 summarizes patient data.

Table 1:

Patient Demographics Characteristics for the Entire Cohort of Men with Low Testosterone.

Variable (N=997)
Age Median (IQR), years	64 (59, 70)
Race (%)
White	78%
Black	12%
Asian	4%
Other	6%
Comorbidities (%)
Hypertension	58%
Dyslipidemia	58%
Obstructive Sleep Apnea	39%
Diabetes	24%
Coronary Artery Disease	9%
Comorbidities Median (IQR)	2 (1, 3)
3 or more comorbidities (%)	33%
Total Testosterone Median (IQR), ng/dL	202 (107, 256)
Testosterone ≤200 ng/dL (%)	49%
Testosterone ≤100 ng/dL (%)	24%
Free Testosterone, Median (IQR), ng/dL (N=753)	5.8 (4.3, 7.5)
LH Median (IQR), mU/mL (N=741)	4.4 (2.8, 8.1)
Estradiol Median (IQR), pg/mL (N=726)	16 (11, 20)
PSA Median (IQR), ng/mL (N=932)	0.02 (0.02, 0.77)
Cancer (%)
No Cancer	26%
Prostate	53%
Bladder	4%
Testicular	3%
Other	14%
Prostate Cancer Treatment (%)
Radical Prostatectomy	46%
Radiation therapy	39%
Androgen Deprivation Therapy	28%
Time on Androgen Deprivation Therapy Median (IQR), months (N=274, 27%)	7 (3, 26)
Androgen Deprivation Therapy ≥ 6 months (%)	52%
Androgen Deprivation Therapy ≥ 12 months (%)	39%
Androgen Deprivation Therapy ≥ 24 months (%)	26%
History of Unilateral Radical Orchiectomy (%)	4%
History of Chemotherapy (%)	8%
Medication Exposure (%)
Proton Pump Inhibitor	17%
Corticosteroids	8%
Selective Serotonin Reuptake Inhibitors	6%
Thiazolidinediones	1%

LH = Luteinizing Hormone, PSA = Prostate Specific antigen, IQR = Interquartile Range.

Bone Density Loss and Predictors:

In terms of bone density outcomes, 36% met criteria for BDL. Among these individuals, the vast majority (89%) had osteopenia, the remainder osteoporosis (11%). A sub-analysis was performed to compare the prevalence of BDL between men with and without ADT exposure. A significant difference was observed: among men with positive exposure, the prevalence of BDL was 42% (osteopenia 81%, osteoporosis 9%), vs. among men without exposure, the prevalence of BDL was 34% (osteopenia 84%, osteoporosis 16%), Figure 1. Predictors of BDL on univariable logistic regression analysis included: older age - OR 1.61 (p < 0.001); the presence ≥3 comorbidities - OR 1.56 (p = 0.002); lower total T levels (as a continuous variable) - OR 1.14 (p < 0.001) for lower total T levels (continuous variable); total T ≤ 200 ng/dL - OR 1.43 (p = 0.006); total T ≤ 100 ng/dL - OR 1.71 (p < 0.001). Additional predictors of BDL included: lower free T level - OR 1.11 (p = 0.001); higher LH levels - OR 1.04 (p < 0.001); exposure to ADT - OR 1.42 (p = 0.015); exposure to proton pump inhibitors - OR 1.54 (p = 0.012); exposure to corticosteroids - OR 2.0 (p = 0.003). On multivariable analysis, three predictors remained significant: lower total T levels (OR = 1.09, p = 0.024), older age (OR = 1.59, p < 0.001), and the presence of ≥3 comorbidities (OR = 1.60, p = 0.002). Tables 2 and 3 summarize the findings from the univariable and multivariable logistic regression analyses. We also conducted an analysis excluding ADT cases while using these predictors. In the univariable analysis, we found that lower total T levels (OR = 1.12, p = 0.024), older age (OR = 1.48, p < 0.001), and the presence of comorbidities (OR = 1.58, p = 0.007) were significant predictors. However, in the multivariable analyses, only older age (OR = 1.49, p < 0.001) and a higher number of comorbidities (OR = 1.66, p = 0.003) remained significant predictors of BDL. A summary of the analysis can be found in Supplementary Table 1.

Figure 1.

Prevalence of BDL between men with and without ADT exposure. Abbreviation: ADT, androgen deprivation therapy; BDL, bone density loss.

Table 2.

Univariable Predictors of Bone Density Loss

Variable	OR	95% CI	p-value
Age (per 10-year increase)	1.61	1.36-1.90	<0.001*
Number of Comorbidities ≥3 (yes)	1.56	1.18-2.08	0.002*
Total Testosterone (per 50 ng/dL decrease)	1.14	1.07-1.22	<0.001*
Testosterone ≤200 ng/dL (yes)	1.43	1.11-1.86	0.006*
Testosterone ≤100 ng/dL (yes)	1.71	1.28-2.31	<0.001*
Free Testosterone (per 1 ng/dL decrease)	1.11	1.04-1.18	0.001*
Estradiol (per 1 pg/mL increase)	0.98	0.96-1.01	0.059
LH (per 1 mU/mL increase)	1.04	1.02-1.06	<0.001*
Androgen Deprivation Therapy exposure (yes)	1.42	1.07-1.89	0.015*
Time on Androgen Deprivation Therapy (per 6m increase)	1.02	0.99-1.06	0.229
Androgen Deprivation Therapy ≥ 6 months (yes)	1.42	0.88-2.31	0.151
Androgen Deprivation Therapy ≥ 12 months (yes)	1.26	0.77-2.06	0.354
Androgen Deprivation Therapy ≥ 24 months (yes)	1.08	0.62-1.86	0.793
History of Radical Orchiectomy (yes)	0.80	0.40-1.61	0.537
History of Chemotherapy (yes)	0.94	0.59-1.50	0.786
Medication exposure: Proton Pump Inhibitor (yes)	1.54	1.10-2.16	0.012*
Medication exposure: Corticosteroids (yes)	2.00	1.27-3.17	0.003*

OR = Odds ratio, CI = Confidence Interval, LH = Luteinizing Hormone

Table 3.

Multivariable Predictors of Bone Density Loss

Variable	OR	95% CI	p-value
Total Testosterone (per 50 ng/dL decrease)	1.09	1.01-1.17	0.024*
Age (per 10-year increase)	1.59	1.34-1.88	<0.001*
Number of Comorbidities ≥3 (yes)	1.60	1.19-2.14	0.002*
Androgen Deprivation Therapy exposure (yes)	1.08	0.78-1.49	0.657

OR = Odds ratio, CI = Confidence Interval

DISCUSSION

In this study, BDL was notably common among men with low T levels, with more than 1/3 of patients showing evidence of BDL. Alarmingly, nearly 10% of those affected exhibited severe BDL, which is consistent with a diagnosis of osteoporosis. These findings highlight the significant impact of compromised skeletal health in men aged≥50 years with low T and a history of cancer, and emphasize the importance of regular bone health assessments in this population. Several factors were identified as significant predictors of BDL, including lower total T levels, advanced age, and a higher number of comorbidities. Understanding these risk factors is essential for identifying patients at increased risk and guiding timely interventions. It is critical for us as men’s health providers to be aware of the significant prevalence of BDL in men with low T in this group of patients. Early identification may enable targeted counseling, the initiation of preventive or therapeutic strategies, and referrals to appropriate specialists, ultimately aiming to improve bone density outcomes and reduce the risk of fractures and other serious skeletal complications.

It is well known that age can be associated with BDL. Gold et al. analyzed the prevalence of osteoporosis in a cohort of men, reporting that the prevalence increases with age: 0.6% for ages 50-60 years, 4.3% 70-80, 10.4% 80-90, and 22.6% for those 90 years old or older.¹¹ A prospective study of 507 men aged 45-92 years found that significant predictors of osteoporosis were age ≥75 years, lower body mass index (<24 kg/m²), current smoking, and no physical activity.¹² Another population-based study of 1122 men aged 70-97 years reported that older age, kidney disease, low vitamin D levels, and medications such as anti-androgens, thiazolidinediones, and loop diuretics were significant predictors of BDL.¹³

Regarding low T, Ferlin et al reported 40% prevalence of osteoporosis in men with early onset of low T due to Klinefelter’s syndrome compared to men at the same age.¹⁴ Ondrusova et al. reported on a cohort of testicular cancer survivors, with a median follow-up of 4 years after diagnosis and treatment. They found that the prevalence of BDL was 43% after radical orchiectomy alone, 45% after orchiectomy combined with chemotherapy, and 51% after orchiectomy with radiation therapy. These occurrences were significantly more common in men with low T levels.¹⁵ Behre et al. studied a cohort of 35 men with secondary hypogonadism and performed bone density analysis by assessing lumbar spine bone mineral density using quantitative computed tomography. They reported a BDL prevalence of 43%.¹⁶

Gotthardt et al. assessed the prevalence of BDL in a group of 144 men undergoing long-term opioid maintenance therapy at a European addiction center. This group was compared to a healthy control group of 35 men. Bone mineral density was evaluated using densitometry. Among men aged 40 years and older, 29% of those with long-term opioid dependency were found to have osteoporosis, while 48% had osteopenia. In contrast, none of the men in the control group had osteoporosis, and 35% were found to have osteopenia. In this study, age was not a significant predictor of BDL in this population. In contrast, these authors reported that lower free T was significantly associated with BDL at the lumbar spine, but not at the hip. Other significant predictors were lower or higher body mass index and lower albumin.¹⁷

Wen et al. conducted a study to assess the prevalence of osteoporosis and osteopenia in a cohort of 4,707 men aged 40-79 years. Low T was defined as a serum total T level of ≤350 ng/dL. Bone mineral density was evaluated using DEXA, focusing on the lumbar spine (L1–L4) and the femoral neck. The findings indicated that the prevalence of osteopenia among men with low T reached 20%, while osteoporosis was observed in 5% of this population. Among men with normal T levels, the prevalence of osteopenia was 17%, while osteoporosis was 3%.¹⁸ Fink reported a prevalence of osteoporosis in a cohort of 2447 men, ≥65 years of age. Low T was defined as total T ≤200 ng/dL. The prevalence of osteoporosis was 12% in the low T group vs 6% in the normal T group.¹⁹ Also, these authors reported that the prevalence of osteoporosis was much higher in men with low estradiol levels (defined as Estradiol ≤ 10 pg/mL), 15% vs 3%, compared to men with normal estradiol levels. In our study, estradiol level was not a significant predictor of BDL.

Our findings align with those from large-sample studies but yield lower estimates than those from small-sample investigations. Directly comparing our results with previous publications is difficult due to variations in analytical criteria, including the definition of low T, and to limitations in many existing studies, which use very small sample sizes. Lower T was not a significant predictor when cases of ADT were excluded. This could be due to our inclusion of only men with low T levels. To strengthen the analysis, it would be beneficial to include men with normal T levels in future research steps.

This study is limited by its retrospective design, particularly the potential for unmeasured confounding and the lack of detailed information on medication exposure associated with BDL, including the timing of medication use. Another limitation of our study is the population we included; this is a relatively older group, and most participants have an oncological history, which may complicate extrapolating our findings to the general population. Including men under 50 years of age would have significantly complicated the multivariable model analysis due to the use of DEXA Z scores in this group.

However, several strengths help to offset these limitations. Notably, we utilized a large, systematically audited database, which enhances the reliability and generalizability of the data. Additionally, total T and free T levels were measured using LC-MS and equilibrium dialysis, respectively; these are the current gold-standard assays that ensure high analytical accuracy. Furthermore, bone mineral density was assessed following a standardized scanning protocol, improving the consistency and validity of our skeletal outcome measurements in men with low T.

CONCLUSION

In our group of men ≥50 years of age, most of them with oncological history, and with low T, approximately 1/3 were found to have BDL, and about 1 in 10 of those men with BDL had osteoporosis. Significant predictors of BDL included lower T levels, older age, and the presence of more comorbidities. This information is critical, particularly for clinicians assessing bone density in a similar group of men presented in this study, especially those with profoundly low T levels.