|Home | About | Journals | Submit | Contact Us | Français|
Some men presenting with diffuse musculoskeletal pain have testosterone values below the normal range for their age. While scant evidence suggests that functional status in such men may improve with testosterone replacement therapy, there are currently no published studies evaluating a potential correlation between diffuse musculoskeletal pain and male hypogonadism. In a cohort of 45 hypogonadal men with diffuse musculoskeletal pain we found a negative correlation between the duration of pain in years and baseline total testosterone values. Although advancing age would theoretically predispose individuals to lower testosterone concentrations, age was not independently associated with baseline testosterone values, nor were any other variables we examined. At this time generalised screening of gonadal function for such men is not indicated. A prospective study would more clearly evaluate a potential relationship between diffuse musculoskeletal pain and testosterone values in men and might determine if testosterone replacement leads to any measurable improvement.
Men often present to rheumatology clinics with complaints of diffuse musculoskeletal pain for which an aetiology is difficult to elucidate. Diffuse pain in men is less likely to fit the diagnostic criteria for fibromyalgia, if based on the original published criteria, requiring the presence of a defined set of tender points.1–3 Rheumatologists have made anecdotal observations that some men presenting with diffuse musculoskeletal pain have testosterone concentrations below the normal range for their age. Currently there are no published studies evaluating a potential correlation between diffuse musculoskeletal pain and hypogonadism in men. There are scattered case reports of hypogonadal men with arthralgias enjoying symptom improvement after testosterone replacement therapy.4,5 There are also variable reports of testosterone replacement improving strength in males,6 but at this time it is unclear if any increases in strength improve quality of life.7 There are however, reports of testosterone supplementation improving myocardial ischaemia and angina severity in males with coronary artery disease.8,9
Several factors have been identified to be associated with testosterone values. It is known that testosterone values naturally fall with age,10 with free testosterone considered to be a more sensitive indicator of androgenic status.11 Additionally, obese men are known to have lower testosterone values than age-matched lean men.12,13
Patients with mental illness commonly have multiple somatic complaints, including pain.14 Likewise, patients with multiple somatic symptoms are at greater risk for developing mental illness.15 In fact, testosterone supplementation has been used for treatment of refractory depression in hypogonadal men with promising but variable results.16–18 There are also reports linking psychiatric medications themselves to low testosterone. Typical antipsychotic medications have been implicated as a cause of hypogonadism by secondarily depressing testosterone concentrations through elevation of prolactin, and appear to have a dose–effect relationship,19–21 with testosterone values in men taking these medications generally within the normal range.22,23 The atypical antipsychotic drugs are not associated with sustained increases in prolactin24 and only mild elevations in prolactin have been reported with tricyclic antidepressants.25 Tetracyclic antidepressants have not been shown to alter prolactin values26 and selective serotonin reuptake inhibitors have not been shown to decrease testosterone concentrations.27,28 Benzodiazepines have mixed effects on serum testosterone values, with brief use appearing to have no effect or slightly increasing concentrations,29–31 while chronic use may decrease concentrations.29 Buspirone, an anxiolytic medication unrelated to the benzodiazepines, increases serum prolactin,32 and thus may theoretically cause a secondary decrease in testosterone. For these afore mentioned reasons, and as psychiatric medications are often used in treatment of chronic pain, it is important to identify any relationship between hypogonadism and such medications.
If hypogonadism is found to be associated with non-specific, diffuse musculoskeletal pain it may be appropriate to test men with such complaints for hypogonadism. If subjective pain complaints of hypogonadal men who undergo testosterone replacement therapy improve, further studies should be done to determine a possible cause and effect relationship as well as a prospective randomised controlled trial with active treatment and placebo arms. With this case series we seek to describe the presentation of these hypogonadal men suffering from diffuse musculoskeletal pain. We will examine potential associations between baseline total and free testosterone with demographic and health related characteristics and psychiatric medication usage.
We identified 49 men presenting between March 2000 and September 2004 to the University of Iowa Hospitals and Clinics (UIHC) rheumatology clinic who presented with persistent, non-specific, musculoskeletal pain and were subsequently found to have abnormally low testosterone values. The medical record was reviewed and the following information was abstracted: demographic data, including age and race; clinical data including body mass index (BMI), presence of pain, presence of weakness, presence of fatigue, pre-testosterone replacement pain score, post-testosterone replacement pain score, whether there was evidence of improvement in pain after testosterone replacement, whether the patient continued on testosterone replacement therapy at 3 months after starting, tobacco use, alcohol use, duration of pain in years, presence of hypovitaminosis D (25-hydroxyvitamin D < 30ng/ml), presence of hypothyroidism (both treated and untreated), frequency of pain medication use, use of psychiatric medications, history of trauma, presence of malignancy, diagnosis of fibromyalgia or other rheumatologic disease, whether the patient was referred to and seen by an endocrinologist, whether the patient was lost to follow-up, and laboratory data including: total testosterone (280–800 ng/dl), testosterone adult male (20–39 years: 400–1080 ng/dl; 40–59 years: 350–890 ng/dl; 60 years: 350–720 ng/dl), free testosterone (47.0–244.0 pg/ml), and prostate specific antigen (PSA) both at diagnosis of hypogonadism and at most recent visit.
Total testosterone was obtained through electrochemiluminescence immunoassay at UIHC. Testosterone adult male and free testosterone were obtained through outside commercial laboratory, by chemiluminescence immunoassay. The concentration of free testosterone is derived from total testosterone, protein binding constants, and the concentration of the testosterone binding proteins. Men were classified as having a low testosterone value if their baseline serum total testosterone or testosterone adult male values were below the lower limits of normal at our laboratory, as described above. Subjects with missing baseline serum total testosterone values (n=4) were excluded. One of these four excluded subjects had both missing total and free testosterone values at baseline. Of the 45 remaining subjects, nine were missing baseline free testosterone values. Demographic information was described with the mean, standard deviation, and range for each continuous variable and as a percentage of the whole for each categorical variable. χ2 tests were performed to determine any differences between those subjects excluded and those included. Spearman correlation matrices were performed to determine the relationship between continuous variables of interest and baseline serum free total testosterone values. The Kruskal–Wallis analysis of variance was performed to evaluate the relationship between categorical variables and baseline serum total and free testosterone values. Psychiatric medication use was divided into categorical type (selective serotonin reuptake inhibitor (SSRI), serotonin-norepinephrine reuptake inhibitor (SNRI), benzodiazepine, tricyclic antidepressant (TCA), and other) and compared with mean baseline free testosterone values with two-tailed Wilcoxon p values, as were categorical health related variables.
Only 10 of the 45 subjects with baseline serum free and total testosterone values had pain scores recorded both at initial clinic presentation and after testosterone replacement therapy. Of these 10 subjects, two did not continue on testosterone replacement therapy for at least 3 months. Of the remaining eight subjects, a paired Student’s t test was performed to evaluate changes in self reported pain scores between presentation and post-testosterone therapy initiation.
The Institutional Review Board at the University of Iowa approved this study. All analyses were performed using SAS statistical software, version 9.1 (SAS Institute, Cary, North Carolina, USA).
Demographic data for all 45 included subjects is displayed in table 1. None of the subjects had been previously treated with testosterone replacement therapy. The majority (79.6%, N=35) of men meeting criteria for non-specific, diffuse musculoskeletal pain and hypogonadism were referred to endocrinology for possible hormone replacement therapy, with only one of these subjects choosing not to undergo evaluation by endocrinology; 33.3% (N=15) were lost to follow-up with no information available after the initial diagnostic visit. No subjects had baseline or increases in PSA >4.0 reported (0.22–3.68) after testosterone replacement therapy, though there were significant numbers of missing PSA values (Nmissing=20/45). Vitamin D status was measured in only eight subjects, with four of these demonstrating low serum concentrations and four with adequate vitamin D stores. The four subjects with missing baseline serum total testosterone values who were excluded from analysis had slightly fewer years of pain, were mostly white, were more frequently smokers, used more alcohol, used pain medications more frequently as compared to included subjects, and had no diagnosed hypothyroidism, malignancies, history of trauma, or fibromyalgia. There were no significant differences in age, baseline free testosterone concentrations, BMI, presence of weakness, fatigue, or comorbid rheumatologic diseases between groups.
Spearman correlation coefficients were determined to show the relationship between variables of interest and baseline serum total and free testosterone values (table 2). Duration of pain was negatively correlated with baseline total (−0.32, p=0.05), but not free testosterone values (0.10, p=61). No other variables reached statistical significance.
Psychiatric medication use divided into categorical type (SSRI, SNRI, benzodiazepine, TCA, and other) is reported with associated mean baseline free testosterone values (table 3). Five of the 16 subjects (31.3%) using any psychiatric medication were in the lowest quartile of total testosterone with four (25%) of these subjects also in the lowest quartile for free testosterone. There were no significant differences between quartiles of free or total testosterone in subjects using psychiatric medications. Association with mean baseline testosterone values and use of benzodiazepines (p=0.07) and psychiatric medications in the “other” category (p= 0.06) approached, but did not reach, statistical significance. Only four subjects were taking more than one category of psychiatric medication consisting of the following combinations: SSRI, SNRI, and benzodiazepine; SSRI, benzodiazepine, and TCA; benzodiazepine and other; and SSRI, TCA and other. The “other” category included bupropion (N=1), St John’s Wort (N=2), zolpidem (N=1), nefazodone (N=1), gabapentin (N=1), and chlorpromazine (N=1). Two of the six subjects taking medications in the “other” category were on more than one psychiatric medication in the following combinations: SSRI, TCA, and other; and benzodiazipine and three medications in the “other” class.
Only 10 of the 45 subjects with baseline serum free and total testosterone values had pain scores recorded both at initial clinic presentation and after testosterone replacement therapy. None of these 10 subjects met clinical criteria for fibromyalgia. No significant correlations between initial pain score for these 10 men at time of diagnosis and the variables tested were found. Of these 10 subjects, two did not continue on testosterone replacement therapy for at least 3 months. Of the remaining eight subjects, a paired t test was performed demonstrating no significant changes in self reported pain scores (mean=0, p=1.0).
In our case series of hypogonadal men with diffuse pain, we found a negative correlation between the duration of pain in years and baseline total testosterone values, but no significant correlation with baseline free testosterone values. While this inverse correlation is intriguing and suggests individuals with diffuse musculoskeletal pain and hypogonadism with longer duration of pain are more likely to have testosterone concentrations at the lower extremes, this was not reproducible when performing the same analysis on the subset of individuals with baseline free testosterone values and is possibly a function of total testosterone being a less sensitive indicator of androgenic status than free testosterone values.11 Additionally, advancing age would theoretically predispose individuals to lower testosterone concentrations10 and could therefore explain this finding; however, in our series age was not independently associated with baseline total or free testosterone values. Additionally, though few individuals were tested for vitamin D stores, one-half of those that were tested proved to be deficient. Unfortunately, the low number of subjects with serum 25-hydroxyvitamin D values available did not allow reliable interpretation of this variable. This finding deserves further investigation in this subgroup as hypovitaminosis D has been variably linked to pain.33–35
We failed to show any significant association between self reported initial pain score and severity of hypogonadism. We also did not find a significant change in self reported pain score after hormone replacement therapy; however, many of these patients were lost to follow-up or did not have post-replacement pain scores, thus data are not sufficient to fully evaluate the impact of testosterone replacement therapy on pain. Additionally, pain scores reflect only the patient reported pain level at the time of a clinic visit, not their overall pain level over time, and therefore may not accurately reflect severity of chronic pain.
The use of benzodiazepines and psychiatric medications in the ‘other’ category approached, but did not reach statistical significance in relation to baseline total testosterone values, but not with free testosterone values. Overall there was no significant difference in testosterone concentrations between users and non-users of psychiatric medications. There is limited evidence that men taking commonly used psychiatric medications have lower testosterone.19,20,22–32 In our review, we found that 31.3% and 25% of the men using any psychiatric medication were in the lowest quartile of total and free testosterone, respectively; however, the relationships between total or free testosterone and degree of hypogonadism failed to reach statistical significance.
Benzodiazepine use may have mixed effects on serum testosterone concentrations depending on duration of therapy.29–31 Our review did not include analysis of the duration of psychiatric medication use; therefore we do not know the chronicity of benzodiazepine use in each patient. Of the psychiatric medications in the “other” category only one of the six individuals was using a medication known to be associated with decreased testosterone values. Additionally, due to the small numbers of subjects taking psychiatric medications in this cohort, effects of multiple medication usage could not be determined.
Rheumatologists have observed that a number of men presenting with diffuse musculoskeletal pain have low testosterone values. In reviewing 45 such cases, we were unable to elucidate reproducible correlates between baseline total and free testosterone concentrations and the variables of interest, or demonstrate subjective improvement in pain complaints after testosterone replacement therapy. Our study had several limitations that deserve mention. Foremost, our retrospective case series may have been insufficiently powered to detect some correlations. We had few subjects who were evaluated after at least 3 months of testosterone replacement therapy, and we did not differentiate between subjects based on their post treatment testosterone concentrations. Also, this short time period may be insufficient to detect changes in perceived pain levels. There was also no standard timing of laboratory testing or of return visits which could have skewed results based on timing of testosterone replacement dosing and peak effect on serum testosterone values. A prospective study would evaluate more clearly a potential relationship between men presenting with diffuse musculoskeletal pain and testosterone concentrations as well as the potential for improvement in pain complaints with testosterone replacement therapy, and could provide guidance in laboratory evaluation and treatment for these individuals.
This work was supported by a grant from the Arthritis Foundation Iowa Chapter.
Competing interests: None.