Please Scroll Down to See Forums Below 
lanky
Well-known member
Robert I. McLachlan and Carolyn A. Allan
Prince Henry’s Institute of Medical Research and Department of Obstetrics and Gynecology Monash University Monash Medical Centre Clayton, Victoria 3168 Australia
Address all correspondence and requests for reprints to: Robert I. McLachlan, Prince Henry’s Institute of Medical Research, P.O. Box 5152, Clayton, Victoria 3168, Australia. E-mail: [email protected].
The physical and cognitive declines noted by men as they age have historically been regarded as an integral part of the normal aging process, yet have also long been the target of empirical therapies aimed at restoring youthful vigor. In recent years, there has been a great deal of medical, social, and commercial interest in whether a decline in serum testosterone (T) underscores some of these changes and would thus be amenable to T replacement therapy (TRT). Various terms have been popularized such as "andropause" or "partial androgen deficiency of the aging male," yet there is no consensus on their definitions (1, 2). It is clear that total T levels decline by 1–2% per annum beginning at ages in the fourth decade, and consequently an increasing proportion of the aging male population has serum T levels below the reference range for healthy young men. For example, age-stratified prevalence data from the Baltimore Longitudinal Study of Aging showed that 19% of 60- to 69-yr-old men had total T levels below those of normal young men (3), raising the possibility that large numbers of men, simply by virtue of falling below an arbitrary threshold, are potential candidates for TRT. However, the nonspecific symptoms consistent with, but not diagnostic of, androgen deficiency (AD) that are prevalent in aging men may be partially or wholly due to comorbidities (4). Despite lack of evidence of benefit over risk (1), clinical practice has evolved rapidly in this area, fuelled by patient expectations of improved physical function and quality of life and by the availability of convenient forms of T (such as the hydroalcoholic gels), leading to a 29% increase in the prescription of TRT between the years 2001 and 2002 (1), mainly in 46- to 65-yr-old men.
The diagnosis of AD demands a synthesis of both clinical features and serum T measures particularly in aging men, as neither is consistently reliable in isolation. With unambiguous clinical features in men with established testicular or pituitary pathology and a serum T level less than 200–250 ng/dl (6.9–8.7 nM), the diagnosis of AD and the merits of treatment are well grounded. In clinical practice, however, most men present with relatively nonspecific symptoms, have confounding medical diagnoses (obesity, chronic illness), and maintain serum T levels in the low-normal range. Moreover, biochemical evaluation is complicated by interindividual variability in normal serum T (extending over a 4-fold range), poor assay performance characteristics, and inappropriate reference ranges (see below). Although attractive and certainly convenient, use of single-value cut-offs to decide clinical treatment is best described as "endonumerology!"
The article by Araujo et al. (5) in this issue, for the first time, attempts to estimate the prevalence and incidence of AD in a prospective longitudinal study using operational diagnostic criteria based on clinical and biochemical parameters, reflecting the recommendation of the Second Annual Andropause Consensus group (2). The Massachusetts Male Aging Study (MMAS) data set includes eight of the 12 clinical features proposed by the consensus group, along with serum total T (determined by a robust assay method) and calculated free T values. Their data are a considerable advance on prior studies reporting biochemical thresholds alone and provide a prototype for developing a more rational algorithm for the diagnosis of AD and for identifying cohorts for interventional clinical research studies.
However, the authors readily acknowledge that their data have limitations, particularly in regard to the definition of AD. For example, subjects needed three or more positive responses to the eight questions for the diagnosis of AD. Most of these variables were assessed on a 1–4 scale, but the thresholds for these subscales are not provided other than for lethargy and erectile dysfunction. For lethargy, if a subject answered "some or a little" to the question "I could not get going" (in the past week), a positive response was recorded, yet many eugonadal men could be expected to respond in this way. For libido, a response that sexual activity was considered only once daily was recorded as a positive symptom of AD. The authors acknowledge that their choice of the three positive responses threshold was arbitrary, stating that two would not have been enough and eight would have been unreasonably stringent.
Herein lies the problem with symptom questionnaires: symptoms and signs suggestive of AD are nonspecific and readily accounted for by comorbidities. Lethargy, reduced concentration, sleep disturbance, irritability, and depressed mood may relate to physical illness (and side effects of treatment), obesity, and/or lack of physical exercise and other lifestyle issues (such as alcohol or drug abuse), relationship difficulties, and occupational or financial stresses. Even low libido, a classic feature of AD, is commonly related to such comorbidities. Finally, erectile dysfunction in most instances has a neurovascular rather than an endocrine origin: indeed, MMAS data using multivariant analyses show hypertension, hyperlipidemia, diabetes, smoking, and limited physical exercise to be independent predictors, whereas serum T is not (6, 7).
In essence, existing screening tools for AD lack adequate specificity and sensitivity to direct clinical treatment. It follows that the value of the operational diagnostic criteria used in the Araujo et al. study is limited, a fact they acknowledge by saying "it is doubtful that the manner in which we use the signs/symptoms available to us can be reliably employed in screening for androgen deficiency." The lack of sensitivity of their symptom scores is revealed by the fact that 37–47% of subjects with total T levels less than 200 ng/dl (6.9 nM) at the T2 and T1 time points, respectively (their Table 3), failed to record even three positive symptoms as required for the diagnosis of AD; yet such a low serum T finding strongly suggests AD and certainly warrants further consideration for TRT. One possible explanation for this is that some hypoandrogenic men lack a clear appreciation of a eugonadal level of function as a result of the insidious decline in their serum T. It is also intriguing that there was no correlation between the symptoms and either free or total T at the initial (T1) time point, which brings into question the validity of this screening tool. The authors raise the interesting notion that symptoms might be weighted in favor of those with greater likelihood of having AD as their cause (such as low libido in comparison to lethargy), although this hypothesis has not been further explored. Recent data pertaining to hypoandrogenic men highlight the consistency of an individual’s symptomatology when their T levels fall, but also the inte-individual variability in symptom patterns and serum T level threshold values [ranging from 30–430 ng/dl (1–15 nM)] (8).
Most critically, we are not provided with the information gathered by the MMAS about comorbidities in their T1 or T2 populations, specifically whether the onset of AD during follow-up was affected by the onset of other diseases. Their previously published data have shown that comorbidities themselves suppress serum T levels (9, 10) and so the apparent hypoandrogenemia may simply reflect their severity and point to the need to address these problems rather than triggering the use of TRT. Indeed, undue emphasis on serum T levels may distract from this proper course of action. Although the authors have excluded 28 "seriously ill" patients from the original 1709 respondents for the T2 assessment, many others would be expected to have experienced a worsening, or new development, of less severe comorbidities that nonetheless directly cause both the positive symptoms complex and the reduced serum T level, yet at present they would be recorded as having AD.
The authors report that the prevalence of AD doubled over the 8.8 yr of follow-up and was predictably more common with increasing age. In addition, it also appeared that the prevalence within age-stratified groups showed an increase over the follow-up period, suggesting a secular change. Although it is not discussed, the most likely reason is the rising incidence of obesity in Western populations, which has more than doubled since 1980 with 30% of adult American men over 40 yr now being obese (body mass index of >30 kg/m2) (11). Serum total T levels fall with increasing body mass due to declining SHBG levels (12); this effect is seen across the overweight to obese range and is striking in massive obesity with a 50% or greater reduction in total T levels.
The lack of a clear biomarker of sex hormone sufficiency in men, such as regular menses in women, means that the laboratory evaluation of androgen status in aging men is vital, but this in turn is complicated by many factors:
1) Age-related changes are variably seen within the hypothalamo-pituitary-testicular axis; in some men a rise in LH supports primary testicular failure, whereas in others serum LH does not rise despite low serum T, suggesting a degree of hypogonadotropic hypogonadism (13). It is disappointing that serum LH data were not provided in the paper by Araujo et al. as a clearly elevated LH strongly supports primary testicular failure and, in turn, the diagnosis of AD.
2) There is uncertainty about the best serum T measure: the rise in SHBG and fall in total T levels with age results in greater reduction in free T; this is proposed by some to represent the biologically important tissue-available parameter. Such an assumption leads to a wider net for the definition of AD in older men (e.g. 19 vs. 34% for men in their 60s; Ref.3). There is little direct evidence for this free hormone hypothesis; for example, a lack of clinical data to support a superior correlation of free T measures with clinical features of AD. It is interesting to note that Institute of Medicine recommendations (1) for biochemical inclusion criteria in future studies of TRT are based upon total T measurements. There is also debate about the usefulness and practicality of the many methods for free T by direct measurement (equilibrium dialysis, bioavailable T, the widely discredited direct T analog assays) or indirectly calculated using total T and SHBG levels such as the Sodergard equation, as used by Araujo and colleagues (14, 15).
3) The performance and reporting of total T assays are particularly vexing. The late Dr. Longcope was a leader in hormone assays, and it is clear that great care was taken in the Araujo et al. study to control for issues such as circadian rhythm, repeat sampling, pooling and storage of samples, the use of a well-validated RIA with excellent performance characteristics, and assessment for loss of immunoactivity during storage. Unfortunately, most clinicians are at the mercy of commercial kit manufacturers and laboratories using high-throughput automated analysers with an emphasis on short turn-around times and cost reduction rather than accuracy. The performance of commercial assay platforms has recently been evaluated against a quantitative mass spectroscopy method and all showed substantial biases that varied between platforms and across the working range, especially at the lower end of greatest importance to diagnosis of AD (15, 16).
The clinician’s problems are compounded by reference ranges derived from populations in whom reproductive disorders have not been specifically excluded by clinical assessment and semen analysis (even the Coronary Artery Risk Development in Young Adults study failed to perform this crucial evaluation). We recently evaluated eight autoanalyzers using 124 samples taken from such young men and noted striking disagreement between the observed vs. the laboratory-reported reference intervals with the discrepancy in the lower reference value varying from –4.8 to +2.6 nM (–140 to +75 ng/dl) (unpublished observations). Platform-specific ranges are a potential quick fix, but kit manufacturers must address the need for methods calibrated against quantitative standards. All clinical algorithms include threshold values, but one’s confidence in their veracity is severely eroded by knowledge of their current poor performance, putting the clinician in an invidious situation.
The authors note that their prevalence estimates of AD are lower than that described by prior studies and speculate they have underestimated the figure due to the MMAS population being predominantly white and of higher socioeconomic status and/or that there was attrition bias with those with AD having been less likely to be seen at the T2 time point. However, one could propose that these prevalence figures substantially overestimate the number of truly AD men because the diagnostic criteria are based on an arbitrary number of non-specific symptoms and biochemical thresholds that lack stringency. The majority of men labeled as AD had serum T 6.9–13.9 nM (200–400 ng/dl), a range that would likely contain a substantial number of all older men (depending upon the specific T assay being used), and thus their diagnosis relied heavily on the calculated free T. Their threshold level of 8.9 ng/dl (0.3 nM), as determined in men in their 20s, is significantly higher than the range of the four other studies cited with a median value of 5.2 ng/dl (0.15 nM). Again, we suffer from the lack of validated reference range for Sodergard values. One wonders what effect dropping the criteria to this lower figure may have on the proposed prevalence and incidence rates that have been extrapolated. Finally, although Araujo et al. derived a mean serum T from two samples collected on one day, within an individual there is considerable week-to-week variability such that he may not consistently meet the diagnostic criteria for AD (14).
The recent Institute of Medicine (2003) report (1) highlighted the limited data supporting substantial benefits from TRT in older men with low-normal T levels and identified the need for research to better define the extent of benefit from TRT administered as a therapeutic agent to address specific problems (frailty, impaired cognition, quality of life) and that there was no case for its use as a general preventative measure. The reader is referred to recent reviews of published randomized, controlled trials (1, 17) that highlight the uncertainty of tangible benefits of TRT. Specifically, the Institute of Medicine advocate studying men over the age of 65 yr with "clinically low" T levels and one or more symptoms suggestive of hypoandrogenemia. Whether the outcomes from such results will be directly applicable to the 40- to 69-yr-old men highlighted by Araujo et al. is unclear. These men have a lower background rate of diseases that may be exacerbated by TRT (e.g. cardiovascular and prostate disease) but will have a much greater period of TRT exposure.
Greater reflection regarding the assumption that TRT in aging men should aim to restore older men’s serum T levels into the middle of the young male reference range is needed. Almost all studies, explicitly or implicitly, suggest this is the goal of TRT, the assumption being that this provides the optimum benefit/risk to TRT. However, one could argue to treat to the target T levels of fit, nonobese, sexually active, disease- and medication-free age-matched controls. If the latter elite group are doing so well with a median total T levels of around 14 nM (400 ng/dl) (as reported for healthy nonobese 60 yr olds) (18), why should one aspire to take a group of symptomatic men and move them to 20+ nM (580+ ng/d) range?
At some point, the use of exogenous T in aging men with borderline normal T levels will blend inexorably from a physiological into a pharmacological strategy. Recent data from Bhasin and colleagues (19) has shown that body compositional changes continue across physiological to supraphysiological T doses to a similar degree in both young and aging men. Thus, a gain of muscle and loss in fat can be achieved in eugonadal older men if the dose of T is sufficient, a fact well known by some athletes who seek performance benefits regardless of dose or safety concerns. To what extent does the current use of TRT outside more stringent criteria for AD represent pharmacology rather than physiology?
Alas, given these flaws in existing symptom scores and laboratory parameters, it is not surprising that a combined operational definition of AD still has major limitations. In truth, the diagnostic criteria in aging men will remain unreliable until clinical research better defines the phenotypic features of men shown to benefit from TRT. In other words, a post hoc evaluation of placebo-controlled, randomized clinical trials will be needed to begin identifying characteristics (clinical, endocrine, possibly genetic) that predict tangible benefits (and at acceptable risk) from TRT. This will assist the clinician to identify such men among the large number of potential TRT candidates. In the interim, any attempt to estimate the prevalence of AD is open to criticism.
In summary, this interesting longitudinal study sets the framework for further attempts to improve the diagnostic algorithm, to inform clinical trial development to define target groups most likely to benefit from TRT, and to power such studies for side effects. Yet we express caution about their suggestion that there are 481,000 new cases annually of AD in the United States. This may be taken by some to indicate that the men meeting their diagnostic criteria are all worthy candidates for TRT. We have pointed to the limitations of the raw data underlying the prevalence estimates and to the paucity of evidence supporting the use of TRT in aging men with nonspecific symptoms and borderline normal serum T levels. These men may be willing recipients for TRT as a convenient and well-marketed "magic-bullet" solution, but this approach may well be inappropriate; rather a vigorous assessment of comorbidities that may underscore the lead symptoms should be undertaken and specifically addressed. In the mean time, individual physicians must educate their patients about the complexities of these issues and their need to decide for themselves where the "T goal posts" lie in the overall context of the man’s health status
Prince Henry’s Institute of Medical Research and Department of Obstetrics and Gynecology Monash University Monash Medical Centre Clayton, Victoria 3168 Australia
Address all correspondence and requests for reprints to: Robert I. McLachlan, Prince Henry’s Institute of Medical Research, P.O. Box 5152, Clayton, Victoria 3168, Australia. E-mail: [email protected].
The physical and cognitive declines noted by men as they age have historically been regarded as an integral part of the normal aging process, yet have also long been the target of empirical therapies aimed at restoring youthful vigor. In recent years, there has been a great deal of medical, social, and commercial interest in whether a decline in serum testosterone (T) underscores some of these changes and would thus be amenable to T replacement therapy (TRT). Various terms have been popularized such as "andropause" or "partial androgen deficiency of the aging male," yet there is no consensus on their definitions (1, 2). It is clear that total T levels decline by 1–2% per annum beginning at ages in the fourth decade, and consequently an increasing proportion of the aging male population has serum T levels below the reference range for healthy young men. For example, age-stratified prevalence data from the Baltimore Longitudinal Study of Aging showed that 19% of 60- to 69-yr-old men had total T levels below those of normal young men (3), raising the possibility that large numbers of men, simply by virtue of falling below an arbitrary threshold, are potential candidates for TRT. However, the nonspecific symptoms consistent with, but not diagnostic of, androgen deficiency (AD) that are prevalent in aging men may be partially or wholly due to comorbidities (4). Despite lack of evidence of benefit over risk (1), clinical practice has evolved rapidly in this area, fuelled by patient expectations of improved physical function and quality of life and by the availability of convenient forms of T (such as the hydroalcoholic gels), leading to a 29% increase in the prescription of TRT between the years 2001 and 2002 (1), mainly in 46- to 65-yr-old men.
The diagnosis of AD demands a synthesis of both clinical features and serum T measures particularly in aging men, as neither is consistently reliable in isolation. With unambiguous clinical features in men with established testicular or pituitary pathology and a serum T level less than 200–250 ng/dl (6.9–8.7 nM), the diagnosis of AD and the merits of treatment are well grounded. In clinical practice, however, most men present with relatively nonspecific symptoms, have confounding medical diagnoses (obesity, chronic illness), and maintain serum T levels in the low-normal range. Moreover, biochemical evaluation is complicated by interindividual variability in normal serum T (extending over a 4-fold range), poor assay performance characteristics, and inappropriate reference ranges (see below). Although attractive and certainly convenient, use of single-value cut-offs to decide clinical treatment is best described as "endonumerology!"
The article by Araujo et al. (5) in this issue, for the first time, attempts to estimate the prevalence and incidence of AD in a prospective longitudinal study using operational diagnostic criteria based on clinical and biochemical parameters, reflecting the recommendation of the Second Annual Andropause Consensus group (2). The Massachusetts Male Aging Study (MMAS) data set includes eight of the 12 clinical features proposed by the consensus group, along with serum total T (determined by a robust assay method) and calculated free T values. Their data are a considerable advance on prior studies reporting biochemical thresholds alone and provide a prototype for developing a more rational algorithm for the diagnosis of AD and for identifying cohorts for interventional clinical research studies.
However, the authors readily acknowledge that their data have limitations, particularly in regard to the definition of AD. For example, subjects needed three or more positive responses to the eight questions for the diagnosis of AD. Most of these variables were assessed on a 1–4 scale, but the thresholds for these subscales are not provided other than for lethargy and erectile dysfunction. For lethargy, if a subject answered "some or a little" to the question "I could not get going" (in the past week), a positive response was recorded, yet many eugonadal men could be expected to respond in this way. For libido, a response that sexual activity was considered only once daily was recorded as a positive symptom of AD. The authors acknowledge that their choice of the three positive responses threshold was arbitrary, stating that two would not have been enough and eight would have been unreasonably stringent.
Herein lies the problem with symptom questionnaires: symptoms and signs suggestive of AD are nonspecific and readily accounted for by comorbidities. Lethargy, reduced concentration, sleep disturbance, irritability, and depressed mood may relate to physical illness (and side effects of treatment), obesity, and/or lack of physical exercise and other lifestyle issues (such as alcohol or drug abuse), relationship difficulties, and occupational or financial stresses. Even low libido, a classic feature of AD, is commonly related to such comorbidities. Finally, erectile dysfunction in most instances has a neurovascular rather than an endocrine origin: indeed, MMAS data using multivariant analyses show hypertension, hyperlipidemia, diabetes, smoking, and limited physical exercise to be independent predictors, whereas serum T is not (6, 7).
In essence, existing screening tools for AD lack adequate specificity and sensitivity to direct clinical treatment. It follows that the value of the operational diagnostic criteria used in the Araujo et al. study is limited, a fact they acknowledge by saying "it is doubtful that the manner in which we use the signs/symptoms available to us can be reliably employed in screening for androgen deficiency." The lack of sensitivity of their symptom scores is revealed by the fact that 37–47% of subjects with total T levels less than 200 ng/dl (6.9 nM) at the T2 and T1 time points, respectively (their Table 3), failed to record even three positive symptoms as required for the diagnosis of AD; yet such a low serum T finding strongly suggests AD and certainly warrants further consideration for TRT. One possible explanation for this is that some hypoandrogenic men lack a clear appreciation of a eugonadal level of function as a result of the insidious decline in their serum T. It is also intriguing that there was no correlation between the symptoms and either free or total T at the initial (T1) time point, which brings into question the validity of this screening tool. The authors raise the interesting notion that symptoms might be weighted in favor of those with greater likelihood of having AD as their cause (such as low libido in comparison to lethargy), although this hypothesis has not been further explored. Recent data pertaining to hypoandrogenic men highlight the consistency of an individual’s symptomatology when their T levels fall, but also the inte-individual variability in symptom patterns and serum T level threshold values [ranging from 30–430 ng/dl (1–15 nM)] (8).
Most critically, we are not provided with the information gathered by the MMAS about comorbidities in their T1 or T2 populations, specifically whether the onset of AD during follow-up was affected by the onset of other diseases. Their previously published data have shown that comorbidities themselves suppress serum T levels (9, 10) and so the apparent hypoandrogenemia may simply reflect their severity and point to the need to address these problems rather than triggering the use of TRT. Indeed, undue emphasis on serum T levels may distract from this proper course of action. Although the authors have excluded 28 "seriously ill" patients from the original 1709 respondents for the T2 assessment, many others would be expected to have experienced a worsening, or new development, of less severe comorbidities that nonetheless directly cause both the positive symptoms complex and the reduced serum T level, yet at present they would be recorded as having AD.
The authors report that the prevalence of AD doubled over the 8.8 yr of follow-up and was predictably more common with increasing age. In addition, it also appeared that the prevalence within age-stratified groups showed an increase over the follow-up period, suggesting a secular change. Although it is not discussed, the most likely reason is the rising incidence of obesity in Western populations, which has more than doubled since 1980 with 30% of adult American men over 40 yr now being obese (body mass index of >30 kg/m2) (11). Serum total T levels fall with increasing body mass due to declining SHBG levels (12); this effect is seen across the overweight to obese range and is striking in massive obesity with a 50% or greater reduction in total T levels.
The lack of a clear biomarker of sex hormone sufficiency in men, such as regular menses in women, means that the laboratory evaluation of androgen status in aging men is vital, but this in turn is complicated by many factors:
1) Age-related changes are variably seen within the hypothalamo-pituitary-testicular axis; in some men a rise in LH supports primary testicular failure, whereas in others serum LH does not rise despite low serum T, suggesting a degree of hypogonadotropic hypogonadism (13). It is disappointing that serum LH data were not provided in the paper by Araujo et al. as a clearly elevated LH strongly supports primary testicular failure and, in turn, the diagnosis of AD.
2) There is uncertainty about the best serum T measure: the rise in SHBG and fall in total T levels with age results in greater reduction in free T; this is proposed by some to represent the biologically important tissue-available parameter. Such an assumption leads to a wider net for the definition of AD in older men (e.g. 19 vs. 34% for men in their 60s; Ref.3). There is little direct evidence for this free hormone hypothesis; for example, a lack of clinical data to support a superior correlation of free T measures with clinical features of AD. It is interesting to note that Institute of Medicine recommendations (1) for biochemical inclusion criteria in future studies of TRT are based upon total T measurements. There is also debate about the usefulness and practicality of the many methods for free T by direct measurement (equilibrium dialysis, bioavailable T, the widely discredited direct T analog assays) or indirectly calculated using total T and SHBG levels such as the Sodergard equation, as used by Araujo and colleagues (14, 15).
3) The performance and reporting of total T assays are particularly vexing. The late Dr. Longcope was a leader in hormone assays, and it is clear that great care was taken in the Araujo et al. study to control for issues such as circadian rhythm, repeat sampling, pooling and storage of samples, the use of a well-validated RIA with excellent performance characteristics, and assessment for loss of immunoactivity during storage. Unfortunately, most clinicians are at the mercy of commercial kit manufacturers and laboratories using high-throughput automated analysers with an emphasis on short turn-around times and cost reduction rather than accuracy. The performance of commercial assay platforms has recently been evaluated against a quantitative mass spectroscopy method and all showed substantial biases that varied between platforms and across the working range, especially at the lower end of greatest importance to diagnosis of AD (15, 16).
The clinician’s problems are compounded by reference ranges derived from populations in whom reproductive disorders have not been specifically excluded by clinical assessment and semen analysis (even the Coronary Artery Risk Development in Young Adults study failed to perform this crucial evaluation). We recently evaluated eight autoanalyzers using 124 samples taken from such young men and noted striking disagreement between the observed vs. the laboratory-reported reference intervals with the discrepancy in the lower reference value varying from –4.8 to +2.6 nM (–140 to +75 ng/dl) (unpublished observations). Platform-specific ranges are a potential quick fix, but kit manufacturers must address the need for methods calibrated against quantitative standards. All clinical algorithms include threshold values, but one’s confidence in their veracity is severely eroded by knowledge of their current poor performance, putting the clinician in an invidious situation.
The authors note that their prevalence estimates of AD are lower than that described by prior studies and speculate they have underestimated the figure due to the MMAS population being predominantly white and of higher socioeconomic status and/or that there was attrition bias with those with AD having been less likely to be seen at the T2 time point. However, one could propose that these prevalence figures substantially overestimate the number of truly AD men because the diagnostic criteria are based on an arbitrary number of non-specific symptoms and biochemical thresholds that lack stringency. The majority of men labeled as AD had serum T 6.9–13.9 nM (200–400 ng/dl), a range that would likely contain a substantial number of all older men (depending upon the specific T assay being used), and thus their diagnosis relied heavily on the calculated free T. Their threshold level of 8.9 ng/dl (0.3 nM), as determined in men in their 20s, is significantly higher than the range of the four other studies cited with a median value of 5.2 ng/dl (0.15 nM). Again, we suffer from the lack of validated reference range for Sodergard values. One wonders what effect dropping the criteria to this lower figure may have on the proposed prevalence and incidence rates that have been extrapolated. Finally, although Araujo et al. derived a mean serum T from two samples collected on one day, within an individual there is considerable week-to-week variability such that he may not consistently meet the diagnostic criteria for AD (14).
The recent Institute of Medicine (2003) report (1) highlighted the limited data supporting substantial benefits from TRT in older men with low-normal T levels and identified the need for research to better define the extent of benefit from TRT administered as a therapeutic agent to address specific problems (frailty, impaired cognition, quality of life) and that there was no case for its use as a general preventative measure. The reader is referred to recent reviews of published randomized, controlled trials (1, 17) that highlight the uncertainty of tangible benefits of TRT. Specifically, the Institute of Medicine advocate studying men over the age of 65 yr with "clinically low" T levels and one or more symptoms suggestive of hypoandrogenemia. Whether the outcomes from such results will be directly applicable to the 40- to 69-yr-old men highlighted by Araujo et al. is unclear. These men have a lower background rate of diseases that may be exacerbated by TRT (e.g. cardiovascular and prostate disease) but will have a much greater period of TRT exposure.
Greater reflection regarding the assumption that TRT in aging men should aim to restore older men’s serum T levels into the middle of the young male reference range is needed. Almost all studies, explicitly or implicitly, suggest this is the goal of TRT, the assumption being that this provides the optimum benefit/risk to TRT. However, one could argue to treat to the target T levels of fit, nonobese, sexually active, disease- and medication-free age-matched controls. If the latter elite group are doing so well with a median total T levels of around 14 nM (400 ng/dl) (as reported for healthy nonobese 60 yr olds) (18), why should one aspire to take a group of symptomatic men and move them to 20+ nM (580+ ng/d) range?
At some point, the use of exogenous T in aging men with borderline normal T levels will blend inexorably from a physiological into a pharmacological strategy. Recent data from Bhasin and colleagues (19) has shown that body compositional changes continue across physiological to supraphysiological T doses to a similar degree in both young and aging men. Thus, a gain of muscle and loss in fat can be achieved in eugonadal older men if the dose of T is sufficient, a fact well known by some athletes who seek performance benefits regardless of dose or safety concerns. To what extent does the current use of TRT outside more stringent criteria for AD represent pharmacology rather than physiology?
Alas, given these flaws in existing symptom scores and laboratory parameters, it is not surprising that a combined operational definition of AD still has major limitations. In truth, the diagnostic criteria in aging men will remain unreliable until clinical research better defines the phenotypic features of men shown to benefit from TRT. In other words, a post hoc evaluation of placebo-controlled, randomized clinical trials will be needed to begin identifying characteristics (clinical, endocrine, possibly genetic) that predict tangible benefits (and at acceptable risk) from TRT. This will assist the clinician to identify such men among the large number of potential TRT candidates. In the interim, any attempt to estimate the prevalence of AD is open to criticism.
In summary, this interesting longitudinal study sets the framework for further attempts to improve the diagnostic algorithm, to inform clinical trial development to define target groups most likely to benefit from TRT, and to power such studies for side effects. Yet we express caution about their suggestion that there are 481,000 new cases annually of AD in the United States. This may be taken by some to indicate that the men meeting their diagnostic criteria are all worthy candidates for TRT. We have pointed to the limitations of the raw data underlying the prevalence estimates and to the paucity of evidence supporting the use of TRT in aging men with nonspecific symptoms and borderline normal serum T levels. These men may be willing recipients for TRT as a convenient and well-marketed "magic-bullet" solution, but this approach may well be inappropriate; rather a vigorous assessment of comorbidities that may underscore the lead symptoms should be undertaken and specifically addressed. In the mean time, individual physicians must educate their patients about the complexities of these issues and their need to decide for themselves where the "T goal posts" lie in the overall context of the man’s health status
