| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Perspectives |
1 Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD.
Address correspondence to the author at: Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, 615 N. Wolfe St., Rm E6132, Baltimore, MD 21205. E-mail eplatz{at}jhsph.edu
Two recent papers in Circulation (1) and Journal of Clinical Endocrinology and Metabolism (2) reported that older men with lower serum testosterone concentrations had a greater risk of dying than did men with higher testosterone. This potentially clinically important finding was not observed in 2 other prospective cohort studies (3)(4), and another cohort study reported that low concentrations of a combination of hormones, but not of bioavailable testosterone alone, were associated with death (5). These studies raise important issues related to measuring testosterone for etiologic and clinical use.
motivation for discussing issues in measuring testosterone
An important question in clinical epidemiology is whether low testosterone concentrations are causally associated with premature death in older men. At least 3 scenarios (Fig. 1
) may underlie the observation that older men with low testosterone are more likely to die (1)(2). Scenario 1: Disease states cause low testosterone, which in turn causes premature death, although disease states also may cause death via other mechanisms. In this scenario, either preventing disease states or intervening to increase low testosterone secondary to disease states could, in theory, reduce risk of death. Scenario 2: Disease states cause both premature death and low testosterone. If low testosterone does not cause death, it will nevertheless appear to be a risk factor for death if disease states are not taken into account. In this scenario, only intervention on disease states, their risk factors, and other sequelae could, in theory, reduce risk of death. Scenario 3: Low testosterone causes disease states, and sequelae of disease states influence premature death. In this scenario, low testosterone also may influence death without causing disease states, and not all disease states that influence death are caused by low testosterone. In this scenario, prevention or intervention to address low testosterone could, in theory, prevent disease states and, thus, prevent premature death. The current standard of care is to intervene in disease states or prevent them via behavioral or pharmacologic approaches with a goal of minimizing early death (Scenario 2). Scenario 3 suggests the possibility of another strategy to improve health and longevity.
|
The 2 recent studies (1)(2) support both Scenarios 1 and 3, and help rule out Scenario 2, by taking into account disease states, their risk factors, and sequelae. Both studies excluded men with prevalent chronic diseases and adjusted for mortality risk factors (e.g., adiposity, smoking). Laughlin et al. also adjusted for circulating indicators of risk and sequelae of disease states. In all analyses, low testosterone was associated with higher risk of death (1)(2). Although Scenario 2 is not fully excluded and it is unclear why 3 of 5 recent studies failed to find an association of circulating testosterone with mortality in older men, Khaw et al. (1) and Laughlin et al. (2) provide an impetus to continue evaluating whether testosterone is a useful clinical predictor of death in relatively healthy older men and, thus, is a possible target for intervention. Several issues need to be addressed well before these findings can be translated into the clinical arena. Some of these issues are discussed in detail in (6)(7).
issues in using testosterone for predicting premature death in older men
1. what is the optimal definition of low testosterone in older men?
Many older men have low serum total testosterone and most have low free testosterone concentrations as defined by reference intervals in young men, but the optimal testosterone concentration in older men remains to be determined. To do so, specific health and well-being criteria must be used to define "optimal," e.g. low bone mineral density, libido, or muscle strength; high fat mass, particularly centrally located; frailty; and premature death. Whether older men can expect to have the same testosterone concentrations as young men also must be determined.
Given a lack of consensus and data, no cutpoint has been uniformly adopted; cutpoints of 8.7, 10.4, and 11.3 nmol/L (250, 300 and 325 ng/dL) are common in literature describing the aging male and health effects of low testosterone. Instead of using these cutpoints, some studies (1)(2) divide testosterone distributions into quartiles. This approach improves precision across categories. Khaw et al. reported a relative risk of death equal to 0.59 when they compared the highest to lowest quartile of serum testosterone (comparing lowest to highest quartile: relative risk = 1.69) (1). Laughlin et al. reported relative risk = 1.44 comparing lowest to highest quartile (2). Although etiologically informative, use of quartile cutpoints does not point to the optimal cutpoint for risk prediction.
Laughlin et al. conducted an additional analysis that is potentially clinically informative. They divided the testosterone distribution into deciles and found that, compared with the highest decile, the risk of death appeared to increase from the 4th to 1st (lowest) decile. Men in the 2nd to 5th deciles had risks similar to men in the highest decile; the median testosterone concentration of the 5th decile was 10.4 nmol/L (300 ng/dL) (1)(2). This analysis provides evidence that an optimal cutpoint for predicting premature death may be <10.4 nmol/L (300 ng/dL) if the link is causal.
More work is needed to define a cutpoint for low testosterone that, with good sensitivity and specificity, identifies men who are more likely to die early. This cutpoint may or may not be the same as for that for evaluating treatments for low testosterone to intervene on adverse consequences. After adopting a clinical cutpoint, more work will be needed to determine how many assessments are needed, over what time interval, to be able to make a diagnosis of low testosterone in an older man and to determine at what age a measured low concentration is best done to predict risk of death.
2. what is the optimal assay method for measuring serum testosterone concentration?
Etiologic researchers have moved away from RIA in favor of nonradioactive immunoassays, and in some cases higher sensitivity mass spectrometry–based assays, for measuring serum testosterone concentrations. When mass spectrometry is not used, some advocate using an extraction step to avoid interferences that can occur in immunoassays and chromatographic methods. Admittedly, testosterone concentrations in men likely would be ranked comparably with each method, and thus the association between the lowest (vs highest) quantile of testosterone and death (or at least inferences drawn) would not differ notably among methods. For example, the 25th percentile cutpoints were quite different in the study of Khaw et al. (12.5 nmol/L = 361 ng/dL), which used a chemiluminescence immunoassay (1), and the study of Laughlin et al. (241 ng/dL = 8.35 nmol/L), which used an RIA after extraction (2), yet the studies provided nearly the same inferences (relative risk = 1.69 (1) and relative risk = 1.44 (2)). Moving forward, however, a cutpoint must be identified for evaluating risk prediction.
Use of assay methods with similar analytical characteristics will be essential for combining findings across studies to identify an optimal cutpoint. Indeed, initiatives to resolve the measurement issues are underway at the CDC (http://www.endo-society.org/news/press/2008/Tests-to-Measure-Steroid-Hormones-Not-Up-To-the-Task.cfm, accessed: 26 April, 2008). If measuring testosterone and intervening to increase low concentrations proves effective, testosterone measurements may need to be standardized in a manner similar to standardization to maintain consistency in cholesterol testing in the US (http://www.cdc.gov/labstandards/pdf/crmln/CertProtocolClinLabsMay04.pdf, accessed 26 April 2008).
3. is total or free testosterone the better predictor of death in older men?
Determining whether total or free testosterone is the better predictor of death in older men will require simultaneous measurement of total and free testosterone in the same men. Free testosterone has not been measured routinely in epidemiologic studies of chronic diseases and was not measured in the recent studies on risk of death. Instead many groups estimated the molar ratio of total testosterone to sex hormone–binding globulin or estimated free testosterone from individual total testosterone and sex hormone–binding globulin measurements and individual or population-level albumin. Other investigators adjusted for sex hormone–binding globulin and sometimes also for estradiol, which is also carried by sex hormone–binding globulin. Use of so many different approaches to attempt to capture the small percentage of testosterone that is available to cells limits comparison of findings among studies and comparison of findings for total and free testosterone.
4. what sources of biological variability must be considered?
Two key sources of biological variability in serum testosterone concentration are time of day and age. Testosterone is highest in the morning in men. Yet, some studies on testosterone and death (1) collected untimed blood specimens, likely for convenience and/or because testosterone was not an original study biomarker. In these studies circadian variation in testosterone concentration is likely introduced, which reduces the ability to detect associations. Other studies (2) collected morning fasting samples (likely to measure meal-dependent analytes such as insulin and glucose); intraindividual testosterone variation should be decreased in such samples.
Most cohort studies collect only 1 blood specimen for feasibility reasons. What cannot be captured with a sample collected at one point in time is the decline in total and free testosterone with age and individual man-to-man variability in the slope of decline. Thus, no large-scale work has been done to determine whether the slope of the decline with age is associated with risk of death; a rapid decline in testosterone might be hypothesized to be associated with a higher risk of early death.
closing thoughts
If circulating testosterone measurement issues are resolved and it is established that low total or free testosterone predicts risk of death, the next challenge will be to determine whether interventions to increase serum testosterone are effective in decreasing the rate of early death in older men. Trials would be needed to determine whether screening for low testosterone, along with subsequently intervening to increase it when it is low, reduces early deaths. Such trials are likely to be problematic. Population-level screening for low testosterone is not recommended at present, nor is testosterone replacement therapy for older men with low concentrations recommended, except possibly in individuals with pronounced symptoms (http://www.endo-society.org/publications/guidelines/final/upload/AndrogensMenGuideline053006.pdf, accessed 26 April 2008). The array of benefits and risks of testosterone replacement in older men is unknown, making the ethics of such trials unclear at present. Additional careful work is needed to further define the health risks of low testosterone, to elucidate the mechanism of the associations of low testosterone with death described in the recent reports (1)(2), and to determine which form of testosterone is most useful to measure and improve measurement methods.
Acknowledgments
Grant/Funding Support: Maryland Cigarette Restitution Fund at Johns Hopkins.
Financial Disclosures: None disclosed.
Acknowledgments: I thank the members of the Hormone Demonstration Program, a multidisciplinary collaboration at the Johns Hopkins Bloomberg School of Pubic Health, the Johns Hopkins School of Medicine, and the Sidney Kimmel Comprehensive Cancer Center, for thoughtful discussions on the role of sex steroid hormones in men’s health. This Program is funded by the Maryland Cigarette Restitution Fund at Johns Hopkins.
References
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  A. M. Traish, F. Saad, R. J. Feeley, and A. Guay The Dark Side of Testosterone Deficiency: III. Cardiovascular Disease J Androl, September 1, 2009; 30(5): 477 - 494. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
