Professional guidelines recommend total testosterone (TT) as the initial screening test to evaluate hypogonadism in men, with a confirmatory test for free testosterone (FT) performed only in patients with TT levels close to the lower limit of normal. However, just how well TT concentrations predict FT has not been studied extensively. Researchers recently examined this question, and their findings are presented in this issue of Strategies.

Male hypogonadism is a relatively common condition that affects about 10% of men older than age 30, and as many as 40% older than age 70. However, there are challenges in diagnosing it, including lack of agreement on the clinical presentation, as well as a host of issues related to measurement of both TT and FT. In addition, little research has been undertaken to examine how well TT levels predict FT, the biochemical confirmation of hypogonadism. Recently researchers at the University of Washington and Veterans Administration (VA) Puget Sound Health Care System in Seattle sought to explore how well TT testing performs in identifying men with normal versus low FT levels (J Urol 2012;187:1369-73).

“The principal question we were trying to answer is how good is total testosterone at predicting a normal or abnormal free testosterone level,” observed lead author, Bradley Anawalt, MD, professor and vice chair of medicine at the University of Washington. “Total testosterone levels have been recommended as the test of choice to start the process of diagnosing male hypogonadism. So we wanted to see how well a low total testosterone level predicts a low free testosterone level, and conversely, how well a normal total testosterone level predicts a normal free testosterone level. As clinicians, we want that initial test to be reasonably consistent with the gold standard.”

Updated guidelines of the Endocrine Society recommend using TT as the initial screening test for hypogonadism, with FT measured only in patients with TT levels near the lower limit of normal and in whom the clinician suspects there may be alterations of sex hormone-binding globulin (SHBG). However, most of the studies examining the relationship between TT and FT have been small or in selected patient populations, according to Anawalt.

Anawalt and his colleagues took advantage of the large patient population served by the VA Puget Sound Health Care System, identifying through the institution’s electronic medical record all men evaluated for hypogonadism with a serum testosterone panel, which includes TT, SHBG, and albumin measurements, calculated FT (cFT), and calculated bioavailable testosterone. In all, the study analysis included 3,672 men.

The researchers generated receiver operating curves (ROC) to see how well TT measurements predicted low cFT. With TT as the continuous variable to predict low cFT at increasing increments from 34-40 pg/mL, they found the ROCs ranged from 0.93-0.92. Although TT generally predicted cFT, the authors found that 61.7% of patients with low TT had normal cFT. They used the ROC to determine the sensitivity and specificity of clinically useful thresholds of TT to exclude and predict low cFT, and found the sensitivity of low TT did not exceed 95% until the threshold for TT was defined as <350 ng/dL. The specificity for low cFT did not exceed 95% until the lower threshold for TT was defined as <150 ng/dL. This contrasts with what the authors had hypothesized, that TT <280 ng/dL would be sufficiently sensitive to rule out low FT, and that TT <150 ng/dL would be adequately specific to predict low FT.

“It was surprising to us that so many men had low total testosterone but normal free testosterone levels, and specifically that you couldn’t be sure that a man had low free testosterone based on total testosterone level until the total testosterone was less than 150 ng/dL,” said Anawalt. “That was a little unexpected, because the normal range with accurate, reliable assays is about 300 to 900 ng/dL, and most clinicians would regard a total testosterone level of 200 or 220 as low enough to indicate that patient is hypogonadal. But our data strongly argue against that. We found that a man’s total testosterone has to be very low before we can be certain he is hypogonadal. Conversely, a man has to have a total testosterone level that’s clearly well above 300 ng/dL before we can be comfortably certain that he is eugonadal.”

Based on their findings, the authors concluded that while a TT level >280ng/dL measured on an accurate, precise assay significantly lowers the likelihood of hypogonadism, a level between 280-350 ng/dL is not sensitive enough to reliably exclude hypogonadism. TT levels must exceed 350 or 400 ng/dL to reliably predict normal FT.

According to Robert Fitzgerald, PhD, DABCC, the authors’ findings correspond with another important analysis published last year, which could become the definitive study of reference ranges in TT and FT in men. That paper, published in the Journal of Clinical Endocrinology and Metabolism, used liquid chromatography tandem mass spectrometry from samples of healthy, non-obese young men to establish reference ranges. “If you compare the reference ranges from the Journal of Clinical Endocrinology and Metabolism paper with Anawalt’s data, his data make a lot more sense. This paper, the best by far on testosterone reference ranges, not only established reference ranges in three populations, but also looked at outcomes related to those ranges. Basically the authors found the lower end of the reference range for total testosterone in men is 348 ng/dL,” he explained. “Likewise, Anawalt found that a total testosterone level of at least 350 ng/mL can be used as the rule-out diagnosis for hypogonadism.” Fitzgerald, who was not involved in either study, is a professor of pathology at the University of California—San Diego. He also is the AACC representative on the Partnership for Accurate Testing of Hormones (PATH), an initiative aimed at achieving standardization of hormone tests, including testosterone.

Both Fitzgerald and Anawalt emphasized the need for standardized testosterone testing. Indeed, the authors point out that testosterone assays are “notoriously variable,” thereby limiting extrapolation of their results to men in the general population. “My gut feeling on Anawalt’s paper is that I like the data, but the interpretation highlights some of the issues the field is having,” said Fitzgerald. “Those issues include the fact that there are no established reference ranges that people agree on, there’s no standardization of the assays, there’s really no gold standard for the definition of hypogonadism, and there are no standard sample collection times, even though we know testosterone concentrations vary by 30 percent within individual from morning to afternoon.”

As these issues are being resolved, Anawalt encouraged laboratorians to ensure that they are using robust TT assays and that they work closely with physicians in interpreting TT results. “I would hope that they would look at their own assays and determine whether they conform to current recommendations. Laboratorians also should develop a reflexive panel so that if a patient’s total testosterone is less than 350 ng/mL, they measure SHBG and calculate free testosterone. In addition, they should inform their local clinicians about the importance of following up any abnormal total testosterone results with an accurate assessment of free testosterone,” he said. “Most clinicians are not very knowledgeable about assays and have to lean on the laboratory medicine experts to do the right thing.”

The June 2012 issue of AACC’s Clinical Laboratory News also provides an overview of challenges involving and progress towards standardizing testosterone measurements.