Put another way: what's the free testosterone cutoff we should use for initiating treatment in potentially hypogonadal men?

Spoiler alert: maybe 70 pg/mL?

Background: a new paper from Anna Goldman, Shalender Bhasin, Frederick Wu, Meenakshi Krishna, Alvin M Matsumoto, and Ravi Jasuja lays out their thoughts on the state of the science in free testosterone measurement. This is important because we currently operate under a theory of "bioavailable testosterone." That is, we think that only testosterone that is unbound by proteins, the most important one called "Sex Hormone Binding Globulin," or SHBG, has any effect on the testosterone receptor. All the testosterone that's floating around attached to SHBG or other proteins is inert.

But like any other hormone, testosterone's binding to SHBG or other proteins is affected by myriad causes, like obesity, other hormonal disorders, and other conditions or medications. Even the temperature at which it's measured. So guidelines from the Endocrine Society suggest measuring free testosterone levels in men whose total testosterone concentrations are low-ish and in men with conditions or medications that make total testosterone measurements less reliable (like obesity). But those same guidelines rely almost entirely on the total testosterone to determine who needs testosterone therapy. From their 2010 guideline:

The panelists disagreed on serum testosterone levels below which testosterone therapy should be offered to older men with symptoms. Depending on the severity of clinical manifestations, some panelists favored treating symptomatic older men with a testosterone level below the lower limit of normal for healthy young men [280–300 ng/dl (9.7–10.4 nmol/liter)]; others favored a level less than 200 ng/dl (6.9 nmol/liter). The panelists who favored treating men who had values less than 300 ng/dl were more influenced by the observation that men who have values below that level often have symptoms that might be attributable to low testosterone. The panelists who favored not treating unless the serum testosterone was as low as 200 ng/dl were more influenced by the lack of testosterone treatment effects in randomized clinical trials when subjects had pretreatment values of 300 ng/dl but suggestions of beneficial effects when the pretreatment values were closer to 200 ng/dl. The lack of definitive studies precludes an unequivocal recommendation and emphasizes the need for additional research. [emphasis mine]

No mention of the free testosterone in there. More research needed, they say. Well, Goldman et al have delivered it, in a way. They didn't do any new investigation, so far as I can tell, but they did review the existing literature pretty thoroughly and come up with a series of conjectures. The most important of these clinically is this (my words, not theirs):

Without a harmonized, replicable normal range between laboratories and methods, we cannot set a clinical threshold for free testosterone levels at which we should initiate treatment.

The authors point out that in a study by Le et al, only about a quarter of labs performing free testosterone assays would even confirm that they had performed validation studies on their assay. But seeking a clinically relevant answer, they go on to point out that In the Framingham Heart Study, the lower limit of the normal range for the calculated free testosterone for men between 19 and 40 (defined as the 2.5th %ile) was 70 pg/mL (242.7 pmol/L). They do not endorse this as a threshold. But given the lack of other published thresholds, I think it is at least a start. 

One more point before I wrap up: whatever value you or your doctor use, whether it's free testosterone, bioavailable, or total, the number is only one piece of data. A free testosterone of 70 pg/mL in a young man with normal body hair, normal muscles, and a normal libido would not cause me to treat him. That same level in a man with a known pituitary issue, enlarging breasts, and fatigue would definitely cause me to treat him. And I tend not to recommend checks of testosterone in elderly men, even those with fatigue, unless they have other compelling evidence of hypogonadism. And my opinion is generally in line with that of the Endocrine Society:

We recommend against a general policy of offering testosterone therapy to all older men with low testosterone levels. (1|⊕○○○)

We suggest that clinicians consider offering testosterone therapy on an individualized basis to older men with low testosterone levels on more than one occasion and clinically significant symptoms of androgen deficiency, after explicit discussion of the uncertainty about the risks and benefits of testosterone therapy. (2|⊕○○○)

Why else do I hesitate to give testosterone? In a series of trials in JAMA and JAMA internal Medicine earlier this year, men aged 65 and older with a total testosterone level <275 ng/dL plus either sexual or physical dysfunction or reduced "vitality" were randomized to get either testosterone gel (lotion that absorbs through the skin) or placebo for 1 year. The men who got testosterone developed more coronary plaque than the men on placebo. (This may or may not mean anything; in other observational studies, the rate of heart problems was lower in men prescribed testosterone than in untreated men). 

Importantly for the men who tend to come to me with suspected androgen deficiency, though, the men in these studies had no improvement in memory or other measures of brain function.

Testosterone did slightly improve bone density and bone strength compared with placebo. And testosterone seems to have improved the hemoglobin level (the weight of the oxygen-carrying protein in our blood) in anemic men by 1 g/dl more often than placebo did.

If all that sounds confusing, it's because it is. When I was an endocrine fellow, one of my faculty told me that in almost every case, if you don't know what to do, the correct path is to do nothing. So that's the direction I lean in borderline androgen deficiency cases, whether determined by total or free testosterone levels. I lean toward withholding therapy.