A Genetic Variation May Hide Steroid Abuse

After the whole Floyd Landis thing, I wrote a long post about the science of detecting steroid abuse. The primary test uses something called the T/E ratio to determine whether the athlete has injected steroids. The World Anti-Doping Agency (WADA) has a maximum T/E ratio of 4. If an athlete gets greater than four on any test, an investigation gets started.

However, researchers in Sweden have just published a paper suggesting that this test has a possibly fatal flaw.
Schulze et al. show that a gene variant present with alarming frequency in the population allows individuals to inject testosterone without a large increase in their T/E ratio.

I went into a lot of background on the science of steroid testing in my older post. I am not going to rehash the whole thing, but there are a couple key points to remember.

Steroid testing for testosterone uses a test called that measures the concentration ratio in urine of testosterone and epitestosterone (the T/E ratio). Epitestosterone is a non-active steroid that is not a metabolite of testosterone. It had been observed in previous research that the T/E ratio is stable in individuals over time and is reliably under around 4 in most individuals. (You can see a distribution of T/E ratios in normal individuals in this chart from the old post.) The World Anti-Doping Agency that does these tests therefore sets a limit of 4 for the T/E ratio. If a tested athlete has a urine test that comes in greater than 4 -- like happened to Landis in the Tour de France -- an investigation begins to confirm that they were in fact abusing steroids. This investigation employs another test that we can talk about later.

Schulze et al. looked at a genetic variation for a gene called UGT2B17 which is involved in the metabolism of testosterone. Testosterone itself is not particularly water soluble, so in order to get it out of the body in the urine it has to be modified in one of many ways. One of these ways is by attaching a sugar molecule on it called a glucuronic acid to it. This process is called glucuronidation. There are two enzymes that do this: UGT2B17 and UGT2B15.

The testosterone that is measured in the T/E ratio test is not testosterone, but rather testosterone glucuronide. This means that it is really important for the accuracy of this test that everyone's UGT2B17 and UGT2B15 enzymes work at about the same level. If some persons enzymes do not work, the testosterone never undergoes glucuronidation. It may be still secreted in the urine (it can be conjugated with other things like sulfates to make it more soluble), but it won't show up on this test. This means that if you have defective versions of these genes you could in theory take a lot of testosterone to bulk up without changing your T/E ratio.

It turns out that there is a deletion mutation in UGT2B17 that follows just this scenario. Individuals who have two alleles of the deleted version of the gene produce significantly less testosterone glucuronide than if they have the wild type version of the gene. Under normal conditions they have lower T/E ratios, in most cases less than 0.4. More importantly, when injected with testosterone, they do not show a significant increase in T/E ratio.

This is illustrated in the figure below (Figure 2 from the paper):

The researchers injected 55 male subjects with 500 mg of testosterone. Then they took urine samples on successive days to measure the T/E ratio. They also examined the genotype for the subjects at the UGT2B17 locus.

The "ins/ins" group is the wild-type group with two functional versions of UGT2B17. The "ins/del" group are heterozygotes with one functional version. The "del/del" group have two deleted forms of the gene. As you can see, the "ins/ins" and "ins/del" show large increases in their T/E ratio in the days following the injection. The "del/del" group, on the other hand, did not show a reliable increase in the T/E ratio. 40% of that group did not go over the 4 criterion that WADA sets for detecting doping.

This is important because this gene variant is relatively prevalent. The researchers note that approximately 2/3 of Asians and 10% of Caucasians carry the gene deletion. That is a big problem for WADA because these athletes could abuse steroids with relative impunity under our current testing regime.

This is a really nice experiment, and it has important anti-doping policy implications.

So what's a WADA regulator to do? How can we fight steroid abuse better knowing this result?

Well, there are a couple of options.

First, you could genotype all the athletes out there. You would have to set up different T/E ratio criteria based on the athlete's genotype. However, this would probably not work because it would be prohibitively expensive.

Second, you could employ the other more accurate test for steroid abuse. I talked about this more in my other post. The test that they use when an athlete fails the T/E ratio test is one that looks at the carbon-13/carbon-12 isotope ratio in the testosterone in the athlete. Briefly, carbon-13 and carbon-12 are two naturally occurring, stable isotopes of carbon. Because 13 is slightly heavier than 12 (having another neutron), it is incorporated into plants in the photosynthetic pathway at a slightly lower rate. (It is a bit more complicated than that. The plants that discriminate between carbon molecules are the ones that use mostly the C3 pathway.) Because the testosterone injected by athletes is from plant sources, it shows up as having a different carbon-13/carbon-12 ratio than the testosterone made in their own body. This allows the test to distinguish between testosterone that was artificially injected and testosterone that was not.

It might be possible to start using this test, but that would be pretty expensive and time consuming as well. If you remember from the Floyd Landis thing, it took a while to get the results of that second isotope test.

The final solution would be to follow the T/E ratio within individuals over their entire athletic life to develop a baseline. I posted a figure in my older post about this method. (It is from this paper, Figure 5, click to enlarge).

The authors of that paper followed the T/E ratios of athletes at baseline over long periods. They found that while the ratio may vary between individuals, it does not vary much within individuals.

It is certainly possible through very regular urine samples to establish a baseline ratio for the athlete. Any swing above baseline could trigger the more accurate isotope test. Schulze et al. do establish that subjects with two deleted genes do not increase their T/E ratio much, they do increase it a little. I think that might be the only cost effective way to keep using the T/E ratio as a test.

In any case, it is important to keep evaluating the accuracy of these tests. Anti-doping is an arms race of sorts, and it is important to know the limitations of the techniques we are using.

Kevin Beck has more on this story here.

Hat-tip: NYTimes (whose coverage of this is also quite good)

Schulze, J.J., Lundmark, J., Garle, M., Skilving, I., Ekstrom, L., Rane, A. (2008). Doping Test Results Dependent on Genotype of UGT2B17, the Major Enzyme for Testosterone Glucuronidation. Journal of Clinical Endocrinology & Metabolism DOI: 10.1210/jc.2008-0218