Floyd Landis, most recent winner of the Tour de France, has tested positive for testosterone use:

Landis denied cheating and said he has no idea what may have caused his positive test for high testosterone following the Tour’s 17th stage, where he made his comeback charge last week. But he aims to find out.

“All I’m asking for,” he said Thursday via teleconference, “is that I be given a chance to prove I’m innocent. Cycling has a traditional way of trying people in the court of public opinion before they get a chance to do anything else.”

Now the cycling world will wait for results from a backup sample which, if negative, will clear Landis. If ultimately proven guilty, Landis could be stripped of the Tour title and fired from the team.

The Switzerland-based Phonak team will ask that the backup sample be tested in the next few days, manager John Lelangue said. The team suspended Landis after the International Cycling Union notified it Wednesday that he had an unusual level of testosterone/epitestosterone” when his test was taken last Thursday, the day he staked his comeback in the Alps.

I thought I would do some research and clarify some of the medicine behind this because the science of how we detect anabolic steroid use is often a litte obscure.

First of all, you should know that testing for anabolic steroids can sometimes be difficult because many of the steroids and hormones that athletes use are naturally occuring in the body. One of these hormones that is regularly abused is testosterone, the primary androgenizing hormone in humans. Testosterone abuse causes increases in muscle mass with or without exercise. How would we test athletes for use of testosterone? Well the straightforward solution would be to just measure their testosterone levels in blood or urine, but that actually doesn’t work because there is a relatively large variability between individuals in basal levels of testosterone.

Instead what anti-doping authorities use is the testosterone to epitestosterone ratio. Epitestosterone is a breakdown product of other hormones. It has never been really associated with a particular biological activity (although some research suggests that it is has an antiandrogenic effect). Until recently it wasn’t even known what enzyme produced epitestosterone, but recent research suggests that it is made from DHEA using an enzyme called 17 alpha hydroxy-steroid dehydrogenase, mostly in the liver.

Here is a diagram of the hypothesized mechanism of production:

What epitestosterone does and why it is there doesn’t really matter though. What does matter is that the ratio in an individual between testosterone and epitestosterone is relatively constant over time and does not show large variability between individuals. Therefore when testosterone is added artificially, the ratio between the two would go up and this can be detected. For Caucasians the ratio is 2 or less with Asian people having a slightly lower ratio. The arbitrary limit that anti-doping agencies set for what would constitute an unacceptable ratio is 6 4 (Ed: This was corrected after several commenters observed that the UCI protocol lists 4.) . This works only because it is assumed that there is little interconversion between the two compounds and that their rates of excretion are relatively similar. Under normal conditions this is the case:

Detection of exogenous substances means identifying the parent compound or at least one metabolite. Nevertheless, with substances that are produced endogenously, such as testosterone, the presence of the substance alone cannot be considered to be an offence by itself. Moreover, a cut-off value for testosterone concentration cannot be used because of large observed interindividual and intraindividual urinary concentrations of the steroid. However, intake of testosterone causes characteristic changes in the pattern of steroids excreted in the urine. Based on studies of athlete populations, the IOC adopted in 1983 a ratio of testosterone to epitestosterone glucuronides (T/E) with an authorised upper limit of 6.0 as a criterion for the administration of testosterone. Since epitestosterone is only a minor product of the metabolism of testosterone and does not increase after testosterone administration, the resulting effect is an increase in the T/E ratio. In several studies, the distribution of results in Caucasian athlete populations shows generally a mean T/E ratio less than 2.0 whereas in Asian populations the mean T/E ratio is significantly lower. The IOC rules clearly indicate that a T/E ratio greater than 6.0 constituted an offence unless there was evidence that this ratio is due to a physiological or pathological condition–for example, low epitestosterone excretion, androgen producing tumour, and enzyme deficiencies. In addition, it has been observed that hepatic metabolism of steroid hormones may be altered by administration of substances as ethanol with the resulting effect of modifying significantly the T/E ratio. (Citations were removed.)

Floyd Landis has failed one test of T/E ratio. This presumably means that his ratio was above 6 4.

OK so what now? What is going to happen about to Floyd Landis? Well, the authorities have indicated that he has other samples that will be retested. This could be to confirm the other result was not contaminated with something. The authorities may also want to take more samples to see if this result reflects a normal baseline ratio for him — the ratio should be consistent over time and in some cases a pathological state could result in a high ratio.

Another test that they could use — and that I have read some news stories that they may — is isotope measurement. The testosterone in your body is made from cholesterol. Cholesterol comes in part from the food that you eat. The food that you eat was created using energy originally derived from the Sun. It turns out that the process of photosynthesis is selective for certain isotopes of carbon over others. Therefore, it is possible to look at the ratios of these isotopes to see whether testosterone is naturally or synthetically derived. Synthetically derived testosterone should not have the same ratios of carbon isotopes.

There is a test where you take the testosterone from someone’s body and measure its isotope ratio to see if it is synthetic or all-natural. In all likelihood they will do this test as well. Here’s more on that test:

Even if longitudinal study gives good quality information on the potential steroid profile manipulation, there is a lack of definitive proof for the exogenous application of natural steroids. One possible way of solving this problem is the ratio of the two stable carbon isotopes 13C/12C, which can allow the differentiation of natural and synthetic steroids. As exogenous testosterone or precursors contain less 13C than their endogenous homologues, it is expected that urinary steroids with a low 13C/12C ratio originate from pharmaceutical sources. Endogenous steroids are produced from cholesterol in the body. Cholesterol is derived from an average of a wide variety of feed vegetal and animal precursors or synthesised from precursors of feed origin.

In plant tissue, the main source of variation in 13C/12C isotopic ratio is derived from the different photosynthetic pathways for carbon dioxide fixation. Plants incorporate carbon dioxide via photosynthesis by three different mechanisms: the Calvin cycle (C3) pathway, the Hatch-Slack (C4) pathway and the crassulacean acid metabolism (CAM) pathway. The C3 pathway results in a large change in the carbon isotope proportions relative to atmospheric carbon dioxide and hence discriminates more strongly against the heavier isotope 13C compared with the C4 pathway. Main representatives of C3 group are wheat, rice, potato, barley, grape, oats, and sugar beet, whereas maize, sugar cane, millet, and pineapple are the important species of the C4 group. The difference in the 13C enrichment of food products in the diet and even in the food chain is caused by different contribution of naturally 13C-enriched constituents. Because maize, millet, and sugar cane (C4 plants) are the common food ingredients in some areas of Africa, it is expected that the basic 13C enrichment of the body store will be high for local populations.34 It is known that urine samples collected from a country such as Kenya have a higher content of 13C in steroids than western or oceanian countries.

The method for determining the isotopic composition of the relevant analyte includes gas chromatography, a subsequent combustion to CO2, and finally, mass spectrometric analysis of this gas in a special multi-collector mass spectrometer (gas chromatography/combustion/isotope ratio mass spectrometry, GC/C/IRMS). The 13C/12C value of testosterone or that of its metabolites will be measured and compared with that of urinary reference steroids within the sample to take into account variation in an athlete’s diet. In addition, it should be emphasised that the 13C/12C value of these endogenous reference compounds should not be affected by steroid administration. The result will be reported as consistent with the administration of a steroid if a significant difference is observed between the 13C/12C values of testosterone metabolites and the endogenous reference compound. (Citations were removed.)

There is a very legitimate question related to the timing of the positive test. Floyd Landis tested positive on day 17 of the Tour de France — right before he made a miraculous comeback to retake the lead. Some people may speculate that testosterone may have helped him do this. This is unlikely to be the case. Testosterone is used during training periods to gain muscle mass. It is not a stimulant like amphetamines. It is not likely that testosterone would improve performance on such a short time scale, nor has it ever been shown to have that effect.

Finally, there is this issue of Floyd Landis’s existing medical condition, osteonecrosis of the hip. Osteonecrosis of the hip is when there is a loss of blood supply to the head of the femur as it sits in the hip joint. This can be caused by injury — actually it is a very common injury in elderly women with osteoporosis. It is also a side-effect of steroid abuse in some cases. Some doctors have speculated that his hip problem might be indicative of steroid abuse:

He suffers from osteonecrosis, or avascular necrosis, in his right hip. The condition occurs when a bone or joint loses its blood supply, causing the bone to wear down. Landis said that in his case, the condition was brought about by a previous cycling crash.

That’s not unusual. Bone fractures are the biggest risk factor in developing osteonecrosis. However, what is unusual is the positive drug test, which adds a new twist to his story because osteonecrosis is also linked to steroid use, according to the National Osteonecrosis Foundation and other sources.

For reasons still unclear, people who take steroids — including anabolic steroids — appear to have a greater chance of coming down with the condition. Landis has admitted to taking corticosteroids, which, since they are used to treat inflammation and aren’t related to performance, are not banned and wouldn’t throw off tests for other steroids, doctors say.

All steroid use leading to or exacerbating osteonecrosis is a “well proven cause and effect,” said Dr. Mark D. Miller, an orthopedic surgery professor at the University of Virginia, who said that it’s possible Landis’ condition may have been caused or worsened by anabolic steroid use. Dr. Lewis Maharam, a New York sports medicine specialist, echoed that sentiment, but emphasized that osteonecrosis is just as likely to be caused by an injury.

It is really too early to tell whether these allegations are true or not. Frankly, I would believe either way. The science of detecting doping is improving but it still isn’t perfect. I just thought I would summarize the medical aspects so that when you read more about it you willl know the background.

UPDATE: I just heard Landis’s doctor on NPR claiming that Landis had a stiff shot of whiskey on day 16 and maybe that altered his test. It is true that ethanol can alter T/E ratio, but I would speculate the mechanism of that to be that ethanol use alters the levels of the enzymes in the liver that metabolize DHEA into epitestosterone. These enzymes do not change their levels rapidly. A comparable change is the risk that alcoholics get from using Tylenol. Acetaminophen is metabolized into something toxic in high doses in individuals who abuse alcohol becaue they have higher levels of a particular enzyme used to metabolize alcohol (this enzyme also metabolizes acetaminophen). Raising the enzymes levels requires regular alcohol use.

I find it highly unlikely that one shot of whiskey the night before the failed test would be sufficient to raise the enzymes in the liver sufficiently to produce a higher T/E ratio. It just does not happen that fast.

UPDATE: OK scratch that. The mechanism may be due to changes in excretion. However, in a paper related to this, I found the amount that you would have to drink to raise your T/E ratio 40% on average: about 2 g per kilogram body weight. Here is that scoop:

It was shown in a pilot experiment that ingestion of ethanol in amounts lower than 1 g per kilogram body weight had only small or negligible effect on the urinary testosterone/epitestosterone ratio. In this experiment, for seven subjects the ratio between testosterone and epitestosterone in urine was 1.00 +/- 0.07 before the ingestion of 0.5 to 1.0 g of ethanol per kilogram body weight. For urine samples collected 12 h after the ingestion, it was 1.13 +/- 0.07 (P >0.05, Student’s unpaired t-test). Ingestion of ethanol in amounts exceeding 1 g per kilogram body weight, however, always resulted in a significant increase in the ratio between testosterone and epitestosterone. Figure 1 shows the results of a carefully controlled experiment in which four subjects ingested 2 g of ethanol per kilogram body weight. After 14 h, the mean ratio had increased to 38% and after 22 h to 41%. The increase ranged from 30% to 90% in the different subjects we studied. The increase after 14 and 22 h
was statistically significant (P <0.05, Student's unpaired ttest).

Say he weights 70 kg (about 150 lbs). That would be 140 g of alcohol. Say a shot contains 20 grams of alcohol (I am being generous). Then he would have to have a hell of a lot more than a double to change his T/E ratio. He would have had to have more like 7 or 10.

So the question is: Did Landis get soused in the middle of the Tour de France?

UPDATE: Commenter Jeff asks:

John Eustice of ESPN is reporting that Landis’s high ratio was caused by a low epitestosterone (the denominator) level, not a high level of testosterone (the numerator). In other words, he doesn’t have a high testosterone level, just a high ratio because of a low denominator. How is it possible to get a low epitestosterone level and why? Would that help performance? What is going on?

The question is not so much whether he has a low epitestosterone overall, but rather why he would have a low urine epitestosterone. This issue is addressed in the review I cited about epitestosterone, but they draw few firm conclusions.

For example, older men with benign prostatic hypertrophy can have low excretion of epitestosterone:

In a group of men with benign hypertrophy of prostate in the age of 61-70 years a subnormal excretion of epitestosterone 24.5 +/- 12.8 μg per day compared with that of age-matched healthy men 42.8 +/- 14.3 μg per day was observed.

With respect to individuals who naturally have low levels of epitestosterone total or excreted there are some cases — and in some cases this may have to do with race:

The testing is, however, valid only under the assumption that the clearance of both epimers is similar, that testosterone administered exogenously is really not metabolised to epitestosterone, and that the ratio of both epimers is not influenced by racial or individual variations. Some exceptions have recently been observed [16, 50 and 52] in individuals with very low epitestosterone excretion, as have racial differences, e.g. between Japanese (T:E 1.99) and Ainu (T:E 2.77). The testosterone:epitestosterone ratio is also influenced by intake of various external compounds, e.g. by high doses of alcohol, more in females than in males. Andostenedione administration increases epitestosterone excretion while decreasing that of its putative precursor. Dehydroepiandrosterone supplementation can increase the testosterone:epitestosterone ratio. (Citations have been removed.)

I tried to look up the case studies of the individuals who had low epitestosterone excretion. Only one was available because they are relatively old (by Internet standards). That one is a case study, but they don’t explain why the person had low epistestosterone, only that he had it and did not have evidence of a tumor.

It appears that there have been cases of individuals with low epistestosterone excretion naturally. Whether Floyd Landis is one of these is unclear. I can’t off the top of my head think of any other reason why his epitestosterone would be naturally low.

UPDATE: I would add too that within physiological ranges testosterone levels do not correlate with performance. Only when testosterone is abused does it tend to improve performance. Actually, I found this article that suggests that they have even tested this assertion for both cortisol and testosterone in cyclists and found that the two are not correlated.

In the literature the use of plasma levels of cortisol and the testosterone and testosterone: cortisol ratio for training management is encouraged. Decreased levels of testosterone and increased levels of cortisol are suggested to be indicative for a disturbance in the anabolic-catabolic balance, which may express itself in decreased performance. The purpose of the study was to examine if the acute hormonal response to a bout of exercise and the resting levels of testosterone, luteinizing hormone (LH) and cortisol are correlated to performance in cyclists. In addition, the effect of training on this correlation was studied. Ten professional cyclists participated and measurements took place before and after a defined period of training. Maximum workload (Pmax), determined on a cycle-ergometer with a slowly increasing protocol, increased by 30 watt (p < 0.001). Workload at a lactate level of 4 mmol/l (P4) increased by 18 watt (p < 0.05). Post training, resting testosterone levels decreased from 28.8 +/- 74 nmol/l to 24.6 +/- 90 nmol/l (p < 0.05). Resting cortisol levels increased from 272 +/- 110 nmol/l pre training to 379 +/- 242 nmol/l post training (p < 0.05). These results indicate an increased catabolic state. The acute hormonal response and the resting levels of LH were not changed post training. The resting levels of testosterone and cortisol and the acute response to exercise showed no correlation with performance pre and post training. In spite of an increased catabolic state post training there was an increase in performance. These results suggest that in endurance trained cyclists, decreased testosterone levels, increased cortisol levels and a decreased testosterone: cortisol ratio does not automatically lead to a decrease in performance or a state of overtraining. (Emphasis mine.)

With respect to epitestosterone, no physiological action for it has ever been shown in humans, so I doubt that naturally low levels of it would affect performance.

UPDATE: Quitter, the issue of the spread of normal values is a good one, so I looked it up here. Click on the icon below and it shows a chart of the normal range of T/E values for a large cohort of male athletes:

Click on the second chart to see the variations in several athletes over several years:

What you see is that while the majority of individuals — greater than 99% 95% — have values less than 6 4, some do not. What they emphasize in the paper is that with individuals who have natural values greater than 6 4, you have to look at historical testing, because those individuals should have stable values over time (as indicated in the second chart). These individuals appear to also have somewhat greater variability in their values, but in general if they had high values once they should always have them.

I have no idea what Landis’s historical or recent tests are, but I agree that they need to check that not only was his value high but it hasn’t been before. I guess I would speculate that it wasn’t or we would have heard about this before. As to whether it would help, your guess is as good as mine. Not being a cyclist, I wouldn’t know.

David Winter, I read the bit about testosterone aiding muscle recovery, and I can’t find evidence on Pubmed for it either. The only trials I can find for supraphysiological doses of testosterone look at overall muscle strength. I think that this muscle recovery story must come from trials for hormone replacement in men with low testosterone, so it is not clear that they are comparable. I guess I would speculate that considering the typical time scale for testosterones action, it is unlikely that measurable effects would be observable that quickly. All the other studies with testosterone show effects over the course of several (10ish) weeks.

UPDATED: So I checked where I looked up the cutoff value for legal action, and I can’t find where I read 6. It says 4 in both the UCI and the WADA protocols. Hmmm. Sorry guys. I don’t know why I got confused.

Considering that the cutoff is 4, it highlights Quitter’s argument that the standard may be too strict. If you look at that bell curve, there are a bunch of people who have values over 4.

UPDATED: Floyd is, as they say, busted. He failed his isotope test. More here.