From the Wall Street Journal:

Apologies if this is posted already.

THE NUMBERS GUY
By CARL BIALIK

Floyd Landis's Alcohol Defense
August 2, 2006

One evening nearly two decades ago, four Swedish men in their mid-thirties gathered to quaff about 10 alcoholic drinks over six hours. Two weeks ago, American cyclist Floyd Landis says he drank two beers and "at least" four shots of whiskey after the worst day of his professional career.

Besides a taste for the bottle, these five men have something in common: The day after drinking, their urine showed an elevated "T/E ratio" of testosterone to epitestosterone, hormones that occur naturally in the body.

For Mr. Landis, the test result was bad news: It may cost him the Tour de France title, as the elevated ratio is indicative of the use of banned performance-enhancing substances that raise testosterone levels. On the other hand, that Swedish night on the town -- part of a body of research on alcohol's effect on testosterone levels -- might help him clear his name.

Testosterone and epitestosterone generally are in balance in the body, but some athletes inject steroids or other substances to artificially raise their testosterone levels, which can help long-term muscle building. (Though it generally takes more than a single day for any muscle-building effect to appear.) The day after his drunken night, Mr. Landis's T/E ratio was found to be 11-to-1, well above the 4-to-1 limit set by international cycling. But athletes' testosterone levels vary widely; for example, a test of saliva in U.S. university students this year found an eight-fold range of the hormone. If Mr. Landis's T/E ratio is normally toward the high end, a night of drinking could have raised it dramatically, putting him above cycling's limit.

That's the theory, anyway.

While Mr. Landis wanted to forget a rough day at the office, the Swedes were drinking at the behest of researchers in the department of clinical chemistry at Huddinge Hospital -- researchers investigating the case of a Swedish athlete whose T/E ratio had ebbed and flowed with his alcohol intake. In the study, the T/E ratio increased among the four male volunteers by a factor of 10% to 50%.

"Our interest here was just to demonstrate that we would see an effect," said Ingemar Bjorkhem, a co-author of the study and now a professor of clinical chemistry at Karolinska University Hospital in Stockholm. "We expected to see an effect."

The study was published in the journal Clinical Chemistry. As Dr. Bjorkhem recalls, he was one of the four study participants, though when asked if it was hard to find volunteers, he replied, "maybe not." It's also, as far as he remembers, his only opportunity to drink alcohol for a research study.

A handful of other studies conducted in Europe have confirmed this effect, but they've generally been limited to just a few participants, and they've found differing results. A 1996 study by researchers at the German Sport University in Cologne found an average increase in T/E ratio of 300% to 400% among six female volunteers and an average increase of 50% to 100% among five males. (The men's results were all over the place, ranging from a decrease of 40% to an increase of 300%).

"The influence was statistically significant," Mario Thevis, professor for preventive doping research at German Sport University, told me. (I wasn't able to read the study myself, as it's not available online.) Dr. Bjorkhem said that the late German drug-testing expert Manfred Donike also confirmed the result several times, but didn't publish all of these findings.

A 2001 review of the literature conducted by Simon Davis, then a postdoctoral student at Lawrence Berkeley National Laboratory in Berkeley, Calif., found alcohol-induced increases in the T/E ratio ranging from 30% to 277%. Excerpts from Dr. Davis's report, which was prepared confidentially on behalf of a U.K. athlete accused of doping and wasn't peer-reviewed, surfaced online last week (see it at Dirt Rag Forums) and was cited by Bloomberg. "I'm surprised it got out, really," Dr. Davis told me. "It was an internal document about the doping case." (He declined to identify the athlete who'd been under suspicion.)

This is one case where several studies don't necessarily yield a reliable conclusion, however.

"The literature isn't completely clear," Richard Hellman, president-elect of the American Association of Clinical Endocrinologists, told me, adding that the studies tested "just a few people" and found "so much variance" in the change in T/E ratio.

"The information is suggestive, but it's not certain," Dr. Hellman said.

Cycling union doctor Mario Zorzoli told me by telephone that Mr. Landis's test was the only positive one on the tour, out of six to eight tests for each stage, and at least five other tests for Mr. Landis on the days he was the overall race leader. That suggests that only once during the race, out of at least six tests, was Mr. Landis's T/E ratio 4/1 or higher.

The alcohol studies have been referred to in the press and cited by Mr. Landis's doctor, Brent Kay, since the T/E test result became public last week. When CNN's Larry King asked him about it last week, Dr. Kay said that "there are, in fact, a number of studies that show that alcohol definitely can have an effect on this … we're not speculating that that was the cause, but certainly there's documentation in the scientific literature." In a press conference, Dr. Kay put the number of scientific studies at five, according to NPR. The possible linkage of alcohol to elevated T/E ratios was also reported by the print Journal, Time, the Boston Globe, the New York Times and the Associated Press.

A colleague of Dr. Kay's at O.U.C.H. Medical Center (O.U.C.H. stands for Occupational Urgent Care Health/Sports), a group of sports medicine specialists in Temecula, Calif., told me Monday that Dr. Kay was no longer answering press calls about the test result, and was now focusing on Mr. Landis's planned surgery to replace his degenerative right hip.

Doctors and drug-testing specialists say the possibility that alcohol may profoundly affect the T/E ratio is one of several drawbacks with the test that's caused Mr. Landis so much trouble. "Not only is the analytical validity of the test in question, but the premise that a ratio of 6/1 infers a doping offence is also unsafe," Dr. Davis wrote in 2001. (Since then, cycling officials have lowered the threshold for test failure to 4/1.)

Dr. Hellman said blood should be tested as well as urine, because urine contains only the waste products of the hormones, rather than the hormones themselves. (Gary Wadler, a member of the World Anti-Doping Agency and an associate professor at New York University School of Medicine, told me he disagreed, because levels fluctuate more widely in blood.) Dr. Hellman also suggested more-frequent testing of athletes, including before events, to establish baselines, since T/E ratios vary so widely among different people.

"If someone's career is hanging in the balance, you don't want to say, 'There's an 85% chance that this test is right,' " Dr. Hellman told me.

Mr. Landis's situation is further clouded by a dearth of numbers from the tight-lipped International Cycling Union, which hasn't officially released Mr. Landis's test results -- so it's not known whether his testosterone level was elevated, or epitestosterone suppressed, or both.

A new wrinkle: The New York Times, citing an unnamed person at the cycling union, reported Monday night that a follow-up test confirmed Mr. Landis's urine contained some testosterone not produced naturally by his system. This result would have been determined by analyzing the atomic makeup of the testosterone molecule, via a technique called mass spectrometry. (Read more about this on Wikipedia.)

Such an analysis is based on the phenomenon that atoms of a given element, such as carbon, can have differing weights, depending on which subatomic particles make them up. The Isotope Ratio Mass Spectrometry reportedly used on Mr. Landis's urine would measure the ratio of Carbon-13 to Carbon-12 in his testosterone, and compare that to the ratio in his cholesterol. Natural testosterone in the body is derived from cholesterol, and so it should have the same ratio of C-13/C-12, but synthetic testosterone would have a different profile.

Dr. Thevis, from the Cologne sports lab, told me this method is reliable: "There have been a lot of studies showing that differentiation is absolutely reliable and reproducible." Dr. Zorzoli declined to confirm the Times report, but, speaking generally, said, "If the case is on the evidence of exogenous testosterone, alcohol intake doesn't create exogeneous testosterone in body."

But a more-cautious note was sounded by Dr. Davis, who is now the technical director for Mass Spec Solutions Ltd., a Wythenshawe, U.K., maker of mass-spectrometry devices. "Quite regularly there are errors in the isotope tests," he said. "It's a very difficult analytical technique."

In response to the report, Dr. Kay told the Times the test may have been inaccurate, adding that a wide range of factors, including alcohol consumption, may have been responsible for the test result.

The apparent sensitivity of the testosterone test's numbers to alcohol consumption, and the announcement of partial test results without full disclosure by the cycling union, has created a milieu for cyclists that is "almost Kafka-esque," Dr. Davis said. "The phrase often bandied about is 'chemical McCarthyism'."