The Science Of Testing Athletes For Doping Isn’t As Easy As You Think

Another #31writenow post….

In my former life, I was a chemist. I have a BS and MS in Chemistry, and worked in industrial labs until September 2012, when I changed careers. In my last technical role, I ran an analytical chemistry lab, and a large part of my job was method development, both for QA/QC but also for unknowns identification. While my training was in polymer chemistry, I had an affinity for analytical chemistry, specifically chromatography.

If you read/watch/listen to the news, then you know that the sports world is highly consumed with the idea of doping, or using performance enhancing drugs (PEDs) for athletes to gain an edge. This week, MLB announced numerous players have been suspended for using PEDs, most notably Alex Rodriguez. Other sports, like track & field and cycling, are extremely concerned with doping, and are often in the news for suspending athletes who use PEDs. The way the news presents these stories, it seems that athletes use these drugs, get tested, and then get caught. In reality, it doesn’t happen that way at all.

And now, it’s time for a science lesson.

Let’s pretend you’re in the lab, and you’ve been asked to analyze samples and determine if someone is doping or not. How would you go about that? Most people assume that you run a test, the equipment spits out an answer, and voila! you can say definitively whether someone is doping. I wish it were that simple – I’d probably still be a chemist if it was.

So you’ve got this sample, right? The first thing you need to know is what you need to test for – what’s the analyte(s) of interest? You can’t say “everything” because you’ll be running that sample until the end of time. You need something specific to look for – it could be a designer pharmaceutical, it could be a hormone like testosterone, etc. Once you’ve identified what you want to look for, now you need to understand what level of testing you need – does it need to be qualitative or quantitative? Qualitative testing will tell you if the analyte is present in the sample, but it won’t tell you how much is there. Quantitative testing will tell you if the analyte is present in the sample and quantify the amount. Therefore, quantitative testing generally requires more work because you need some type of method of determining the “how much” part, but we’ll get into that a little later. It may seem like you’d always want to choose quantitative testing, but it’s not necessary all the time. For example, let’s saying you’re testing for an illegal drug, and the only way for it to be present in a person’s sample is if they are using the drug (meaning a non-user would have none of the drug present in their sample). In this case, a qualitative test that simply tells you if the analyte is present or not would be sufficient, because all you need to know is if the drug is present or not. It doesn’t matter how much is there – any trace amount is bad. Now there are other times were quantitative testing is essential, like when athletes are using testosterone. Because that’s a natural hormone that everyone has in their body at some level, it’s crucial to figure out how much a person has in their sample. If you simply test to see if it’s present or not, everyone would be in trouble because everyone has testosterone in their body. Get it?

Alright, so you’ve figured out what analyte(s) you want to look for and what type of testing you’ll need to do. For the sake of this example, let’s say you need to do quantitative testing, so you’ll need to report how much of the analyte is in the sample. Now you need to determine what technique you’ll use to analyze the sample. Generally this involves some type of chromatography, which is the science of separating a mixture into individual components and detecting them. There are numerous chromatography techniques, based on medium (gas, liquid, ionic, etc) and even more detection methods (UV, FID, MS, etc). I won’t go into too much detail on this portion, but it’s vitally important to tailor your chromatography & detection technique as much as possible to the sample. The key to chromatography is the ability to separate your analyte(s) of interest from the rest of the sample, which can be achieved in various ways – by size of the molecule, the polarity of the molecule, the ionic charge of the molecule, etc. You also want think about other characteristics of the analyte(s) – does it fluoresce under a UV light, does it volatilize (go from a liquid to a gas) easily, is it similar to other compounds in the sample. All these factors must be considered when determining what type of chromatography method to try. With a brand new analyte that isn’t an industry standard, there’s a lot of trial and error at this phase, but with an analyte that’s common, it’s possible to find information on the best testing method in scientific journals or various websites.

Let’s assume we’ve found our chromatography method and it’s going to be HPLC, which stands for high pressure (or performance) liquid chromatography. In this method, you have a liquid called the mobile phase which is constantly moving through the entire system via a pump, which will pump at a set flow rate. You inject a small amount of your sample into the flow path, where the mobile phase carries it to the column(s), which do the work of separating your analyte(s) from the rest of the sample. The mobile phase then moves your analyte(s) to the detector, where the presence of your analyte is recorded, and then flows out to the waste container.

Simple HPLC Schematic

I’m sure you’re asking yourself “but how does this tell me how much of my analyte is in the sample?”. Great question! The short answer doesn’t. Not empirically. The equipment has no concept of units of measure, which is why you have to supply them. This is where we use what’s called a calibration curve. A calibration curve is a set of standard samples, that contain a known amount of the analyte you’re testing for, that you run before running your sample of interest. Generally calibration curves are made on a linear scale that correlates to the range of possible values for your analyte in the sample.

But before we can make our calibration curve, we have to determine two different things – the limit of detection (LOD) and the limit of quantification (LOQ). Both of those values are essential to our testing scheme, and important to understanding PED testing overall. LOD is the smallest amount of the analyte that can be detected by our chromatography method – basically, what’s the smallest amount of analyte that will register a peak that isn’t just background noise. LOQ is the smallest amount of the analyte that the chromatography method can detect but also calculate a quantity based on the area of the peak. It may seem that LOD = LOQ, but that’s not true – this is another reason why the distinction between qualitative and quantitative testing is so important. We may be able to detect the presence of an analyte at low levels, but we may not be able to quantify it until the level is higher. When you’re analyzing for trace amounts, this can be a big challenge.

Once we’ve determined our LOD and LOQ, next we need to determine our calibration curve range. The calibration curve range should be set so that our analyte of interest comes out near the middle. Because we’re making a linear calibration curve, we’ll need to have at least 2 points (remember from geometry class that 2 points make a line) but to be more accurate and precise, a good rule of thumb is to use 4-5 points. Each point is a specific concentration of the analyte in a medium similar to that of our sample. If we’ll be diluting our sample in a solvent, the calibration standards must also be made using the same solvent, so that the samples will respond in the same way. The other crucial part of calibration standards is the source of the analyte used in the calibration standards. For the standards to be accurate, a pure sample of the analyte should be used, preferably one from a chemical manufacturer or standard body, which certifies its purity for use as a calibration standard. For known analytes, this is pretty easy, but for a new chemical compound, finding a standard can be almost impossible. We’ll come back to this a little later.

Example of a calibration curve

Where do the points come from? Well first each of the calibration standards are analyzed using the HPLC. For each sample, there will be a chromatogram (that’s the graph that results) with a peak corresponding to the analyte. If you run the calibration standards in ascending order, you’ll see the size of that peak get bigger with each run, which makes sense considering that the amount of the analyte in each sample is bigger than the previous one. Because we made these standards, we know the exact amount of analyte in each of these samples, and we can then give that information to the software program. We can tell it that the first sample with the smallest peak area has 10 micrograms of analyte, the second sample has 50 micrograms of analyte, the third has 100 micrograms of analyte, etc. We then tell the software to fit a line to those points, and presto, we have our calibration curve. This line can now be expressed in y=mx+b format (remember that from algebra?) and now the next time we run a sample, the software will use that equation to determine how much of the analyte is in that sample.

Alright so let’s recap. We figured out what we want to test for, how we’re going to test, and even made a calibration curve. Now all we have to do is run our samples and get an answer, right?

If only it was that simple – and this leads me to the real point of this piece.

Analytical testing is pretty straight forward, if you know what you’re looking for AND they are known compounds. The trouble with PED testing in athletics is that most of the time, the testing community hasn’t kept pace with the scientists, trainers and athletes who are designing and using these drugs. Obviously the users aren’t going to volunteer samples of the chemicals they are using to give them an edge, so it requires that scientists synthesize their own versions for use in the lab when developing testing methods. With a new analyte, there is a lot of trial and error to determine the optimal method of testing for the analyte – what technique to use, what specific parameters (temperature, pressure, mobile phase, etc) to use, etc. Method development is a long arduous process, and even more so when you have to start from scratch. If a compound is new and testing methods are not available to analyze for it, it will be missed in screening tests. Remember, these tests don’t identify everything that is in a sample – they are looking for specific analytes.

Another concern is the capability of the instruments themselves. Remember our talk about limit of detection (LOD)? LOD isn’t a static value – as technology gets better, scientists are able to detect smaller and smaller amounts of analytes. Remember, not being able to detect an analyte is not the same as the analyte not being present. If the analyte is present, but the quantity is lower than the LOD, the instrument won’t be able to distinguish it from background noise. A perfect example of this was illustrated in the 30for30 special on the 1988 Summer Olympics, where several track athletes used PEDs. Several athletes were caught at the time, and others were suspected but had clean tests. Those samples were saved, and when they were retested years later, on more advanced equipment, those samples also tested positive for banned substances, they simply had levels that were below the LOD for the equipment used in 1988. Technology continues to get better, but someone will always be able to figure out how to use just enough of a PED to gain an edge, without being detected by the current testing technology.

Think about the last few doping scandals that have taken place – how many of these cases resulted in positive test results? Not many, especially in the MLB case. These cases aren’t being made by testing, because the testing technology isn’t keeping pace with the abundance of PEDs out there. These cases are being made because someone with insider info is blowing the whistle. It’s not a simple “well test them and get the answer” situation – testing is a huge undertaking. If athletic governing bodies are truly committed to rooting out PEDs, they would do well to make a serious investment in upgrading testing technology, which includes collaborating with researchers and equipment manufacturers. Until then, the science will always be one step behind.

  • I love everything about this post. Nerds rule! This reminds of malware detection in the tech field. Producers of malware are usually one step ahead.

    What has bothered me most about PED testing in MLB they’ve been going after the high profile people to make it appear as it they’re tough on testing. Most of the players caught are household names in their own house. Dave Zirin had an interesting take on it.