Interval Training and Biomarkers

Bill Parsons winning the 1967 Nevada City Bike Race. I think this counts as an all-out sprint.

High Intensity Interval Training (HIIT) has been all the rage for some time now, promising a large increase in fitness for a small amount of time. A recent publication¹ supporting this promise has gotten a lot of press over the last few days, primarily because it claims to be one of the more rigorous studies on this topic. The authors argue in this publication that three 20 second sprints provide the same health benefits as 45 minutes of moderate effort cycling; I could obtain all the benefits of my MAF Test workout in one minute!

That is an overstatement, of course. The HIIT workout needs to include the two minute recovery time between the 20 second intervals as well as the five minutes of warmup and cool down it shares with the moderate intensity workout. In total, the HIIT workout takes 10 minutes, the moderate intensity workout, 50 minutes. And of course there is other fixed overhead that further diminishes the difference; the time required to get to and from the workout, to dress for cycling, to shower afterwards, etc. When you factor in that the 45 minutes of moderate effort are rather pleasant whereas the 60 seconds of intervals are sheer agony, the choice between the two workouts may not be as clear as it first seems.

All of that aside, I'd like to focus in on what I see as the core issue; has this study really demonstrated that a workout centered on three 20 second intervals provides the same long term health benefits as a workout centered on 45 minutes of moderate effort? How to evaluate these kinds of studies has been a recurring theme on this blog. I have previously noted, for example, that a single study, or even a series of studies from a single research group, needs to be interpreted with caution. At best, this recent study is just one more brick in the wall. I have also discussed the problems with observational studies. If the authors of the study had compared athletes who, on their own, had chosen a HIIT workout to those who, on their own, had chosen a more conventional, moderate exercise workout, their conclusions would be weakened by an alternative explanation; that preexisting differences between the two groups might explain any difference observed. For example, people who are naturally more athletic and as a result have better long term health prospects might be more attracted to HIIT, whereas those less gifted by nature might choose the comfortable workout. But the authors of this study avoided that problem by utilizing the gold standard² for this kind of study, a randomized trial. To be able to do so, however, they were forced to create a different set of problems for themselves. To make a randomized trial possible, they could not measure long term health, which is what they were really interested in, but instead had to measure a proxy. This kind of proxy is known as a biomarker³.

What the authors of this study are really interested in knowing is if HIIT provides the same increase in lifespan and improvement in health as the conventionally-recommended 150 minutes a week of moderate intensity exercise. However, to answer that question directly in the context of a randomized trial, they would have needed to persuade people to follow assigned, rigid exercise protocols, either HIIT or moderate exercise based on the flip of a coin and to stick to these protocols for the rest of their lives. This would never happen, and even if it did, results of the study would only become available after decades. So, instead of studying what they are really interested in, the authors chose to study something they could measure quickly which they believe correlates with a long and healthy life, a biomarker. Persuading volunteers to follow a randomized exercise protocol for 12 weeks is a lot easier than persuading them to follow it for 40 years. Thus, instead of measuring lifespan, lifetime incidence of diabetes, and heart disease, they measured VO₂max⁴, body weight, percent body fat, glucose tolerance, and muscle mitochondria⁵.

Why did they measure these particular things? Firstly, body weight and percent body fat are not related to the conclusions of this study, but I thought they would be of interest so included these results. Secondly, the reason exercise improves lifespan and long term health is currently not known. What is known is that exercise also affects VO₂max, glucose tolerance, and muscle mitochondria. Are these other effects the reason exercise affects long term health? It is plausible that they might, but this hypothesis remains to be tested.  So what are VO₂max, glucose tolerance, and muscle mitochondria?

• VO₂max is the maximum rate at which a particular person can use oxygen during intense exercise. Typically, VO₂max increases with exercise. When VO₂max is increased due to moderate aerobic exercise, at least part of this increase in VO₂max is due to increased heart capacity. It is not yet known if increases in VO₂max are always accompanied by increased heart capacity; it is possible that HIIT increases VO₂max without increasing heart capacity, for example.
• Glucose, like most chemicals, is normally maintained at a fixed level in the blood. If one increases blood glucose, either by eating or (as in the case of this study) by having glucose injected, glucose will rise but then eventually return to normal levels. Glucose tolerance is a measure of how quickly a person can return elevated blood glucose to normal levels. Glucose tolerance has been shown to be a strong predictor for adult onset diabetes. 
• Muscles can generate the energy they need either aerobically (by using oxygen) or anaerobically (without using oxygen.) To use oxygen, muscles use their mitochondria. For fuel, they can use glucose or fat. One effect of moderate aerobic exercise is to switch muscles from being more anaerobic, glucose-based to more aerobic, fat-based, and this switch is accompanied by an increase in the number of mitochondria in muscles. It is thought that using oxygen and using fatty acids are a sign of fitness and good health, but this is still speculative. It is also thought that when muscles use oxygen, they contribute to improved glucose tolerance. If this latter speculation turns out to be true, it may be that glucose tolerance and muscle mitochondria are not independent, but rather that increased muscle mitochondria causes an increase in glucose tolerance. This would mean that these apparently different measurements may, in fact, be measuring the same thing.

So, at long last, what was the result of this study⁶? The study involved three groups of men, with 6 to 10 men in each group. All the men were between 19 and 35 years of age, were not doing any exercise, and were overweight. One group was the control, they didn't do anything. One group rode three MAF tests a week, 45 minutes of cycling at moderate effort. One group rode three sets of intervals a week, where the exercise part of each set consisted of three 20 second sprints, riding as fast as possible. At the end of 12 weeks, none of the three groups had lost any weight. As expected, the no exercise group had about the same percent body fat as they did at the beginning. However, both exercise groups had a significant⁷ decrease in percent body fat; they had not lost weight, but they had converted fat to muscle. Similarly, the no exercise group had no change in VO₂max, glucose tolerance, and muscle mitochondria, but both exercise groups had similar increases in VO₂max, glucose tolerance, and muscle mitochondria. The conclusion the authors drew from this study is that HIIT provides the same benefits as moderate exercise in much less time.

Am I convinced that HIIT provides as much benefit as moderate exercise in extending longevity and improving health? Not yet. Don't get me wrong, I think this is a great study, I think that it adds a lot to our understanding of how different kinds of exercise affect the body, and that it makes a strong argument for the benefits of HIIT. I also would not argue with anyone who, based on current evidence, decided to build their weekly exercise plan around HIIT rather than moderate exercise. Not being convinced that this hypothesis is true is not at all the same as being convinced that it is false; rather, I remain to be convinced one way or another. The reason I am not convinced is two-fold. In the first place, I am not convinced that changes in VO₂max, glucose tolerance, and muscle mitochondria are the only reasons exercise improves long term health. It is highly probable that they contribute, but the extent of that contribution remains to be demonstrated. In the second place, I am not convinced that HIIT and moderate exercise have the same long term effects on these variables. After twelve weeks, HIIT and moderate exercise produce the same changes in VO₂max, glucose tolerance, and muscle mitochondria, but would these changes be equally maintained if the experiment were extended to a year or ten years? This is the problem with biomarker studies, they are only as good as their biomarker, and judging the relevance of a biomarker is a research project all of its own.

So is there no hope, is it never possible to know anything? Not at all! True, science never gives us absolute certainty, but as more and more evidence is collected, it is possible to be sure beyond a reasonable doubt. When one group publishes a biomarker study demonstrating the value of HIIT, we note this results with interest but also with skepticism. When a second group comes to the same conclusion using an observational study, our interest goes up and our skepticism down. As more and more experiments reach the same conclusion, each done independently by different groups of scientists, each looking at the same question from different angles, using different techniques, we become more and more confident of that conclusion. Next post I will take a broader view of this topic and share my intuitions. Stay tuned.

---

1) "Twelve Weeks of Sprint Interval Training Improves Indices of Cardiometabolic Health Similar to Traditional Endurance Training despite a Five-Fold Lower Exercise Volume and Time Commitment" by Jenna B. Gillen, Brian J. Martin, Martin J. MacInnis, Lauren E. Skelly, Mark A. Tarnopolsky, Martin J. Gibala of McMaster University, Hamilton, Ontario, Canada published in the April 26, 2016 issue of PLOS ONE (DOI:10.1371/journal.pone.0154075) 
2) Actually, the gold standard would have been a double blinded, randomized trial, but blinding is impossible in this case, so I will use this term to indicate they did the best that was possible.
3) The National Institutes of Health defines a biomarker as “a characteristic that is objectively measured and evaluated as an indicator of normal biological processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention.”
4) What they actually measured was VO₂peak rather than the more familiar VO₂max, but the two are equivalent for the purposes of this blog post.
5) To determine the number of mitochondria, they measured the level of two enzymes in leg muscle - citrate synthetase and hydroxyacyl CoA dehydrogenase.
6) Read the actual article to get a definitive explanation of how they did their study and what the results were. What I present here is a simplified summary and may contain errors that I introduced.
7) I use the word "significant" in this post to mean two different things; both must be true before I will use the word significant. First, statistical analysis must demonstrate that any difference seen is unlikely to be due to random chance. Second, the difference has to "matter". That is, if a decrease in body fat is seen, the amount of that decrease must be enough that the medical community believes it would contribute to better health.