Recently I’ve been diving back into the world of exercise physiology research, especially as it relates to running performance. I got behind on following ex phys in grad school (too busy following biomechanics!), so it’s been a few years since I’ve caught up on the latest research.
This weekend I was reading about capillarization: the growth of new, tiny blood vessels–capillaries–that run between and around individual muscle fibers. Capillaries are super important for aerobic performance, since they’re the place where oxygen diffuses out of red blood cells and into muscle fibers.
Here’s a (real!) cross-sectional slice of a quadriceps muscle sample under a microscope, with some staining to highlight a few key features (from Groen et al 2014). The black “holes” aren’t really holes; they’re the fast-twitch muscle fibers. The green blobs are slow-twitch muscle fibers, the blue areas are the cell walls, and the magenta-pink dots are the capillaries. The direction of blood flow is into/out of your screen.
Pink areas are capillaries - more of these per muscle fiber is a good thing!
If training could induce the growth of new capillaries, each muscle fiber would, on average, have more capillaries “touching” the fiber–a higher capillary to fiber ratio–and as a result, fibers would have easier access to oxygen for aerobic energy production and an easier outlet for lactate. That would be great! And it turns out training can indeed induce “capillarization.” So, naturally, we’d want to know what kind of training optimizes this adaptation?
What’s the best kind of training to increase capillary density?
A solid body of scientific research suggests that the main trigger for inducing capillary growth is the mechanical stretching that capillaries experience when blood pulses through them during exercise. Building on these mechanistic findings, a string of small studies (example here, 9 subjects) in the 2000s and 2010s made a case that the best way to improve capillarization was moderate-intensity, continuous exercise: steady running at 50-80% of VO2max, for example. This exposed the capillaries to long bouts of stretching, while avoiding a potentially capillarization-inhibiting hormonal response that is generated during fast intervals. Excellent, another win for high mileage!
Except…the party was spoiled a bit by a study that came out last year: instead of a single study, it was a meta-analysis that aggregated data from 38 different studies, with over 400 subjects in total. The main findings: compared to low-intensity training (<50% VO2max), moderate-intensity training (50-80% VO2max) increased capillary to fiber ratio by 21%, but higher-intensity intervals (80-100% VO2max) increased capillary to fiber ratio by 54%!
Now, those results were only for the sedentary subjects, who composed about 90% of the subject pool (no results were significant for already-fit athletes, probably on account of the small sample size). But I’m less concerned about the specific findings here, and more concerned about the general problem of what to do when you have a pretty reasonable theory that gets walloped by a large, comprehensive meta-analysis.
Believing all the studies vs. some of the studies
The reason I’m conflicted is because there are meta-analyses in my “home field” of running injury biomechanics that come up with findings that I find somewhat implausible, like this one that found that one very specific ground reaction force parameter (vertical loading rate) was related to one specific type of injury, while other very similar parameters (vertical impact peak) was not.
To be clear, I’m not saying that the analysis is wrong, just that pooling together many studies won’t help if the individual studies themselves are flawed. As a more extremely-implausible example, here’s a perfectly reasonable-looking meta-analysis showing that homeopathy is an effective treatment for ADHD (critique here).
At the same time, it’s very, very easy to be led astray by small studies that seem to support your theoretical constructs. A few papers on a handful of animals with incorrectly-done statistics were what kicked off the biomechanical obsession with impacts and loading rates in the first place, after all. So trusting a few studies doesn’t seem much better than trusting all of them, especially if you’re not in a position to really dig into the fine details of the included studies.
How I read exercise science papers when research is conflicting
After six years of doing exercise science research, the best solution I have is to be a “sloppy Bayesian”: combine some reasonable “priors,” or beliefs about what is plausible given your previous knowledge, with a sober assessment of the new evidence in front of you. A strong prior against the claimed mechanism behind homeopathy, plus the relatively weak studies included, is why I find the ADHD meta-analysis above implausible, for example.
In the case of capillarization research, what kind of priors should I have? Here’s the kind of thought process I went through on this topic:
Well, I should have some modestly-sized priors in favor of the prevailing theory, but I shouldn’t be too sure, because I haven’t deeply studied that area of physiology. I should probably weaken those a bit since they’re based on a few experiments with small sample sizes. Nine college-aged men is pretty small, and that study was a cross-over trial that didn’t take order effects into account. I should also be careful of confirmation bias here, since my own personal bias is in favor of continuous moderate-intensity training (“high mileage”).
I should put more strength in the aggregated findings across many studies, which should be less affected by cherry-picking. Perhaps weaken them a bit, though, because the meta-analysis included many different kinds of exercise (cycling, running, kayaking) and made some semi-arbitrary categorization choices (like 80% being the cut-off for “moderate” vs. “high-intensity”).
Though publication bias is quite prevalent in exercise science and experimental research more generally, it would probably affect both views (few studies vs. all of them) to a similar-ish degree.
I’m not a meta-analysis expert but the forest plots seem reasonable. Four hundred-plus subjects is also pretty good, and the findings of the meta-analysis were similar regardless of whether the outcome was capillary density or capillary to fiber ratio. These seem like pretty reliable measurements, so I’m not super concerned with measurement error. Some of the other findings are counterintuitive though, e.g. the lack of any bigger effect for longer interventions, so perhaps weaken these beliefs because of that finding.
So, combining the evidence and my priors, I’m more willing to entertain the idea that interval work at 80-100% of VO2max could be a useful way to increase capillarization than I was before reading this study, especially for a new runner or someone coming off a break from running.
This sounds and feels awfully “unscientific,” but part of learning to read science is also learning to form priors (and priors about your ability to form priors…) and weigh or update them them with new evidence. I go through a similar process when I’m trying to make sense of any exercise science topic that seems murky, unclear, contentious, or rife with the potential for publication bias and irreproducible results.
Main takeaways on the science of capillary growth
What are my actual takeaways on how to train to optimize capillary growth? Here’s the best I can do:
- I’m a bit more willing to see early interval work as useful for newer runners as a way to “jump start” their fitness (and not just because of capillarization; there are other low-hanging fruit to claim with intervals that makes later training much more effective).
- I’m still unsure about the best way to boost capillarization in experienced runners, but I’m willing to hedge my bets a bit with a combination of long strong runs at 70-80% VO2max and medium to long intervals at ~90-95% VO2max.
- I’m more certain that very slow running does not have much benefit for capillarization.
- I definitely want to keep in mind that increased capillarization is just one of several improvements to the aerobic system that we’re trying to make in training.
Maybe that adds to the confusion, but here, some level of confusion is probably more appropriate than certainty!
Postscript: An approximate rule for %VO2max
By the way, if you want some hard numbers on these %VO2max values, the pace you can sustain for about 15 minutes tends to elicit 95-98% of VO2max in most people.
These numbers depend a bit on how you measure VO2max, and vary from person to person–among ~15:00 5k runners, %VO2max at 5k pace can range from 93-100% (according to this study which measured VO2 at 5k pace). Intensities as slow as ~90% of VO2max will eventually elicit true VO2max if you sustain them long enough--though that phenomenon (the slow component of VO2 kinetics) is a topic for another day!
Thanks for addressing a rather frustrating topic. I am a physical therapist and athletic trainer, and I get very frustrated when seeing meta-analyses that lump together all studies about a clinical question regardless of the study rigor. Generally, however, I have been impressed with the systematic review process in most of the physical therapy literature. Too often, however, the conclusion is that there aren't enough rigorous studies from which to draw definitive conclusions. You make a good point about prior experience, which is part of what PTs and others call Evidence Based Practice. If a study conclusion seems off when compared to my experience is when I take a more critical look at it....often finding that the subject group is too small or the way in which the intervention was applied is not like it would be applied in the clinic. Thanks again.