Shownotes

Maximal glycolytic flux and metabolic thresholds

00:56 -

• The first study is titled Maximal glycolytic flux modulates metabolic thresholds independent of maximal oxygen uptake, and it investigated whether VLamax can shift an athlete's threshold once VO2 max is controlled for.
• For anyone new to the concept, the idea originates with Alois Mader and his colleagues.
• You have your aerobic engine, which is VO2 max, and your glycolytic engine, which is VLamax, your maximal glycolytic rate.
• The two work almost as antagonists. A higher VO2 max results in a higher threshold, while a higher VLamax results in a lower threshold.
• This is where the theory begins rather than proven results, but the reasoning is that a more glycolytic athlete is drawing on faster fibres earlier and is therefore less able to hold a steady power output or pace for a long period.
• The extension of that theory is that lowering VLamax should raise your threshold without touching VO2 max at all.
• Taken further, a large enough reduction in VLamax could still produce a net gain in threshold even if VO2 max drops slightly along the way. 
• What makes this study interesting is that it is the first to empirically demonstrate one of the central claims of that theory.
• For a fixed VO2 max, a higher VLamax is associated with a lower threshold, and a lower VLamax with a higher threshold.
• Fifty trained cyclists took part, twenty of them women, which is an unusually good ratio. VLamax was measured in the laboratory with a 15 second all out seated sprint on an ergometer, using the athletes' own bikes.
• Lactate was measured at baseline before the sprint and then every minute for nine minutes, to make sure the highest accumulated value was captured.
• From that the lactate accumulation rate can be calculated.
• The study refers to VLapeak rather than VLamax, since what is measured is the peak value rather than a true maximum. VO2 max, ventilatory thresholds and a range of other variables were measured alongside it.
• Beyond the correlation itself, the researchers built regression models to see how much of threshold could be explained by each variable.
• The base model used only VO2 max and sex. Adding VLapeak improved that model considerably, with the R squared rising by something in the region of ten to twenty percent depending on which threshold measure was used, though that figure is from memory and may not be exact.
• The more revealing part is what did not help. Adding raw sprint power made almost no difference, and adding sprint power relative to lean body mass produced only a very small improvement. It was the lactate derived measure that carried the explanatory weight.
• That makes sense once you consider that a large share of a 15 second sprint is fuelled by the phosphocreatine system, which is not glycolytic at all.
• If you want to assess the glycolytic contribution rather than the phosphocreatine contribution, you have to look at the lactate dynamics rather than the power number.
• My own thinking had always been that a 30 second or one minute test might give a more accurate picture of glycolytic capacity.
• The discussion section addresses this, and the consensus appears to run the other way.
• Either 15 seconds, or something slightly shorter at around 10 to 12 seconds, is preferred. I have not looked deeply enough into the reasoning to explain clearly why that is.

What the study proves, and what it does not 

07:17 -

• The finding is correlational, which means it does not establish that a more glycolytic engine causes a lower threshold.
• That leaves an obvious opening for further testing. It is also notable that no link was found between VLamax and fat oxidation, which is not what you would expect given that substrate utilisation should differ between athlete types. 
• The debate was loud a few years ago and has quietened somewhat since, largely because the concept has spent a long time being described theoretically and demonstrated in single-cell work rather than in real athletes. 
• Single cell studies have their advantages, but at some point you need to see the effect appear in a whole organism, which is a complex system rather than a theoretical model of cellular physiology. This study is a good step in that direction.
• The next step would be a training intervention study. In principle you could manipulate VLamax in either direction, but for endurance athletes the typical scenario is an athlete with a medium to medium high VLamax trying to lower it in order to raise threshold.
• That could be split into two questions. First, which training interventions actually lower VLamax. Second, if you succeed in lowering it, does threshold rise proportionally when VO2 max is unchanged, with any VO2 max changes controlled for statistically.
• The correlation exists in a snapshot in time. Whether it holds longitudinally under training interventions aimed at VLamax is the open question.
• The time course of adaptation matters here too. Muscle fibre typology can produce a similar profiling of athletes, and that is neither quick nor easy to change.
• Whether VLamax can be shifted meaningfully within a season, or whether it takes multiple seasons, would tell us a great deal about how usable it is as a metric.
• Anecdotally, from what Sebastian Weber has said, six to eight weeks is enough to produce some change.
• Those would be the more functional changes, with structural changes taking longer, but a six- to eight-week window is short enough to make a pilot intervention study feasible. Nobody is going to fund a study requiring a year of training with no guarantee of a result.
• I have seen changes over similarly short windows, though not measured directly with lactate.
• Software that models and estimates VLamax shows similar movement, and you can argue about how accurate that is against a lactate-based measurement.
• Lactate has its own drawbacks, including a fairly high coefficient of variability, but taking nine samples removes a lot of that potential error compared with something like lactate-guided training.
• On that basis, I consider the protocol of an all-out sprint with a sample every minute to be a valid experiment.
• I have seen it done live. Dan Lorang came to Lisbon to run a seminar for triathlon coaches in Portugal, some time around 2018, and did exactly this with a volunteer from the audience.
• All-out sprint on an indoor trainer, lactate samples every minute, plotted on a whiteboard.
• From memory, the maximum value did not arrive until around seven minutes after the sprint, producing a nicely parabolic curve.
• That is the practical lesson if you want to run this test yourself. You could probably get away with sampling every 90 seconds or every two minutes, but you need to extend the sampling window far enough to capture the peak.
• The longer the gap between samples, the greater the chance you miss the true maximum, though the amount by which you miss it is unlikely to be large enough to matter much.

Whether VLamax changes how you actually train 

14:31 -

• You would not give an Ironman athlete the kinds of sessions that promote VLamax in the first place, so for many athletes this is a refinement of existing training principles rather than a change of direction.
• I went through a period of using it a great deal more than I do now, and that period genuinely sharpened my coaching, but I have not used it directly in a long time.
• The concepts stayed with me. The way I profile and view athletes now is based more on fibre typology, which is qualitative and subjective rather than quantitative, and there is nothing measured about it at all.
• For an Ironman athlete and specific training, you would not be doing VLamax work regardless.
• General training is a different matter. One pattern I have found is that athletes with a very low VLamax struggle to benefit from high-intensity interval training aimed at VO2 max.
• In those cases it has sometimes been useful to build VLamax first, which gives them enough of a buffer to complete the sessions that raise VO2 max.
• The volume side still has to be there, but that is an example where the concept earns its place.
• I would not recommend that athletes go out and start doing sprints with nine lactate samples. The value is in the principle.
• You cannot fully trust your own feel about whether you are glycolytic, because a strong sprint may be coming from the phosphocreatine system rather than from glycolytic capacity.
• What you can do is place yourself roughly, at least at the extremes. You probably know whether you are clearly not very glycolytic, or clearly not very low in that respect, and you may well be somewhere in between.
• That is enough to prompt some useful what-ifs. If you have not been responding to your training, one possible explanation is that the training is not well suited to your physiology because you are more glycolytic, or less, than you assumed.
• This is also a case where research is catching up to what coaches were already doing. The value is not only in verifying it.
• Seeing the effect controlled in a laboratory gives the concept explanatory power that helps other people understand it.

How much running is too much 

19:31 -

• The second study is titled How much running is too much, and it asked whether a particular metric predicts a higher likelihood of injury in runners.
• It was published in July 2025 as a cohort study using 18 months of Garmin training data from 5,200 runners, which makes it a rather different proposition to a standard study design because it uses real athlete data.
• Three exposures were tested. The first compared a single run, the longest one, against the athlete's longest run in the previous 30 days.
• The second compared one week of relative running load against the previous three weeks. The third used a week-to-week ratio, which is essentially a test of the familiar rule that you should not increase running load by more than ten per cent per week.
• It was the first metric that stood out. Compared with the longest run of the previous 30 days, a run that was 10 to 30 per cent longer came with a 64 per cent higher chance of injury.
• A run 30 to 100 per cent longer carried a 52 per cent increase.
• Runs of more than double the previous 30-day longest, which some athletes were genuinely doing, came with a 128 percent increase in self-reported injuries.
• Worth clarifying on the terminology, because it is easy to misread. The injuries were self-reported, and the risk was then calculated from them.
• The athletes were not estimating their own injury risk. That said, self-reported injury data will always carry some incompleteness compared with athletes being assessed by a sports medicine doctor or in a laboratory.
• The wider point is that acute-to-chronic ratios are the usual tool for this, and something appears to be hiding inside them.
• An athlete might do a very long run and then shorten the other runs that week, which leaves the weekly totals looking unremarkable on a TrainingPeaks dashboard while the actual risk has already been taken.
• Plenty of runners are not thinking about their long run in these terms at all. They are thinking about the demands of the race, and as the race approaches they make large jumps to close the gap.
• This is observational data, so it shows an association rather than proof of cause, and other factors may have been at work in the background.
• Even so, checking your planned long run against the longest run of the previous 30 days and keeping the increase within about ten per cent of that distance is a usable rule.

Long run progression versus weekly volume 

24:02 -

• The long run is where I would be more careful, and that is where the study points as well.
• Running injuries tend to be bone stress, tendon or muscular, and the risk comes from a large exposure to impact over a long duration.
• Looking back at athletes who have been injured, those occasions are what stand out, whereas smaller doses appear easier to absorb.
• That leads to what I consider the practical takeaway, particularly for a runner or triathlete with an injury history.
• If you are preparing for a marathon or an Ironman and you have a longest run in mind, say 34 kilometres three or four weeks out from a goal marathon, that is not something you start thinking about six or seven weeks beforehand.
• It has to be planned well in advance so the progression can be built properly. When you map out the training for that block, the run progression is where most of your attention should go.
• Reducing the step size between one long run and the next, whether in distance or duration, matters more than which sessions you are doing or what your total volume looks like.
• The picture changes for runners who never go especially long. If your longest run is 15 or 16 kilometres, I would be considerably less concerned even about a larger jump.
• Once you approach 20 kilometres, and certainly beyond half marathon distance, this becomes something to take seriously.
• One caveat on the earlier point about risk hiding inside weekly totals.
• The fact that a big long run can be masked by shorter runs elsewhere in the week is not an argument for keeping the weekly total high while you increase the long run. 
• Adequate recovery after the long run still matters, and total running load still needs managing. It is not irrelevant, it is secondary to managing the longest run.

Acute to chronic ratios and training stress balance 

27:56 -

• The comparison windows used in this study, one week against one week and one week against three weeks, are similar but not identical to what TrainingPeaks does with training stress balance, which compares seven days against 28 days with the seven days included inside the 28.
• It is also a differential rather than a ratio. The underlying concept is the same, comparing what you have done acutely against a longer look back, and the two probably track each other reasonably well.
• What matters is that the acute to chronic workload ratio was not the measure that best explained running injuries.
• I have been saying for years that I am not much of a believer in the utility of the performance management chart and training stress balance, and this is one of the areas where I would trust it least. 
• Trying to control running load through a metric like training stress balance lets you fool yourself, because the averaging smooths over exactly the spikes of injury-provoking stimulus you are trying to catch, whether those come from volume or intensity.
• The same caution applies to the older rule about not increasing total running mileage by more than ten per cent, which also failed to correlate strongly with injuries here.
• What is left is a metric that not many people use deliberately. Some athletes have effectively been applying it by feel, avoiding large jumps in their long runs without formalising it.
• This is a case for putting it in the training log and measuring it, and for not being pressured into group runs that are far longer than anything you have done in the last 30 days.

Tracking the longest run by week 

30:50 -

• The chart I look at most in my own TrainingPeaks dashboard is the one showing longest run by week. It is a simple bar chart, 20 kilometres one week, then 22, then 18, then 22, then 20, then 24, and you can set the window to the last 90 or 180 days.
• A quick glance and a hover over the bars tells you the actual distances of recent weeks and whether the progression is appropriate.
• Practically any training software will let you build the same chart, and I would recommend doing it. The point is simply that here is something shown to raise injury risk, so make sure you are looking at it.

Dose response, training load and biomarkers in elite triathletes

31:52 -

• The third study is Dose response to endurance training, training load and biomarkers in elite triathletes, published this year by Cejuela and colleagues.
• The athletes are a Spanish group, and many of them are at a very high level, with some having reached podiums in World Triathlon Championship Series races.
• They all train together under the same coaches, and some of those coaches collaborate with researchers, which produces work of unusual ecological validity because the data comes from the athletes' actual training.
• The limitation is obvious from the outset. There is no control group, and nothing is tightly controlled, because everything is optimised for the athletes' performance rather than for collecting clean data at the moments a researcher would want it.
• That is also the strength. It happens in the field, which avoids the familiar problem of a laboratory result that does not transfer into practice.
• Every few weeks, the athletes gave a blood sample to measure urea and creatine kinase.
• Urea, taken from blood, serves as a marker of protein turnover and is used as a proxy for metabolic load. Creatine kinase is more of a marker of muscular damage.
• Both were then correlated against the training load the athletes were carrying in different periods. Training load was not only quantified objectively.
• Subjective measures were also quantified, and the ratio of objective to subjective load was analysed as well, which opened up a number of interesting comparisons.
• Performance testing was done on the bike only, which is a shame, but it measured VO2 max along with VT1 and VT2, and those results were correlated against both the blood markers and the training load figures.
• Urea tracked objective training load well. Creatine kinase barely tracked it at all.
• Creatine kinase rose at the start of a training period following time off, which is what you would expect from an unadapted body, then dropped quickly and flattened out, after which it did not distinguish between heavy, medium and light training periods.
• Urea also tracked gains in VO2 max and in power at VT2. A larger rise in urea went with smaller gains in those measures.
• The authors framed this as a dose-response relationship. Where urea rose too far above an athlete's own baseline, the training response did not appear.
• The final analysis looked at the ratio of objective to subjective training load against the blood markers and performance changes.
• Athletes with a higher objective-to-subjective ratio, meaning more work for the same perceived effort or the same work at a lower perceived effort, showed a smaller rise in urea.
• Urea in turn was associated with performance improvements, with lower urea going alongside larger gains.
• What you cannot do is chain those two findings together and conclude that the ratio therefore predicts performance improvement. That link was not demonstrated in the study.

What to take from the biomarker findings 

37:15 -

• None of this is dramatic or changes how training should be viewed. What it does is confirm in the field what we already believe. These athletes were all training hard and training a great deal, and a dose-response relationship still applies to them.
• There is a level at which markers of stress rise sharply, and those markers were associated with the size of the performance gains.
• Building a proper dose response model from this would require a lot more controlled work, but there is very little research of this kind on elite triathletes at this level.
• The conclusion is not that you should be taking blood tests every few weeks, and the authors did not argue for that either.
• What they did argue is that a single blood test read against population reference ranges for urea or creatine kinase tells you almost nothing.
• The value comes from measuring longitudinally. This study ran for four full years, with athletes moving in and out, so most contributed one, two or three seasons and only a few were present for the whole period.
• My main takeaway as a coach reinforces something I already do, which is to assess the athlete's subjective workload against the objective work they are producing. In practice, that is what an athlete provides when they log an RPE for a session in TrainingPeaks.
• As a coach, you can then see that the RPE for a session sits higher than usual while the power over recent weeks has been lower.
• A single instance means nothing beyond a check in the next day, and usually the next day is better. Two days puts me on the lookout. Three days is a warning flag and the point at which I want to investigate immediately.
• That fits the wider pattern. When we look at metrics for assessing load and understanding what it is doing to the body, no single metric ever carries much explanatory power on its own.
• It is always the combination of subjective and objective scores in relation to each other that tells you something.
• The authors also reported substantial inter-individual variability, particularly in urea. If you were going to pursue a metric like that, you would need to build your own baseline and understand what it means for you.
• It is not a plug-and-play measure where you ask your doctor for a blood urea value and draw conclusions about the quality of your training. It takes time.
• The absence of a relationship between creatine kinase and training load is the other finding worth sitting with.
• From experience with trail runners in particular, the eccentric loading in running would suggest creatine kinase should be a meaningful marker of muscle damage and of an inability to absorb training. 
• Apparently not, at least here.
• That has implications beyond this study, because creatine kinase is frequently used as an outcome measure when comparing training approaches as harder or easier on the body.
• With elite triathletes at least, it may not be sensitive enough to serve that purpose.

Limitations and directions for further research 

42:12 -

• All the triathletes in this study were male. They also all followed the same training under the same coach in the same group, so how the findings extend to other training philosophies or other groups is unknown.
• Testing across swimming and running as well as cycling would add a great deal, and something like cortisol, or the cortisol to testosterone ratio, could be useful on the biomarker side.
•  One of their papers covers two years of training from Fernando Alarza, a two-time Olympian who was one of the strongest athletes on the short course scene around 2018 and 2019, which makes for an interesting case study. 
• This study also showed their periodisation model, which turned out to be very traditional periodisation. They have a number of these in the field studies with elite triathletes and are a good group to follow.

Supersets versus traditional sets in strength training 

44:12 -

• The fourth study is titled Less time, same gains, and it compared supersets against traditional set training for muscular adaptation.
• There were 43 participants, male and female, all of them strength trained rather than beginners.
• The superset group paired two exercises together.
• A lower body exercise such as squats combined with an upper body exercise such as bicep curls, with the second movement performed immediately after the first rather than after a long rest. 
• The traditional group completed one exercise before moving on to the next. The intervention ran for eight weeks with two strength sessions per week.
• The outcome was the same across the board. Muscular strength, hypertrophy, body composition and local muscular endurance all improved equally in both groups.
• The difference was time. With the six exercises and the superset structure used, the superset group achieved the same adaptations in 36 per cent less time.
• In absolute terms, that is 69 minutes per session for the traditional group against 44 minutes for the superset group, so 25 minutes saved per session, twice a week, which comes close to an hour saved per week.
• That is nothing, particularly for triathletes who routinely skip strength training or avoid it entirely because of the time it takes.

Applying supersets to endurance athletes 

46:30 -

• This is a strong practical study and the result makes sense, though the pairings used were clear agonist and antagonist groups.
• Most endurance athletes, whether triathletes, runners or cyclists, follow a lower body dominant strength programme with few upper body exercises, if any, so how to combine movements within that structure is the open question.
• The way I prescribe strength training helps here. Rather than fitting four or five heavy lifts into a session, I am happy with two, with the remainder made up of one or two plyometric exercises and then work on the athlete's specific weaknesses or improvement areas, which is often injury prevention.
• That makes it straightforward to pair one of the heavy lifts with a core exercise or a supplemental movement, whether that is crab walks, side planks, tibialis raises or whatever the athlete's history calls for.
• You still get the recovery time between heavy efforts even though the pairing is not a true agonist and antagonist combination.
• What would be interesting to know is how far the principle can be pushed. If you wanted to keep compressing the session, could you pair a squat with a deadlift, or a leg press with a leg curl. That combination would be challenging, but the question is worth asking.
• There is also a related protocol worth knowing about, from a study on rest pause training.
• Instead of three sets of six repetitions with a consistent break between them, athletes performed as many repetitions as they could, paused for 20 seconds, and repeated that process until they reached 18 repetitions in total.
• Same total volume as three by six, same adaptations.
• Here it is lifts at a given load rather than intervals, but it is the same concept and the same answer, with a very different distribution producing equivalent results. 
• It does not follow that you could do one lift, rest five minutes, do the second, and stretch the 18 lifts across another 90 minutes, but the finding is a good one.
• The practical point is to have these tools available. You do not have to superset every exercise in the session.
• Pairing a few of them and doing the rest traditionally still saves time in the gym, which is a reasonable way to keep strength training in the programme when time pressure is what usually pushes it out.

Making it work in a commercial gym 

51:30 -

• The one complication is setting up supersets in a commercial gym where you do not control the equipment. It can still be made to work.
• I have a home gym now, but when I trained in a commercial gym I was always opportunistic about it.
• I would be doing squats, finish a set, notice the elastic bands were free and go and use them, then find someone had taken them by the time I came back from the next squat set, because I was not going to hoard equipment while I was mid session. 
• Then something else would be free, maybe the seated calf raise machine, and I would cross reference what was available against my list.
• It is not the most efficient way to organise a session and it got scattergun at times, but it got the work done in a short amount of time.
• For the main lifts, squats, deadlifts and so on, this approach still lets you perform them at the same quality you would achieve with full seated rest, which is exactly what the study demonstrates.
• If you have been doing something like this and worrying that it compromises the quality of your work, it does not.

LINKS & RESOURCES

• Follow Daniel Rowland and Endurance: Ideas + Implementation on his website, Substack and Instagram.
• The full podcast episode archives (including category filters) can be found here.
• New Research on Torque Training, Ketones, Carb Loading and Pacing | Daniel Rowland | EP#698
• Is VO2max more important than lactate threshold? | Research roundup | EP#705
• Is Optimised Workout Design Pointless (and Time in Zone What Really Matters)? | Q&A | EP#708
• Maximal glycolytic flux modulates metabolic thresholds independent of maximal oxygen uptake - Meixner et al. 2026
• How much running is too much? Identifying high-risk running sessions in a 5200-person cohort study - Schuster Brandt Frandsen et al. 2025
• Dose-response to endurance training: a longitudinal study of training load and biomarkers in elite triathletes - Cejuela et al. 2026
• Less time, same gains: Comparison of superset vs traditional set training on muscular adaptations - Burke et al. 2024
• Strength and Muscular Adaptations After 6 Weeks of Rest-Pause vs. Traditional Multiple-Sets Resistance Training in Trained Subjects - Prestes et al. 2019