Ed Coyle's paper about LA delta efficiency is a fraud.

[acoggan]

Would you have felt comfortable submitting that paper for peer review?

Me? No, at least not the way it was written/presented. But, my standards for myself are higher than average.

 

[ScienceIsCool]

I picked Sassi's paper at random. Very interesting, but I'm not sure it supports your central hypothesis. First paragraph of the discussion:

"GE and EC, despite of their in-subject seasonal waves, did not show changes related to those of other maximal and sub-maximal parameters of aerobic fitness" That seems odd to me. You think they would track nicely if the GE changes were due to training. Doping, perhaps?

As for the GE and EC changes themselves: the authors cannot exclude the possibility that the improvement over a season wasn't due to their methodology. They changed from measuring at 78% PPO (Peak Power Output) to 73% PPO in later measurements. <Paragraph 3 of the discussion> They even give a proposed mechanism for why that change would affect their measurement of GE and EC.

I could go on, but I have to admit that I'm kind of disappointed. In a more formal setting, I would even be tempted to use this paper to refute the idea that GE can be improved through training when using a trained subject and the sport they are accustomed to.

Did I completely miss something?

John Swanson

 

[acoggan]

I picked Sassi's paper at random. Very interesting, but I'm not sure it supports your central hypothesis. First paragraph of the discussion:

"GE and EC, despite of their in-subject seasonal waves, did not show changes related to those of other maximal and sub-maximal parameters of aerobic fitness" That seems odd to me. You think they would track nicely if the GE changes were due to training.

Not if different physiological mechanisms were involved.

As for the GE and EC changes themselves: the authors cannot exclude the possibility that the improvement over a season wasn't due to their methodology. They changed from measuring at 78% PPO (Peak Power Output) to 73% PPO in later measurements. <Paragraph 3 of the discussion> They even give a proposed mechanism for why that change would affect their measurement of GE and EC.

I could go on, but I have to admit that I'm kind of disappointed. In a more formal setting, I would even be tempted to use this paper to refute the idea that GE can be improved through training when using a trained subject and the sport they are accustomed to.

Did I completely miss something?

John Swanson

Yes: the fact that gross efficiency tends to plateau at high power outputs, such that the difference between 78% and 73% of peak power output would not matter.

Anyway, keep scrounging...I'm sure you'll come up with a way to dismiss all of the other papers as well.

 

[acoggan]

If you wish to whittle the list to just those studies using trained cyclists, it would look like this:

Paton and Hopkins J Strength Cond Res 2005; 19:826-830
Sassi et al. Appl Physiol Nutr Metab 2008; 33:735-742
Hopker et al. Med Sci Sports Exerc 2009; 41:912-919
Hopker et al. Appl Physiol Nutr Metab 2010; 35:17-22
Sunde et al. J Strength Cond Res 2010; 24:2157-65

Whoops, how could I forget:

Santalla, Naranjo, and Terrados Med Sci Sports Exerc 2009; 41:1096-1101

And to make the list complete, at least w/ respect to studies mentioned in this thread:

Hopker et al Int J Sports Med 2012; 33:789-794

So, by my count that's now 7 studies of trained cyclists demonstrating an improvement in efficiency over time (8 if you count Coyle's original paper).

Here's a review for anyone interested in digging into the literature for themselves:

http://www.ncbi.nlm.nih.gov/pubmed/19941249

 

[Le breton]

I cited that study simply because it is a longitudinal investigation of changes in gross efficiency resulting from endurance exercise training, i.e., in that regard it is similar to the others. If you wish to whittle the list to just those studies using trained cyclists, it would look like this:

Paton and Hopkins J Strength Cond Res 2005; 19:826-830
Sassi et al. Appl Physiol Nutr Metab 2008; 33:735-742
Hopker et al. Med Sci Sports Exerc 2009; 41:912-919
Hopker et al. Appl Physiol Nutr Metab 2010; 35:17-22
Sunde et al. J Strength Cond Res 2010; 24:2157-65

Plus the recent Hopker et al. paper, as well as a few others published between 2010 and now.

(Note the correction in the volume number of the first reference.)

In contrast, here's a list of all the longitudinal studies (of untrained or trained individuals) that I have been able to locate reporting that efficiency does not change:

Roels et al. Med Sci Sports Exerc 2005; 37:138-146

As for the role of biomechanics vs. biochemistry, it doesn't really matter to the question at hand, i.e., either efficiency is changeable, or it is the immutable "Holy Grail" that Ashenden claimed it was.

Finally, gross efficiency increases with increases in absolute exercise intensity, as the impact of the intercept of the VO2-power relationship becomes less and less - this is what the authors meant by "PO-dependency".

This reply is not particularly addressed to you Andy, but to everybody involved in the discussion.

What was of interest to me in Ed Coyle's paper was not the possible increase in efficiency of the subject, and in fact I am willing to believe that because LA's fiber type distribution evolved post cancer towards a higher fraction of slow twitch (Andy also seems to suggest that the use of pharmaceuticals - such as AICAR - can improve the aerobic capabilities of fast twitch fibers) his gross efficiency went up a bit, it had been very low before.

Maybe Ed. Coyle as suggested by some tried to give a helping hand to LA knowing that Ashenden (for SCA) would grasp at the efficiency straw and didn't pay too much attention to the other data he was revealing in that paper.

No, what I really found interesting was the fact that LA was just like the rest of us regular cyclists : with one liter of oxygen per minute he can only produce about 80 watts, he is no better than Boardman or me, may be worse than Andy, who knows.

To see that look at the data gathered when LA produces about 400 watts.
Now Andy and I had this exchange in 2010 on this forum

Originally Posted by Le breton
The following calculation, was posted on topica in 2004 and you did not challenge it at the time, I have modified it slightly so readers of this forum can follow more easily:

(the contributor) just got privileged information from a spectator on the
AdH TT who reads this forum (topica/wattage) but does not write.
Point 1 ( split time in l’Equipe newspaper) was really at km 1.7 ( as I thought), hence at
~725 m.a.s.l.
Point 2 was 50m after hairpin No 7, hence at ~ 1395m asl, km
9.15.
.....
Distance = 7 450m
Elevation 670m
time 21:03 = 1263 sec
slope =0.0899
v = 7450/1263 = 5.899m/s
Air density 28°C 1100m altitude -> 1.03
CdA 0.4 m^2
Crr = 0.0036
-----
70 kg cyclist + 8 kg equipment ( bike, shoes,... 1/2
bottle)

RESULT = 464 watts
+ 2.5% transmission losses -> 476 watts over 21 min

476/70 kg = 6.79 Watts/kg

There was no wind on that section of AdH that day.
(Higher up the last racers had to contend with a very noticeable wind)
Using only that lower part between 725 and 1395 m one does not need to take the physiological effects of altitude into consideration : effect on performance being almost imperceptible below 1500m (although Andy can find – no doubt – counterexamples).

Speed variations negligible.
Crr has been measured by extrapolation to V=0m/s on a similar French mountain road. Not very different from the 0.004 default figure in analyticcycling.com calculations.

With those parameters :

Gravity accounts for 405 watts, uncertainty under 1% (altitude gain) = 5.79 W/kg
Air resistance accounts for 42 watts = 0.6 W/kg
Rolling resistance for 16.5 watts = 0.23 W/kg

I consider changes in kinetic energy have a totally negligible effect until proven wrong.

Total 464 watts/kg = 6.63 W/kg

The uncertainty on gravity is about 0.5 to 1%, ie at most 0.06 W/kg
The uncertainty on Air resistance is at most about 10%, ie at most about 0.06 W/kg.
The uncertainty on rolling resistance at most 10%, ie at most 0.02 W/kg.

So the total uncertainty is about 0.09 watts/kg – 1.6%

Transmission losses, according to Ed Kyle are about 2.5%, although under ideal conditions they could be as low as 1.5%.

Therefore, on his time trial up AdH in 2004, LA developed 464 + 2.5 % = 476 watts (+/- 6 watts), or more precisely 6.8 W/kg (+/- 0.1 W/kg) during 1263 seconds in the section starting at 1700m and ending at 9150 m.

People seem to like to muddle the issue by introducing microscopic effects which play at a very low level of usually less than 1%, sinking to below 0.1% on a steady climb.

CONCLUSION

EVEN WHEN INCLUDING THE UNCERTAINTY ON TRANSMISSION LOSSES WE END UP WITH A TOTAL UNCERTAINTY ON THE CALCULATION OF THE ORDER OF 2%!
How are you going to do to bring that up to 11% Andy? The total uncertainty arrived at above is based upon their starting assumptions - my generic calculation (i.e., not aimed at the performance of Armstrong or anyone else in particular) was based upon different values for, e.g., CdA and more importantly by assuming a 2% margin of error in each of the variables and performing a standard propogation-of-error analysis.

In any case, thanks for dredging that up - I was going to point out to people that 1) contrary to what many here might believe, air resistance (and not, e.g., rolling resistance or drivetrain efficiency) is still the 2nd most important factor in determining the power required when climbing at such speeds (such that you need to account for, e.g., wind and drafting effects), and 2) the power required to overcome gravity and gravity alone, which is fairly easy to estimate accurately, is well within the range of plausibility from a physiological perspective. Hence, what is ultimately responsible for "putting someone over the line" are the estimates for the power required to overcome air resistance and, secondarily, rolling resistance...which is where almost all of the uncertainty lies.

---------------
At the time I thought LA could have been only 70 kg, now I have learned that under oath he stated that he never raced as 72 kg or less, so his weight could have been 73 kg.

However that hardly affects his watts/kg value over those 21 minutes, it only means that his power was 6.79 times 73 = 496 watts, not 476 watts.

Now, we can assume without much risk that over those first 21 minutes of the climb where there was no wind effect (contrary to higher up, above Le Ribot where it slowed down the racers) he was below VO2 max, roughly in the same regime as when doing a test at 400 watts in Ed Coyle's lab.

Therefore during those 21 minutes he was using 496/80 = 6,2 liters O2 /min.
that is 6,2 / 73 = 84.9 ml/min.kg.

Since he knew he had another 20 minutes of effort ahead of him past that time check point in Le Ribot, it is quite probable that he was at about 90% of VO2 max, which would put his VO2 max after many stages of the Tour de France at 6.9 liters /minute or 94.3 ml/min.kg.

Now you are free to play with those numbers if you so wish*, but remember this was after many days of racing and at an average altitude of 1060 meters that he was able to muster 6.9 liters/min while at a younger age in 1996, as documented in the July issue of Scientific American he could at best reach 6.2 liters/min ( And I would certainly not give much credit to anybody claiming LA was clean in 1996) possibly at an altitude of 1800 meters.

So, when I first read the full Ed Coyle article ( Thanks to Andy, no, actually it must have been Robert Chung, not Andy), in 2005 or 2006) my first thought had been that Ed Coyle wanted to show everybody that LA was doping. Instead of making the type of calculation I just did, Ashenden concentrated on the efficiency bit, dumb.

Now, some of you might remember that around that time Jonathan Vaughters publicly estimated that climbing the Alpe d'Huez that fast required a VO2 max of 94 ml/min.kg (ie 6.86 liters/min for LA). not far from the 94.3 ml/min.kg calculated above.

If somebody should know it would be Jonathan Vaughters as his memorable Ventoux climb in the 99 Dauphiné TT in 56:51 required exactly the same kind of VO2 max ( he had nicely defeated LA who presumably was not yet on the fully fledged TdF programme.

* Andy has problems with compounding uncertainties coming from various sources so he seems to use a programme instead of common sense and gets ridiculously high values of those uncertainties.

I will post without rereading as I have not been saving what I wrote.
If i need to I will come back for corrections.

I am now changing the wattage values by 2.5% as I forgot to include the transmission losses in what I just wrote.

Edit 3 Robert Chung, not Andy

 

[the big ring]

much debate regarding the possible ability to increase GE.

is that diiferent to: much debate regarding the ability to increase GE?

curious that the review relies on metholodical inadequacies for its opening argument given the methodological inadequacies used in Coyle's.

if we added up all the citations to the studies in this review, would you accept that as evidence of the soundness of said studies, given you appear to imply being cited 56 times somehow indicates soundness? if you meant something else, would you apply that thing to those studies based on the same measure?

i thought the debate was about an increase in GE for well-trained athletes; armstrong being the classic example of well-trained.

this review seems to imply the debate is any level of cyclist, which seems strange to me as newbies tend to move inefficiently - rocking or bobbing shoulders, as an example. how would you control for something like increased core strength from novice to pro cyclist when determining efficiency due to training?

 

[ScienceIsCool]

Not if different physiological mechanisms were involved.



Yes: the fact that gross efficiency tends to plateau at high power outputs, such that the difference between 78% and 73% of peak power output would not matter.

Anyway, keep scrounging...I'm sure you'll come up with a way to dismiss all of the other papers as well.

Can I ask what those physiological mechanisms are?

To your second point, I thought the authors very clearly disagreed; that the slow VO2 something-or-other would make the 78% vs73% difference matter. They very clearly made the point that they cannot rule out that this was the cause for the measured change in GE.

Believe it or not, I'm not scrounging. I'm looking at the citations you're offering with an open mind. I'm just not making the same conclusions you are. I was more than a bit surprised, actually. You seem very confident that the literature supports your claim about training and GE. As it pertains to trained athletes in their respective sport, I just don't see it.

Could this be my lack of training in your field? Oh, heck yeah. Which is why I'm glad that you're answering questions.

John Swanson

 

[ScienceIsCool]

http://www.ncbi.nlm.nih.gov/pubmed/19941249[/url]

Oh, Christ. "Results of both experiments in this study demonstrate training related increases in GE, but not V˙O2max."

I wasn't able to access the entire study, but came across the full abstract. Really? The cyclists trained, but didn't increase VO2max!? Were these trained or untrained cyclists? Athletes, or "off the couch" performers.

I swear that I am taking only random samples of your citations.

It's incredible. Every citation I research is quite damning.

Dr. Coggan, what am I missing? Please, give a semi-layman's (scientist in a different field's) explanation of why GE in a trained athlete changes with training. Please give *relevant* citations.

I swear that I remain open-minded and cognizant that this is not my field. Not impressed so far. I only have so much free time, so will only take random samples of your citations until I make my way through them. If you have any citations I should focus on, I would welcome the suggestion.

As an aside, abstracts are useless. Full papers or GTFO.

John Swanson

 

[Random Direction]

I swear that I am taking only random samples of your citations.

I work in the policy field. When someone comes to me waving a series of studies and saying 'trust me, I am xyz', my reaction - lobbying to influence outcome. My action? Find out for myself. Dig in to the questions asked and the assumptions, look who funded the work, test with another model, see if I get the same results, or something entirely different.

Unfortunately acoggan's approach here seems to be one of showing credentials and waving a bunch of studies. When ScienceIsCool and others dig in and try o find out for themselves, they get criticized for not being expert in the field. Given this reaction, my intuition is that it is not science that is being debated, but instead 'science' is being improperly used to back up a predetermined world-view.

 

[sittingbison]

forgive me if I'm completely wrong, because I'm not a scientist and don't know what I'm talking about. But isn't all this a big stinking pile of BS?

Because again forgive me if I'm wrong, but wasn't Armstrong doping since a teenager with Carmichael, and admitted to taking Test, EPO, steroids, insulin and some other stuff in 1996? So how the frak can a physiological study (purportedly about normal elite athletes (if there is such a thing)) of Armstrong when he was juiced to the gills have ANY relevance to anything ?

And I didn't even mention Ashenden et al having misgivings about the methodology (I gather they don't see eye to eye). Such as LAs weight and height being misreported etc.

 

[D-Queued]

forgive me if I'm completely wrong, because I'm not a scientist and don't know what I'm talking about. But isn't all this a big stinking pile of BS?

Because again forgive me if I'm wrong, but wasn't Armstrong doping since a teenager with Carmichael, and admitted to taking Test, EPO, steroids, insulin and some other stuff in 1996? So how the frak can a physiological study (purportedly about normal elite athletes (if there is such a thing)) of Armstrong when he was juiced to the gills have ANY relevance to anything ?

And I didn't even mention Ashenden et al having misgivings about the methodology (I gather they don't see eye to eye). Such as LAs weight and height being misreported etc.

1. Ever since he lost the swim meet

2. Nothing like a controlled experiment

Dave.

 

[DirtyWorks]

Unfortunately acoggan's approach here seems to be one of showing credentials and waving a bunch of studies. When ScienceIsCool and others dig in and try o find out for themselves, they get criticized for not being expert in the field. Given this reaction, my intuition is that it is not science that is being debated, but instead 'science' is being improperly used to back up a predetermined world-view.

Winner! Winner! Turkey dinner! Then Acoggan resorts to a personal attack, labeling you a conspiracy theorist. http://forum.cyclingnews.com/showpost.php?p=1034627&postcount=382

The bad news here is the fundamental scientific process is sound, yet here's another instance where it seems Acoggan has used the process to meet some other non-scientific goal.

 

[acoggan]

Can I ask what those physiological mechanisms are?

I'm not sure I'm clear on your exact question, but you acted surprised that VO2max, LT, and efficiency did not change roughly in parallel. My point is that you wouldn't expect them to unless there were some common mechanism responsible.

For example, let's consider training-induced changes in VO2max and efficiency. The former is dependent upon changes in maximal cardiac output (stroke volume, actually, since maximal heart rate tends to go down w/ training) and a-vO2 difference (influenced by changes in capillarization, etc.), and plateaus relatively rapidly after someone takes up endurance sports (e.g., I started racing at age 15, had a VO2max of 80 mL/min/kg when I was first tested at age 18, and could still hit that figure when I was age 32).

The mechanism(s) responsilble for changes in efficiency, OTOH, appear (emphasis on appear, because this is still a rather open question) to be linked to changes in skeletal muscle biochemistry/characteristics, e.g., changes in myosin expression/contractile properties, uncoupling protein levels, calcium handling proteins, NOS enzyme activity, etc.). These particular adaptive mechanisms could very well follow a different time course, and/or have a higher potential for improvement, than changes in VO2max. If so, you would not expect VO2max and efficiency to always track one another.

 

[acoggan]

Winner! Winner! Turkey dinner! Then Acoggan resorts to a personal attack, labeling you a conspiracy theorist. http://forum.cyclingnews.com/showpost.php?p=1034627&postcount=382

How is calling someone else's theory that Coyle's paper was part of a conspiracy to provide cover for Armstrong a "conspiracy theory" a personal attack?

The bad news here is the fundamental scientific process is sound, yet here's another instance where it seems Acoggan has used the process to meet some other non-scientific goal.

Now who's resorting to personal attacks?

 

[acoggan]

you appear to imply being cited 56 times somehow indicates soundness?

On the contrary: my point was that Coyle's paper has been rather middling in its direct impact (at least compared to the rest of his work), and seems to continue to be cited at about the same frequency after as before all the Letters-to-the-Editor.

 

[the big ring]

On the contrary: my point was that Coyle's paper has been rather middling in its direct impact (at least compared to the rest of his work), and seems to continue to be cited at about the same frequency after as before all the Letters-to-the-Editor.

So if the studies in the review are cited more often, that would make them more sound, yes?

 

[ScienceIsCool]

I'm not sure I'm clear on your exact question, but you acted surprised that VO2max, LT, and efficiency did not change roughly in parallel. My point is that you wouldn't expect them to unless there were some common mechanism responsible.

For example, let's consider training-induced changes in VO2max and efficiency. The former is dependent upon changes in maximal cardiac output (stroke volume, actually, since maximal heart rate tends to go down w/ training) and a-vO2 difference (influenced by changes in capillarization, etc.), and plateaus relatively rapidly after someone takes up endurance sports (e.g., I started racing at age 15, had a VO2max of 80 mL/min/kg when I was first tested at age 18, and could still hit that figure when I was age 32).

The mechanism(s) responsilble for changes in efficiency, OTOH, appear (emphasis on appear, because this is still a rather open question) to be linked to changes in skeletal muscle biochemistry/characteristics, e.g., changes in myosin expression/contractile properties, uncoupling protein levels, calcium handling proteins, NOS enzyme activity, etc.). These particular adaptive mechanisms could very well follow a different time course, and/or have a higher potential for improvement, than changes in VO2max. If so, you would not expect VO2max and efficiency to always track one another.

That makes a lot of sense, thank you. I was looking at it from the opposite direction. If training/racing is responsible for the other listed changes/improvements, then why wouldn't the GE track with it? What is it about that particular training that would not cause GE to change? Unless of course, the GE does not change with training. Or it takes a very specific type of training to improve GE. Or as you state, it is a relatively long term process.

So if you're correct and it takes some long term physiological changes to the fibers themselves then it would take a very long term study to show this. Am I correct? Am I also correct that this should be of intense interest, because of the implications? Understanding and manipulating GE through a specific training regime could give an athlete a very large advantage in an endurance sport like cycling.

John Swanson

 

[acoggan]

To your second point, I thought the authors very clearly disagreed; that the slow VO2 something-or-other would make the 78% vs73% difference matter. They very clearly made the point that they cannot rule out that this was the cause for the measured change in GE.

Clearly the authors were just being cautious in acknowledging this as a theoretical limitation (for which I commend them). In reality, it is quite unlikely to explain their observations, because gross efficiency was measured at a power output (i.e., 337 +/- 44 W) well below the power at "threshold" (in this case, quantified as the respiratory compensation point) (i.e., 465 +/- 43 W), meaning that the slow component to VO2 would be essentially non-existent.

 

[the big ring]

So if you're correct and it takes some long term physiological changes to the fibers themselves then it would take a very long term study to show this.

The new study showed a 1 percentage point increase (19.6 - 20.6) in 6 weeks of training.

 

[the big ring]

Clearly the authors were just being cautious in acknowledging this as a theoretical limitation (for which I commend them). In reality, it is quite unlikely to explain their observations, because gross efficiency was measured at a power output (i.e., 337 +/- 44 W) well below the power at "threshold" (in this case, quantified as the respiratory compensation point) (i.e., 465 +/- 43 W), meaning that the slow component to VO2 would be essentially non-existent.

The power at threshold was between 422 and 508W?

 

[acoggan]

Oh, Christ. "Results of both experiments in this study demonstrate training related increases in GE, but not V˙O2max."

I wasn't able to access the entire study, but came across the full abstract. Really? The cyclists trained, but didn't increase VO2max!? Were these trained or untrained cyclists? Athletes, or "off the couch" performers.

"14 endurance-trained competitive male cyclists (mean±SD: 34±8 years, 74.3±7.4 kg, Wmax 406±43 W, V˙O2max 59.5±3.8 mL · kg − 1 · min − 1) were selected from local clubs and professional teams to take part in this investigation. All cyclists had a minimum of 3 years competitive cycling experience (national to international level of competition)"

IOW, a further training-induced improvement in VO2max would not necessarily be expected.

Dr. Coggan, what am I missing? Please, give a semi-layman's (scientist in a different field's) explanation of why GE in a trained athlete changes with training. Please give *relevant* citations.

I have, and I have.

I swear that I remain open-minded and cognizant that this is not my field. Not impressed so far. I only have so much free time, so will only take random samples of your citations until I make my way through them. If you have any citations I should focus on, I would welcome the suggestion.

As an aside, abstracts are useless. Full papers or GTFO.

GTFO yourself. I've given you a list of references; it isn't my fault if you can't access them.

 

[acoggan]

The power at threshold was between 422 and 508W?

"Thirteen subjects (mean ± SD: age 26 ± 4 years,
body mass 67.4 ± 7.2 kg, height 178 ± 5 cm, _V O2 max
74.7 ± 4.9 mL
kg–1
min–1) were involved in the study: 8
professional road cyclists (3 of whom finished in the top
5 places of the Giro d’Italia or Tour de France overall
final ranking) and 5 mountain bikers competing at the international
level (with 4 out of 5 being current members
of the Italian National Team)."

 

[acoggan]

So if the studies in the review are cited more often, that would make them more sound, yes?

No, not necessarily. Papers are cited (or not cited) for many reasons, not just whether they are methodologically sound.

 

[the big ring]

"Thirteen subjects (mean ± SD: age 26 ± 4 years,
body mass 67.4 ± 7.2 kg, height 178 ± 5 cm, _V O2 max
74.7 ± 4.9 mL
kg–1
min–1) were involved in the study: 8
professional road cyclists (3 of whom finished in the top
5 places of the Giro d’Italia or Tour de France overall
final ranking) and 5 mountain bikers competing at the international
level (with 4 out of 5 being current members
of the Italian National Team)."

mmmm 6.8W/kg threshold for 80 ml/m/kg. looks efficient.

 

[acoggan]

So if you're correct and it takes some long term physiological changes to the fibers themselves then it would take a very long term study to show this. Am I correct?

Again, not necessarily. For example, efficiency might change by a rather small-but-still-measurable amount (assuming careful attention to methodological issues) rather rapidly, but then continue to improve at a very slow rate for a long time, such that the long-term gain would be much greater.

Am I also correct that this should be of intense interest, because of the implications? Understanding and manipulating GE through a specific training regime could give an athlete a very large advantage in an endurance sport like cycling.

You're assuming that it takes some sort of specialized training regimen, rather than simply happening as a matter-of-course as an athlete "matures".

In any case, though, it is true that changes/differences in efficiency have significant implications w/ respect to performance ability - see Horowitz, Sidossis, and Coyle. Int J Sports Med 1994; 15:152-157.