Thursday, January 23, 2020

Does the FTP relate to the MLSS - Yes, No, Maybe?

Several paper published recently have tried to elucidate the differences between the MLSS and FTP.  Are the tests equivalent in evaluating current performance status as well as fitness improvement after training.  Since the conclusions of these studies were quite different, i thought it might be interesting to see why.
This post will attempt to point out the definitions of each test, what they measure and the use of these values for training zone boundaries.  In addition, the use of the FTP (or equivilant) for fitness monitoring will be reviewed. 

First, what does each test measure?
The FTP (functional threshold power) is a near constant interval time trial with the average power defined as the FTP for that duration.  So the FTP20 would be your best constant effort of power for 20 minutes.  Therefore, this test is to exhaustion and very stressful.  In addition the performance would be heavily dependent on the subjects motivation, existing fatigue, recent training and other myriad factors.  If you had a poor night sleep or your legs were sore from some intense efforts done recently, your FTP may or may not be affected.
On the other hand, the MLSS (maximal lactate steady state) is generally defined as the best constant power over 30 minutes with little further lactate rise (above 1 mmol) from minute 10 to 30.  The centerpiece of this test is thus metabolic stability (lactate), where the FTP is an all out (potentially non stable) effort aiming for no reserve at the finish.  Technically, the MLSS can go on for a variable time above 30 minutes but the FTP can't.  

Here are a couple of photos illustrating the difference (yes, that's Lance in both):

This is the FTP - all out, maximal effort

The MLSS - generally done in a laboratory, a 30 min lactate steered interval with several repeats

Given the metabolic disparity mentioned (exhaustion vs steady state), various formulas have been proposed to reconcile the difference.  A popular method is using 90-95% of the FTP to approximate the MLSS.  Since many training regimes and zone 3 prediction methods rely on knowing the MLSS (or RCP - respiratory compensation point), it is valuable to getting this number right.  Practical note - Given the measurement issues in obtaining RCP/LT2, one may be better off doing zone 3 efforts well above MLSS so as not to bleed into zone 2 by mistake.  

Now for the 2 basic questions: 
  • Does the MLSS relate to the FTP20 by some constant (90% or 95%)?
  • Can one track the change in fitness over time using either MLSS or FTP?

FTP20 vs MLSS:
Several studies have looked at the relation between FTP and MLSS.  Several months ago a well done study was published showing that both a direct 90% x FTP and alternate formula were correlated with MLSS.  Here is the testing protocol:

Functional Threshold Power and Correction Factors. After standardized warm-up (5-minute pedaling, 80 and 90% of VT1 intensities), subjects performed a TT20 test using the
software provided by the ergometer (Rouvy; Cycleops), with 5%of slope simulation(1). Subjects were asked to produce the highest MPO possible for 20 minutes using their own pacing strategies, cadence and the gear ratio.
Note - virtually no warm up!  This is not a criticism, but it will become important later on.

They found reasonable correlation between FTP and MLSS but the correction factor needed was higher in the superior performing athletes:
To factor this observation into an equation, they proposed the following:


Bottom line:
  • 90% of the FTP20 power or using the alternate formula above (.7488 x FTP power + 43) is very close to MLSS

Another study showed excellent correlation between FTP20 and MLSS with no correction needed!  Here are some details:
Thus, 7 cyclists were classified as trained (T; V̇ O2max 55 − 64.9 mL/kg/min); and 8, as
well trained (WT; V̇ O2max 65 − 71 mL/kg/min).  

To investigate the concurrent validity between FTP20 and MLSS, the cyclists performed
in this order: an incremental test, the FTP20 protocol, and several tests to determine the MLSS.
The riders were asked to refrain from strenuous exercise in the 48 h preceding each test.
Participants were given at least 2 and a maximum of 4 days between visits and all tests were completed within 3 weeks. 

The incremental test was started at 100 W, with increments every 3 min of 30 W until maximum voluntary exhaustion

The warm up: 

Note - the warm up included a 5 minute time trial!

The results:
  • Excellent correlation of the MLSS and FTP with no correction needed.
  • The 5 minute time trial power was equivilant to the Maximal aerobic power.  One of the recommended validation methods to VO2 max testing is to perform a 5 minute constant TT at the power achieved during the last stage of testing.  It's nice to see this work out here as well.
  • Why was the FTP equal to the MLSS?  I feel fairly certain this was due to the warm up, that included the 5 minute MAP time trial.  An effort of that magnitude will have lasting effects in the short term on the next longer FTP interval.  This type of maneuver should be differentiated from a brief sprint to initialize some "fast start" strategy effect.  A 5 minute bout at MAP will introduce lactate elevation, local and central fatigue, obviously affecting the later FTP interval.
  • This does not invalidate the study, but does illustrate the need to read the details of the methods section carefully.
  • The first study had minimal warmup and needed substantial correction of the FTP20.

The next study was done by the group lead by Dr Murias, who has done excellent pioneering work on muscle O2 kinetics as well as other areas.  In this paper the additional question was asked whether a performance change after a cycling season is measurable in either the MLSS or FTP.  

A total of 18 participants (values in mean [SD]; 12 males: 37 [6] y, 180[6]cm,79[8]kgand6females:28[6]y,171[6]cm,68[9]kg) volunteered and provided written informed consent to participate in this study. 
The study was separated into 2 separate parts with identical testing procedures. The first part included all participants (n= 18), whereas the second part included 10 returning participants (9 males and 1 female; 39 [5] y, 178 [8] cm; PRE 76 [10] kg, POST 76 [11] kg). For these 10 participants, the first and second parts corresponded to before (PRE) and to the end (POST) of a 7-month cycling season.

FTP test and warm up
FTP20 Test. The Velotron 3D software (Racer Mate, Seattle, WA) was used for the FTP20 during which the participants controlled the gearing of the ergometer. Participants were familiarized with the gearing system prior to the test.The test was preceded by an 8-minute baseline at 80 W. For the FTP20 test, the participants were familiar with the goal of achieving the highest average PO possible across the 20 minutes, and no verbal encouragement was provided. During the test,participants were blinded to the PO,but they were allowed to see time and cadence to allow for individual pacing strategies.

Note - warmup was only 8 minutes at 80w and subjects did not see their power output.

Although the results of this study indicate that 88.5% (4.8%) of the FTP20 is more likely to reflect the PO at MLSS, the large amount of variability in the agreement for these measures (limits of agreement=9 to −44 W) prevents the use of this percent value with any confidence as a superior approximation of MLSS. 

  • Again we see that with a short, low power warmup, FTP20 is higher than MLSS as in the first study presented above.
  • With an additional correction to near 90% of FTP20, the match to MLSS is close.

Do either MLSS or FTP change with fitness improvements through the cycling season?
This is complicated and is a topic that is valuable to discuss.  
First, how do we measure improvement in aerobic fitness?  The authors did VO2 max testing in the Pre and Post groups but failed to see a change:
For the 10 participants who completed both phases of the study, no increase in VO2max was observed from PRE (4.32[0.53] L·min−1, 56.6 [4.3] mL·kg·min−1) to POST (4.37 [0.60] L·min−1, 57.7 [7.9] mL·kg·min−1)
However, the MLSS power improved from Pre to Post which is a better parameter of race success than pure VO2 max change:

  • The MLSS Pre power was 252w and increased to 264w after training.  This indicates that on average the subjects were able to increase their power by 12w during the 30 minute MLSS test without change in lactate parameters.
  • However the FTP20 power did not change (286 to 288w).  Does this mean that the FTP20 is incapable of registering the metabolic improvement seen in the MLSS?  A further discussion is warranted.

To better appreciate the FTP20 effort from a cardiovascular standpoint in both Pre and Post training tests, we need to look at additional physiologic parameters.  One metric that would have been helpful would be to actually measure the VO2 (O2 usage) during the FTP tests.  Since the FTP is an effort/motivation/fatigue affected index it is plausible that these athletes were in "better shape" but just were not as physically fresh as before the season.  Cyclists will know this feeling all too well after months of intense, frequent riding.  There does not appear to have been any tapering or post intense training preparation for the Post test.

The cyclist below was possibly in excellent cardiovascular condition, but the consequences of continued high intensity would lead to a poor maximal test.

Although we don't have a measure of VO2, we do have the heart rate to power relationship to go by.  Garmin (Firstbeat) uses this formula to calculate a "Performance Index" that I have discussed before.  Although prone to error with changes in ambient temperature, humidity, altitude, looking at the performance index in these subjects does show a fitness improvementAlthough the Post FTP power was the same as Pre, the heart rate was less (174 to 170 bpm).

Therefore although the conclusion was a negative in regards to using the FTP as an index of fitness/performance, it seems to be as valid as the MLSS, although in a different way.  

From the conclusion:
Furthermore,the results demonstrated that the POs from the FTP95% are not sensitive to small but meaningful and significant changes in fitness level, and thus its use as a tool for monitoring training may be limited
True, power did not change, but the overall picture indicates an improvement in fitness:
  • The ability to cycle at the same power but with a lower heart rate does signify a fitness change.  Whether this is related to better muscle O2 extraction, fiber type shifts, capillary enhancement, mitochondrial change is unclear.  Whatever the reason, in my view this does indicate that the FTP20 can be used to monitor fitness change.  However, it is also monitoring both physical, mental fatigue and potential over reaching as well.
  • The lack of Post test power improvement seems to be more of a fatigue related issue.
  • Perhaps if the subjects tapered in an optimal fashion before the Post test, the FTP20 would have been higher (they would have been able to reach the same HR as in the Pre).
  • The other possible wrinkle is the lack of power feedback during the test.  If the rider was able to see their current and running average power/heart rate, they may have pushed a bit harder.  From personal experience, knowing what power is possible for a given interval is very helpful in pacing strategy and could override elements of fatigue.

Some final thoughts on FTP and longer maximal tests:
Almost 20 years ago an interesting study was published looking at lactate levels, power and gas exchange thresholds in cyclists.  The subjects did a 30 minute time trial with power and lactate values measured throughout.  The results showed that the average power was very near the RCP (respiratory compensation point or VT2):
The average self-chosen work intensity during the ITT30 corresponded to 88% of the subjects’ VO2max (234M11W) and was not statistically different from the energy demand eliciting VT2
This certainly makes sense since both the MLSS and RCP are felt to be at similar intensity.
In essence, this test is basically an FTP30 equivilant.  This finding is also in line with the study above where the FTP20 was similar to the MLSS (without correction) as long as a 5 min MAP TT was in the warm up.  That makes sense if we consider that part of the FTP30 test is encompasses the 5 minute MAP warm up of the FTP20.

5 min MAP warm up then FTP20 = MLSS
Light warm up then 90% of the FTP20 power = MLSS
FTP30 power = MLSS

Here is the interesting graphic form the study:
This is a plot of power and lactate over the 30 minutes.  

  • A key point here is that although power was relatively stable there was some variation in pacing strategy (some higher at the end or beginning).  
  • The metric of most interest is the lactate tracking.  Several subjects had a gradual rise after the 10 minute section and there was significant variation of absolute lactate levels.  The OBLA (onset of blood lactate) of 4 mmol would have substantially different meaning in each of the above  participants.  There was almost a twofold difference in the lactate values during the 30 minute cruise!
  • As a parallel to the FTP20 test we can see differences in metabolic stability during the 30 min TT.  Some but not all subjects have a stable lactate, some continue to rise and a few even have a small decline.  The mean/average response is relatively flat, masking the individual variations of the group.
  • This plot reinforces the key difference between MLSS and FTP testing.

Summary points:
  • The MLSS is a metabolically defined test looking at stability of lactate from minute 10 to 30 of a time trial.
  • The FTP is a maximal effort (presumably to exhaustion) and therefore subject to factors such as motivation, fatigue and pacing strategy.
  • Multiple studies have been done looking at the equivalence of performing an FTP20 in lieu of the MLSS.  
  • The disagreement in equivalence seems related to the "warm up" protocol.  Lack of a significant warm up effort will yield FTP values well above MLSS and correction formulas are needed.  However, incorporating a 5 min TT at the MAP warm up seems to be sufficient to lower the FTP20 to the MLSS power.
  • The FTP20 may not show improvements in average power with positive fitness change.  This may be related to either fatigue, issues in pacing (if power readings are blinded) or even over reaching.  Comparing the heart rate:power relation of the FTP test, pre and post training should help to show a performance improvement (al la Garmin "performance index").
  • The MLSS is not a trivial enterprise to perform since multiple "test runs" are usually needed.  However, the FTP20 test is also physically taxing, given the exhaustive nature of a 20 minute maximal effort.  I personally have done neither.
  • Alternate surrogates for determining the RCP/LT2/MLSS are therefore very attractive and necessary.  Continued investigation into using muscle O2 kinetic breakpoints to estimate MLSS will be a focus of this blog.

Further reading


  1. you might enjoy this discussion w/ San Millan and Attia - they get into pretty good detail on lactate production and metabolism.

  2. Yes, I've seen some of the pseudo science around some of those claims. The idea of using metformin as a training aide is preposterous (I am a board certified endocrinologist and am very familiar with the drug). One of the reasons behind my doing a blog was to keep to real science, rather than generating click bait articles. Sure, zone 2 may have some benefits (if that is all you do), but if one spends too much time in there, the accumulated fatigue will interfere in HIT/zone 3 work and eventually result in over training. As far as mitochondria, it's complicated - see my post on that.

  3. I think its a bit disingenous to have a warmup protocol at such a high intensity close to someone's MAP lasting 5 minutes before the 20 min test is conducted and then claiming that an arbitrary number such as 95% will work to get 60 minute submaximal power. However you twist and turn the protocols around, I think the FTP20 method is akin to the 180-Age formula for HR max, doesnt work for majority of people plus you end up with a number that could be unnecessarily high as a sustainable 60 minute effort.

    Thanks, good read.

    More similar thoughts here, where I explore a more fundamental aspect of the very definition of FTP :