Friday, March 1, 2019

Determining MLSS with long intervals

A concern about using ramp associated breakpoints is the issue of what protocol was used.  For instance, one may do a power escalation of 25w/min or 25w/3min and get totally different results.  One could argue that the ideal way to look for patterns and breakpoints would be longer, steady state sessions.  The problem with that of course is that it may take a very long overall time, as well as inducing fatigue toward the end.  However, after looking over my data, I feel that some intriguing insights can be obtained by looking at NIRS and respiratory data of intervals near the MLSS.  The laboratory determination choices of MLSS are controversial and subject to a given amount of error.  According to a review on the subject, at least 2 trials are needed and even so the the coefficient of error can be as high as 3-4%.  
In comparison to HRmax and VO2max, the range of CVs
reported for threshold measurements is large, spanning
1.5–10.4 % for AerT and 1.2–11.9 % for AnT (Table 1).
This may be partly attributed to differences in protocol and
study design, including, in some cases, the reliability with
which investigators are able to identify threshold measurements
by visual inspection [118, 119]. It is also
apparent that the reliability of threshold measurements
varies according to whether the threshold is reported as a
workload, an HR, a VO2, or a blood lactate concentration.
Of six examples [109, 111, 112, 115, 120] of a threshold
reported according to the corresponding HR, speed or
power output, and VO2, four found threshold HR to be the
most reliable with CVs of 1.5–3.8 %
Here is the summary table of studies:

The lactate strips themselves can introduce an element of random error since no two readings will be exactly alike.  In addition, depending on what parameter (Heart rate, power, lactate, running speed, percent of VO2) is indexed, further error can occur.  Since many athletes use this as a fitness status benchmark or a training workload target, significant error can lead to erroneous strategy and conclusions.
While there is a strong theoretical basis for using
threshold-based exercise prescription, the challenges of
determining thresholds in practice may partially explain
why many researchers continue to favor the use of
%VO2max, %HRmax, %VO2R, or %HRR. For instance,
when derived from a blood lactate curve, neither the AerT
nor the AnT can be assumed to pinpoint the true thresholds
of metabolic response in all individuals without
verification. Verification of threshold measures on an
individual basis would require two or three additional visits
to the laboratory and is highly uncommon. Nevertheless,
failure to verify threshold measurements may create the
same individual variation in blood lactate accumulation for
which %VO2max and %HRmax have been criticized. It
follows that VO2max and HRmax, which can be measured
and verified within a single laboratory visit, have a definite
practical, if not theoretical, advantage over threshold
measurements for prescribing exercise intensity.

What is generally agreed upon though, is that once lactate accumulation becomes non steady state, a compensatory ventilation increase will occur that will continue to rise.  
As is well known, different exercise
intensity domains are associated not only with a shift in
blood lactate responses but also with changes in ventilation
[44], oxygen uptake kinetics [45], and catecholamine
responses [46, 47]. For example, constant-intensity exercise
within the ‘‘severe’’ exercise intensity domain ([AnT)
is characterized by a continuous increase in ventilation and
VO2, progressive acidosis, and metabolite accumulation,
whereas constant-intensity exercise equal to or below the
AnT is associated with a physiological steady state
Further investigation into the NIRS behavior of both the RF and VL at the MLSS and above comes from Dr Murias' group again.  This was covered in a previous post, but let's look at this from a different perspective this time.

They were looking at long sessions (30 min) of cycling at the MLSS as well as MLSS+10w in regards to NIRS changes in HHb dynamics of the RF and VL along with cardiac-respiratory parameters. Given the differences in recruitment, fiber characteristics between sites, they felt that a change would be apparent in the MLSS+10 tracings.

The subjects cycled for 30 minutes at MLSS and MLSS+10w.
Some observations-
Ventilation and Heart rate:
It was plainly seen that the ventilation rate remained relatively stable in the MLSS group, but progressively rose in the MLSS+10 trials.  Despite the buildup of lactate and higher locomotor+respiratory load in MLSS+10, the heart rate rose, but to a fixed degree above MLSS.  The VO2 response did mirror the heart rate as well (which makes sense).  A take home lesson is that heart rate by itself seems a poor metric to demarcate MLSS, as opposed to ventilation rate (if available) which is a good marker.

HHb of VL vs RF:

In the left column, both RF and VL HHb values remain stable throughout the 30 minute trial (the end >30 minutes is a time to exhaustion sprint).  The right sided figure is the HHb pattern during MLSS+10.  It is a bit difficult to see which muscle group rises more from the figure but here is a table breakdown:
It seems that the RF HHb value (footnote c) does have a statistical change from minute 10 to 30 in the MLSS+10 where the VL does not.  During MLSS, both RF and VL HHb values remain stable.
Some of their comments:
Concomitantly with the accumulation of lactate, a greater
ventilatory response when exercising at MLSSp+10 was also
observed. Multiple factors might have contributed to this exacerbated
response, such as augmented metabolic acidosis,
body temperature, respiratory muscle fatigue, and perception
of effort. The greater increase in V.E, along with the accumulation
of lactate, seems to be a defining characteristic of
exercising slightly above MLSS (compared to “at” MLSS).
So the Ventilation response was felt to be a good indicator of loads above MLSS. 

After exercising in the severe-intensity domain at a PO
substantially above MLSS a reduction in performance is to
be expected. The findings from the present study are notable
as they demonstrate that even a very small increase in PO
(+10 W, ~3%-5%) above MLSS, reflected in a ~100 mL/min
increase in V.O2, results in a progressive rise in [La−]b and
disproportionately impairs subsequent exercise performance.
This is an important observation, as most methods utilized to
estimate the PO associated with MLSS, and similar thresholds
or critical intensities (supposedly eliciting stable [La−]b
and V.O2 responses) generally have an error in their estimate
that is greater than 10w
What they are saying is that it relatively easy to misjudge the MLSS (even with good testing), but exercising at MLSS+10 leads to severe reduction in performance later on.  In other words, setting your pace according to "measured" MLSS may lead to disappointing results with endeavors such as time trialing (if the measurement is off by just 10w).  Remember that a part of their study was devoted to showing the significance performance hit incurred by a MLSS+10 pace.

Overall, these data corroborate previous evidence highlighting
the diverse relative contribution of each of the muscles
(or muscle portions) engaged in the exercise task
the whole-body arteriovenous O2 difference which
arises from the heterogeneity in motor unit recruitment patterns, fiber-type
expression, and the resulting vascular
dynamic controls that characterize the active muscles.
the different dynamics in the behavior of the
[HHb] signal confirms that the regulation of the delivery
and utilization of O2 is different across the quadriceps muscles,
and may reveal that metabolic perturbations of exercising
in the severe-intensity
domain may be greater in
some muscles compared to others 
The RF behaves differently both in a ramp and at steady state at moderate to high power outputs.  Given the difference, it would seem that the RF would be a much better "target" for observation in distinguishing the MLSS transition.


Practical value of the above:
Expanding on the concept of looking for the point where steady state is no longer present let's look at a 10 min session comprised of 5 min of cycling about 15w below my measured MLSS (or 4 mmol threshold) immediately followed by a 5 min segment boosting power to 15w higher than MLSS.  Note that the "measured" MLSS could be over or under whatever the true MLSS is.  
What I would like to consider here is the use of NIRS measurements of various locations as surrogate markers of MLSS.  The costal muscle O2 response would be expected to track with ventilation rates, making that an attractive metric.  However, as seen in the last post, other muscle areas have similar breakpoints and could be helpful.  Yes, I have generally been skeptical of NIRS for MLSS or breakpoint monitoring but perhaps there is an effective use case scenario if done with long segments.  

Here is the overall breakdown from
On this day, sensors were on the costal, deltoid, biceps and calf areas.

To "verify" loss of steady state after the power shift, a look at ventilation data should be helpful:
I drew a blue line through the relatively stable Vent rate during the first 5 minutes, then a gradual rise over the next segment.

As a side note, I did have both the Moov sweat heart rate sensor and the Hexoskin recording to separate devices, heart rate correlation was excellent:
  • It is interesting that although heart rate did rise during the second segment, there was not a continuous rise as there was in ventilation.  This was also seen in Dr Murias' paper.

Costal HHb and saturation:

  • It seems pretty clear that both costal (serratus) HHb and O2 sat remain stable during the first segment (below MLSS).  
  • It is also apparent that the desaturated hemoglobin rises, with a fall in O2 saturation during the second half (above MLSS).
If one was training or racing, it would not be particularly difficult to track this in real time.  The absolute value is not critical, however the 5 minute curve shape is.   Acidosis leads to a respiratory rate/ventilation rise, increasing respiratory muscle activity along with costal O2 usage and extraction.  Absolute change in HHb is substantial.

Are these changes present in other sites?
According to the ramp testing, they should be.


  • The deltoid response is very similar to the costal despite that location having little to do with respiration.  This is presumably cardiac output redistribution, as the legs and now respiratory system use a larger piece of what cardiac flow is available, restricting flow elsewhere (to preserve blood pressure), increasing extraction.  
  • There is a stable HHb pattern below MLSS, with a continuous rise above it.


  • Very similar to deltoid, for the same reasons.
  • HHb stable below MLSS.
  • HHb continually rises above it.


This was a surprise.  I record the calf on many occasions but usually don't look at the data since it's generally not a recognized site for cycling.  In addition, the previous ramp data (for me) did not show a clear breakpoint.  
  • In this case, there is a clear demarcation of both HHb and O2 sat at the point of wattage change above MLSS.  HHb is stable below MLSS.
  • As in the other tracings, the HHb continues to rise throughout the MLSS+15 power.

Session summary of differences between MLSS-15w vs MLSS+15w (approx):
  • Mild elevation of HR that does not continue to rise despite rising ventilation in MLSS+15.
  • Initial stable ventilation during MLSS-15, then take-off of ventilation rate closely following change in power that continues throughout the MLSS+15 interval.
  • Stable costal O2 during MLSS-15 power, then continued rise in costal (serratus) HHb, fall in O2 sat after power boost to MLSS+15.
  • Stable calf, deltoid, biceps HHb/O2 sat during the MLSS-15w, then all rise continually after MLSS+15w.

Absolute HHb changes (approx):
                                         Costal                Biceps             Deltoid                 Calf         
        Total change         1.7 (9.0-7.3)      2.5 (9.3-6.8)     1.1 (9.1-8.0)       1.3 (10.3-9.0)

  • The biceps seems to have the best absolute change in this comparison followed by the costal.  The calf is just a bit higher than the deltoid and has a very nice looking curve.  
  • Despite the lack of a noticeable breakpoint on the ramp (possibly from the Hex sensor smoothing), the calf seems like a valid site for lactate change after all!

Another day with different sites:
Sensors were placed on the costal, biceps, forearm and rectus femoris.  The first 5 minutes were done at 240w with the next 5 minutes at about 275w.  So lets call them MLSS-10w and MLSS+25w (even though the MLSS may not be exactly 250w).

Ventilation/Hexoskin HR response:
 Ventilation only:

  • After a rapid rise, the ventilation and heart rate remain stable toward the mid to end of the MLSS-10 interval.  
  • The MLSS+25 results in a steady rise in ventilation throughout.  Heart rate rises but stabilizes during the last half of that interval.

Costal HHb and O2 sat:

  • During the MLSS-10, costal HHb and saturation remain stable.
  • During the MLSS+25, the HHb rises, with the O2 sat falling in a similar fashion to the MLSS+15.
  • The absolute increase in HHb is slightly higher than the MLSS+10 shown above this, but this could be related to placement rather than effort of course.  I don't think anyone would dispute the curve slope shifting (which occurs rather quickly).

Biceps HHb and O2 sat:

  • During the MLSS-10, the biceps HHb and O2 sat remain stable.
  • Shortly after MLSS+25, biceps HHb rises and O2 sat fall progressively as in MLSS+15.  Absolute HHb rise is higher than MLSS+15 (same disclaimer as in costal).

Forearm (inside/volar surface):

This is an interesting location since on the ramp test, it had a breakpoint at a lower power than the other sites. 
  • Although I may be over reading, the HHb and O2 sat seem to be somewhat less than stable on the initial MLSS-10.  Perhaps the last stable power range is below MLSS-10.
  • Certainly after MLSS+25, the curve markedly changes slope.
  • There seems to be an overshoot of HHb rise after pedaling stops.  The significance is unclear, but this has occurred after other high load intervals.

RF HHb O2 sat:

  • The rectus femoris response is in keeping with the concepts outlined above (higher recruitment with load), as well as the ramp tests done in a prior post.
  • The HHb, O2 sat remain stable in the MLSS-10 segment.
  • Both HHb rise and O2 sat fall steadily with the MLSS+25 interval.
  • There is a bit of an overshoot in HHb after pedaling stops.

Absolute HHb changes (approx):
                                         Costal                Biceps             Volar Forearm           RF         
        Total change         2.3 (9.8-7.5)      2.7 (10.9-8.2)      3.7 (10.3-6.0)       2.0 (9.8-7.8)

The forearm (which did have the earliest ramp takeoff) did have the largest absolute HHb rise.  Absolute values of HHb change were higher than in MLSS+15 testing.

Riding at MLSS and MLSS+18
Another 5-6 minute pair of intervals closer to the MLSS:
In this session, the sensors were on the costal, RF, VL and calf areas, giving a comprehensive view of both respiratory and locomotor areas in one session.  The first 5 minute segment was done close to the MLSS (247w) with the next part (6 minutes) done about 18w higher.  Considering the MLSS may not be exactly 250w, the first section may already exceed this, however the second part definitely should.

Ventilation, Power and Hexoskin heart rate:

Here things get a bit more interesting.  The first segment done at just under MLSS did have a stable ventilation rate.  However the second segment of MLSS+18 was not done at a steady power output.  The first part was done at 270w, then dipping down to an average of 255w during the last 2 minutes before coming back up (this was not intentional).  This lead to an increase in ventilation that peaks during the higher power section, but falls as power is cut back.  
  • This is a great example of how sensitive ventilation is to ongoing power fluctuations.  The question will be if this is seen in the NIRS tracings.

Costal HHb and O2 sat:

  • The MLSS pace seems to produce a stable costal HHb and O2 sat.  This actually looks even more stable than the raw ventilation rate curve.
  • The higher power segment is associated with a rising HHb that has a decrease in slope around the time of power decline.
  • The O2 sat generally follows HHb pattern (in reverse).

Calf HHb and O2 sat:

  • Calf HHB seems to remain steady during the MLSS pace.
  • During the higher power segment there is a steady rise in HHb that appears to follow the fluctuation in cycling load later on.  Whether this is real or random is unclear.  There is no doubt though that the HHb is rising in MLSS+18.
  • The O2 sat generally follows HHb pattern (in reverse).

VL HHb and O2 sat:

  • The HHb seems stable in the last part of the first section of MLSS power.
  • There seems to be a small but steady rise in HHb during the MLSS+18 segment.
  • However this is of a small absolute magnitude and there is no slope change associated to the power dip/rise.
  • The O2 sat generally follows HHb pattern (in reverse).

RF HHb and O2 sat:

  • The RF HHb response seems stable during the MLSS phase.
  • There is a steady rise in HHb during MLSS+18.
  • There is a HHb slope change associated with cycling power fluctuation toward the end of MLSS+18 (like the calf).
  • The O2 sat generally follows HHb pattern (in reverse).

Absolute HHb changes (approx):

                                         Costal                Calf                     VL                 RF         
        Total change         1.1 (7.9-6.8)      .9 (9.4-8.5)         .4 (7-6.6)       .7 (8.2-7.5)

On this day, the costal and calf sites still had decent absolute range of value change that could be distinguished in real time, but the VL did not have changes that would be noticeable while riding outside on the road.   

What about a longer interval at the MLSS?

Here is a session composed of an initial 30 seconds at 400w (well above maximal aerobic power) then a steady 253w for 10 minutes.  This type of effort would presumably accumulate a fair amount of lactate initially, then allow a slow rise (while riding at the MLSS+3).   Sensors were on the costal, VL and RF areas.  The RF placement was different than the above, hence the difference in saturation numbers.

 Ventilation and Heart rate:

 Close up of Ventilation only:
  • Although the rise is not as sharp as in some of the above tracings (MLSS+25), there is a steady rise throughout.
  • Heart rate remained stable past midway.

Costal HHb and O2 sat:

  • There is a steady rise in HHb, fall in O2 sat (parallel to ventilation) starting about minute 3.

Vastus HHb and O2 sat:

  • There is a rise in VL HHb, fall in O2 sat that is of a much smaller magnitude than the Costal site.
  • Although there is a trend, the limited dynamic range may obscure any practical use in real time on the road.  
  • This makes sense in view considering the VL has a blunted desaturation response at high loads, therefore not much should happen to the HHb in this zone

Rectus Femoris HHb and O2 sat: 

  • There is a steady rise in HHb, fall in O2 sat past the 3 min mark of a reasonable magnitude.
  • This seems a better locomotor site (than the VL) to look for MLSS associated change, or worsening acidosis in general. 

Summary of absolute change in HHb ((nadir at about 2-3 min) - (end value)):

                                         Costal                        VL                         RF         
        Total change         1.1 (8.2-7.1)         .35 (7.25-6.9)         1.1 (7.6-6.5)

Absolute value change is not much different to the set of figures immediately above this one.  The VL again shows poor absolute range in saturation and HHb magnitude.

Although lactate was not measured, given the initial, then steady state power, the gradual rise in ventilation, this interval probably was done somewhat past MLSS.  The absolute amount over is less important than the behavior of each metric.  In either road training, racing, time trialing, one won't be able to perfectly modulate their pace to maintain an exact MLSS power output.  If ventilation was available as a viewable value (with graph), that would give a good indication whether a lactate buildup was occurring, or if MLSS was exceeded.
  • In the absence of ventilation data the next best measure seems to be the various muscle HHb/O2 sat measures reviewed above.  Unfortunately, heart rate is not very helpful and remains stable (and submaximal, although higher) at MLSS+10 or even MLSS+25 (my data).

  • Measurement of the true MLSS is quite difficult and prone to error.  Multiple separate studies are needed.  
  • The MLSS (even if noted exactly) can fluctuate over time depending on training and conditioning status.
  • Consequences of exercise minimally above the MLSS are significant.  This includes impairment of performance later in the session.  Therefore, unintended exercise time above MLSS can be counterproductive in a race.
  • Ramp testing using NIRS of locomotor muscles has been used to estimate MLSS.  Unfortunately, this has a substantial error, well above the 10 watts shown to adversely affect performance. 
  • Longer steady state intervals have the potential to identify the MLSS power with proper sensor placement.
  • Many MLSS ramp tests use the VL since it is "active" in cycling.  But, it's ramp breakpoint curve is blunted at that intensity.  Therefore, this site is not optimal for steady state MLSS identification.
  • The RF NIRS response has been shown to have potential to indicate power above the MLSS using long interval observation.
  • Ventilation rates have also been noted to be useful in the demarcation between steady lactate versus accelerating rise in lactate values.
  • Alternate NIRS sites including the respiratory muscles and non locomotor areas have potential to indicate power above MLSS.

Practical Considerations:
Although multiple studies have shown some correlation between NIRS breakpoints and MLSS, most suffer from the following:
  • Use of the VL which has a blunted HHb response at high load.  This in turn will limit the dynamic range of HHb/O2 sat values making trend analysis difficult.
  • If one were to use a locomotor area, either the RF or perhaps the calf would be better choices given the ramp behavior and net change in HHb/O2 sat.
  • Usual method to determine MLSS was a ramp test.  Depending on the ramp protocol and interpretation of NIRS curves, substantially different results can be claimed.  Ramp tests are not practical for the average user.  In addition, even small changes in sensor placement may have effects on the values obtained.
  • Little to no exploration of HHb/O2 sat behavior with longer, steady state intervals.
  • Sensor placement on locomotor muscles only.  In view of the excellent correlation of MLSS with ventilatory rate change, the respiratory muscle seem a logical choice as a marker of respiratory effort.  Other non locomotor sites (deltoid, forearm, biceps) may also be useful in sports that do not generally engage these muscles.  In these areas, cardiac output redistribution with resultant blood flow reduction would be expected to lead to higher net O2 extraction.  My (anecdotal) data does indicate the potential usefulness of these sites particularly in view of the large absolute saturation/HHb changes that occur.
  • Instead of short discontinuous intervals (moxy 5-1-5), athletes should consider doing their own longer, continuous steady state, HHb/O2 sat trials.  A limited number of segments may be all that is needed to see where NIRS values begin to lose a stable pattern.  Given individuals may have different insights depending on their own best site for sensor placement as well as interval time.  "One size may not fit all" for both site and length of interval required.
  • During a long segment of exercise, tracking the direction of the O2 saturation slope (not the absolute values) may provide important feedback in regards to not exceeding MLSS.  Limiting long time periods above MLSS may allow one to have better performance and reserve power at the end of a long session.

     Lactate related posts


    1. ounds like a Costal is a better alternative , but in Rowing or AirBike, maybe a issue , like forearm you describe when engaged muscle.
      Multiple long steady states to find MLSS , sounds best alternative, but do it every single long sessions sounds not practical

      1. Certainly not recommended to do this on every session - only when you want to estimate your second threshold