The ramp test is done using a linear increase in power, but the physiologic metrics of heart rate, cardiac output, lactate, VO2 (O2 usage) and muscle O2 change are not always linear. One of the purposes of the ramp is to investigate "breakpoints" in these and other parameters. A breakpoint represents a departure from whatever curve tracing was previously present. For instance, ventilation may increase in a relative similar linear fashion as power rises up to the zone where acidosis occurs and a compensatory hyperventilation begins. The BSX Insight was originally designed and advertised as a device to measure the lactate threshold, presumably from O2 saturation kinetics of the calf muscle. In some previous posts, the concept of using a breakpoint for this purpose was reviewed. Although in theory it's a somewhat attractive method, the reality is that errors from the true threshold can be quite large. Even though the exact breakpoint in the desaturation (or HHb) curve may not correspond to a given physiologic state change, the order of breakpoint change may be of great value to us in real time road race pacing. For instance, let's say we knew that at a particular site the O2 sat breakpoint would occur at below the lactate threshold, but at different site the breakpoint is near the lactate threshold, and a third site has a breakpoint at above the respiratory compensation point. If one was monitoring these sites it is conceivable that we could have a window into an integrated whole body intensity index. Yes, a cycling power meter (with heart rate) may come close, but for other sports without power or in cycling situations where power is erratic, the breakpoint data may be helpful.
Before getting into some of my ramp data of various sites, I'd like to review a paper published about 2 years ago that addressed the breakpoints of leg, respiratory muscles and cerebral O2. In the following discussion, the authors pointed out that HHb (desaturated Hb) was a better parameter to follow since it was less affected by blood volume under the NIRS probe. The commercial devices do not report this as a raw figure but it can be calculated if one is interested. I mention this since the vast majority of my tracings are using raw O2 sat as reported by the BSX or Humon sensors, but the following "ideal" ramp breakpoint curves will use HHb or oxy Hb.
Here is an example of the locomotor muscle ramp, power vs HHb and O2 Hb:
The GET is the gas exchange threshold (GET) and has been used an an index of anaerobic threshold because it can be measured noninvasively. GET is estimated "from a breakpoint in breath by breath values of carbon dioxide uptake (Vco2) and oxygen uptake (Vo2) obtained during a progressive exercise test".
The authors also commented on the sluggish HHb response at low levels of exercise load:
Several mechanisms have been proposed to underpin the sluggishAnother very appropriate comment is on the shape of the curve and a potential mechanism why this is so:
increase in HHb at the onset of incremental ramp exercise
(Fig. 1b, phase 1). It can be explained by the mechanical
effects of muscle contraction (i.e., muscle pump and rapid
vasodilation) at the onset of exercise and/or by changes in
sympathetic-parasympathetic balance in cardiovascular
control (Ferreira et al. 2007).
It is logical that these specific fiber type characteristics in combinationThis model incorporates the various fiber populations that will be recruited as power/load rises. The fiber type usage pattern will probably be different across muscle groups as well as across individuals (running/cycling technique) making exact comparisons difficult. In addition, individual athletes will have different proportions of type 1 vs 2 fibers further complicating the picture.
with the sequential recruitment of slow-twitch to fast-twitch
fibers, which is typical of incremental ramp exercise, could
result in a sigmoidal profile of fractional O2 extraction, and
thus locomotor muscle HHb. As the incremental ramp exercise
proceeds, progressively more fast-twitch fibers will
be recruited, leading to a substantially greater increase in
fractional O2 extraction and consequently in HHb
Lastly, the limit of the curve:
the HHb response shows a plateau at the end ofAnd:
the incremental exercise (Fig. 1b, phase 3), indicating that
fractional O2 extraction reaches it limits.
This is supported by studiesThe breakpoint correlation with physiologic metrics was presented as such:
showing that the mechanical constraints of heavy muscle
contractions (Kowalchuk et al. 2002) and redistribution
of the cardiac output to other parts of the body, such as
the respiratory muscles (Legrand et al. 2007) might limit
blood flow to the locomotor muscles.
Some preliminary comments from the paper are interesting:
These data points at substantial degree of intra-individual variabilityAnd:
between the thresholds which questions the equivalence
among these data. Given this discussion it remains
to be elucidated whether these breakpoints and landmarks
of exercise intensity are mechanistically linked.
Additionally, the high level of heterogeneityAnother comment was made in regards to the kinetics of response vs the amplitude:
within and among muscles in oxygenation pattern
(see above) (Koga et al. 2007; Chin et al. 2011; Koga et al.
2011; Okushima et al. 2015) questions the possibility to
accurately determine a landmark of exercise intensity from
measurements of the oxygenation pattern at a single location
during an incremental cycle test, as a replacement for
the traditionally determined thresholds from whole-body
Analysis of the amplitude of the NIRS response
in 64 subjects (Boone et al. 2016) with a heterogeneous V
O2peak showed that the amplitudes of both totalHb and
HHb were positively correlated to ˙VO2peak, indicating that
subjects with the highest aerobic fitness levels showed
the highest change in totalHb and HHb (Fig. 2). Also the
few studies that have addressed the impact of a training
intervention on the HHb response show an increase in the
amplitude of the HHb-response following the training program
(Prieur and Mucci 2013; Takagi et al. 2016
it seems likely
that an increased capillary-to-fiber ratio is one of the possible
peripheral adaptations in subjects with a higher aerobic
fitness status (Proctor et al. 1995). This seems a reasonable
explanation for a more pronounced increase in totalHb
in subjects with a higher ˙VO2peak
Respiratory muscle kinetics
The next site of interest reviewed was the respiratory muscles. This is a very important site given the large amount of cardiac output devoted to the respiratory system at high loads (>15% of cardiac output) and the existence of the metaboreflex which may cause vasoconstriction of locomotor muscles at high ventilatory rates.
The breakpoint pattern was described as such:
Here we see that the nose dive in O2 Hb (or increase in O2 desat) occurs near the respiratory compensation point probably related to the increase in ventilation (respiratory muscle usage) associated with acidosis and/or build up of metabolites.
The NIRS probes were generally felt to work best at the serratus anterior muscle (what I term costal).
One interesting fact that was new to me was that since a ramp is usually quite short, it is not sufficient to induce the metaboreflex:
Romer et al. (2007) observed that incrementalAnother observation looking at the above tracing is that the breakpoint in O2 Hb is relatively sharp and steep (less so with the HHb). This is in contrast to the locomotor ramp curve which has blunted curve shifting. Although in a lab environment this may not be an issue, on a practical note, a more obvious change is needed on the road given the multiple causes of data value variation.
exercise did not elicit diaphragm fatigue and since unloading
of the respiratory muscles had no impact on incremental
exercise duration, it can be suggested that the duration of
heavy intensity cycling (>85 % ˙VO2max) is too restricted to
induce the respiratory muscle metaboreflex.
The NIRS changes at the level of cerebral tissue are much different than either locomotor or respiratory areas. The idealized tracing is below:
Here we have both an increase in O2 Hb (oxygenated Hb) and a gradual rise in HHb (de-oxygenated Hb. How can this be? The reason is the much higher flow into the CNS with exercise and thereforeboth the oxy and deoxy Hb can rise. Both the oxygenated and de oxygenated portions rise until near the RCP when hyperventilation results in a lowering of CO2 levels in the blood (respiratory alkalosis), which apparently diminishes cerebral perfusion. At that point, since perfusion drops, the HHb rises sharply as well as the O2 Hb now dropping (lower flow, high usage with net higher O2 extraction).
Here is a plot of cerebral blood volume (Cbv) and oxygenation index (Cox) measured by NIRS with a drop off near the RCP:
The falloff is quite sharp and if the technology was available could provide meaningful insights to performance status.
Apparently, cerebral oxygenation patterns are somewhat different in more vs less fit (by VO2 max):
Aerobic fitness status has an impact on the patternSo a very complex situation, but one that could provide information about fitness, training zones and high intensity domains.
of cerebral oxygenation as well (Rooks et al.
2010). In trained subjects (i.e., with an average ˙VO2peak
of 62.7 ± 7.9 mL min−1 kg−1) cerebral O2Hb levels
increased less at low to hard intensities compared
to untrained subjects (i.e., with an average ˙VO2peak of
40.3 ± 10.4 mL min−1 kg−1) whereas they showed no (or
only a small) decrease in cerebral O2Hb at maximal intensity.
Cerebral HHb, however, showed a more pronounced
increase from moderate to very hard intensity in trained
versus untrained subjects
Finally, the authors put the whole system together, integrating the work rate (Power) with
- local muscular recruitment effects, O2 extraction affecting locomotor HHb
- acidosis, ventilation, cardiac output redistribution affecting respiratory HHb
- low plasma CO2 from higher ventilation, higher CNS O2 usage affecting cerebral HHb
An excellent schema and summary!
Several parting comments are worth focusing on:
Although these breakpoints in oxygenation and thresholds
determined from ‘whole-body’ responses occur in
close correspondence at an intensity between 75 and 90 %
˙VO2 max and there might be physiological mechanisms
linking these points (see above), it is highly questionable
whether the breakpoints in oxygenation pattern at the different
body regions can be used to demarcate intensity
zones and/or whether they are equivalent to critical power
and the thresholds determined in pulmonary gas exchange
and blood lactate concentration.
Whether measurements of local oxygenation patterns
are a better alternative is highly questionable for several
reasons. First, it seems very unlikely that oxygenation
measurements of only small portions of one muscle or
body site reflect whole-body responses to different exercise
intensities. Critical power can be considered as a ‘performance
threshold’, derived from the power-duration relationship.
Maximal lactate steady state and respiratory compensation
point are ‘physiological thresholds’ derived from
physiological measurements. In both the ‘performance’
and ‘physiological’ thresholds several (or all) physiological
processes (intermuscular, pulmonary, cardiovascular, and
metabolic) are integrated. One the other hand, breakpoints
in the oxygenation patterns at one location in the body are
derived from only limited portions of the physiological
responses. Therefore, extrapolating these breakpoints as if
they reflect a whole-body response to exercise would be too
simplistic in our view
Second (and in line with
our first point), there is a large heterogeneity in the oxygenation
pattern among and within the locomotor muscles (but
also the respiratory muscles) (Chin et al. 2011; Okushima
et al. 2015), in a large part related to muscle activation pattern
but also in relation to different vascular and metabolic
control properties of fiber type populations. In this regard
also the penetration depth of the NIR light will have an
impact on the oxygenation response to exercise, given the
difference in muscle fiber distribution between deep and
superficial muscle portions (Okushima et al. 2015). Therefore,
the detection of the breakpoint will largely depend,
within an individual upon the specific characteristics of
the region under the probe, and between individuals upon
the fiber type distribution of the studied region.
Oxygenation responses at level of the locomotor and respiratory
muscles, and the brain are the result of a complex
interaction of several physiological mechanisms determining
˙QO2 to and ˙V O2 in these different tissues. Breakpoints
in these oxygenation responses have been identified (by
means of visual inspection or curve fitting) to ‘quantify’
the response and simplify the interpretation of complex
phenomena. Therefore, it is highly likely that transfer of
these oxygenation responses to practice (e.g., guidance of
athletes) will result in an oversimplification of the response
and its underpinning physiology
Finally, it should be
noted that little information is currently available on the
reliability of the breakpoints in the oxygenation responses.
To our knowledge the only study reporting test–retest values
(Miura et al. 1998) shows limits of agreement (95 %
confidence interval) of >25 % for the second breakpoint in
locomotor muscle O2Hb. This indicates that the reliability
might be poor limiting the application of these breakpoints
I decided to try a ramp with a start power of 85 watts, increasing 30 watts every minute. Sensors were placed on the L calf, L rectus femoris, L costal (serratus), L deltoid. The Hexoskin shirt provided both ventilation data as well as a single lead EKG. A 35 minute warm up was done first. The session was done on an indoor trainer (doing this outside on the road would be near impossible as well as dangerous).
Here is a quick overview with lap power averages in yellow:
Let's take a closer look at power, heart rate and ventilation:
The power is in blue, heart rate is in red and ventilation (L/min) is in green. I really couldn't see any breakpoint or curve deviation in heart rate past the 130w stage. There did appear to be a flat ventilation response until about 135w then a possible breakpoint near 300w. That's above my MLSS and RCP, but given the shorter nature of the test and failure to reach an equilibrium, not unreasonable. We will keep the 300w figure in mind though when looking at the NIRS data.
The max heart rate and vent metrics are in line with attaining my VO2 peak/max.
Below is a combo of all 4 sensor locations looking at Deoxy hemoglobin. This is a simple calculation of (100-O2 sat) x (total Hb):
I did not attempt to look for breakpoints yet, but it is interesting to see that deoxy Hb peaks first in the RF, costal areas, later in the Deltoid and last in the Calf. The fall in deoxy Hb early on in the Deltoid (purple) perhaps consistent with a muscle pump effect. This phenomenon is supposedly from venous return enhancement from leg muscle activity which in turn boosts stroke volume. We will look at the RF HHb in a bit since that is an interesting muscle group to discuss. The calf sensor was the Humon Hex, which although gives consistent results across sessions, does have a very smooth response. Maybe too smooth - and I have wondered if there is a smoothing algorithm in the device (I never see a spiky trace).
Since the above paper discussed using the oxygenated hemoglobin and deoxy hemoglobin to graph looking for breakpoints, let's take a look at all 4 sensor locations. The color key is seen below, the vertical line is the end of the ramp.
Individual curve shapes:
Here is the Calf HHb:
Perhaps there are some mild curve inflections but I did not want to over read patterns where there are none.
The huge increase in HHb post ramp is perhaps another situation to look into. I wasn't sure if this was a time delay in the Humon sensor, and there simply was a breakpoint 30 seconds back. I will need to try this with the BSX sensor for comparison (especially since it was designed for the calf). The main reason I have been using the Humon Hex on the calf is that this sensor is very light tolerant so no shielding is needed (which is hard to do on the calf).
Addendum: according to tech support at Humon, the readings are subject to about 20 sec of smoothing. Whether or not this affects the breakpoints is unclear.
Here there is a better demarcation (although this is debatable), with a sharper drop off starting near 235w and even more sharply at 325w. Of course, if the ramp was done with different timing (longer) the watt values would be different. It is different than the sinusoidal model noted above. However, Dr Murias' group has shown that the EMG activity in the RF does not fully engage until higher power domains are reached. This may lead to the mild early desaturation with a steep falloff at high intensity:
This does resemble the HHb curve below with a slow rise, a breakpoint at high intensity.
Here is my HHb curve:
- The breakpoint is indeed about 80-85% of peak power, which is what the quoted paper indicated.
- This value is quite close to the ventilatory rate breakpoint derived from the Hexoskin
No major change until about 235w, then a dropoff with another shift at 325w. This seems like an excellent 2 breakpoint tracing. The first break comes just before the MLSS.
- The costal does seem to have a significant HHb rise even after the ramp ends, perhaps signifying continued high extraction from flow redistribution away from this area.
Somewhat similar to the deltoid, the costal/serratus anterior seems to have 2 breakpoints in the oxy hemoglobin tracing. They also appear to occur at about the same time/power values. Of course, alternate ways of drawing the lines is possible, but the above does fit nicely.
- The first curve shift does occur near the MLSS where acidosis related higher ventilation would occur. The second is past the RCP and into a severe intensity domain.
- It does appear that the higher ventilation rate occurs just before the costal oxy hemoglobin curve shifts.
This agrees with the O2 Hb plot, depending on where you draw lines.
- The post ramp HHb quickly reverses, unlike the deltoid. Although O2 usage is still high, flow is high enough to compensate (as opposed to the deltoid).
- This does resemble the data shown By Dr Boone's paper (figure on the right):
While riding on the road, the display of either a Garmin device or Ipbike for android shows O2 saturation, not oxy or deoxy hemoglobin. How do those curves look?
They appear to mirror the oxy Hb, both in breakpoints and post interval recovery. That is very convenient and helpful. The calculations of HHb, O2Hb are not difficult, but not easily incorporated into the real time experience.
- The recovery delays discussed above are present in this tracing as well (no delay in costal, RF, moderate delay in deltoid, longer delay in calf).
- Most of the "action" as far as rapid desaturation is concerned, occur above the RCP (from the vent rate plot)
There appears to be a
- Steady fall in calf THb with acceleration at the higher loads (higher intramuscular pressure)
- Slight rise in RF THb with a dip at the end (high intramuscular pressure)
- Slight fall in costal/deltoid then leveling out. Absolute magnitude is small.
- Except for the calf, there is trivial change in THb, making simple O2 saturation a valid surrogate for HHb or oxy Hb.
- Total hemoglobin should not be confused with overall flow.
On a different day, I did a similar ramp (30w/min) with sensors on costal, left Vastus Lateralis, L inner forearm and L biceps. Unfortunately, the Hexoskin heart rate signal was poor and not shown. The respiratory data was good and ventilation rates did reach max values at the end. Power at high intensity was comparable to the first ramp.
Here are the breakdowns in power intervals:
HHb O2 Hb tracings:
Let's look at individual tracings in isolation.
This does resemble the hypothetical tracing from the published review (the left one below):
- Although I have never been able to measure my GET, the RCP inflection point (264w) is probably close to my current figure (with MLSS at 250). Of course other observers could draw the lines differently but there does seem to be a reasonable similarity between the two.
There does seem to be 2 breakpoints, one about 233w and the other 300w. These are similar to the figures noted in the first ramp (MLSS and RCP?). The last curve slope is very steep.
This is very similar to Ramp #1 above.
Costal O2 Hb
Comparison of costal vs VL at high loads:
The costal O2 Hb mirrors the HHb plot with a sharp dip at about 300w marker. I wanted to contrast the costal vs VL patterns at high intensity:
- The curve slopes of oxy hemoglobin are very different at high loads, making costal monitoring an attractive possibility to avoid near VO2 peak loading. A small rise in power (past a certain point) produces large drops in costal O2 change.
- There appear to be 2 breakpoints in this as well, at similar locations to the costal. The first breakpoint is at a lower power which will be looked at further on the post.
- The post ramp HHb persistence is present but less than the biceps below.
- The biceps HHb ramp tracing is pretty flat until near 300 watts when there is a marked departure and very steep slope upward.
- The post ramp HHb stays elevated for close to 1 minute. This may be related to the cardiac output redistribution with continued reduced flow.
Biceps and Forearm Oxy Hb
- There is a relatively level beginning then a slight fall in O2 Hb in both Biceps and Forearm, followed by a steep fall off. The inner forearm seems to fall off sooner (265w) than the biceps(300w).
- On the other hand, if one draws a intermediate breakpoint in the forearm, the steepness occurs closer together.
- No matter which curve choice you make, it is reasonable to conclude the biceps and inner forearm will have a sharp decline in oxy Hb at higher loads, especially above the RCP.
Costal and VL:
I wanted to double check that the curve slope was equally distinct with the costal curve very sharply negative:
Definitely seems very different at high load.
Biceps and Inner forearm O2 saturation:
There does seem to be an earlier drop in saturation in the forearm than the biceps (264 vs 300w). The biceps recovery appears to be delayed as well. This is similar to the oxy and deoxy Hb figures.
In both cases, the desaturation accelerates after 300w.
There are some small changes in total Hb with:
- Steady forearm
- Slight fall in biceps
- Fall in costal then a small rise as intensity picks up.
- Steady rise in VL, even at the end with high muscular effort. This was a surprise for me since the RF had the (expected) fall at very high power from muscular compression.
- Rebounds in all sites after the ramp
- There is trivial change in THb, making simple O2 saturation a valid surrogate for HHb or oxy Hb at these sites.
- Total hemoglobin should not be confused with overall flow
Summary and points of interest:
- It is possible (for the amateur) to do a ramp on an indoor trainer looking at heart rate, ventilation and O2 NIRS kinetics. Both deoxy and oxy hemoglobin can be derived from O2 sat and total hemoglobin if desired.
- Your personal ramp curve may or may not resemble "classic" models. In my case, the VL was spot on in shape and perhaps breakpoints. I did see good resemblance in shape in the costal (serratus) plot. There does not appear to be much literature on the other non locomotor sites I measured to compare. The RF curve was similar to that reported and did have a different shape from the VL.
- Depending on an individual's muscle fiber predominance, recruitment pattern, depth of NIRS measuring, ramp protocol, curve pattern, the breakpoint values may be quite different.
- Drawing breakpoints can be highly subjective. There are mathematical models, but given the less than ideal data fluctuation, the human eye may be better at ignoring certain anomalous areas. Coupled with variation in daily sensor location, the validity of a given "model", using breakpoints as a surrogate of MLSS or other physiologic measures is problematic. Yes, the BSX was designed for this, but we saw that in the calf (for me) there was no sharp breakpoint.
- The desaturation kinetics of the costal, forearm, deltoid and biceps may be a more reliable metric to follow in the intense exercise domain. Demarcation of high intensity zones appears to be sharper than locomotor breakpoints. With this sharpness comes a higher absolute range in saturation value, making visual inspection easier.
- If one is to watch for a sharp area of desaturation, the RF seems a better choice than the VL. The RF accelerates it's hypoxic pattern (more recruitment) but the VL levels off in the high intensity zone.
- From a physiologic standpoint, the accessory respiratory muscles would be expected to extract more O2 (from high usage) as ventilation rises with locomotor intensity (especially post RCP). This should lead to high HHb, lower oxy Hb and dropping O2 saturation which is what was seen. With experience, an athlete may be able to modulate their effort more efficiently, avoiding decompensation with a better net overall result.
- Alternate sites such as deltoid, inner forearm and biceps appear to desaturate sharply in the higher intensity zones. Although the forearm appears to be affected earlier, this may not be the case in other subjects.
- These non respiratory, non locomotor sites are probably victims of cardiac output redistribution with resulting lower flows and higher O2 extraction. There is no major shifting in total Hb indicating muscular compressive effects from sudden use.
- This cardiac redistribution will also affect the gut, limiting water and nutrient absorption. Making sure there are solid time periods of normal O2 saturation at these non active sites may help avoid dehydration, nutrient and gut issues.
- Finally, each individual may want to run their own ramp at these sites to determine the ordering of breakpoints. The exact watt value is less important than the relation of which site will break first or last. In addition, a shallow breakpoint curve may not be very noticeable in real time. Sites with sharp, high magnitude breakpoints may be easier to decipher during real time racing or training.
I would also like to thank the groups of Drs Boone and Murias for their excellent publications and work in this field.
Next: MLSS determination with NIRS and long intervals