In recent years I have been very interested in the utility of continuous blood glucose monitors (CGM) for measuring both performance and health metrics. Last year, I tested a CGM for two weeks while living a normal life as a junior doctor – some commuting, working across days and through night shifts, fitting in exercise, and eating! – and concluded that, while CGMs such as the Freestyle Libre are revolutionary devices for individuals with diabetes, there are multiple reasons why CGM data has limited use for otherwise healthy and fit people, let alone people who want to push performance.
Thus, it was with extreme interest that I read the slick adverts for Supersapiens that had infiltrated my newsfeed and blogs and various websites I read. The Supersapiens package comprises a CGM produced by Abbott (the manufacturer of the Freestyle Libre that I tested, while Supersapiens uses the Libre Sense), a special Supersapiens dressing to hold the CGM when placed onto the arm, and the Supersapiens app – a package prepared for the most hardcore swimmers, runners, cyclists or Ironman athletes. They’re big news – the most recent headline is of its exclusion from the UCI pro peloton, which adds to the product’s appeal!
Marketed at endurance athletes of all calibres, it promises to make us ‘go faster longer’, and to ‘master fuelling’ by monitoring real-time blood glucose to derive information to ‘manage your fuel levels, push your limits longer, [and] get bigger gains’.
The new technology asserts that optimal glucose levels are crucial for both intensity and duration of performance at ‘threshold’ (assumed to mean the lactate threshold, a point where the production of lactate by working muscles during exercise starts to increase rapidly – you’re working hard in training or racing). This ties into a Supersapiens concept of ‘ON-hours’, where glucose monitoring can help to identify and keep an individual within the Glucose Performance Zone (GPZ) that facilitates exercise performance.
In contrast, ‘OFF-hours’ reflects normal living or the post-exercise recovery period. During these times, blood glucose monitoring is suggested to help an individual understand what foods may promote ‘stable energy’ levels throughout the day, while a more athlete-specific application may be to chase glucose levels, using a metric termed the ‘Trailing Average Glucose’ (I’m unfamiliar), most conducive to glycogen replenishment.
That’s a lot of output for a device that monitors only one input of glucose. Of course, you could argue that there is useful extra information from the raw glucose measurement; the absolute glucose value, its variability and its rate of ascent or descent would be some extra data that could be derived – as I tried in my previous blogs.
So does it work?
I don’t know because I haven’t used the Supersapiens product – but from my experience with the Freestyle Libre, which is similar to the Libre Sense used by Supersapiens, I can’t see that the information derived from the sensor being game-changing for a recreational athlete, or even a professional athlete.
My first issue will be with the user-interface. Although I haven’t used the Supersapiens app, my experience with the Freestyle Libre app was lacking. This was because the amount of data that was collected was immense, but the ability to overlay significant events in the day onto my glucose reading was clunky and thus underused. I struggled to identify the exact time of me hopping onto my bike to commute to work, meaning that the corresponding changes in my blood glucose were inappropriately labelled or forgotten about. When I did do a workout, the blood glucose wasn’t paired to other performance metrics – like the weight used, rest taken, RPE, or when training for endurance, my heart rate and power. The Supersapiens product integrates with Garmin, so blood glucose can be read when on the bike (and simulatenously with power/ heart rate if used), so perhaps there will be future availability to display both glucose and other performance metrics side-by-side after the workout. An even better scenario would be integration of these performance metrics with algorithms to more closely relate glucose to power, and any insights this could provide. This would mean I wouldn’t have to trawl through Excel to draw my own insights from the raw data, like the AUC measurement I derived in my previous post.
But wait –
The success of the app relies on the success of the product. How do we know the data produced by the device is actually useful (accurate, precise?)
One of the strengths of the Freestyle Libre I tested is its ability to detect highs and lows – essential for individuals with diabetes, for whom elevated glucose levels may contribute to premature development of chronic diseases, while ‘low’ hypoglycaemic episodes may be acutely disabling, damaging and/or fatal. On the other hand, the precision with which the monitor can delineate between small degrees of difference in blood glucose in the normal range is questioned – the range most likely to be inhabited by a healthy athlete. For example, the accuracy of these monitors is summarised by MARD – the Mean Absolute Relative Difference, or from the Abbott website ‘on average, how far away the glucose sensor reading is from a blood glucose reading’. The Freestyle Libre CGM has a MARD value of 11.4%, so any measurement, on average, may be ~11% above or below the true reading. It’s not huge, but if you’re not chasing the extremes of glucose but are looking for nuances within the ‘normal range’ after you eat a snack, this might make a difference to your ability to make valuable conclusions.
What about the integration of the CGM with our intrinsic physiology? Or, is glucose a high-yield and useful metric to measure?
To quote Supersapiens: ‘Manage your fuel levels, push your limits longer, [and] get bigger gains’
Fuel levels are a tricky thing – because unlike our vehicles and devices (fuel tanks and battery metaphors abundant), we aren’t limited to one fuel source (or two if being pedantic and thinking of hybrid cars). Glucose, while an abundant fuel, is accompanied by fatty acids from food and our own fat stores and by ketones, derived from fat metabolism. Fat, too, can fuel muscles. Ketones can fuel tissues such as muscles, but also extends its reach into the brain (which often is thought and said to be solely fuelled by glucose). Metabolism is wonderful and builds in endogenous redundancy to make sure that we stay alive – every new day is a new PB!
So, as at any point your body may be deriving its fuel from a varied balance of three sources, there is room to suggest that monitoring only one source may not provide the most complete insights into energy availability to working muscles and brains.
The fuel your body might choose at rest or during exercise is based on many things – for example, the diet you eat (hypothetically, would ‘keto – or fat-adapted’ athletes benefit less from a Supersapiens glucose monitor if they require less glucose?), the training you do (athletes metabolise more fat at lactate threshold than do people with obesity, thus may be expected to have different glucose readings, too) and the current intensity of exercise (at threshold or efforts >80 of VO2max. glucose utilisation by working muscle undeniably increases with a reciprocal decrease in fat oxidation, but ‘fat- or keto- adapted’ athletes can maintain a large proportion of utilised energy from fat).
So, assuming a high amount of inter- and intra-individual variation regarding fuel sources during these different states surrounding exercise, can we derive meaningful insights into energy metabolism from a sole metric?
My suggestion is ‘probably not, but I’m curious all the same’.
What about energy depletion during exercise? Can we look at a downward trend of blood glucose to suggest that we refuel? Is it as simple as watching the ‘battery’ run out?
This is something that I would love to see in practice – if this works, it could be an amazing corollary to exercise intensity and duration.
E.g. ‘Oh no, I’m at 56% charge and am less than halfway through my ride – I’d better top up!’
But, would it work? For the reasons above, it’s already difficult to know truly what proportion of energy is being derived from glucose at threshold, or any other exercise intensity. So if we don’t know our glucose utilisation with respect to other fuels, how can we extrapolate the consumption and depletion of our glycogen stores?
I would expect that within a session of steady exercise intensity, glucose levels would remain constant until they drop. Due to the body’s ability to break down glycogen to liberate glucose (glycogenolysis), the supply of glucose should be constant. Even when glycogen starts to deplete, the body can use fats and proteins to create glucose (gluconeogenesis). The body can even convert lactate into glucose. In fact, rather than being my lowest readings, intense Wattbike or HIIT sessions produced some of my highest blood glucose readings using the Freestyle libre. My point is – there are multiple systems in place to make sure that glucose availability is resilient and maintained. I would expect that these systems would not allow the blood glucose concentration to reflect glycogen status, but again, would be happy to be shown to be wrong.
What about ‘Off-Hours’
To quote Supersapiens: ‘Glucose Loading … Know whether you are ready to go by seeing where your Trailing Average Glucose is’.
I’m trying to imagine what a lay-person imagines when reading this statement. Is it that your ‘trailing average glucose’ (what?) reflects the status of filling of glycogen stores? If that’s so, then that’s problematic – because in health, blood glucose shouldn’t rise drastically or persistently in response to dietary intake of carbohydrate. Instead, multiple intrinsic control mechanisms closely govern blood glucose concentrations. after eating to shuttle glucose into glycogen and fat stores. After exercise, this process is aided by a reactive upregulation of extra transporters that move glucose from the blood into tissues. All of these mechanisms achieve stable glucose with little variation – if blood glucose doesn’t appear stable, then some of these mechanisms have been overwhelmed (e.g. in type 1 diabetes, where insulin secretion is inadequate, or in type 2 diabetes, where over-filled energy stores and insulin resistance limit the passage of glucose from the blood into tissues for storage or utilisation).
Looking to blood glucose trends to gauge one’s energy stores is akin to waiting until a dam has burst to know that it’s full. It takes a lot of glucose or energy to break the dam – before this point, insulin secretion increases, insulin resistance increases, the rate of fuel utilisation by the mitochondria increases – all in an attempt to maintain stable glucose levels. I don’t think I can emphasise this enough, and will be gladly corrected if wrong, but unless Supersapiens have an incredibly sensitive algorithm working behind the scenes to predict glycogen stores from blood glucose in the absence of measuring insulin levels, I can’t see this metric being accurate.
I love the idea that we can monitor glycogen replenishment, and can see how this would be useful for an elite athlete. My question is – how does a CGM monitor do this? Or, what data are these claims based upon?
The Abbott website references a research thesis by Olssen, studying the glucose trends of four (4) elite swimmers. From the abstract only, the paper concluded that a relationship between meals ingested and blood glucose existed only when assessing 6 days, rather than on a daily basis, and asserted that due to only 4 participants, this data can only encourage future work. Can we derive glycogen from this? No. Other studies? Not proven elsewhere yet. In glycogen storage diseases ? Sure, but how many professional athletes have an inborn error of metabolism?
Personally, I’m excited by the prospect of wearable medical technology being available to a wide range of consumers, athletes and the curious included.
I was really excited to use the Freestyle Libre, but drowned in a sea of data that I concluded might be redundant (due to the MARD..) and a clunky app. As the Supersapiens appears to use the similar hardware (the Libre Sense), I can’t comment on how valid the data will actually be.
Much of my difficulty with the Freestyle Libre was in the data analysis – Excel spreadsheets are complicated and clunky and provided a substantial barrier to analysis (especially realtime analysis!)
If the Supersapiens app has improved its interface, can integrate with performance metrics, and/or has AI features to gain better insights into correlations between glucose and hard measures of performance, more power to it.
But, I can’t escape the fact that a CGM is but one small window into a dark cavern of inner physiological workings. Glucose is an important variable in disease, health and performance, but relying on glucose as a sole metric for these things neglects that blood glucose is affected by many intrinsic (and extrinsic) factors. Blood glucose is even affected by the very factors it tries to draw conclusions upon! Furthermore, the use of CGMs and sports performance are largely untested in the literature- and so while the statements made by companies asserting that CGMs are the next big thing in sports performance, they’re not yet based on available data.
What do you think? Have you tried a CGM, and have you tried the new Supersapiens device? What insights has it given you regarding your training and fuelling?