Response to “Discrepancies Between Blood Glucose and Interstitial Glucose—Technological Artifacts or Physiology: A Reply”

Sirs,

We highly appreciate the letter to the editor by E. Biermann from Munich as a reply to our publication. ¹ Subsequently, we are writing this response letter to address the comments he made one by one.

• Comment 1: “Observations occurring especially under extreme situations like cycling marathon or repetitive hypoglycemia”

Clearly a marathon is an extreme situation; however, many patients with diabetes are interested in exercise, also in strenuous exercise. Differences between glucose levels measured in blood (BG) and interstitial fluid (ISF) show up during or after such physiologically challenging situations to a maximal extent. By means of an intravenous bolus of glucose an extreme situation can be introduced, with maximal glucose differences between both compartments. However, also under not so demanding conditions they show also up, but to a smaller extent. Clearly, such a gap will last for some minutes only. A regular meal will not produce such a temporary disequilibrium.

Like we’ll outline below (and as we have mentioned in our publication), there is a need to evaluate the extent of such differences under controlled conditions in the framework of clinical studies. Only such studies will provide us with clear information about how large the differences between the two compartments are under different physiological and nonphysiological conditions. Without such a detailed evaluation it is not easy to understand what the magnitude and cause for the observed differences is, these might even be different under varying conditions.

• Comment 2: “Caution is required for interpreting the causes without scientific evidence and deriving solutions for therapy planning”

As long as we have no good data about these differences; also between and within patients, this comment is correct; however, visual inspection of real-time (rt) continuous glucose monitoring (CGM) profiles from different patients provides interesting insights into the impact that therapeutic and lifestyle factors have on glucose changes in the human body over time. It is worthwhile to highlight, that with rtCGM for the first time we can see changes in glucose profiles in larger number of patients under different daily life conditions. Such a response letter is not the right place to discuss in detail our thoughts about the physiological background for differences between the two compartments. It was the aim of our article to describe our observations and to initiate a scientific discussion. It appears as if we achieved the goals of our publication.

• Comment 3: “Many studies were published including sampling from deep and superficial lymphatic system draining muscle under basal and super physiologic insulin levels with and without glucose tracers, clamp-studies and mathematical modeling”

It is correct that a number of respective studies have been performed in the past decades with a variety of techniques. However, the number of studies with rtCGM systems with a high analytical performance (= accurate measurements) in humans and/or patients with diabetes focusing on the relevant physiological aspects during dynamic changes in glucose levels, for example, during/after strenuous exercise, is quite small. Dynamic aspects, rapid decrease or increase in glucose levels in one of the two compartments are most often not systematically studied in such studies.

• Comment 4: “Concentration differences between plasma and ISF are dominated by diffusion . . . big differences shown by the authors have not been found”

Biermann is focusing on the absolute concentration difference between plasma and ISF. While the numbers stated are correct, as outlined they were obtained during steady-state conditions and not during dynamic changes in glycemia. Without such changes, no differences show up. These differences most probably depend on a number of factors and might also differs between organs. However, it might also be, that nonphysiological factors are of clinical relevance, for example the algorithms build-in into all commercially available rtCGM systems. Unfortunately, little is known about what these algorithms are doing in detail as these are proprietary knowledge of the manufacturer of the rtCGM systems. In clinical studies it would be worth to differentiate between the contributions of the different factors.

• Comment 5: “Difference between BG and sensor current . . . the correct estimation of the calibration-factor . . . the calibration procedure, done at rest, is potentially corrupted during or after a marathon because of the sensor-tissue-interface”

Changes in body fluids during exercise might have an impact on the calibration factor; however, our knowledge about such changes is scarce. In general, there is a need for a careful assessment of CGM profiles to check the quality of the calibrations. The examples provided in our publication - and also other published data - indicate that calibrations should preferably be performed in situations with low glucose dynamics, that is, without rapid changes in glucose levels. This is also the recommendation in the instructions for use by the manufacturer of rtCGM systems. The reasoning for such recommendations is obvious from our point of view. It would be of interest to study if the calibration factor and / or the calibration procedure is reliable during, for example, strenuous exercise.

• Comment 6: “Interpretation of the authors should be reviewed critically”

While we agree with the critical approach proposed by Biermann; we disagree with his negative statements about the observations we reported. Just because such data do not fit with one’s own expectations, this not a good reason to deny their potential existence. Others should try to falsify the observations described. Until we are proven wrong by a systematic evaluation, no conclusive statements can be made and the possibility that our “hypothesis” is correct is there (and we believe in it obviously).

• Comment 7: “The recommendation that therapeutic decisions should be based on CGM-values and not on BG values . . . appears to be premature”

To our understanding the FDA has accepted rtCGM data for nonadjunctive usage based largely on modelling studies and only partly on clinical studies (these can’t be performed for all clinically relevant questions). In clinical practice patients with diabetes rely on the data that are displayed on their rtCGM handhelds. From our point of view this is fine, assuming that the calibration procedure is performed adequately and a good rtCGM system is used. Patients treated by us in daily practice report better glucose control when they base their therapeutic decisions on rtCGM data (and not on BG data). Clearly these are anecdotal reports; however, as long as Biermann cannot provide data demonstrating that we impair the safety of patients, opinion stands against opinion. Hopefully studies with different rtCGM systems (from different companies) will be performed sooner than later to clarify this interesting topic. Outcome of these studies will also help to make clear if we should change the content of, for example, training courses for patients or of continuing medical education for physicians (and nurses) accordingly.

• Comment 8: “According to those arguments, accuracy would be secondary or even needless”

To be honest we don’t understand the argument of accuracy made by Biermann in the circumstances discussed here. Our point is not about the analytical performance of either measurement system (both are good to our understanding), but the timing of glucose changes when dynamic changes in glucose levels show up. We clearly disagree with his statement; without accurate rtCGM systems the discrepancies could not be observed.

• Comment 9: “It is very important that outliers in extreme situations are reported by frequent users in order to be verified via a scientific analysis”

This last statement shows the fundamental difference in way of thinking; it is not easy to define what are “Outliers” and what “extreme” situations in daily life are, are clinically relevant differences between blood and ISF already observed with extreme meals (is a Big Mac with a Classic Coke such a meal?) or is strenuous exercise “needed” to induce these? As stated above, as long as we have no data from clinical studies that shows that our observation is not correct, we suggest that patients with diabetes using rtCGM systems rely their therapeutic decisions on these data and not on BG data, especially in conditions with rapid changes in glycemia.