Search

independent predictorof diabetic vascular complications [1]. This conclusion isderived essentially from retrospective epidemiological ana-lyses with hard (microvascular) endpoints [2] and fromobservational and interventional studies with surrogateendpoints (oxidative stress and endothelial function, assessedwith flow-mediated dilation) [3,4]. Clearly, the correlationfound between glycaemic variability and microvascularcomplications/oxidative stress does not necessarily indicateacause–effect relationship; furthermore, the effect ofglycaemic instability on macrovascular disease is not known.As a consequence, whether glycaemic variability is causallylinked to diabetic complications and, in particular, to macro-vascular dysfunction, remains an open question.In their elegant paper, Horvath et al. [5] investigatedboth the effect of glucose fluctuation on endothelialfunction (a surrogate marker of macrovascular complica-tions) and the possible mechanisms linking glucose‘swings’ to vascular damage. In particular, they comparedthe effect of glycaemic variability on endothelial dysfunc-tion in four rat models: poor glycaemic control (PGC),good glycaemic control, untreated diabetic (UD) and non-diabetic. Interestingly, although the ‘global glycaemicexposition’, mathematically expressed by the AUC of their48 h glycaemic profiles and correlated with 48 h meanglucose level [6], was significantly higher in UD rats thanin PGC rats, endothelial dysfunction was greater in thePGC rats. This finding supported their conclusion that ‘theoscillation of blood glucose level itself is a more importantcontributor to the loss of endothelial function in diabetesthan the average level of mean blood glucose level per se’.In our opinion, this conclusion is not completely supportedby the data presented. In fact, the different variability of theglycaemic profiles of PGC versus UD rats is onlypresumptive and is not supported by an objective math-ematical analysis, i.e. comparison of one or more indices ofvariability (mean amplitude of glycaemic excursions[MAGE], standard deviation, continuous overall net gly-caemic action [CONGAn] and others) [6] between the twogroups. If the indices of variability for PGC and UD ratswere similar, then this would indicate that the betterendothelial function of UD rats might be related to anunexplored variable. Therefore, the conclusion should stateonly that mean blood glucose is not enough to completelyexplain vascular dysfunction.

Eight insulin-treated diabetic patients in good glycaemic control were studied as out-patients with frequent determinations of stable glycosylated haemoglobin (HbA1c) before, during and after 1 week of induced poor glycaemic control. Stable HbA1c was determined by cation exchange chromatography after elimination of the labile fraction by incubation in saline (0.15 mol/l). The increase in mean blood glucose was significant on the first day of reduced insulin therapy and greatest after 1 week (6.9 +/- 3.9 mmol/l above basal values). Stable HbA1c increased significantly on day 7 of the reduced insulin treatment. The increase represented, on average, 0.009% of total haemoglobin per mmol/l increase in mean blood glucose per 24 h during the period of induced hyperglycaemia. After restoring insulin therapy, a significant decrease in blood glucose was achieved on day 1 and after 2 days, the blood glucose level was similar to before the study. There was no significant decrease in stable HbA1c within the first 2 weeks of improved glycaemia.