Abstract
Hypoglycemia remains a risk of exogenous insulin (and insulin secretagogue) therapy for people with insulin-deficient diabetes. Impaired awareness of hypoglycemia (IAH) similarly remains the major modifiable risk factor for severe episodes. 1 Associated with all three phenomena are defects in endogenous counter-regulation and delayed and diminished neurohumoral and symptom responses to induced hypoglycemia, without equivalent lowering of the glucose value associated with cognitive impairment, 2 which is at least partially reversible by avoidance of exposure to glucose values <54 mg/dL. 3,4
Improvements in insulin delivery have often been associated with reduced risk of severe hypoglycemia, although less reliably with restoration of awareness. According to current hypotheses, any therapy associated with reduced exposure to glucose concentrations <54 mg/dL should be associated with improved awareness and improved counter-regulatory responses. Failure of this to happen has been attributed to incomplete avoidance of the key variable and/or longer duration of the insulin deficiency, although it is unlikely that earlier studies of reversibility of IAH completely eliminated low glucose episodes and some included people of diabetes duration of 30 years or more. 3
Against this background, data from the type 1 diabetes clinic exchange reported last year that 30% of people using hybrid closed loop therapy continued to report IAH, and 9% severe hypoglycemia. 5 Even allowing for over-representation of people with problematic hypoglycemia among early accessors of new technology, these are unexpected and somewhat disappointing outcomes.
In this edition of Diabetes Technology and Therapeutics, Flatt et al. report a mechanistic study of the impact of 18 months of automated insulin delivery (AID, at least including AID suspension when the controlling algorithm predicted the occurrence of hypoglycemia based on continuous glucose monitoring [CGM] data), plus “intensive provider support,” our best yet approach to hypoglycemia avoidance short of beta cell replacement therapy, in people with IAH, recurrent severe hypoglycemia, and a mean (type 1) diabetes duration of >30 years. 6 (This commentary uses IAH in preference to “hypoglycemia unawareness” sometimes used in the article, as awareness is variable and complete unawareness is not required for adverse associations.)
The AID was associated with clinical improvements in hypoglycemia, especially reduction in severe hypoglycemia, as well as reduced time spent <56 mg/dL from just over an hour a day to just >4 min and low blood glucose index by >60%. Responses to induced hypoglycemia were measured in a scrupulously conducted clamp protocol, with controlled avoidance of glucose extremes immediately before study, making the hypoglycemic challenge both modest and reproducible.
The magnitude of pancreatic polypeptide and epinephrine responses to the induced hypoglycemia had improved by 6 and 18 months, respectively: associated with reduced peripheral glucose uptake and increased symptom scores by 18 months only. Strictly, the study was negative, in the sense that restimulation of endogenous glucose production (EGP) in response to hypoglycemia was the chosen primary end point and did not occur. In passing, we might use this opportunity to consider the logic of depending on a predecided single outcome for judging the value of a research study. Clearly this study was not, in any important sense, a negative one.
As with all good research, this study answers some questions and raises others. From a clinical perspective, the first positive outcome is further evidence that AID plus health care professional support for people with problematic hypoglycemia (defined as IAH plus at least one severe hypoglycemia in a year) is effective therapy to reduce severe hypoglycemia, including in people with long diabetes duration.
The novelty of the study lies in the mechanistic investigations. Use of technology reduced time spent, including time spent in level 2 hypoglycemia, especially at night and this reduced exposure was perhaps responsible for the recovery of pancreatic polypeptide and eventual improvement of epinephrine responses to experimentally induced hypoglycemia. It remains puzzling that time spent <54 mg/dL on CGM has not been found a reliable marker of hypoglycemia awareness status. 7
Inadequate reduction has been offered as an explanation, but if antecedent hypoglycemia is the cause and maintainer of IAH, then a relationship between time below range and awareness status would be expected. Notwithstanding this study, percentage time spent <54 mg/dL cannot be used as a proxy for awareness status at present.
There were no significant changes in times in or above range, but these were within consensus-recommended targets from baseline. Glucagon responses remained absent throughout. In terms of glucose kinetics, a greater suppression of peripheral glucose uptake was seen by 18 months but no other changes. Stronger autonomic symptom responses were also reported in the clamp only at 18 months (their absence at 6 months was probably an important driver for this study).
Importantly, however, there was no significant improvement in clinical awareness. Clarke scores did improve but this was driven by the avoidance of severe hypoglycemia (items 3 and 4 of the 7-item score). When only the other items, related to subjective awareness, were assessed, there was no significant improvement. 8 This means that protection from severe hypoglycemia by AID is dependent upon functional devices being worn.
Given the changes that occurred between 6 and 18 months, it is possible that even longer use of hybrid closed loop will have incremental effects, but we probably need to look for other solutions. The inclusion of glucagon in pump therapies is of course under investigation and remains a possible avenue for more complete success. 9 Meanwhile, several recent studies show persistence of IAH in ∼30% of technology users 1,5,10 and the exchange clinic registry data need a more urgent solution.
Restimulation of EGP during hyperinsulinemic hypoglycemia is a desirable goal in the management of problematic hypoglycemia, but the present data show this is unachievable with current technology. Failure to restimulate EGP during insulin-induced hypoglycemia in type 1 diabetes is well established, as is its association with absent glucagon responses. 11 Both have been associated with increased c peptide negativity 12 and recovery has been demonstrated by Rickels' group only with islet cell replacement, 13 in keeping with current hypotheses that the alpha cell requires communication from viable beta cells and/or an active counter-regulatory role of c peptide.
Five people (50%) did report at least one severe hypoglycemia in the second 6 months of follow-up, one of whom required medical intervention for two episodes. No treatment is likely to be 100% effective, and we do not know whether these events were associated with malfunction or absence of the technology. However, it is of interest that the authors also report a severe hypoglycemia in a participant who could not engage with the technology. This is an important observation, especially given the high engagement with the technology by those who successfully completed the study.
People who stand to gain benefit from technology, in terms of hypoglycemia avoidance, do not always find it easy—in the RESCUE study, in which CGM was added to insulin pump therapy for people largely with IAH, ∼20% of participants stopped using the CGM, quoting issues such as alarm fatigue and technical problems. 14 In a recent trial of a novel therapy for people with treatment-resistant problematic hypoglycemia, a significant proportion of participants were not using technology at recruitment, despite most stating they had been offered or had tried it. 15 Cognitive barriers to hypoglycemia avoidance such as prioritization of hyperglycemia avoidance and underestimation of the impact of hypoglycemia may be inhibiting success with current technology solutions. 16,17
Could such human factors be underlying the persistence of therapeutic failure in current efforts to eliminate IAH and severe hypoglycemia? The congruence between the 2022 percentage of people experiencing severe hypoglycemia while using hybrid closed loop and the percentage of people contributing most of the severe hypoglycemia in a clinical service back in 2004 18 strongly suggests that more than current educational and technological strategies are needed. Addressing how people think about their hypoglycemia risk and experience may be able to help. 19
What questions then remain to be addressed? The one most pertinent to the hypothesis investigated is what drives the persistent failure of EGP responses to hypoglycemia in people with near optimally managed blood glucose concentrations. Why does an apparently universal persistent failure of elements of counter-regulatory mechanisms allow severe hypoglycemia only in some? Why does not the marked reduction of exposure to glucose concentrations <54 mg/dL achieved with AID here restore clinical awareness of hypoglycemia?
Is it possible that how hypoglycemia avoidance is done is as important as the extent to which it is achieved? The old studies of awareness restoration depended on active engagement of thinking and behavior, although less deliberately than current psychoeducational approaches. We certainly need urgently to define the role of health care professional support in the successful deployment of technologies to prevent hypoglycemia—and if it is essential, work out who needs it and how we are going to provide it to all who do. Meanwhile, will adding glucagon to the therapeutic regimen address the remaining deficits Flatt et al. describe, or do we have to restore all physiological beta cell connectivity?
Will providing extra options—glucagon perhaps and/or help to address cognitions around hypoglycemia—eliminate the residual percentage of people not avoiding hypoglycemia with AID or will we ultimately need cell replacement therapy and a cure for the missing beta cells? HypoRESOLVE (
Footnotes
Author's Contribution
S.A.A. wrote the article and guarantees the integrity of it.
Author Disclosure Statement
S.A.A. has served on advisory boards for NovoNordisk and Medtronic and has spoken at educational events sponsored by NovoNordisk and Sanofi. She is a coinvestigator on the EU-IMI HypoRESOLVE project.
Funding Information
No funding was received for this article.