HbA1c reduction following flash monitoring commencement is not independently associated with adverse diabetic eye disease outcomes in type 1 diabetes

Introduction Intensification of therapy has been associated with early worsening of retinopathy prior to subsequent risk reduction. We sought to assess whether glycated hemoglobin (HbA1c) reduction, following flash monitoring, was associated with early worsening. Research design and methods An observational study in 541 individuals with type 1 diabetes and paired HbA1c and eye assessment prior to and following flash monitoring commencement. Results Change in HbA1c was −4 mmol/mol (IQR −9–1) (−0.4% (−0.8–0.1)) and 25% achieved a fall in HbA1c of ≥10 mmol/mol. The occurrence of the composite end point (panretinal photocoagulation, macular laser or anti-VEGF therapy) was associated with baseline HbA1c >75 mmol/mol (9.0%) (HR 4.0 (95% CI 2.0 to 7.9), p<0.001) but not with fall in HbA1c of ≥10 mmol/mol (0.9%) (HR 1.6 (95% CI 0.8 to 3.2), p=0.203) over a follow-up period of 615 days (527–863). In multivariate analysis, diabetes duration (p=0.035) and prior retinopathy (p<0.001) were most predictive of the composite end point. Baseline HbA1c was the strongest predictor of worsening retinopathy (p=0.002) or new retinopathy (p=0.002) in multivariate analysis whereas change in HbA1c was not independently associated with either (p=0.930 and p=0.830, respectively). Conclusions Progression of eye disease is associated with baseline HbA1c, diabetes duration and previous retinopathy and such individuals should be monitored during intensification of glycemic therapy. Reassuringly, the extent of glucose lowering does not appear to be an independent risk factor for early worsening of eye disease in this context.


INTRODUCTION
Flash glucose monitoring provides users with an interstitial glucose value only on scanning a glucose sensor with a reader device or compatible mobile phone. In other respects, it is similar to conventional continuous glucose monitoring (CGM) in providing a 24 hours glucose trace and trend arrows. 1 We have previously shown that flash monitoring is associated with clinically important reduction in glycated hemoglobin (HbA1c) in type 1 diabetes, particularly in people with above target HbA1c at baseline. 2 Flash monitoring is also known to reduce hypoglycemia and glucose variability in people with HbA1c ≤58 mmol/mol (7.5%) prior to commencement. Intensification of glycemic control reduces the long-term risk of microvascular complications in type 1 diabetes, 3 although, in the Diabetes Control and Complications Trial (DCCT), an early worsening of diabetic retinopathy (EWDR) was observed before longterm, sustained risk reduction accrued. 4

Significance of this study
What is already known about this subject? ► Glycated hemoglobin (HbA1c) lowering has been associated with early worsening of diabetic retinopathy before risk reduction occurs.
What are the new findings?
► Baseline HbA1c and diabetes duration were associated with subsequent need for intervention for diabetic eye disease and with the development or worsening of diabetic eye disease. ► Change in HbA1c was not greater in those with subsequent panretinal photocoagulation, worsening of retinopathy or new development of retinopathy. ► When stratified based on HbA1c response, there were no differences in any eye outcomes when adjusted for baseline HbA1c.
How might these results change the focus of research or clinical practice?
► People with markedly elevated HbA1c, prior retinopathy and long diabetes duration require careful monitoring after commencement of flash monitoring, but these data offer reassurance that extent of HbA1c lowering may not be a major contributor to risk.

Clinical care/Education/Nutrition
It is possible, however, that the method of intensifying glycemic control may influence the risk of early worsening of diabetic eye disease. An analysis of recent CGM randomized controlled trials suggests that intensification using CGM is associated with a significantly lower risk of hypoglycemia compared with self-monitored blood glucose (SMBG). 5 Improved glycemic control, in the context of flash monitoring, is likely to be associated with lower rates of hypoglycemia and less glucose variability than was previously possible. 6 Consequently, we hypothesized that improvement in glycemic control, following commencement of flash monitoring, may not independently predict early worsening of diabetic eye disease. To test this, we prospectively assessed the need for diabetic eye disease intervention (panretinal photocoagulation (PRP), macular laser or intravitreal antivascular endothelial growth factor (VEGF) therapy) or progression/development of diabetic eye disease in a cohort of individuals commencing flash monitoring, with particular reference to achieved fall in HbA1c.

PARTICIPANTS AND METHODS Study design and participants
We conducted a prospective observational study of the first 589 individuals with type 1 diabetes commenced on National Health Service (NHS) funded flash monitoring (Freestyle Libre, Abbott, Witney, UK) in a University hospital clinic (Royal Infirmary of Edinburgh) during February and March 2018. Glycemic outcomes for this cohort (including a group from another clinic within our center) have previously been reported, where the baseline characteristics are described in detail. 2 In this current study, 14 individuals were excluded due to death or moving from the hospital catchment area, 14 were excluded due to absence of paired HbA1c data and a further 20 were excluded due to absence of paired eye outcome data, leaving a total cohort of 541 individuals (online supplemental figure). Two hundred twelve (39.2%) individuals had self-funded flash monitoring use prior to 2018 and provided their commencement date on a questionnaire. The study was entirely observational (with no deviation from standard clinical care) and ethics approval was not required.

Outcomes
The primary outcome was an assessment of factors associated with the development of a composite end point comprising PRP, macular laser therapy or anti-VEGF therapy. Additional outcomes of interest were the individual components of the composite outcome, new onset of retinopathy, new onset of maculopathy and worsening of retinopathy. Retinopathy was classified as: none, mild background, intermediate (any eye disease between mild background and proliferative, including all non-proliferative retinopathy and people with previous PRP and stable eye disease) or proliferative. Worsening retinopathy was defined as at least one step up through these categories, with the most advanced category being recorded for individuals when a discrepancy existed between eyes. Retinopathy and maculopathy data were obtained from the national diabetic retinopathy screening programme, 7 which is accessible via our national clinic database system, SCI-Diabetes. In individuals attending specialist eye clinics, data on severity of eye disease and treatments administered were obtained from our hospital's electronic health records. All individuals had a first eye assessment following flash monitoring with a further 'final' assessment available in 397/541 (73.4%) (online supplemental figure). Change in HbA1c was defined as the difference between HbA1c prior to commencement of any flash monitoring and the next available value after the flash monitoring education session (and change from baseline to the final available HbA1c). We also report the proportion of individuals achieving the Scottish HbA1c target (<58 mmol/mol (7.5%)). HbA1c was measured by ion-exchange high performance liquid chromatography using the Arkray Adams A1c automated platform (A. Menarini Diagnostics) and is typically measured every 6 months in people attending our clinics. Mode of insulin delivery (multiple daily injection (MDI) or continuous subcutaneous insulin infusion (CSII)), smoking status and urinary albumin status were obtained from SCI-Diabetes. Self-reported hypoglycemia data were available in 356/541 (65.8%) individuals from clinic questionnaires (including Gold score and a modification of the Clarke assessment 8 ) completed within a year of commencing flash monitoring.

DISCUSSION
We have demonstrated that HbA1c lowering, in the context of flash monitoring commencement, is not independently predictive of clinically important changes in diabetic eye disease in the short-term. HbA1c prior to commencement of flash monitoring, as well as diabetes duration and pre-existing retinopathy were all independently associated with progression of eye disease.   Intensification of glycemic management in type 1 diabetes is unequivocally associated with reduction in the development and progression of diabetic retinopathy, 3 9 10 but those randomized to intensive glycemic control in the DCCT experienced EWDR prior to accruing substantial and sustained benefit in the longer term. 4 A number of potential pathophysiological mechanism have been posited to explain this phenomenon, including perturbations of the somatotropic axis, increased retinal concentration of VEGF and changes in other angiogenic growth factors. 11 However, the evidence in support of these proposed mechanisms is, at best, mixed.

Clinical care/Education/Nutrition
We hypothesized that advances in the management of type 1 diabetes since the early 1980s (eg, modern CSII, insulin analogues, aggressive blood pressure management and ACE inhibitor prescribing) and specific features of intensification relating to flash monitor use may have reduced the risk of early worsening of eye disease following HbA1c reduction. Randomization to intensive glycemic control in DCCT was associated with a threefold increase in severe hypoglycemia while we did not, as previously reported, observe an increase in severe hypoglycemia following commencement of flash monitoring. 2 RCT evidence attests to reduction in hypoglycemia (and glucose variability) during flash monitoring use 6 in individuals with on-target HbA1c at baseline and similar findings have emerged from a large French observational study with respect to severe hypoglycemia. 12 Analysis of data from two recent real-time CGM studies, in MDI users, has suggested a significant attenuation of the increased risk of hypoglycemia as HbA1c levels fall, in contrast to those using SMBG. 5 glycemic variability is another potential contributor towards risk of diabetes complications 13 and has been associated with structural damage to the neuroretina in type 1 diabetes based on CGM measures of variability, 14 while visit-to-visit variability in HbA1c has been independently associated with retinopathy progression in adolescents with type 1 diabetes. 15 Lending support to the potential importance of reduced glucose variability and hypoglycemia, early worsening of retinopathy has not been reported in the context of HbA1c reduction following islet 16 17 and pancreas transplantation. 18 Islet transplantation is associated with both reduced rates of hypoglycemia and lower glucose variability. 19 Similarly, no significant progression of retinopathy was observed in a study of people with type 1 diabetes following CSII commencement. 20 The cohort described in our study is different from the DCCT cohort in a number of important respects: older age (46 vs 27 years), longer duration of diabetes (23 vs 6 years) and higher prevalence of prior diabetic eye disease intervention (12.9% vs none in DCCT). Baseline Clinical care/Education/Nutrition HbA1c (63 mmol/mol (7.9%) vs 76 mmol/mol (9.1%)) was lower in our study, although baseline HbA1c in those with a 10 mmol/mol (0.9%) or greater fall in HbA1c (72 mmol/mol (8.7%)) was similar to the DCCT cohort. Our cohort is broadly representative of typical clinical practice in the UK, although, as previously described, it is slightly skewed toward lower than average HbA1c, younger age, CSII use and lower socioeconomic deprivation than our total clinic population. 2 The study is open to the usual criticisms of observational methodology particularly the potential influence of unmeasured confounders. As a 'real-world' assessment, the timing of HbA1c measurement and eye assessment was not uniform and reflected the expected variation in normal clinical practice. A minority of individuals did not have an eye assessment within 1 year of flash monitoring commencement, meaning some early worsening may have been missed. Lack of uniformly timed HbA1c and eye assessments limits our ability to comment on the potential influence of rate of change in HbA1c. However, we have no reason to suspect variation in follow-up intervals introduced any systematic bias. An advantage of this 'real-world' methodology is its likely generalizability, in the absence of stringent exclusion and inclusion criteria, to modern diabetes clinic populations in the UK and beyond. The DCCT 3 represents a landmark in methodologically robust evidence gathering in type 1 diabetes but the treatment options and risk factor management for the condition have evolved substantially since the 1980s and, to our knowledge, our study represents the largest assessment of factors associated with early worsening of eye disease in the context of modern intensification of glycemic control. A significant limitation of our study is the absence of detailed Gold Standard evaluation of retinopathy, typically considered to be the Early Treatment Diabetic Retinopathy Study (ETDRS) classification system. 21 We relied on clinical data from our national screening programme 7 and electronic health record entries from specialist eye clinic evaluation, which do not provide the same level of granularity as the ETDRS classification system. It is conceivable, therefore, that we have under-reported subtle changes in retinopathy. However, by relying on clinical data, we feel it is unlikely that we have missed clinically important changes and have also reported unambiguous hard end points (PRP, macular laser and anti-VEGF therapy). It is possible that we have missed an association between HbA1c lowering and EWDR because the degree of HbA1c reduction observed in this cohort is below the threshold at which this effect occurs, although this seems unlikely as none of the reported multivariate analyses of HbA1c reduction came close to approaching statistical significance. However, these data cannot exclude the possibility that more extreme reduction in HbA1c increases the risk of EWDR. It could also be argued that our results reflect a type 1 error, in refuting the independent association Clinical care/Education/Nutrition between glucose lowering and early EWDR, due to insufficient cohort size or events. However, in multivariate analysis of the composite end point, new development of retinopathy and worsening of retinopathy, the absence of association was clear, consistent and not close to approaching statistical significance. The study benefits from comprehensive follow-up data and a wide range of associated clinical parameters derived from our national diabetes database, however, we were not able to report information on ACE inhibitor prescribing (and other antihypertensives) and pregnancy which would have been of interest in the context of diabetic eye disease.

CONCLUSIONS
Elevated HbA1c, longer duration of diabetes and preexisting retinopathy are all significant risk factors with respect to diabetic eye disease in people with type 1 diabetes commencing flash monitoring. However, in the short-term, subsequent change in HbA1c does not appear to independently predict the risk of retinopathy development or progression nor the need for interventions to treat advanced eye disease in a representative cohort of flash monitoring users. These single-centre observational findings, while offering a degree of reassurance, clearly require corroboration from larger national datasets and randomized controlled trials of novel glucose-lowering technologies. Understanding the risk of abrupt reduction in HbA1c, as well as the potential moderating influence of hypoglycemia and glucose variability, will become increasingly important as we approach the era of closedloop insulin delivery.
Contributors All authors were involved in the study design. Data collection and study design was by RA. Data analysis and interpretation was by FWG. The document was drafted by FWG and all authors were involved in subsequent revisions and final approval. The study guarantor is FWG.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests FWG has received speaker fees from Abbott Diabetes Care. ARD has received speaker fees from Abbott Diabetes Care.
Patient consent for publication Not required.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.
Open access This is an open access article distributed in accordance with the Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits others to copy, redistribute, remix, transform and build upon this work for any purpose, provided the original work is properly cited, a link to the licence is given, and indication of whether changes were made. See: https:// creativecommons. org/ licenses/ by/ 4. 0/.