Background: Hemoglobin A1c (HbA1c) is an unreliable glycemic marker in the dialysis population, and alternative methods of glycemic monitoring should be considered. Continuous glucose monitoring (CGM) measures interstitial glucose, an indirect measure of plasma glucose, and allows for estimating mean sensor glucose, glucose variability, and time in ranges. Thus, CGM provides a more nuanced picture of glycemic variables than HbA1c, which only informs about average glucose and not variation in glucose or hypoglycemia. Summary: In non-dialysis patients with type 1 and type 2 diabetes, CGM metrics are increasingly used to estimate glycemic control and are associated with improvements in glucose levels. Although a clear link has not yet been established between some CGM variables and the development of late diabetic complications, CGM use could be an important step forward in improving glycemic control in patients receiving dialysis. The ability to detect and prevent hypoglycemia while optimizing glucose levels could be particularly valuable. However, long-term CGM use has not been evaluated in the dialysis population, and the practical burden and cost associated with CGM use may be a limitation. We discuss the strengths and limitations of using CGM in the dialysis population with type 1 and type 2 diabetes. Key Messages:CGM circumvents the pitfalls of HbA1c in dialysis patients and provides detailed measures of the mean sensor glucose, glucose variability, and time in ranges. Guidelines recommend a minimum of 50% time spent in the target range (3.9–10.0 mmol/L) and less than 1% below range (<3.9 mmol/L) for patients receiving dialysis but remain to be evaluated in the dialysis population. CGM can be a valuable tool in reducing overall glucose levels and variations while detecting hypoglycemia, but the practical burden of CGM use and cost may be a limitation.
Hemoglobin A1c (HbA1c) was first identified more than 50 years ago and correlated to the mean plasma glucose of the previous 3–4 months [1, 2]. For both the dialysis and the general population, HbA1c is widely used for glycemic monitoring and achieving a glycemic target set out by the clinician. However, for patients receiving dialysis, studies have established that HbA1c underestimates plasma glucose levels when compared with plasma glucose assessed by continuous glucose monitoring (CGM), glycated albumin, or fructosamine [3‒5]. This has led to recent guidelines from Kidney Disease: Improving Global Outcomes (KDIGO), emphasizing HbA1c’s unreliability in the dialysis population and recommending the use of CGM, particularly for people treated with glucose-lowering medication with a high risk for hypoglycemia .
A second challenge for using HbA1c in the dialysis population is the undetermined optimum for HbA1c in terms of limiting the progression of late diabetes complications. Major studies exploring the association between HbA1c levels and mortality in the dialysis population with diabetes show conflicting results [7, 8]. KDIGO guidelines provide only an HbA1c target for chronic kidney disease stages 1–5 but state these targets do not apply to the dialysis population as the optimal HbA1c target is unknown .
CGM represents a different approach to glycemic monitoring where glycemic levels are measured continuously and provide mean sensor glucose, including metrics on glucose variability and time in ranges [9, 10]. In recent years, CGM technologies have undergone rapid development with improvements in accuracy and usability. This paper aimed to review recent CGM studies of the general diabetes population and the dialysis population to discuss if the use of CGM can improve glycemic control in patients receiving dialysis.
CGM and Glycemic Control
CGM measures interstitial glucose levels by a subcutaneous sensor measuring glucose levels every 5–15 min. Interstitial glucose equilibrates with plasma glucose within minutes, whereby sensor data becomes an indirect measure of plasma glucose. The sensor is attached to a transmitter that sends sensor data to a receiver unit (Fig. 1) either in real-time (open CGM) or assessed retrospectively (blinded CGM). Some models contain alarms for high or low glucose levels, aiding the user in the early detection of glucose excursions. Not all models are approved for use in the dialysis population by the manufacturers (Table 1) as their validity in dialysis patients has not yet been examined [11‒17]. CGM sensor data are provided in a user-friendly format (ambulatory glucose profile report), presenting the mean sensor glucose, glucose variability (coefficient of variation and standard deviation), and time in ranges (Fig. 2). Time in ranges is defined as the percentage of time spent in each of five ranges: time below range level 1 (3.0–3.8 mmol/L) and level 2 (<3.9 mmol/L); time in the target range (3.9–10.0 mmol/L); and time above range level 1 (10.1–13.9 mmol/L) and level 2 (>13.9 mmol/L).
CGM is widely used in type 1 diabetes and, to a lesser extent, in type 2 diabetes. Studies in patients with type 1 diabetes have demonstrated that using real-time CGM to follow plasma glucose improved glycemic control with fewer hypoglycemic events than blood glucose monitoring (BGM) measured by finger prick . CGM use in type 2 diabetes is less thoroughly investigated. A recent trial (n = 175, follow-up of 8 months) of patients with type 2 diabetes treated with basal insulin randomized participants to use either CGM or BGM. CGM use was, after 8 months, associated with a mean 59% time in target range compared with 43% in the BGM group (p < 0.001). This improvement for the time in the target range was not associated with an increase in severe hypoglycemic events, with just one incident occurring in each group . These findings correspond with a retrospective study (n = 36,080) where CGM use was associated with better glycemic control and lower incidence of hypoglycemia-related hospitalization than diabetes patients not using a CGM .
Studies on the correlation between various CGM metrics and late diabetic complications are limited and warrant further research. A post hoc analysis of the Diabetes Control and Complications Trial study (n = 1,440) explored the association between time in the target range (3.9–10.0 mmol/L) and the development or progression of retinopathy and microalbuminuria . The mean time in target range was 41% assessed by a seven-point profile from BGM. The study found that for each 10% reduction of time in target range, the hazard ratio for progression in retinopathy was increased by 64% (95% confidence interval 51–78) and for microalbuminuria by 40% (95% confidence interval 25–56). Similar findings were found in a 12-month prospective study of patients with type 1 diabetes and elevated albuminuria, randomized to a sensor-augmented insulin pump, where an increase in CGM-measured time in the target range was significantly associated with a decrease in albuminuria .
Experience with CGM in the Dialysis Population
Whether CGM use can improve glycemic control in the dialysis population remains undetermined. Two small studies have used CGMs to adjust the antidiabetic treatment and reported changes in mean sensor glucose. The first study used CGM for 54 h at baseline in 28 hemodialysis patients, and readings were used to adjust insulin treatment. At 3-month follow-up, without using CGM, mean sensor glucose values significantly decreased from 9.9 (±1.9) to 8.9 (±2.1) mmol/L (p = 0.05) . The second study, with 15 patients receiving hemodialysis, used BGM for a 6-week period (measured three times daily), followed by 6 weeks with an intermittent CGM (measured 5 days every 2 weeks). Treatment was adjusted according to the measures from the BGM or CGM readings. The main finding was that CGM, unlike BGM, was associated with an improvement in mean sensor glucose level. Mean sensor glucose was 8.3 (±2.5) mmol/L at baseline, 8.2 (±1.6) mmol/L at the end of the BGM period, and 7.7 (±1.6) mmol/L at the end of the CGM period (p < 0.05) . In both studies, the reduction in mean sensor glucose was not associated with an increased risk of hypoglycemic events. These two studies indicate that the CGM may aid the clinician in improving overall glycemic levels in dialysis patients.
An important quality of CGMs is their ability to detect episodes of hypoglycemia. Studies in patients with diabetes (not on dialysis) indicate that hypoglycemia may have proarrhythmogenic effects on the heart via sympathoadrenal stimulation and hypokalemia . However, the interplay between hypoglycemia and cardiac arrhythmias is not yet understood, and whether the use of CGM can prevent potentially lethal cardiac arrhythmias as well as other cardiovascular complications in the dialysis population is yet to be determined.
Current Guidelines for CGM in the Dialysis Population
In 2019, an international consensus report was published outlining targets for the time in ranges both for the general diabetes population (including type 1 and 2 diabetes) and a second group including older and high-risk diabetes patients . This group includes patients with a high risk of complications, requiring assistance, or having comorbid conditions (including renal disease, cognitive deficits, joint disease, osteoporosis, fracture, and cardiovascular disease). In this high-risk group, which includes dialysis patients, the goal is to maintain:
more than 50% time within the target range (3.9–10.0 mmol/L) (>12 h per day)
less than 1% time below 3.9 mmol/L (<15 min per day)
less than 50% time above 10.0 mmol/L (<12 h per day)
less than 10% time above 13.9 mmol/L (<2 h and 24 min per day)
For the general diabetes population, the aim is to achieve more than 70% time in the target range, with up to 4% allowed below 3.9 mmol/L. Compared with the general population, the CGM-based targets for the high-risk group have focused on reducing time below the range and accepting less time in the target range, consequently meaning an increased mean plasma glucose level. The purpose of these recommendations is to avoid hypoglycemia. The guidelines argue that older and high-risk individuals are at increased risk of hypoglycemia due to greater prevalence of hypoglycemia unawareness, including physical and cognitive impairment . However, these guidelines represent a compromise in terms of minimizing the risk of hypoglycemia by accepting higher mean sensor glucose and have not been specifically evaluated in patients undergoing dialysis. Whether a minimum of 50% time spent in the target range provides sufficient glycemic control to prevent late diabetic complications is unknown.
Clinical Implementation of CGMs in the Dialysis Population
Overall, the use of CGM metrics in dialysis patients is a relatively new approach, and recommendations are not well-established in daily clinical practice. CGM comprises a range of practical challenges. Ideally, the user applies and replaces the sensor after each period (7–14 days) and interprets and acts appropriately to sensor data. However, handling of the sensor, transmitter, and software, on, for instance, a smartphone, may constitute a problem for some of the most fragile dialysis patients. These practical issues could limit the clinical implementation of CGM in the dialysis population and conducting large-scale CGM studies. Alternatively, periodic use of CGMs could be an approach that would reduce the practical burden on dialysis patients and reduce costs. Whether the periodic use is inferior or similar to daily use of CGMs is undetermined but could represent a realistic approach to wider CGM use. For example, a feasible approach could be periodic CGM use every 4–6 months, where dialysis staff could apply CGM with subsequent analyzes of CGM readings and adjustments of the antidiabetic treatment by qualified medical staff. For some dialysis patients, daily use may be appropriate. Both daily and periodic use could likely aid in avoiding hypoglycemia and reduce glucose variability while improving mean sensor glucose.
Limitations of CGMs
CGMs carry an economic burden with prices varying between CGM models and countries. As an example, daily sensor use for 1 year of Dexcom G6 (Fig. 1) in Denmark is 1,822 EUR. The annual CGM cost will be a limitation for wider implementation. Second, the practical challenges of sensor insertion and removal are also a likely limitation. Future studies are warranted and should assess if daily or periodic CGM use will improve glycemic control effectively, compared with standard-of-care, and if CGMs can aid in reducing the incidence of hypoglycemia and diabetic complications in the dialysis population.
HbA1c underestimates plasma glucose in dialysis patients and provides no information on glucose variability or hypoglycemia. CGM gives a detailed analysis of the glucose profile, including metrics on mean sensor glucose, glucose variability, and time in ranges. In non-dialysis patients with type 1 and type 2 diabetes, CGM use is shown to lower mean glucose levels while also reducing the incidence of hypoglycemia. Minor studies in dialysis patients also indicate a beneficial effect of CGM use. Implementation of CGMs, either daily or periodically, could improve glycemic control in the dialysis population. Concerns do remain on the practical burden imposed by CGM use on the most fragile dialysis patients and the cost associated with wider CGM implementation.
Statement of Ethics
An ethics statement is not applicable because this study is based exclusively on published literature.
Conflict of Interest Statement
P.R. received funding for his institution from AstraZeneca for participating in the steering committee for DAPA-CKD. He has received funding to his institution for advisory boards from AstraZeneca, Sanofi Aventis, and Boehringer Ingelheim; from Bayer, Gilead, and Novo Nordisk for steering committees; from Novo Nordisk, Bayer, and Eli Lilly for lectures; and has received grants from Novo Nordisk. K.N. received funding for her institution for participating in advisory boards from Medtronic and Novo Nordisk and for lecturing from Sanofi, Novo Nordisk, Medtronic, and Dexcom. Her institution received funding for studies she performed from Zealand Pharma, RSP Systems, Novo Nordisk, Medtronic, and Dexcom. M.H. received funding for participating in advisory boards from Bayer and AstraZeneca. Remaining authors declare no conflict of interest.
Tobias Bomholt and Dea Kofod drafted the paper and Kirsten Nørgaard, Peter Rossing, Bo Feldt-Rasmussen, and Mads Hornum all contributed significantly to the paper revision.