Time to Plan for Continuous Glucose Monitoring in Dialysis-Dependent Kidney Failure

More than half of individuals with dialysis-dependent kidney failure have diabetes, yet assessing their glycemic control is challenging.¹ Dialysis-related factors render traditional measures of glycemia, such as glycated hemoglobin (HbA1c), unreliable.² As dialysis-dependent kidney failure increases susceptibility to severe hyperglycemia and hypoglycemia, both of which associate with adverse outcomes, more accurate measures of glycemia are needed. In the general diabetes population, continuous glucose monitoring (CGM) improves glycemic control and reduces hypoglycemia. CGM is standard of care for type 1 diabetes, and the American Diabetes Association (ADA) now recommends CGM for all individuals treated with insulin, irrespective of diabetes type.³ Standardized CGM metrics, such as time in range (TIR), correlate with HbA1c. The ADA recommends targeting ≥70% TIR between 70 and 180 mg/dl for most adults living with diabetes because it aligns with a HbA1c target of approximately 7%.³ Because these metrics provide a more accurate assessment of glycemia in settings where HbA1c is unreliable, they may be particularly valuable in the setting of dialysis-dependent kidney failure. In fact, a recent CGM pilot study revealed substantial undiagnosed hyperglycemia among individuals receiving dialysis with “burnt-out diabetes” (HbA1c <6.5% without glucose-lowering therapy for >6 months).^2,4 This study highlights the potential of CGM for detecting unrecognized glycemic variability in individuals receiving dialysis, but more population-specific evidence is needed.

In this issue of JASN, de Boer et al. present findings from the BLOod Sugar Sensing On Maintenance dialysis (BLOSSOM) study, the largest prospective study of CGM in people treated with dialysis to date.⁵ The BLOSSOM study obtained CGM data from 420 adults with dialysis-dependent kidney failure (hemodialysis, n=365, peritoneal dialysis, n=55) with or without diabetes. All individuals receiving maintenance dialysis at a Northwest Kidney Center–operated clinic with the capacity to provide informed consent were eligible to participate. Outcomes were assessed by dialysis modality and across three diabetes subgroups: no diabetes, untreated diabetes, and treated diabetes.

The findings revealed a wide range of glycemic control and a high prevalence of CGM-detected hypoglycemia regardless of diabetes status. Time in tight range (TITR; 70–140 mg/dl) was used to evaluate glycemic control among the 157 (37%) participants without diabetes, as glucose levels >140 mg/dl should be uncommon in this group. Only 68% achieved ≥70% TITR, with lower rates observed in peritoneal dialysis (31%) compared with hemodialysis (75%). By contrast, 99% of the reference population, composed primarily of dialysis clinic personnel without diabetes, achieved TITR ≥70% (mean TITR was 97%), underscoring the profound impact of kidney failure on glucose homeostasis independent of prior diabetes status. Notably, participants without diabetes experienced the most sustained hypoglycemia among all subgroups, though the clinical implications of this finding are uncertain.

Among the 88 (21%) participants with untreated diabetes, the majority achieved the ADA-recommended TIR, with an average TIR of 71% for hemodialysis and 70% for peritoneal dialysis. In this subgroup, the CGM-derived mean blood glucose was approximately 160 mg/dl, (corresponding to an estimated HbA1c approximately of 7%), suggesting reasonable glycemic control. However, measured HbA1c substantially underestimated glycemia (average HbA1c 5.7%), reinforcing the inaccuracy of HbA1c in the dialysis population.

By contrast, glycemic control was poor among the 175 (42%) participants with treated diabetes. The mean TIR was 44% for individuals receiving hemodialysis and 33% for individuals receiving peritoneal dialysis. Only 25% and 8% of people receiving hemodialysis and peritoneal dialysis, respectively, achieved TIR ≥70%. It is worth noting that the ADA suggests a less stringent target of ≥50% TIR for older adults, but few people receiving dialysis with treated diabetes achieved even these more liberal targets. Despite an average HbA1c of 7.0% for individuals receiving hemodialysis and 7.4% for those receiving peritoneal dialysis, TIR ≥50% was achieved by only 40% and 25% of people receiving hemodialysis and peritoneal dialysis, respectively. Even though the high glucose content of peritoneal dialysis fluid is known to contribute to hyperglycemia in people with diabetes, the extent and severity of hyperglycemia observed is alarming. These findings underscore both the importance of close glycemic monitoring among individuals receiving peritoneal dialysis and the limitation of HbA1c as a marker for glycemic control in the dialysis population. Moreover, they highlight the urgent need for improved diabetes care strategies in the dialysis setting.

Strengths of the BLOSSOM study include the large sample size, pragmatic design, and inclusion of people across diabetes status and dialysis modality. When applying the findings to the nonstudy setting, it is important to consider that the US dialysis population includes a higher proportion of Black and a lower proportion of Asian individuals compared with the population studied.⁶ Similarly, regional differences in diabetes care and lifestyle may influence glycemic patterns, necessitating additional investigation into glycemic control in more diverse dialysis cohorts. Furthermore, uncertainty about CGM accuracy remains because of limited information about the correlation between CGM-detected glycemic excursions and capillary blood glucose measurements. Only a small subset of BLOSSOM participants (n=12) performed self-monitored capillary blood glucose testing, and information about the timing of measurements relative to dialysis sessions was not provided.

Despite these uncertainties, accumulating evidence, including that from the BLOSSOM study, supports the overall accuracy of CGM in the dialysis population^7,8 and suggests that routine CGM use may be indicated for at least some individuals receiving dialysis. Although consensus glycemic targets for the dialysis population are lacking, observational data show associations between extreme hyperglycemia and higher risks of cardiovascular and all-cause mortality.¹ In BLOSSOM, CGM-measured glycemic control was poor among participants treated with glucose-lowering medications. As such, CGM could play an important role in identifying individuals treated with glucose-lowering medications who might benefit from intensified diabetes management. This may be particularly important in the setting of peritoneal dialysis, with CGM providing early warning that prescription adjustment or modality change may be needed. Second, individuals at risk for life-threatening hypoglycemia and/or diabetic ketoacidosis, such as individuals with type 1 diabetes, may benefit from CGM's close surveillance. Individuals with type 1 diabetes were enrolled in BLOSSOM, however, their data were not analyzed separately. Dedicated studies in this subgroup are warranted. However, the extreme risk of hypoglycemia in people receiving dialysis who are treated with insulin renders it reasonable to consider adopting the ADA recommendation of use of CGM in all people using insulin regardless of diabetes type.

While the BLOSSOM research team successfully applied CGM to participants in dialysis clinics, the transferability of this approach to the nonresearch setting should not be assumed. Successful CGM implementation in the clinical setting requires infrastructure, such as support for prior authorizations, detailed documentation of need, and prescription through durable medical equipment suppliers. In endocrinology clinics, where CGM is most often prescribed, specially trained personnel perform these tasks. CGM users also require device training and support for device issues, resources available in endocrinology but not dialysis clinics. However, <10% of people with diabetes and dialysis-dependent kidney failure receive care from endocrinologists,⁹ posing a potential barrier to CGM access. Encouraging endocrinology referral is one strategy for overcoming this challenge, but it would introduce more medical providers and appointments to individuals with already burdensome care. As such, CGM prescription and application in dialysis clinics may be necessary to achieve widespread uptake of CGM.

Implementation studies that consider the dialysis context will be critical to the success of CGM. A recent Danish qualitative study found that CGM use increased insulin adjustment during hemodialysis and enhanced patient engagement.¹⁰ The eight participating hemodialysis nurses cited the centrality of staff and patient training to the successful integration of CGM into routine dialysis care.¹⁰ Elucidating the perspectives of US dialysis medical providers and clinic personnel on CGM and their roles in the prescription and monitoring of CGM is essential. Patient acceptability must also be examined. CGM systems alarm for hyperglycemia and hypoglycemia. The alarms are clinically important, but they can increase stress and burden, particularly for people with persistent hyperglycemia, as observed in BLOSSOM's peritoneal dialysis subgroup. Another important consideration when implementing CGM in the dialysis setting are the data that CGM generates. As BLOSSOM demonstrates, CGM will inevitably reveal problematic hyperglycemia, but who should be responsible for this information? Innovative diabetes care models, such as dialysis clinic–embedded diabetes care and education specialists or virtual endocrinology consults, may be necessary. Although the need to improve diabetes care in the dialysis population has long been acknowledged, CGM could serve as the catalyst for improved collaboration between nephrology and endocrinology.

The message is clear: It is time to bring CGM to individuals with dialysis-dependent kidney failure. Additional research to validate CGM accuracy and determine the people who are best-suited for the technology is needed, but dedicated focus on implementation should not be delayed. Implementation research that considers the perspectives of clinicians and patients, tests strategies to overcome CGM-related operational and training challenges, and examines delivery models and protocols for responding to CGM-detected dysglycemia will be critical to facilitating sustained adoption of CGM. The evidence provided by BLOSSOM and others offer hope that CGM may be the technology that finally inspires long-overdue innovation in diabetes care for the dialysis population.

Disclosures

Disclosure forms, as provided by each author, are available with the online version of the article at http://links.lww.com/JSN/F291.

Author Contributions

Conceptualization: Jennifer E. Flythe, Klara R. Klein.

Supervision: Jennifer E. Flythe.

Writing – original draft: Klara R. Klein.

Writing – review & editing: Jennifer E. Flythe, Klara R. Klein.

Funding

None.