Diabetes in Dialysis: Closing the Loop

#NephJC Chat

Tuesday August, 17, 2021  at 9 pm Eastern

Wednesday August, 18, 2021  at 9 pm Indian Standard Time 

Wednesday August, 18, 2021  at 9 pm BST

Nat Med. 2021 Aug 4. doi: 10.1038/s41591-021-01453-z. Epub ahead of print. PMID: 34349267.

Fully automated closed-loop glucose control compared with standard insulin therapy in adults with type 2 diabetes requiring dialysis: an open-label, randomized crossover trial 

Charlotte K. Boughton, Afroditi Tripyla, Sara Hartne, Aideen Daly, David Herzig, Malgorzata E. Wilinska , Cecilia CzerlauC, Andrew Fry, Lia BallyC and Roman Hovorka1

PMID: 34349267 

Introduction

Diabetic nephropathy is one of the major microvascular complications of Diabetes mellitus (DM). Half of patients with DM for more than 20 years develop diabetic nephropathy. According to both the United States Renal Data System (USRDS) and European Renal Association - European Dialysis and Transplant Association (ERA-EDTA), diabetes accounts for about 30% of end-stage kidney disease (ESKD). The frequency can vary from as high as 66.7% in Malaysia to as low as 12% in Albania. This wide range is thought to be due to variations in genetic susceptibility, exposure to environmental risk factors, and care of DM.

Chronic kidney disease (CKD) is a state of insulin resistance and glycemic control tends to deteriorates with decreases in glomerular filtration rate. The 2020 KDIGO guidelines suggest insulin, sulphonylureas with low hypoglycemic propensity (eg Glipizide, Glimepiride, Gliclazide), DPP-4 inhibitors, and alpha glucosidase inhibitors for management of DM in ESRD patients (See NephJC summary). But insulin is most widely used as there are wide variations in blood glucose levels in patients undergoing dialysis (Abe et al, Nat Rev Nephrol 2015). Both hypoglycemia and hyperglycemia are associated with increased mortality risk in DM patients undergoing dialysis. Patients with ESKD have increased risk of hypoglycemia due to a lack of gluconeogenesis from the kidneys, inadequate nutrition, decreased insulin clearance, and removal of glucose from dialysis. Post dialysis, due to the mechanisms similar to Somogyi effect and insulin resistance, there is also rebound hyperglycemia in these patients causing glycemic disarray. Uremic toxins in ESKD patients contribute to increased insulin resistance as well.( Abe et al, Nat Rev Nephrol 2015 and Wathen et al, Am. J. Clin. Nutr. 1978).

DM is managed on a day-to-day and hour-to-hour basis by self monitoring of blood glucose levels with the help of finger sticks and glucose-monitor (capillary blood glucose measurement) or by real-time continuous blood glucose monitoring (CGM). There is no long term outcome data on the usage of CGM by patients on dialysis, especially for type 2 diabetes. A small pilot study, the DIALYDIAB  (Joubert et al, Diab Res Clin Prac 2015) showed that CGM in dialysis patients was associated with more frequent changes in anti-hyperglycemic drug prescriptions, better glycemic control and reduced risk of hypoglycemia. CGM was found to be particularly useful while altering the current hypoglycemic prescription to avoid hypoglycemia. 

Closed loop insulin delivery system extend the utility of CGM alone (see Horovka Nat Rev Endocrin 2011 for more). It consists of a CGM, an insulin pump, and an computer that continuously modulates the subcutaneous insulin delivery according to the glucose measurement. This system has been used for Type 1 DM. However, there is little data on patients with Type 2 DM who are on dialysis (Bally et al, Kidney Int 2019).

Figure from Horovka, Nat Rev Endo 2011

The present study addresses this gap in the evidence, examining the effect not just of a CGM, but a closed loop delivering system, in type 2 dialysis patients, on hemodialysis. 

The Study

Study design

This was a multinational, two center, randomized, two period crossover trial conducted in Europe from 21st October 2019 to 3rd November 2020. 

Inclusion criteria:

  • Adults (>18 years) with 

  • Type 2 Diabetes mellitus treated with insulin\

  • End stage renal disease requiring dialysis (Hemodialysis or Peritoneal dialysis) 

Exclusion criteria:

  • Type 1 Diabetes mellitus

  • Pregnancy or breast feeding

  • Severe hearing or visual impairment

  • Physical/ psychological disease/ medications likely to interfere with the conduct of the trial 

Intervention

The Device - closed loop fully automated insulin delivery system

The CamAPS HX closed loop application is linked to an Android phone. It receives sensor glucose data from Dexcom transmitter and uses Cambridge adaptive algorithmic model for dose calculation and delivery of insulin.The glucose target in this vulnerable population was set at 7 mmol/L (126 mg/dL).

Extended data, Figure 1 from Boughton et al, Nature Med 2021: CamAPS HX fully automated closed loop insulin delivery system

During the completely automated closed loop therapy, patients received faster acting insulin-aspart based on continuous monitoring of interstitial glucose levels. In the control arm, CGM was performed, but masked, and insulin therapy followed the standard multiple dose daily strategy. The participants were allowed to follow an unrestricted diet and daily activity. Each treatment period was for 20 days followed by a washout period of 2 to 4 weeks before crossover. 

Outcomes

Primary outcome

Time in target glucose range (5.6mmol/L to 10mmol/L which is about 100mg/dl to 180mg/dl)

Secondary outcomes

  • Mean glucose level

  • Standard deviation of glucose

  • Coefficient of variance (CV) of glucose

  • Between days CV of glucose

  • Total daily insulin dose

Statistical analysis

Intention to treat analysis was applied. Mean + SD was used for normally distributed values. Median was used for non-normally distributed values. A two-sample-t test was used for comparison of normally distributed values. A Mann–Whitney–Wilcoxon rank-sum test was used for comparison of non-normally distributed values. Statistical analysis was performed using SPSS version 27 software.

Funding

This study was supported by NIH Research Cambridge Biomedical  Research Centre. Among the authors, C.B. was supported by a grant from The Novo Nordisk UK Research Foundation and L.B. was supported by a grant from the Swiss Society for Endocrinology and Diabetes (SGED/SSED) and a grant from the Swiss Diabetes Foundation and Swiss Kidney Foundation.

Results

67 participants were approached for the study, 28 declined to participate and 12 were not eligible. 27 participants were randomized following which 1 participant withdrew from the study. Hence, a total of 26 participants were included in the study ( 17 men, 9 women). The planned sample size was not achieved due to COVID-19 pandemic and Brexit.

Figure 1a from Charlotte K. Boughton et al, Nat med.2021, Overview of the participant flow. 

The average age of the participants was 68 + 11 years (see figure 1b, which represents the traditional table 1). The average duration of DM was 20 + 10 years. 27 patients were randomized, 13 were allocated to a fully automated closed loop insulin delivery system first  and the other 13 were allocated to standard insulin therapy with masked CGM. The average time since diabetes diagnosis was 20 years, 12 of them on insulin, while the average time on dialysis was 1.5 years. Except for 1 PD patient, all patients were on hemodialysis. The average washout period was 17 + 5 days. The patients were then switched to the other groups. The participants who completed a minimum of 48 hours in each study period were included for analysis.

Figure 1b from Charlotte K. Boughton et al, Nat med.2021, Baseline characteristics of the study participants

The primary endpoint was the time sensor glucose was within the target range (between 5.6 mmol/L to 10 mmol/L or about 100 to 180 mg/dL). It was longer during closed loop compared to standard insulin  (52.8 + 12.5% versus 37.7 + 20.5%) (95% CI 8.0-22.2). (Figure 2, and Figure 3 below)

Figure 3 : Proportion of time when sensor glucose levels were within the target range (5.6 to 10 mmol/L). Closed loop use - Black bars. Standard insulin use - Grey bars.

The mean glucose levels were also lower in the closed loop compared to standard insulin  (10.1 + 1.3 versus 11.6 + 2.8 mmol/L). (Figure 2 below)

image6.png

However, time spent in hypoglycemia (< 3.5mmol/L or 63 mg/dL) was lesser in the closed loop compared to standard insulin (median (IQR) 0.12 (0.02–0.44%) versus 0.17 (0.00–1.11%); P=0.040). 

The coefficient of variance of within day and between day glucose levels were similar in both the interventions. Also, the requirement of daily insulin was similar in both the interventions. (Table 1)

There is no significant difference in glycemic outcomes between dialysis and non-dialysis days, apart from a lower insulin delivery required during dialysis days in the closed loop group (dialysate glucose was 5.5 mmol/L or 100 mg/dL). (Table 2)


Adverse events

Seven serious adverse events - hypoglycemic events were reported. One was during the closed loop non-operational period. The remaining six were due to concurrent illness such as COVID-19, cellulitis and so on, unrelated to the interventions.

Of note, there were three adverse events related to the study device (one - hyperglycemia due to infusion set failure, another two were reactions to infusion sets). Six device deficiencies were noted during the study period (one phone-receiver-related, two errors related to closed loop initiation, three due to sensor malfunction). However, none resulted in a serious adverse event.

Utility evaluation

The experience of the closed loop system was recorded with the help of a questionnaire. All responders were happy with the closed loop system and reported that they would recommend it to others. 92% felt they spent less time worrying about DM management on closed loop device as it was fully automated. 87% were less worried about the blood glucose levels with closed loop. 50% had better sleep quality with closed loop. 8% reported worsened sleep maybe due to wires and device management during bedtime. 

Discussion

The use of this closed loop insulin pump system ensured greater percentage of time spent in target glucose levels, coupled with lower hypoglycemia. The latter aspect, hypoglycemia is probably more important than hyperglycemia since it is associated with increased risk of all cause mortality in this patient population. Moreover, patients on dialysis have hypoglycemic unawareness. 

The closed loop system was able to manage blood glucose fluctuations during both dialysis and non-dialysis days. The requirement of insulin during dialysis days was lesser in closed loop system which could be attributed to the lower dialysate glucose (5.5 mmol/L = 100 mg/dL).


Strengths

  • Multinational randomized cross-over design 

  • Fully automated closed loop system

  • At home setting and during dialysis sessions

Limitations

  • Smaller sample size than planned due to Brexit and the COVID pandemic. 

  • Outcomes are surrogate outcome (glucose levels) and not necessarily clinically relevant outcomes

  • The duration of the study was very short (and hence inadequate to show any clinical outcomes)

  • The device was managed by the study team and not the patient population. 

  • There was no uniform distribution based on the type of dialysis. Only one patient on peritoneal dialysis was included in the study - glucose control in those patients is quite different given the high glucose load in PD fluid

The benefits of a fully automated closed loop system was seen in this small cohort of a vulnerable population. This trial builds on the previous studies that report that CGM can be used in dialysis patients, and extend that to the outpatient dialysis setting and to patients on hemodialysis. Larger and longer studies would be helpful, especially to convince payers for coverage. 

Summary prepared by Mythri Shankar

Assistant Professor

Department of Nephrology

Institute of Nephro-urology

Bengaluru, India

NSMC intern 2021