Insulin Delivery Hardware: Pumps and Pens

Abstract

Introduction

This year, the scope of this article has been expanded to include forms of insulin delivery outside of insulin pumps, including “smart” insulin pens. These devices can be downloaded and combined with glucose data to visualize insulin injection patterns. In the nonrandomized study by Adolfsson et al., use of a connected pen led to improved time in the target range and reduced missed meal boluses. Further randomized studies are required to assess the efficacy and utility of smart connected pen devices.

The race is on to develop an extended wear infusion set for tethered insulin pumps! Medtronic, Capillary Biomedical, and Convatec are all working on pushing beyond 3 days wear. These efforts are revealing more about the factors involved in infusion set failure. Waldenmaier and colleagues report the survival of standard steel and Teflon infusion sets at 7 days. Eisler and colleagues suggest that an angled cannula provides superior characteristics over a 90-degree cannula. Finally, Tschaikner and colleagues report running a commercial continuous glucose monitoring (CGM) wire through the lumen of an infusion set with excellent operation 6 mm beyond the cannula tip. This work reveals that an all-in-one CGM and extended-wear infusion set is feasible.

Girardot and colleagues offer data on the delivery accuracy of insulin pumps at basal rates as low as 0.1 units/hour using novel methodology. Their Kalman filter technique combines data from a flow meter and weight scale. A key finding of the analysis is the methods that several pump manufacturers use to alter basal delivery. Some systems are more reliant on stroke volume and others on interstroke time.

There were many works touching on the use of insulin pumps in type 1 diabetes. With so much heterogeneity in study designs evaluating diabetes technology, Pease and colleagues present a much-needed network meta-analysis to account for these differences. Bergis and colleagues describe the development of the Insulin Pump Attitude Questionnaire to systematically assess perceived benefits, barriers, and handling of insulin pump therapy. Authors found six subscales: Glycemic Control, Flexibility, Impaired Body Image, Technology Dependency, Ease of Use, and Functionality, with total scores correlated significantly with diabetes distress, self-efficacy, diabetes empowerment, psychological well-being, and treatment dissatisfaction. Work by Nauck and Ostrovski highlights the dawn phenomenon and importance of basal testing. Dejgaard and colleagues present results from the Lira pump trial, showing improvements in HbA1c, weight, and total daily dose. Finally, Paterson and colleagues provide data adding to the body of evidence on insulin requirements for high-protein meals.

Although the benefits of insulin pumps and CGMs are well established in type 1 diabetes, relative benefits in type 2 diabetes are less clear. In the meta-analysis by Dicembrini et al., only modest benefits were noted with the use of pump and CGMs in type 2 diabetes compared to multiple daily injections (MDI) and self-monitoring of blood glucose (SMBG), respectively. The treatment of type 2 diabetes has evolved considerably in the last decade, with the introduction of newer agents such as GLP-1 analogues, DPP-4, and SGLT2 inhibitors. Type 2 diabetes can be a clinically heterogeneous condition, so further studies are required to identify subgroups of people most likely to benefit.

Key Articles Reviewed for the Article

Increased time in range and fewer missed bolus injections after introduction of a smart connected insulin pen

Adolfsson P, Hartvig NV, Kaas A, Møller JB, Hellman J

Diabetes Technol Ther 2020; 22: 709–718

A prospective study of insulin infusion set use for up to 7 days: early replacement reasons and impact on glycemic control

Waldenmaier D, Zschornack E, Buhr A, Pleus S, Haug C, Freckmann G

Diabetes Technol Ther 2020; 22: 734–741

In vivo investigation of the tissue response to commercial Teflon insulin infusion sets in large swine for 14 days: the effect of angle of insertion on tissue histology and insulin spread within the subcutaneous tissue

Eisler G, Kastner JR, Torjman MC, Khalf A, Diaz D, Dinesen AR, Loeum C, Thakur ML, Strasma P, Joseph JI

BMJ Open Diabetes Res Care 2019; 7: e000881

Novel single-site device for conjoined glucose sensing and insulin infusion: performance evaluation in diabetes patients during home-use

Tschaikner M, Powell K, Jungklaus M, Fritz M, Ellmerer M, Hovorka R, Lane S, Pieber TR, Regittnig W

IEEE Trans Biomed Eng 2020; 67: 323–332

All insulin pumps are not equivalent: a bench test assessment for several basal rates

Girardot S, Jacquemier P, Mousin F, Rendekeu C, Hardy S, Riveline JP

Diabetes Technol Ther 2020; 22: 476–483

Perceived benefits and barriers regarding CSII treatment: development and psychometric evaluation of the insulin pump attitudes questionnaire (IPA-questionnaire)

Bergis D, Roos T, Ehrmann D, Schmitt A, Schipfer M, Haak T, Kulzer B, Hermanns N

Exp Clin Endocrinol Diabetes 2019 Aug 19. [Epub ahead of print] DOI: 10.1055/a-0899-4980

Twenty-four hour fasting (basal rate) tests to achieve custom-tailored, hour-by-hour basal insulin infusion rates in patients with type 1 diabetes using insulin pumps (CSII)

Nauck MA, Lindmeyer AM, Mathieu C, Meier JJ

J Diabetes Sci Technol 2021; 15: 360–370

Analysis of prevalence, magnitude and timing of the dawn phenomenon in adults and adolescents with type 1 diabetes: descriptive analysis of 2 insulin pump trials

Ostrovski I, Lovblom LE, Scarr D, Weisman A, Cardinez N, Orszag A, Falappa CM, D'Aoust É, Haidar A, Rabasa-Lhoret R, Legault L, Perkins BA

Can J Diabetes 2020; 44: 229–235

Liraglutide reduces hyperglycaemia and body weight in overweight, dysregulated insulin-pump-treated patients with type 1 diabetes: the Lira Pump trial—a randomized, double-blinded, placebo-controlled trial

Dejgaard TF, Schmidt S, Frandsen CS, Vistisen D, Madsbad S, Andersen HU, Nørgaard K

Diabetes Obes Metab 2020; 22: 492–500

The efficacy of technology in type 1 diabetes: a systematic review, network meta-analysis, and narrative synthesis

Pease A, Lo C, Earnest A, Kiriakova V, Liew D, Zoungas S

Diabetes Technol Ther 2020; 22: 411–421

High-protein meals require 30% additional insulin to prevent delayed postprandial hyperglycaemia

Paterson MA, Smart CEM, Howley P, Price DA, Foskett DC, King BR

Diabet Med 2020; 37: 1185–1191

Impact of technology on glycaemic control in type 2 diabetes: a meta-analysis of randomized trials on continuous glucose monitoring and continuous subcutaneous insulin infusion

Dicembrini I, Mannucci E, Monami M, Pala L

Diabetes Obes Metab 2019; 21: 2619–2625

SMART INSULIN PENS

Increased time in range and fewer missed bolus injections after introduction of a smart connected insulin pen

Adolfsson P^1,2, Hartvig NV³, Kaas A⁴, Møller JB⁵, Hellman J⁶

¹Department of Pediatrics, The Hospital of Halland, Kungsbacka, Sweden; ²Institute of Clinical Sciences, Sahlgrenska Academy at University of Gothenburg, Gothenburg, Sweden; ³Data Science, ⁴Medical & Science, and ⁵Digital Health, Novo Nordisk A/S, Søborg, Denmark; ⁶Department of Medical Sciences, Uppsala University, Uppsala, Sweden

Diabetes Technol Ther 2020; 22: 709–718

Background

This Swedish multicenter observational study investigated whether the connected NovoPen 6 could influence insulin regimen management and glycemic control in people with type 1 diabetes (T1D) using a basal-bolus insulin regimen and continuous glucose monitoring in a real-world setting. The connected insulin pen, NovoPen 6, administers insulin in 1-unit dose increments, with a maximum dose of up to 60 units. It can be downloaded, using near-field communication (NFC), allowing both users and healthcare professionals to visualize insulin injection patterns over time combined with glucose data. The electronic display of the pen also shows the last number of units of insulin administered and the time since the last dose.

Methods

This study was a single-arm, prospective, observational, proof-of-concept study. Participants from 12 Swedish diabetes clinics downloaded pen data at each visit. At baseline, participants received a NovoPen 6 for basal and bolus insulin injections. Follow-up visits were scheduled according to usual clinical practice; at each visit, pen and CGM data were downloaded, discussed, and acted upon by the participant and healthcare provider (HCP). CGM data were also uploaded between visits. Pen data could only be downloaded at the HCP's office. Glycemic summary measures and the number of missed bolus dose (MBD) injections were compared between the blinded baseline period and a follow-up period that was defined as any point after at least five HCP visits. An “on-time” dose was defined as when a bolus insulin injection was detected within 15 minutes before and 60 minutes after the start of a meal. Meals were detected when the CGM signal was ≥7.2 mmol/L (≥130 mg/dL), and the rate-of-change was ≥5.3 mmol/L/hour (≥1.6 mg/dL/min) for the last two consecutive readings or ≥5.0 mmol/L/hour (≥90 mg/dL/hour) for two of the last three readings. Outcomes included time in range (TIR), time in hyperglycemia and hypoglycemia, (level 1: 3.0 to <3.9 mmol/L; level 2: <3.0 mmol/L) and insulin doses. Missed meal injections were classified as meals without injection within −15 and + 60 minutes from the start of a meal as detected from sensor glucose signal by the clinically validated glucose rate increase detector (GRID) algorithm. Outcomes were compared between the baseline and follow-up periods (≥5 healthcare professional visits). Data were analyzed from the first 14 days following each visit. Approximately 2500 days' worth of CGM data with acceptable coverage were included.

Results

Data from 94 participants were included in the main analysis. Median (IQR) follow-up period was 196 (167, 260) days. TIR significantly increased (+ 1.9 [0.8; 3.0] 95% CI hours/day; P<0.001) from baseline to follow-up period (+ 8.5% increase), with a corresponding reduction in time in hyperglycemia (−1.8 [−3.0; −0.6] 95% CI hours/day; P=0.003) (−6.2%) and level 2 hypoglycemia (−0.3 [−0.6; −0.1] 95% CI hours/day; P=0.005), and no change in time in level 1 hypoglycemia (P=0.181). While the mean glucose level did not change significantly (−0.34 mmol/L [−0.96; 0.28] 95% CI), the coefficient of variation was significantly reduced by 3.8% [−6.1; −1.6] 95% CI from a baseline level of 35.9% (P=0.001). Eighty-one adults with T1D were included in the missed meal injection analyses, which included 2747 detected meals; a significant decrease of 43% (estimated relative change: −43.1% [−60.5; −18.0] 95% CI, P=0.002) in the average daily number of missed meal injections was observed from baseline (0.74 [0.62; 0.88] 95% CI) to the follow-up period (0.42 [0.30; 0.60] 95% CI, P=0.002). Change in missed bolus injections corresponded to a decrease from 25% to 14% based on the assumption that participants had three main meals/day. In terms of bolus insulin dose (n=81), there was a significant increase from baseline (25.1 U/day [22.0; 28.7] 95% CI) to after five HCP visits of 28% [9.4; 49.5] 95% CI (P=0.002).

Conclusions

This study highlights the potential benefit on glycemic control and dosing behavior when reliable insulin dose data from a connected pen contribute to insulin management in people with T1D.

Comment

The majority of people living with type 1 diabetes are still treated with injections, and a minority reach modern glycemic targets. Unlike insulin pumps, the actual amount of insulin delivered by conventional insulin syringes and pens remains unrecorded. The lack of reliable insulin dosing information has been hypothesized to be a barrier in optimizing MDI therapy. In this study, use of smart pens combined with CGM and HCP discussion led to an 8.5% increase in TIR and 43% reduction in the average daily missed meal injections.

One of the limitations of the current study is the observational, nonrandomized, single-arm design. As such, we cannot be sure that the benefits are solely due to the use of the connected pen. In fact, the benefits observed in this study are likely to be the combined effect of HCP advice around meal bolusing, CGM data analysis, and study participation effect. The mean time between visits was 71 days, which is more frequent than the standard 90–120 days but more realistic than some clinical trials. Nevertheless, this study provides useful insights and shows the potential of this novel technology to improve outcomes and form the basis for future randomized controlled trials.

This study also shows that missing bolus injections can be frequent in type 1 diabetes, and assuming three meals per day, at least 25% of meal boluses were missed at the baseline. In another recent study, using Bluetooth-enabled insulin pen caps, similar levels of missed meal injection were found in 24% of bolus and 36% basal insulin injections (1). Both of these studies highlight the importance of the completeness of meal and snack insulin bolus administration during routine diabetes consultations, especially in those with higher HbA1c.

INFUSION SETS

A prospective study of insulin infusion set use for up to 7 days: early replacement reasons and impact on glycemic control

Waldenmaier D¹, Zschornack E¹, Buhr A², Pleus S¹, Haug C¹, Freckmann G¹

¹Institut für Diabetes-Technologie, Forschungs- und Entwicklungsgesellschaft mbH an der Universität Ulm, Ulm, Germany; ²Ypsomed AG, Burgdorf, Switzerland

Diabetes Technol Ther 2020; 22: 734–741

Background

The justification of infusion set replacement intervals remains vague and largely unchanged since the 1980s. This study investigates 7-day wear for steel and Teflon infusion sets.

Methods

This was a German single-center, open-label, randomized cross-over study. Participants with type 1 diabetes and an HbA1c ≤8.5% wore YpsoPump Orbit micro (steel cannula) and YpsoPump Orbit soft (Teflon cannula) twice for 7 days each. Participants could choose catheter and tubing length, as well as decide on the use of insertion device (Orbit Inserter). Individuals received insulin aspart, Dexcom G5 CGM with Abbott FreeStyle Precision Neo glucometer and blood ketone meter. Participants were randomized to wear one of the two sets for the first 2 weeks and the other for the second 2 weeks. “Occlusion” was defined as: (1) a blood glucose >250 mg/dL for more than 2 hours with a failed correction bolus (correction does not reduce glucose 50 mg/dL within 90 minutes), (2) ketones ≥0.6 mmol/L, or (3) a pump occlusion alarm that could not be solved. Outcomes included infusion set failure and glycemic control over the wear time.

Results

Forty participants wore 160 infusion sets in the abdomen (80%) and hips (20%), with insertion devices used for 93% of insertions. Two infusion sets were replaced immediately after insertion due to poor adhesion; 94% of sets lasted 3 days, 82% lasted 5 days, and 66% lasted the full 7 days. Mean wear time was 6.2±1.5 days (range 0.8–7 days). There was no statistically significant difference in full 7-day wear time for steel (69%) or Teflon (63%). The premature infusion set failures were due to pump occlusion alarm (9% of sets), unconfirmed “occlusion” (4% of sets), adhesive failure (4% of sets), accidental pull-out (4% of sets), unexplained hyperglycemia (3% of sets), ketones (3% of sets), blood in tubing (3% of sets), leakage (2% of sets), and discomfort (2% of sets). There were no reported infusion set infections. Mean glucose, glycemic variability, and daily insulin did not show a substantial change over time.

Conclusions

In this study, 7-day infusion set wear did not appear to have a meaningful impact on glycemic control or insulin requirements, and infection rate was low. The authors suggest that infusion set wear time can be individualized beyond the labeling.

Comment

This study was very similar to the 2014 trial by Patel and Buckingham, but with significantly different results (2). The Patel study focused on comparing the 7-day performance of Quick-Set (Teflon) and Sure-T (Steel) infusion sets. As in the current study there was no significant difference in wear time between Teflon and steel catheter. However, in the Patel study only 36% of sets lasted the full 7 days compared to 66% in the presented study.

Waldenmaier's work illustrates the importance of replication studies. The authors acknowledge the differences in outcomes and implicate the 15% initial failure rate in Teflon catheters due to kinking, which did not occur in their study. Other differences between the studies include sample size (n=20 for Patel vs n=40 for Waldenmaier), baseline demographics (Patel's cohort was younger with less diabetes experience and a higher HbA1c), differences in infusion sets beyond catheter material choice, and choice of catheter length (Patel study fixed at 6 mm).

Likely, all factors contributed to the difference, especially the population selected. Both studies report a significant influence of the individual subject on wear time, although Waldenmaier attributed this effect to a single subject. The patient population in this study had an HbA1c of 7.3±0.7% and were excluded for tape allergies or history of catheter abscess. It may be the case that this population tends to be able to wear sets longer. While ConvaTec manufactures the majority of infusion sets for Medtronic and Tandem, the sets under investigation are manufactured by YpsoMed themselves. This difference may also be attributable to infusion set design; although infusion set design is not specifically discussed in either paper.

In vivo investigation of the tissue response to commercial Teflon insulin infusion sets in large swine for 14 days: the effect of angle of insertion on tissue histology and insulin spread within the subcutaneous tissue

Eisler G¹, Kastner JR^1,2, Torjman MC¹, Khalf A¹, Diaz D¹, Dinesen AR¹, Loeum C¹, Thakur ML³, Strasma P⁴, Joseph JI¹

¹Anesthesiology, Thomas Jefferson University, Philadelphia, PA; ²Endocrinology and Diabetology, Medical University of Graz, Graz, Austria; ³Radiology, Thomas Jefferson University, Philadelphia, PA; ⁴Capillary Biomedical, Inc, Irvine, CA

BMJ Open Diabetes Res Care 2019; 7: e000881

Background

The relationship between insulin absorption and inflammatory tissue response remains unclear. The authors sought to explore the effect of infusion set catheter angle (30° versus 90°) on inflammation, bolus shape, and infusion set tubing pressure.

Methods

Angled or straight infusion sets were inserted every other day for 14 days into 11 pigs. Prior to sacrifice, a 70 μL bolus of insulin and contrast were infused while recording pressure profiles, including a peak (p_max) and area under the curve (AUC). Bolus surface area and volume were assessed with micro-CT. Tissue staining was used to analyze the area of inflammation and inflammatory layer thickness surrounding both cannulas.

Results

The angled cannulas exhibited a larger mean surface area (P<0.001) and volume (P=0.001) with lower area of inflammation (P<0.001), inflammatory layer thickness (P<0.001), p_max (P=0.005), and AUC (P=0.014). Independent of cannula angle, both bolus surface area and volume decreased significantly over wear time. Increases in inflammatory layer thickness correlated with increase in p_max and decrease in surface area and volume.

Conclusions

Thirty-degree angled infusion sets exhibited a number of favorable features over 90-degree sets with regards to local inflammatory response.

Comment

Medtronic has received the CE mark in Europe for their 7-day extended wear infusion set. Studies of extended wear infusion sets are in progress by ConvaTec (NCT03819634), Capillary Biomedical (NCT04398030), and Medtronic (NCT04113694). These new infusion sets have required multiple incremental improvements in the design. The inflammatory response to the cannula, insulin, and preservative can have a significant effect on insulin absorption (3). Inflammatory response also appears to play an important role in infusion set survival, particularly, failures attributable to unexplained hyperglycemia. These events do not consistently trigger occlusion alarms.

The present study demonstrates that insertion angle, which has not been previously explored, may play a role in infusion set survival. It is very exciting to see this scientific rigor applied to infusion sets. Animal data seems to support the inflammatory benefit of angled cannulas, warranting clinical studies of angled infusion set longevity.

Novel single-site device for conjoined glucose sensing and insulin infusion: performance evaluation in diabetes patients during home-use

Tschaikner M¹, Powell K², Jungklaus M¹, Fritz M³, Ellmerer M¹, Hovorka R⁴, Lane S², Pieber TR¹, Regittnig W⁵

¹Department of Internal Medicine, Division of Endocrinology and Diabetology, Medical University of Graz, Graz, Austria; ²Triteq Ltd, Hungerford, UK; ³4A Engineering GmbH, Traboch, Austria; ⁴Wellcome Trust–MRC, Institute of Metabolic Science and the Department of Paediatrics, University of Cambridge, Cambridge, UK; ⁵Department of Internal Medicine, Division of Endocrinology and Diabetology, Medical University of Graz, Graz, Austria

IEEE Trans Biomed Eng 2020; 67: 323–332

Background

Combining CGM and insulin infusion site in a combined single-port device is an important step in increasing acceptance and usability of automated insulin delivery technology.

Methods

This was a pilot feasibility study of an improvised single-site CGM (Dexcom G4 Platinum) and infusion set (Medtronic SOF-Set Micro QR) connected to an Aniams-Vibe insulin pump. Volunteers with type 1 diabetes used the device for up to 6 days in an ambulatory setting and with automated insulin delivery (AP @ Home closed-loop platform) for 1 day as an inpatient. Measurements from the embedded CGM were compared to capillary blood glucose measurements with an Abbott FreeStyle Freedom Lite and another Dexcom G4 Platinum CGM worn by the participant.

Results

Ten participants were enrolled in the trial. Median mean absolute relative difference (MARD) of the combined CGM and infusion set versus glucometer was 13.0% (IQR 10.5%–16.7%). This MARD was not significantly different (P=0.922) from that of the additional CGM at 13.9% (ICR 11.9%–15.3%). The inpatient closed-loop insulin delivery functioned adequately, providing an average TIR of 70.4% (range 58.1%–87.3%), superior (P=0.002) to at-home TIR of 54.8% (45.9%–61.0%).

Conclusion

This trial illustrates the feasibility of open and closed-loop glucose control with a single-site device with a 6 mm infusion set cannula and glucose-sensitive sensor extending 6 mm beyond the cannula end.

Comment

There are currently multiple efforts being directed at extended-wear infusion sets. If an infusion set can be worn for as long as a CGM, the hope is that a combination device may be possible. It remains unclear how far the glucose sensor must be from the site of insulin delivery. Nørgaard reported on the MiniMed Duo in 2015; in this dual design the sensor wire and infusion site were separated by 11 mm (4). Another optical-based glucose sensor system was constructed over a commercial 8 mm steel cannula infusion set. The distance between the glucose sensing element and the insulin egress pathway was 6 mm (5).

Earlier in the year Tschaikner and colleagues reported the creation of an improvised single-port device from commercially available supplies and 3D-printed adapter (6). The glucose sensor extended 6 mm below the cannula opening utilizing an older generation Dexcom with longer wire length. The single-port design integrating existing glucose oxidase-based sensor technology is the next logical step following design of a 7–14-day infusion set.

This proof-of-principle pilot data provides real-world evidence that existing CGM technology can be integrated in a straightforward manner with infusion sets. The minimal distance between insulin delivery and glucose sensing remains to be determined, but larger clinical trials will be required.

DELIVERY ACCURACY

All insulin pumps are not equivalent: a bench test assessment for several basal rates

Girardot S^1,2,3, Jacquemier P^1,2,3, Mousin F¹, Rendekeu C¹, Hardy S¹, Riveline JP^2,3,4

¹Air Liquide SA, Explor Center (Healthcare), Paris, Île-de-France, France; ²IMMEDIAB Lab UMRS1138, Institut National de la Santé et de la Recherche Médicale (INSERM), Paris, France; ³Université Paris Diderot-Paris VII, Paris, France; ⁴Department of Diabetes and Endocrinology, Lariboisière Hospital, APHP, Paris, France

Diabetes Technol Ther 2020; 22: 476–483

Background

Accurate insulin delivery is essential for glycemic control and has been insufficiently studied in insulin pumps, particularly at the low basal rates used in the pediatric population and occasionally delivered with closed-loop systems.

Methods

Four off-the-shelf pumps were assessed: Medtronic MiniMed 640G, Ypsomed YpsoPump, Insulin Omnipod, and Tandem t:slim X2. Pumps were tested at basal rates of 2, 1, 0.5, and 0.1 units/hr. A double-measurement approach combined with Kalman filtering was utilized employing direct mass flow meter and time-stamped microgravimetric test bench. Pump delivery accuracy was evaluated using a mean dose error. Mean absolute relative difference (MARD) of error was calculated at different observation windows. Peak insulin delivery was assessed regarding stroke regularity in terms of frequency and volume.

Results

Mean error was −16%. Mean MARD varied by pump and basal rate: 2 units/hr—11.8% MiniMed, 10.3% Omnipod, 7.9% Ypsomed, 7.4% Tandem; 1 unit/hr—13.3% MiniMed, 21.6% Omnipod, 9.5% Ypsomed, 12.7% Tandem; 0.5 units/hr—35.0% MiniMed, 22.7% Omnipod, 12.9% Ypsomed, 18.7% Tandem; 0.1 units/hr—35.9% MiniMed, 61.3% Omnipod, 44.8% Ypsomed, 22.7% Tandem.

Peak insulin delivery is achieved by frequency modulation for Omnipod, volume modulation for Tandem, and a combination for Ypsomed and Medtronic. At low basal rates delivery was better maintained if the interstroke time was held constant (with volume modulation).

Conclusions

Delivery accuracy is dependent on the pump and the basal rate. Imprecision can result from volume and stroke frequency variability. Some pumps exhibit better delivery accuracy for low basal rates. The clinical implications of these errors require investigation.

Comment

In the present study, Girardot and colleagues employ their Kalman filter technique to combine data from a Bronkhorst BL100 flow meter and weight scale (7). A key finding of the analysis is that the Omnipod has a fixed stroke volume and adapts interstroke time to alter delivery, whereas the t:slim X2 maintains interstroke time and can vary stroke volume. The Medtronic and Ypsomed pumps vary stroke volume for higher basal rates and frequency for lower basal rates. The effects of these decisions on closed-loop control require investigation. Likely, those developing interoperable automated glycemic controllers will need to work around these specifications when issuing basal rate modulation commands.

Under real-world conditions, many other factors may predominate over delivery accuracy. Absorption after delivery can be variable even within the same individual due to lipohypertrophy; scarring; local trauma; and inflammation from insulin, excipients, preservatives, or the cannula (3). A previous study comparing different pumps in adults with type 1 diabetes found no clinically relevant difference in HbA1c reduction (8). Head-to-head clinical studies will be necessary to assess the clinical effects of these deviations in dosing accuracy. Crucially, we would expect the most significant differences in children who have lower insulin needs and basal rates as baseline. Thus, investigation in youth is warranted, particularly for closed-loop technology.

TECHNOLOGY IN TYPE 1 DIABETES

Perceived benefits and barriers regarding CSII treatment: development and psychometric evaluation of the insulin pump attitudes questionnaire (IPA-questionnaire)

Bergis D¹, Roos T², Ehrmann D^2,3, Schmitt A⁴, Schipfer M⁵, Haak T^2,6, Kulzer B^2,3,6, Hermanns N^2,3,6

¹Division of Endocrinology & Diabetes, Department of Internal Medicine 1, Goethe University Hospital, Frankfurt am Main, Germany; ²Research Institute of the Diabetes Academy Mergentheim (FIDAM), Bad Mergentheim, Germany; ³Department of Clinical Psychology and Psychotherapy, University of Bamberg, Bamberg, Germany; ⁴Diabetes Akademie Bad Mergentheim, e.V., Bad Mergentheim, Germany; ⁵Profusa, Inc., South San Francisco, CA; ⁶Diabetes Clinic Mergentheim, Bad Mergentheim, Germany

Exp Clin Endocrinol Diabetes 2019 Aug 19. [Epub ahead of print] DOI: 10.1055/a-0899-4980

Background

Continuous subcutaneous insulin infusion is associated with both positive and negative expectations, attitudes, and experiences. This paper describes the development of a new questionnaire—The Insulin Pump Attitude Questionnaire—to systematically assess perceived benefits, barriers, and handling of continuous subcutaneous insulin infusion therapy.

Methods

Based on the literature review and interview responses with continuous subcutaneous insulin infusion (CSII) users and HCP, 60 survey items were created. For a pilot test, a sample of 70 people with type 1 diabetes and 10 healthcare providers rated these items regarding comprehensibility, relevance, content, and clarity. In addition, item selectivity and reliability of the theoretically derived scales were calculated. Based on these procedures, a total of 28 survey items were kept, covering perceived benefits, barriers, and ease of use of CSII. These items included statements to which the respondents could rate their agreement on a 5-point Likert scale (0=fully disagree to 4=fully agree). IPA was tested in two samples (n=265/452) comprising pump users and non–pump users. The first sample of participants with CSII therapy was used for exploratory factor analysis to establish the factorial structure of the IPA. The primary purpose of the second sample was to assess the validity of the scale: First, by confirming the suggested factorial scale structure with confirmatory factor analysis. Second, by demonstrating criterion validity with associations to previously validated questionnaires; and third, by demonstrating the discriminative ability of the IPA regarding expected differences between pump users and non–pump users.

Results

Exploratory factor analysis revealed 26 items comprising 6 subscales: Glycemic Control, Flexibility, Impaired Body Image, Technology Dependency, Ease of Use, and Functionality. Confirmatory factor analysis verified this factor structure. The IPA sum score correlated significantly with diabetes distress (r=−0.30), self-efficacy (r=0.22), diabetes empowerment (r=0.36), psychological well-being (r=0.16), and treatment dissatisfaction (r=−0.24), supporting criterion validity with small-to-medium effect sizes. Furthermore, the IPA was able to differentiate between pump users and non–pump users, with higher scores for pump users regarding Glycemic Control, Flexibility, Ease of Use, and Functionality, and lower scores for pump users regarding Impaired Body Image and Technology Dependency.

Conclusions

The Insulin Pump Attitude Questionnaire is a reliable and valid new instrument to assess attitudes toward continuous subcutaneous insulin infusion. With six scales, the Insulin Pump Attitude Questionnaire provides a comprehensive analysis of possible benefits, barriers, and handling problems of continuous subcutaneous insulin infusion therapy. In clinical practice, the Insulin Pump Attitude Questionnaire might be used to address the different attitudes in pump users but also in people considering continuous subcutaneous insulin infusion therapy.

Comment

CSII is often considered the more advanced form of insulin delivery compared to MDI. Increasingly, CSII also allows treatment options such as predictive low glucose suspend and hybrid closed-loop automated insulin delivery. Pump therapy is also associated with risks and inconveniences, including unexplained cannula failures, body image issues, and a higher risk of diabetic ketoacidosis. Using the specific subscales, the IPA provides a detailed and multidimensional understanding of the impact of CSII on user experience and expectations. The IPA can help to assess the core elements of user wishes and needs in a structured way, which then can be further explored during patient consultations. By assessing the specific benefits and barriers, these barriers can be addressed by training and education, and the potential benefits can be emphasized in people who might benefit from CSII therapy but are reluctant to start with this treatment. These questionnaires could also be used in the comparison of different types of CSII technologies, for example, traditional tube pumps vs patch pumps, sensor-augmented, and closed-loop pumps.

Twenty-four hour fasting (basal rate) tests to achieve custom-tailored, hour-by-hour basal insulin infusion rates in patients with type 1 diabetes using insulin pumps (CSII)

Nauck MA^1,2, Lindmeyer AM², Mathieu C³, Meier JJ²

¹Diabeteszentrum Bad Lauterberg, Germany; ²Diabetes Division, St. Josef-Hospital, Klinikum der Ruhr-Universität Bochum, Germany; ³Clinical and Experimental Endocrinology, KU Leuven, Belgium

J Diabetes Sci Technol 2021; 15: 360–370

Background

One feature of continuous subcutaneous insulin infusion is the ability to modulate basal insulin delivery over a 24-hour period. For basal insulin used to cover resting endogenous glucose production, a 24-hour fast can be helpful for “basal testing” but is challenging in the ambulatory setting. In specialized German diabetes centers this testing is performed on inpatients to optimize basal profiles.

Methods

Data from a large cohort of adults with type 1 diabetes on pump therapy seen as inpatients and undergoing 24-hour fast at a single German diabetes center between 2008 and 2014 were retrospectively analyzed. During the day prior to inpatient stay, participants were asked to avoid strenuous exercise and alcohol. Starting at 6 p.m., meals were omitted and there was no carbohydrate intake except to compensate for plasma glucose <3.9 mmol/L (<70 mg/dL). Data collected included plasma glucose (at 1800, 2000, 2200, 0000, 0200, 0400, 0645, 0900, 1200, 1400, and 1800 the following day) and hourly basal rate. Strict goal glucose range was defined as 4.4–7.2 mmol/L (80–130 mg/dL). Two periods of special interest were defined: dawn (1–7 a.m.) and dusk (4–7 p.m.).

Results

Data from 339 individuals (age 16–80 years) with type 1 diabetes were retrospectively reviewed. Pump manufacturers included Medtronic (28.6%), Disetronic (16.8%), Cozmo (5.6%), Roche (5.1%), and Insulet (3.2%). Basal rates (mean 0.90±0.02 units/hr) showed peaks at dawn (mean 1.07±0.02 units/hr) and dusk (mean 0.95±0.02 units/hr). Plasma glucose averaged 6.6±0.1 mmol/L (119±1.8 mg/dL) with 53.1% in the strict target range.

Conclusions

Twenty-four-hour fasts can provide valuable information regarding basal rates. Basal rates appeared significantly higher during the dawn period.

Comment

Use of basal insulin varies among providers. Some providers attempt to use it to match resting endogenous glucose production, others use it in whatever manner helps achieve glycemic targets. The latter can include increasing basal rates around meal times or choosing a slightly higher than required basal to continually reduce glucose. A safer method of basal testing is to titrate the basal rate to keep blood sugars stable within 30 mg/dL when there are no carbohydrates or bolus insulin on board (9).

In the study, patients came in with empirically derived basal rates. Heroic efforts were applied to tune basal rates by admitting individuals for a 24-hour fast and analyze basal rates with respect to maintaining glucose within a tight range. These basal rates were often variable, with troughs that preceded peaks at dawn and dusk. Given the extended action profile of insulin, it is difficult to know if glucose perturbation was physiologic or a result of the shifting basal rates.

Twenty-four-hour fasting may be unrealistic for many practice settings. Overnight basal testing, particularly with CGM, is simple and practical. In a research setting, the 24-hour fast might be useful for determining the need for multiple basal profiles. Rather than optimizing glucose to a particular range, it may be more useful to maintain glucose within a tight window. The ultimate question is whether multiple basal rates and this type of tuning have a meaningful clinical outcome.

Analysis of prevalence, magnitude and timing of the dawn phenomenon in adults and adolescents with type 1 diabetes: descriptive analysis of 2 insulin pump trials

Ostrovski I¹, Lovblom LE¹, Scarr D¹, Weisman A¹, Cardinez N¹, Orszag A¹, Falappa CM¹, D'Aoust É², Haidar A^3,4, Rabasa-Lhoret R^2,5,6,7, Legault L⁸, Perkins BA^1,9

¹Lunenfeld-Tanenbaum Research Institute, Mount Sinai Hospital, Toronto, Ontario, Canada; ²Institut de recherches Cliniques de Montréal, Montréal, Québec, Canada; ³Department of Biomedical Engineering, Faculty of Medicine, McGill University, Montréal, Québec, Canada; ⁴Research Institute of McGill University Health Centre, Montréal, Québec, Canada; ⁵Division of Experimental Medicine, McGill University, Montréal, Québec, Canada; ⁶Nutrition Department, Faculty of Medicine, Université de Montréal, Montréal, Québec, Canada; ⁷Montreal Diabetes Research Centre, Montréal, Québec, Canada; ⁸Montreal Children's Hospital, McGill University Health Centre, Montréal, Québec, Canada; ⁹Division of Endocrinology and Metabolism, Department of Medicine, University of Toronto, Toronto, Ontario, Canada

Can J Diabetes 2020; 44: 229–235

Background

The dawn phenomenon may be an important cause of morning hyperglycemia in those with type 1 diabetes. There are no standardized guidelines on how to adjust basal rate and timing to accommodate this dawn phenomenon.

Methods

Data from two prior sensor-augmented insulin studies conducted at McGill University were retrospectively analyzed. Participants in the previous Basal Rate Analyzer Trial (BRAT) performed five fasting overnight basal assessments during which basal rates were tuned nightly. A second confirmatory group was pulled from the CLASS 04 study. Data on basal rate modulation was obtained for the insulin-only intervention in those with glucose ≤7 mmol/L (126 mg/dL). Baseline basal was the midnight basal rate for BRAT participants and total daily basal divided by 24 hours for the CLASS 04 group. Dawn phenomenon was uniformly defined as a >20% increase in insulin requirements from baseline, lasting at least 90 minutes. Among cases, time of onset and percent change in the magnitude of basal delivery were determined.

Results

Among all 20 participants in the BRAT cohort and 13 chosen participants in CLASS 04, 52% experienced the dawn phenomenon. Time of onset was around 3 a.m. The magnitude of the dawn phenomenon was 58.1% (IQR 28.8% to 110.6%) in the BRAT cohort and 65.5% (IQR 45.6% to 87.4%) in the CLASS 04 cohort.

Conclusions

The dawn phenomenon occurs in about half of the patients reported and has a predictable onset but highly variable magnitude. Fasted overnight basal insulin assessments may be needed to optimize glycemic control.

Comment

Ostrovski and colleagues present a secondary data analysis from prior open- and closed-loop studies. As reported previously by Carroll and Schade (10), they find that not all individuals with type 1 diabetes exhibit a dawn phenomenon. For those with the dawn phenomenon there was a highly variable increase in basal needs, with a >20% increase in insulin requirements around 3 a.m. The authors present overnight basal profiles for those with and without dawn phenomenon. Due to the significant fraction without dawn phenomenon, it is important to consider that difference in basal profiles are washed out when results are pooled. The BRAT cohort was dubbed the derivation set, and the basal profiles preceding the increase were much flatter than in the CLASS 04 cohort, which exhibited substantial overnight oscillations. Given the variability in presence and magnitude of increased insulin needs at dawn, it is very important that basal testing be personalized to the individual. Future studies should consider what factors contribute to the substantial differences between individuals with type 1 diabetes.

Liraglutide reduces hyperglycaemia and body weight in overweight, dysregulated insulin-pump-treated patients with type 1 diabetes: the Lira Pump trial—a randomized, double-blinded, placebo-controlled trial

Dejgaard TF^1,2, Schmidt S^1,2,3, Frandsen CS², Vistisen D¹, Madsbad S², Andersen HU¹, Nørgaard K^1,2

¹Steno Diabetes Center Copenhagen, University of Copenhagen, Denmark; ²Department of Endocrinology, Copenhagen University Hospital Hvidovre, Hvidovre, Denmark; ³Danish Diabetes Academy, Odense, Denmark

Diabetes Obes Metab 2020; 22: 492–500

Background

The objective of this study was to investigate the efficacy of adding the glucagon-like peptide-1 receptor agonist liraglutide to CSII in overweight or obese individuals with type 1 diabetes and nonoptimal glycemic control.

Methods

In a single-center, 26-week, randomized, double-blind, placebo-controlled trial, 44 overweight or obese patients with type 1 diabetes were randomized 1:1 to liraglutide 1.8 mg daily or a placebo added to CSII treatment. Patients eligible for inclusion were adults between ages 18 and 70 years and clinically diagnosed with type 1 diabetes ≥12 months before the screening visit. Patients had to be treated with CSII, with an HbA1c >58 mmol/mol (7.5%), and use carbohydrate counting and the insulin pump bolus calculator for all meals. No upper limit of HbA1c was specified. Participants had to be overweight or obese with a body mass index (BMI) >25 kg/m². The treatment was initiated with an injection of 0.6 mg liraglutide/0.1 mL placebo daily, increased to 1.2 mg liraglutide/0.2 mL placebo daily after 1 week, and increased to 1.8 mg/0.3 mL placebo after another week. To reduce the risk of hypoglycemia, the basal insulin rate was reduced by 10%, and the insulin sensitivity factors and insulin-to-carbohydrate ratios were increased by 15% at the initiation of liraglutide treatment. The primary endpoint was the change in HbA1c. Secondary endpoints included change in insulin dose, CSII settings, glycemic variability, body weight, and patient-reported outcome measures. Finally, adverse effects, including hypoglycemic events, were registered.

Results

After 26 weeks of treatment the HbA1c level decreased from 66 mmol/mol (8.2%) to 61 mmol/mol (7.7%) (P<0.001) for persons treated with liraglutide, while no significant change was found with placebo 66 mmol/mol (8.1%) to 68 mmol/mol (8.3%) (P=0.058). Thus, the difference between groups at the end of treatment was 7 mmol/mol (0.7%) (P<0.001). Using blinded CGM, after 26 weeks of treatment persons treated with liraglutide had more time in target range (3.9 to 10 mM [70–180 mg/dL]) compared with the placebo, 57% versus 45% (P=0.044). Percentage of time spent in hypoglycemia level 1 (glucose value <4.0 mmol/L [71 mg/dL]) and hypoglycemia level 2 (glucose value <3.0 mmol/L [54 mg/dL]) was no different between liraglutide and the placebo (6% vs 6% of time spent in level 1 and 1% vs 2% of time in level 2, respectively).

The total daily insulin dose decreased in persons treated with liraglutide compared with the placebo (P=0.008). This difference was mainly due to a difference between groups of 7 units/day of bolus insulin after 13 weeks and 8 units/day at the end of treatment. When adjusted for body weight, no significant change was found in either the total bolus or the total basal insulin doses in the groups during the study.

Mean body weight was reduced by 6.3 kg (P<0.001) compared with placebo. Concomitantly, the total daily carbohydrate intake after 13 weeks decreased by 32 g/day in patients treated with liraglutide compared with the placebo (P=0.031). However, at the end of the study, this difference decreased to 20 g/day, and no difference was found between groups (P=0.188). The insulin sensitivity factors and the insulin-to-carbohydrate ratios between 06:00 and 11:00 h, between 11:00 and 16:00 h, and between 16:00 and 24:00 h (weighted means) did not change within or between groups after 26 weeks of treatment. Gastrointestinal adverse events were reported more frequently in persons treated with liraglutide compared with the placebo (nausea 64% vs 32%, diarrhea 18% vs 9%, and vomiting 18% vs 5%).

Conclusions

Liraglutide treatment reduced HbA1c, total daily insulin dose, and body weight without increasing the risk of hypoglycemia in CSII-treated patients with type 1 diabetes and insufficient glycemic control. Liraglutide may be considered a potential add-on therapy to insulin in this growing subgroup of patients.

Comment

A significant proportion (40% to 50%) of adults living with type 1 diabetes in the Western world are either overweight or obese, with consequent metabolic complications and increased cardiovascular risk (11). When adding liraglutide, 1.8 mg once daily to CSII therapy in this population, authors demonstrated a clinically significant improvement in HbA1c of −7.3 mmol/mol (−0.7%) with concomitant clinically meaningful weight loss when compared with the placebo.

A key difference in the current study and previous studies evaluating liraglutide treatment in type 1 diabetes is in the concomitant reduction in HbA1c in the placebo arm (which was absent in the current study) in previous studies, thus minimizing the effect of liraglutide between groups. In a systematic review and meta-analysis including five trials with 2445 randomized participants with type 1 diabetes, liraglutide provided modest reductions in HbA1c, with liraglutide 1.8 mg producing the most considerable decrease (mean treatment difference in HbA1c [%]=−0.24%, 95% CI −0.32 to −0.16). Significant weight reduction, up to 4.87 kg with liraglutide 1.8 mg, was also observed (95% CI −5.31 to −4.43).

A recent study (12) has evaluated the mechanism of weight loss with liraglutide treatment 1.8 mg daily vs the placebo for 6 months in overweight and obese people with type 1 diabetes. Subcutaneous adipose tissue biopsies; a high-calorie, high-fat meal challenge test; continuous glucose monitoring; dual-energy X-ray absorptiometry; and MRI were performed before and at the end of treatment. Fasting glucagon levels fell significantly, but there was no change in postprandial glucagon levels. In contrast, both fasting and postprandial GLP-1 levels increased. Bodyweight fell significantly in the liraglutide group (−4.2 kg), mostly in the form of fat mass loss (including visceral fat), with no change in lean mass. The change in adipose tissue mass after liraglutide treatment was associated with changes in expression of lipid oxidation and metabolism-related genes in adipose tissue with increased expression of adipose tissue triglyceride lipase, carnitine palmitoyl transferase-1, peroxisome proliferator-activated receptor (PPAR)α, PPARδ, uncoupling protein-2, and type 2 iodothyronine deiodinase in the adipose tissue.

In summary, treatment with liraglutide in type 1 diabetes leads to improved glucose levels and weight loss due to loss of adipose tissue without affecting lean body mass, as well as reduced bolus insulin due to reduced carb intake, without significant changes in insulin sensitivity.

The efficacy of technology in type 1 diabetes: a systematic review, network meta-analysis, and narrative synthesis

Pease A^1,2, Lo C^1,2, Earnest A¹, Kiriakova V², Liew D^1,3, Zoungas S^1,2,3

¹School of Public Health and Preventive Medicine, Monash University, Melbourne, Australia; ²Monash Health, Melbourne, Australia; ³Alfred Health, Melbourne, Australia

Diabetes Technol Ther 2020; 22: 411–421

Background

We now have a range of medical technologies that could be utilized in the treatment of type 1 diabetes. The objective of this paper was to systematically review the literature and perform network meta-analysis to inform clinical practice by comparing available technologies in the management of adults with type 1 diabetes.

Methods

Authors included RCTs of parallel and cross-over study design that was ≥6 weeks duration overall (or each phase of a cross-over study) and adults (≥18 years of age) with type 1 diabetes. Authors considered studies that compared technologies for insulin delivery, glucose monitoring, insulin dosing advice, or MDI and SMBG via capillary testing. The combination of CGM and CSII that facilitated low-glucose suspend, glucose threshold alarms, or any automated adjustment of insulin delivery was defined as an “integrated system.” Smart device applications for insulin dose calculation were merged with the freestanding bolus-calculator node for analysis of participants receiving MDI therapy. Fifty-two parallel or cross-over RCTs met the inclusion criteria, involving 3975 participants eligible for analysis. Respectively, there were 43, 40, 19, and 14 studies included in network meta-analysis for the outcomes of A1c, severe hypoglycemia, nonsevere hypoglycemia, and quality of life (QoL). Overall, the mean sample size of included studies was 78 (± 79), the mean duration of intervention was 8 (± 7) months, and 78% received industry funding or material support. Authors ranked the treatments for A1c, severe hypoglycemia, nonsevere hypoglycemia, and QoL with the surface under the cumulative ranking curve (SUCRA).

Results

The mean difference of A1c values favored integrated systems when compared to MDI combined with either flash glucose monitoring (FGM) (0.96; 95% predictive interval [PrI] 0.04–1.89) or SMBG (0.87; 95% PrI 0.12–1.63). Among integrated systems, sensitivity analyses found that hybrid closed-loop therapy was the primary driver for significant A1c reductions. Ranking technologies by A1c values favored integrated systems (SUCRA 96.4), CSII with standalone CGM (SUCRA 80.0), MDI with CGM (SUCRA 72.5), and CSII with bolus calculators (SUCRA 52.6). Regarding rate ratios for severe hypoglycemia (36 RCTs comprising the same within-study cointerventions, and 2844 person-years), CSII with SMBG was significantly better than MDI with bolus advisors (2.21 95% PrI 1.03–4.74). Ranking technologies by nonsevere hypoglycemia rates alone favored MDI with CGM (SUCRA 94.6), MDI with FGM (SUCRA 67.6), and integrated systems comprising low glucose suspend or hybrid closed-loop therapy (SUCRA 64.1). In terms of QoL (14 RCTs comprising 1499 participants), no intervention was clearly superior. When simultaneously considering A1c and severe hypoglycemia, integrated systems comprising low-glucose suspend or hybrid closed-loop therapy, as well as MDI with CGM, appeared to provide the highest composite ranking in the cluster analysis of SUCRA values. When simultaneously considering A1c and QoL, CGM with either CSII or MDI appeared to provide the highest composite ranking in the cluster analysis of SUCRA values.

Conclusions

Integrated insulin pump and CGM systems with low-glucose suspend or hybrid closed-loop capability appeared best for A1c reduction and the composite ranking for A1c and severe hypoglycemia.

Comment

Given the range of different technologies in the modern armament for treating type 1 diabetes, it will be challenging to test all different combinations of options in a single RCT. Unlike standard meta-analyses and randomized controlled trials (RCTs), network meta-analyses utilize both direct and indirect evidence. The network approach facilitates comparison of devices within and between categories of insulin delivery, glucose monitoring, and insulin advisors or the comparison of devices when direct trial evidence is sparse. To our knowledge, this is the first network meta-analysis to integrate the comparison of technologies for insulin delivery, glucose monitoring, and insulin dose calculations in the management of type 1 diabetes.

Limitations of such network analysis include study heterogeneity. In addition, individual patient-level data were not available in this paper. While the potential effect modifiers of age, diabetes duration, and A1c appeared similar for included studies, important aspects such as the patient preference for technology types and implementation strategies could not be assessed. Authors also report that the certainty of the evidence for the treatment effects was very low.

Accepting these limitations, authors report that integrated CSII and CGM systems comprising low-glucose suspend or hybrid closed-loop algorithms appeared best for A1c reduction, composite ranking for A1c and severe hypoglycemia, and possibly QoL. Exploratory cluster analysis of SUCRA data suggested that integrated systems, as well as MDI with CGM, may provide the best compromise of A1c reduction and severe hypoglycemia prevention. Additional cluster analysis suggested that CSII with standalone CGM or MDI with CGM may provide the best compromise of A1c reduction and QoL. For patients using MDI and SMBG, integrated CSII and CGM with low-glucose suspend or hybrid closed-loop algorithms may be the best overall option for the majority of outcomes, with evidence strongest for A1c improvement. If only one technology is desired or practical, then CGM appears most favorable from the composite ranking of A1c, hypoglycemia, and QoL.

Clinicians treating patients with type 1 diabetes are aware that no single technology is suitable for all patients. Ultimately, the impact and use of medical technology are affected by a range of factors, such as efficacy and safety, patient choice and ease of use, familiarity and views of HCPs, funding availability, and other human factors. In the absence of patient-level data, network meta-regression was also not possible.

High-protein meals require 30% additional insulin to prevent delayed postprandial hyperglycaemia

Paterson MA^1,2, Smart CEM^1,2, Howley P³, Price DA^4,5,6, Foskett DC⁷, King BR^1,2

¹Department of Paediatric Endocrinology and Diabetes, John Hunter Children's Hospital, Newcastle, Australia; ²Hunter Medical Research Institute, School of Medicine and Public Health, The University of Newcastle, Callaghan, Australia; ³School of Mathematical and Physical Sciences/Statistics, The University of Newcastle, Rankin Park, New South Wales, Australia; ⁴Pacific Private Clinic, Gold Coast, Australia; ⁵School of Medicine, Bond University, Gold Coast, Australia; ⁶School of Medicine, Griffith University, Gold Coast, Queensland, Australia; ⁷Insulin Pump Angels, Gold Coast, Australia

Diabet Med 2020; 37: 1185–1191

Background

Despite accurate carbohydrate estimation, postprandial hyperglycemia remains a common clinical challenge affecting glycemic control. Previous studies have shown that in addition to carbohydrate, protein and fat content of meals can impact postprandial glycemia in individuals with type 1 diabetes, causing delayed glycemic excursions and increased insulin demand. The objective of this study was to assess the amount of additional insulin required for a high-protein, carbohydrate-containing meal to prevent postprandial hyperglycemia in individuals with type 1 diabetes using insulin pump therapy.

Methods

In this randomized cross-over study, 26 participants aged 8–40 years, HbA1c <65 mmol/mol (8.1%), received a 50 g protein, 30 g carbohydrate, low-fat (<1 g) breakfast drink over five consecutive days at home. Participants' blood glucose levels and insulin doses were reviewed daily for a week prior to the study by a diabetes educator. Adjustments were made to optimize each participant's insulin-to-carbohydrate ratio and ensure 24 h glucose targets were met.

Participants consumed identical test drinks as breakfast over 5 consecutive days following an overnight fast with varying insulin doses given in randomized order. Participants were instructed to fast (water only) for the following 4 h unless they were required to treat hypoglycemia. If fasting glucose levels were >10 or <3.6 mmol/l prior to the drink, participants were instructed to treat as per usual management and the study day excluded and repeated.

On the control day, a standard bolus of insulin (100%) was delivered using a normal bolus. On the remaining 4 study days, additional insulin doses in increments of 15% were added to the standard dose, giving test insulin doses of 115%,130%, 145%, and 160% and delivering them using a combination bolus. The immediate (upfront) portion of the bolus was standardized to deliver 65% of the insulin-to-carbohydrate ratio. The remainder of the insulin was delivered in the extended portion of the combination bolus over 3 h. Doses were commenced 15 min pre-drink. Postprandial glycemia was assessed by 4 h of continuous glucose monitoring. Each test drink contained 50 g (200 kcal) of protein, 30 g (120 kcal) of carbohydrate, and 0.3 g (2.7 kcal) fat derived from a protein powder. Continuous glucose monitoring (Dexcom G4 Platinum, Inc., San Diego, CA) was used to measure interstitial glucose.

Results

The 100% dosing resulted in postprandial hyperglycemia. From 120 min, ≥130% doses resulted in significantly lower postprandial glycemic excursions compared with 100% (P<0.05). A 130% dose produced a mean (SD) glycemic excursion that was 4.69 (2.42) mmol/l lower than control, returning to baseline by 4 h (P<0.001). From 120 min, there was a significant increase in the risk of hypoglycemia compared with the control for 145% (odds ratio [OR] 25.4, 95% confidence interval [CI] 5.5–206; P<0.001) and 160% (OR 103, 95% CI 19.2–993; P<0.001). Some 81% (n=21) of participants experienced hypoglycemia following a 160% dose, whereas 58% (n=15) experienced hypoglycemia following a 145% dose. There were no hypoglycemic events reported with 130%.

Conclusions

This study suggests that for a large (50 g) protein meal, consumed with 30 g carbohydrate and no fat, adding 30% additional insulin to a standard dose and delivering it using a dual wave/combination-type bolus optimizes the postprandial glycemic profile without significantly increasing the risk of hypoglycemia. Further, higher doses of 45% and 60% extra insulin increased the risk of postprandial hypoglycemia.

Comment

In this study, authors assessed the amount of additional insulin required for a high-protein, carbohydrate-containing meal to prevent postprandial hyperglycemia in individuals with type 1 diabetes using insulin pump therapy, showing the need for an extra 30% of insulin delivered in a 3 h extended bolus.

A daily challenge for people living with type 1 diabetes is the need for accurate carb counting. Even when carb counting is deemed accurate, people living with T1D experience variability of postprandial glucose levels. In addition to the timing and adequacy of meal-related insulin doses, other factors that contribute to postprandial glycemia include the glycemic index of carbohydrates as well as the nature and quantity of other macronutrients such as fat and protein (13 –15). Meals high in fat and protein can be associated with prolonged raised glucose levels, particularly overnight. Common examples of such meals include Indian or Asian cuisine, pizza, and meals like fish and chips. However, marked inter-individual differences exist in the impact of fat, protein, and glycemic index of carbohydrates, making it challenging to come up with uniform advice. Previous studies looking at fat-protein meal bolusing have proposed an equation in which 100 kcal of protein and fat require the same amount of insulin as 10 g of glucose (16). Authors in the current study used this equation to determine the upper limit of increased insulin dose of 60%. The findings from the current study suggest that the algorithm proposed by Pankowska (16) may lead to hypoglycemia in some individuals.

One of the limitations of the study is the use of a liquid meal; a solid meal may need even further prolongation of the extended bolus. Given the challenges in carb counting, some patients may also struggle to count the additional protein in meals, which is required to implement the proposed extended meal bolus. The UK Diabetes Technology Network CSII best practice guide suggests for high protein (>25 g)–high fat (>40 g) to consider increasing meal bolus by 25% to 30% in a combination bolus with 50% to 70% given initially and remainder given over 2 to 6 h. The findings of this study further support this recommendation.

TECHNOLOGY IN TYPE 2 DIABETES

Impact of technology on glycaemic control in type 2 diabetes: a meta-analysis of randomized trials on continuous glucose monitoring and continuous subcutaneous insulin infusion

Dicembrini I^1,2, Mannucci E^1,2, Monami M¹, Pala L¹

¹Department of Diabetology, Careggi Hospital, Florence, Italy; ²Department Mario Serio, University of Florence, Florence, Italy

Diabetes Obes Metab 2019; 21: 2619–2625

Background

The objective of this meta-analysis was to assess the effect of CSII, as well as real time and intermittently scanned continuous glucose monitoring (rtCGM and isCGM), on glycemic control in type 2 diabetes.

Methods

The analysis included randomized clinical trials comparing CSII with MDI in people with type 2 diabetes, as well as studies comparing rtCGM or isCGM with SMBG, with a duration of at least 12 weeks, identified in Medline or clinicaltrials.gov. The principal endpoint was HbA1c at the end of the trial. Mean and 95% confidence intervals (CIs) for HbA1c and Mantel–Haenzel odds ratios for severe hypoglycemia were calculated, using random-effect models.

Results

Five studies with 679 participants (CSII: n=344 and MDI: n=335) were eligible for the meta-analysis comparing CSII vs MDI in type 2 diabetes. These studies showed a significant heterogeneity (I²=90%). Using a random-effects model, the difference in HbA1c between CSII and MDI was not statistically significant (−0.26% [95% CI –0.74; 0.22]; P=0.29). In contrast, a sensitivity analysis using a fixed-effect model suggested a significant difference in favor of CSII (−0.38% [95% CI −0.51; −0.25]). Severe hypoglycemia events were rare, and the difference across treatment arms was not statistically significant (OR 0.75 [95% CI 0.24; 2.33]; P=0.61). Criteria for the definition of nonsevere hypoglycemia differed across studies; however, no trial reported significant differences between CSII and MDI. The only study reporting data on nocturnal hypoglycemia found no significant difference between CSII and MDI. Total insulin doses were available for two trials, showing a significantly lower dose for CSII in comparison with MDI (−19 units/day [95% CI –29;–10]; P=0.0001).

Four studies with 439 participants (CGM: n=224 and SMBG: n=215) were eligible for the meta-analysis comparing CGM vs SMBG. No significant (I² value 0%) heterogeneity was noted. Using a random-effect model, the difference in endpoint HbA1c between CGM and SMBG was significant (−0.28% [95% CI –0.43; −0.13]; P<0.01. Similar results were obtained using a fixed-effect model (−0.28% [95% CI −0.43; −0.13]; P<0.01). Only three of the available trials reported data on severe hypoglycemia, specifying that no such events were observed. Nonsevere hypoglycemia was reported among endpoints, with different definitions, in four studies; no significant differences between CGM and SMBG were observed in any of those studies. Only one trial explored the effect of isCCGM, as compared with SMBG, on HbA1c in type 2 diabetes, finding no difference across groups (at study end: 8.4% ± 0.8% vs 8.3% ± 1.1% with isCGM and SMBG, respectively). Conversely, isCGM was associated with an improvement in the quality of life and with a lower incidence of hypoglycemic events.

Conclusion

Authors concluded that CSII, rtCGM, and isCGM provide only small benefits compared with MDI (on either HbA1c, hypoglycemic risk, or quality of life) in insulin-treated people with type 2 diabetes.

Comment

Results of this meta-analysis suggest a limited benefit of CSII therapy compared to MDI in people with type 2 diabetes. However, these findings need to be taken in the context of significant heterogeneity of the studies involved with a wide range of baseline HbA1c levels. A key question is whether CSII offers an additional benefit in those with high HbA1c after optimization with MDI. A previous 2017 meta-analysis used the same five studies used in the current study (17). Pickup et al. have shown that improvement in HbA1c is closely related to the baseline HbA1c. In that study, expected difference in HbA1c between CSII and MDI increased from 0.15% (2 mmol/mol) with a baseline of 8.0% (64 mmol/mol) to 0.59% (6 mmol/mol) with a baseline HbA1c level of 10% (86 mmol/mol). Similarly, in order to show differences in rates of severe hypoglycemia, patients with high rates of baseline hypoglycemia need to be included in the studies. It is also interesting that we see different results based on the different statistical models used. The difference between groups in endpoint HbA1c was not significant using a random-effect model, whereas it was significant with a fixed-effect model. Since there was evidence of heterogeneity, authors argue that the (more conservative) random-effect model should be preferred.

The treatment of type 2 diabetes has evolved considerably in the last decade with the introduction of newer agents such as daily and weekly GLP-1 analogs, DPP-4 inhibitors, and SGLT2 inhibitors. Therefore, future trials of CSII in type 2 diabetes may need to include the above agents in the comparison. Another consideration is the cost of CSII, making it beyond the reach for many healthcare systems in the world. One would assume the cost and complexity of CSII in type 2 diabetes may come down in the future, making it available for those with type 2 diabetes and high HbA1c and high insulin needs despite optimal care with MDI and other agents.

Footnotes

Author Disclosure Statement

RL has research support from the NIDDK (1K23DK122017) and Stanford Maternal Child Health Research Institute. RL has consulted for Abbott Diabetes Care, Biolinq, Capillary Biomedical, Morgan Stanley, and Tidepool. LL reports having received speaker honoraria from Animas, Abbott, Insulet, Medtronic, Novo Nordisk, Roche, and Sanofi; been on advisory panels for Animas, Abbott, Novo Nordisk, Dexcom, Medtronic, Sanofi, and Roche; and received research support from Novo Nordisk and Dexcom.

References

Munshi

, Slyne

, Greenberg

, et al. Nonadherence to insulin therapy detected by Bluetooth-enabled pen cap is associated with poor glycemic control. Diabetes Care, 2019; 42: 1129–1131.

Patel

, Benasi

, Ferrari

, et al. Randomized trial of infusion set function: steel versus teflon. Diabetes Technol Ther, 2014; 16: 15–19.

Hauzenberger

, Hipszer BR, Loeum

, et al. Detailed analysis of insulin absorption variability and the tissue response to continuous subcutaneous insulin infusion catheter implantation in swine. Diabetes Technol Ther, 2017; 19: 641–650.

Norgaard

, Shin

, Welsh

, et al. Performance and acceptability of a combined device for insulin infusion and glucose sensing in the home setting. J Diabetes Sci Technol, 2015; 9: 215–220.

Rumpler

, Mader

, Fischer

, et al. First application of a transcutaneous optical single-port glucose monitoring device in patients with type 1 diabetes mellitus. Biosens Bioelectron, 2017; 88: 240–248.

Tschaikner

, Simic

, Jungklaus

, et al. Development of a single-site device for conjoined glucose sensing and insulin delivery in type-1 diabetes patients. IEEE Trans Biomed Eng, 2020; 67: 312–322.

Girardot

, Mousin

, Vezubet

, et al. Kalman filter-based novel methodology to assess insulin pump accuracy. Diabetes Technol Ther, 2019; 21: 557–565.

Leelarathna

, Roberts

, Hindle

, et al. Comparison of different insulin pump makes under routine care conditions in adults with Type 1 diabetes. Diabet Med, 2017; 34: 1372–1379.

Lal

, Maahs

. Optimizing basal insulin dosing. J Pediatr, 2019; 215: 7–8.

10.

Carroll

, Schade

. The dawn phenomenon revisited: implications for diabetes therapy. Endocr Pract, 2005; 11: 55–64.

11.

Shah

, Grimsmann

, Foster

, et al. Undertreatment of cardiovascular risk factors in the type 1 diabetes exchange clinic network (United States) and the prospective diabetes follow-up (Germany/Austria) registries. Diabetes Obes Metab, 2020; 22: 1577–1585.

12.

Ghanim

, Batra

, Green

, et al. Liraglutide treatment in overweight and obese patients with type 1 diabetes: a 26-week randomized controlled trial; mechanisms of weight loss. Diabetes Obes Metab 2020 May 19. DOI: 10.1111/dom.14090 [Epub ahead of print].

13.

Bell

, Smart

, Steil

, et al. Impact of fat, protein, and glycemic index on postprandial glucose control in type 1 diabetes: implications for intensive diabetes management in the continuous glucose monitoring era. Diabetes Care, 2015; 38: 1008–1015.

14.

Paterson

, Bell

, O'Connell

, et al. The role of dietary protein and fat in glycaemic control in type 1 diabetes: implications for intensive diabetes management. Curr Diab Rep, 2015; 15: 61.

15.

Elleri

, Allen

, Harris

, et al. Absorption patterns of meals containing complex carbohydrates in type 1 diabetes. Diabetologia, 2013; 56: 1108–1117.

16.

Pankowska

, Blazik

, Groele

. Does the fat-protein meal increase postprandial glucose level in type 1 diabetes patients on insulin pump: the conclusion of a randomized study. Diabetes Technol Ther, 2012; 14: 16–22.

17.

Pickup

, Reznik

, Sutton

. Glycemic control during continuous subcutaneous insulin infusion versus multiple daily insulin injections in type 2 diabetes: individual patient data meta-analysis and meta-regression of randomized controlled Trials. Diabetes Care, 2017; 40: 715–722.