Type 1 diabetes mellitus (DM1) is a lifelong condition caused by autoimmune or idiopathic destruction of pancreatic islet β-cells. This results in an absolute insulin deficiency, which causes subsequent hyperglycemia and, if not corrected, ketoacidosis. DM1 has an approximate prevalence of 2.6–3.4/1000 people in the United States.1 These individuals are subject to potentially chronic hyperglycemia, which can cause dysfunction in multiple organ systems due to tissue glycosylation, oxidative stress, protein kinase C activation, and other factors. Complications are frequently categorized as macrovascular, including cardiac complications, atherosclerotic coronary artery disease, hypertension, and associated diastolic dysfunction, and microvascular, including nephropathy, retinopathy, and neuropathy. Improved glucose control can reduce some of the complications of diabetes; however, strategies aimed at stricter glycemic control increase the risk of hypoglycemia and secondary untoward events including loss of consciousness, seizures, and death.2–4
A high level of patient engagement is required to control hyperglycemia in those with DM1 to reduce both short- and long-term consequences of the disease while avoiding hypoglycemia. Maintenance of optimal glycemic control can be challenging due to numerous patient factors, amplified by hourly, daily, or longer variations in insulin responsiveness due to cortisol cycles, physical and physiological stress, intercurrent illness, frequency and degree of exercise, and diet.5
The Food and Drug Administration (FDA) categorizes artificial pancreas device (APD) systems as “Threshold Suspend,” which reduce or discontinue insulin for low-glucose thresholds, “Control-to-Range,” which adjust infusions for sensed glucose values outside of high and low thresholds, and “Control-to-Target,” which constantly adjust infusions toward a set glucose value. The latter 2 categories of APDs, with control-based algorithms (closed-loop artificial pancreas devices [C-APDs]), can be insulin-only or bihormonal.6 Regardless of category, the goal of C-APD system use is to improve glycemic control with a lower risk of hypoglycemia.7–9 These systems also require less involvement in the hour-to-hour management of diabetes, improving user acceptance.7–10
Perioperative physicians are increasingly likely to encounter patients undergoing inpatient and outpatient procedures who are using C-APDs. An appreciation of the elements of such systems, their respective weaknesses, and insights into how to use these systems safely in the perioperative period forms the basis of this commentary. This will also provide a framework for understanding the likely clinical evolution and future indications for use of C-APDs.
ARTIFICIAL PANCREAS DEVICES
At its core, a C-APD requires an accurate, reliable, robust, and frequently sensing glucose monitor, an algorithm that dictates insulin administration, and an insulin or bihormonal delivery system. The Juvenile Diabetes Research Foundation describes APD systems in 3 generations encompassing 6 stages. First-generation systems (stages 1–3) lack the control algorithms described above and are instead geared toward decreasing hypoglycemic events. The second generation, fourth-stage systems are hybrid closed loops, allowing for manual-assist insulin boluses with meal times. The second generation also includes fifth-stage fully automated closed-loop devices (Figure 1). Finally, the third generation (stage 6) consists of fully automated multihormone closed-loop devices.7
The FDA regulates APDs as class III devices—“high risk of potential injury.” Ideally, however, C-APDs can safely improve glycemic control through more frequent glucose sampling and insulin adjustments than a patient could realistically perform. The literature has reported the safety of C-APDs in both outpatient and modified outpatient (eg, diabetes camp) settings, and finally in the home environment with and without remote monitoring.8–10 The use of C-APDs has improved glycemic control and decreased hypoglycemic events,9 even in poorly controlled adolescents with DM1.10
The continuous glucose monitor (CGM) in a C-APD is typically a subcutaneous enzymatic glucose sensor combined with a transmitter and a receiver. The sensor is inserted under the skin in a fatty area of the body, typically the abdomen, to sense interstitial fluid. An enzymatic reaction with glucose via glucose dehydrogenase or glucose oxidase occurs within the sensor and provides a current that is directly proportional to the glucose concentration. This current is calibrated to finger-stick glucose at least twice daily. Some C-APDs have a second sensor, which may act as a backup in case of primary sensor failure or to improve the accuracy of the device. Of note, there are nonenzymatic CGMs in development that use glucose-responsive polymer microgels.11
Any inaccuracy in the CGM that yields a reported glucose value higher than the actual blood glucose (a positive-sensor deviation) can delay hypoglycemia alarms and thus pose a direct risk to the patient. This is particularly important due to the rapidity with which hypoglycemia can develop, and the fact that the CGM reports glucose values up to 15 minutes behind the finger-stick glucose meter. For example, calibration when the blood glucose is dropping rapidly from 150 to 120 mg/dL over 10 minutes would result in the CGM being calibrated toward the “higher” glucose that remains in the interstitial fluid (150 mg/dL). This would result in delayed hypoglycemia alarms and thus, intervention, if left uncorrected and the glucose continued to drop. The majority of this delay is physiological and due to to systemic-to-interstitial glucose transport.12–14 Intraoperatively, factors such as vasopressor administration or hypothermia could potentially increase this physiologic delay or introduce other inaccuracy, as has been found with capillary glucometry in states of poor perfusion.15 There are added delays for processing and data filtering time as well as for the enzymatic reaction at the sensor tip.12–14
In 2016, Blauw et al16 reviewed the multiple causes of inaccuracy associated with subcutaneous enzymatic sensors. The most prominent CGM error is related to calibration. Simple user error may result in inaccurate reference input; however, errors can also be due to calibration error when the interstitial and blood glucose are not at equilibrium due to delays addressed previously.12–14 Outside of calibration, the accuracy of a CGM can be decreased during hypoglycemic and hyperglycemic periods,17,18 and in the hours after sensor insertion.19 Finally, sensor values can drift over many hours related to “biofouling” caused by inflammatory tissue depositing on the sensor surface.20–22 There are many other sources of calibration error including edema, dehydration, and chemical interferences that can result in error in glucose measurements.23
It is challenging to use traditional statistical metrics to assess CGMs because they are not as accurate as a blood glucose meter yet provide added trend monitoring. Frequently, the mean absolute relative difference (MARD), defined as the mean value of individual absolute relative deviations between the CGM and reference input (Figure 2), is used. However, a reference of 220 mg/dL and CGM reading of 200 mg/dL—an absolute relative deviation of 10%—may not change management or induce risk. Therefore, statistical tools such as consensus error grid (CEG) and surveillance error grid (SEG) analyses are used to determine the degree of risk posed by inaccurate measurements.23–27 The Guardian Sensor 3 (Medtronic, Fridley, MN) was recently evaluated by Christiansen et al,28 who reported a MARD of 8.4%–9.6% compared to reference values. Similarly, Forlenza et al29 reported use of the Medtronic Enlite 2 CGM (Fridley, MN) postoperatively following pancreatectomy with islet cell autotransplantation in adults. Postoperative islet engraftment is improved with strict glycemic control, thereby increasing the risk of hypoglycemic events. The authors reported a MARD of 11.0%. CEG analysis showed 99.4% of pairs in the low risk of error zones A and B, and SEG analysis showed 99.19% of pairs in the no-risk or low-risk zones, and 0.81% of values in the “moderate, lower” risk zone.29 Further, Elder et al30 in 2017 evaluated the Dexcom G4 Platinum CGM device (Dexcom, San Diego, CA) for the same procedure in pediatrics, reporting a MARD of 10.6%, with CEG analysis showing 100% of values in clinically acceptable zones and SEG analysis showing 96% of paired values within clinically acceptable agreement.
While these sensors do offer trend monitoring and therefore may improve anticipation of dysglycemia, it is important to note that fundamental accuracy remains far less than reference laboratory testing and many devices would struggle to meet the current FDA guidance for blood glucose meter accuracy.17,23,24,27,31 The problem of meeting the stringent FDA requirements despite reasonable safety by CEG and SEG analysis is well reviewed by Liang et al.32 Although some studies suggest unchanged accuracy of CGMs in septic shock,33 in grouped-together “critically ill” patients34,35 and in patients requiring norepinephrine,36 for some critically ill patients, the accuracy of a CGM may be inadequate.17 Nonetheless, there is rising argument that added trend monitoring with CGMs improves overall safety.35,37
After the subcutaneous current is sensed and translated into a glucose concentration, the C-APD dictates an adjustment to insulin administration and, with resensing of the glucose, a closed-loop emerges (Figure 1). Control-to-range and control-to-target algorithms are highly challenging to develop due to individual variance in insulin pharmacokinetics between patients and within the same patient over hours during real-world situations (eg, exercise or illness) or as pathophysiological states evolve.38 Further difficulty is encountered due to the delayed peak effect and longer duration of some insulin preparations. These pharmacokinetic challenges are being addressed by using faster-acting insulins, enabling user input in hybrid closed-loop systems, and by adding other sensed inputs.38 For example, adding a heart rate input to the C-APD can decrease the risk of hypoglycemia during exercise.39 In 2016, Trevitt et al40 comprehensively reviewed the cognitive problems encountered in the development of algorithms and the multiple logic strategies applied.
The infusion pump is the most robust portion of the C-APD. Pumps can contain insulin with or without glucagon or amylin. Insulin-only pumps tend to be favored due to the poor stability of glucagon and the simplicity of “insulin-only” C-APDs.
THE MINIMED 670G SYSTEM
In September 2016, the FDA approved the MiniMed 670G C-APD (Medtronic) for marketing to patients 14 years of age and older who have type 1 diabetes. The MiniMed 670G is a second generation, fourth-stage, hybrid closed-loop design C-APD.7 It consists of a Contour Next Link 2.4 Meter (Ascensia Diabetes Care Holdings AG, Basel, Switzerland) with a 1-press inserter, which transmits via a Guardian Link 3 transmitter (Medtronic, Fridley, MN) to the insulin pump, which contains the algorithmic software (Figure 3).41 The MiniMed system algorithm uses a control-to-target strategy, aiming for a glucose concentration of either 120 mg/dL (standard) or, in certain situations, 150 mg/dL.
DISCUSSION: INTRAOPERATIVE UTILIZATION
There is currently no guidance for the use of the commercially available C-APDs during surgical procedures. While C-APDs are not FDA cleared for intraoperative use, we expect that with increasing utilization of these devices, there will be concomitant pressure to study intraoperative use. Two recent studies report safe use of C-APDs in the operating room. The first, by Hirose et al42, detailed the use of a Nikkiso STG-22 C-APD (Nikkiso, Tokyo, Japan) during an insulinoma resection for a 71-year-old patient, using a control-to-range strategy targeting glucose of 80–120 mg/dL. There were no adverse events during the procedure, and glucose control was adequately managed using the device. Next, Mita et al43 reported decreased postoperative creatinine elevations using a Nikkiso STG-55 C-APD system. They used a control-to-range strategy targeting a glucose of 100–150 mg/dL during 19 hepatectomy procedures, compared to 19 controls managed with a traditional insulin sliding scale using intermittent intravenous bolus doses of insulin. Patients undergoing hepatectomies can have profound glycemic alterations due to hypoxia-induced glycogen release caused by Pringle clamping used to limit bleeding during hepatic resection. The authors reported no adverse events, a CEG analysis showing 100% of paired values in the clinically acceptable zones, and less postoperative creatinine elevation with use of the closed-loop system.42
A future joint anesthesiology and endocrinology multidisciplinary guideline for intraoperative use of C-APDs is necessary for safe, clear, and standardized management. We propose the following to guide the discussion toward formal recommendations. First, there must be reasonable understanding of how the system functions, including the insulin suspend feature. On the 670G, this is on the home menu of the device, labeled “suspend delivery.” Next, immediate availability of experts to troubleshoot the device is mandatory. Finally, Medtronic instructs patients to remove their pump, sensor, transmitter, and meter before entering a room with x-ray, magnetic resonance imaging, or computed tomography scanning equipment, so the surgical location also needs to be considered.44
One to 2 days before surgery, the patient should place the CGM sensor in an area as far from expected electrocautery as possible, well away from the surgical field and at no risk of positioning dislodgment or pressure. In 2006, Piper et al45 reported many episodes of CGM system error alarms when using the Medtronic Guardian RT CGM coinciding to timing of electrocautery use.
Generally, the CGM is placed in the abdomen, potentially precluding use during certain procedures. However, Christiansen et al28 reported good correlation between abdominal and arm insertion sites for the Guardian Sensor 3 device,27 and Weinstein et al46 reported the same for the FreeStyle Navigator CGM (Abbott Diabetes Care, Alameda, CA). Furthermore, the device should be far from any inflamed area. The device and surrounding area should be covered with a dressing that allows for adequate preparation, if required, and that allows a provider to check the site throughout the procedure and remove the device, if necessary.
In the preoperative area, the sensor needs to be calibrated to a capillary (finger-stick) glucose meter, and the provider should be aware of how to calibrate the sensor. Although standardized calibration would be at least twice daily (Medtronic suggests 4 times a day calibration), we would suggest increased frequency of calibration intraoperatively; every 1–2 hours. Because the MiniMed 670G uses a control-to-target strategy with a target glucose of 120 or 150 mg/dL, this target should be confirmed. Intraoperatively, the pump should then be allowed to run, titrating the insulin infusion as needed, with an established “exit” plan to turn the device off for significant dysglycemia, error compared to reference, or for any development of a state that would make the device less accurate (eg, development of shock, vasoactive medication requirement, or device positioning compromise). Importantly, the device should be discontinued early for any concern rather than used inappropriately or with any potential risk (Table).
APDs are being developed rapidly. In 2016, Trevitt et al40 reported 18 closed-loop systems representing various C-APD generations under development and in various phases of testing.40 These include single and bihormonal hybrid and fully automated closed-loop systems. Alternative insulin regimens such as fast-acting insulin aspart are also being developed.47 Given the improved safety and glycemic control with C-APDs, with the decreased requirement for user input, we expect the deployment of these devices to increase rapidly. Anesthesiologists should be aware of the components of evolving C-APDs and be active in the development of systems that facilitate intraoperative use. The use of these devices has the potential to improve patient safety through better intraoperative glycemic control, which may translate to enhanced outcomes and limited morbidity.
Name: Micah T. Long, MD.
Contribution: This author helped prepare the manuscript, develop recommendations, and approve the final manuscript.
Name: Douglas B. Coursin, MD.
Contribution: This author helped prepare the manuscript, develop recommendations, and approve the final manuscript.
Name: Mark J. Rice, MD.
Contribution: This author helped prepare the manuscript, develop recommendations, and approve the final manuscript.
This manuscript was handled by: Maxime Cannesson, MD, PhD.
1. Menke A, Orchard TJ, Imperatore G, Bullard KM, Mayer-Davis E, Cowie CCThe prevalence of type 1 diabetes in the United States. Epidemiology. 2013;24:773–774.
2. Nathan DM, Genuth S, Lachin J, et alThe Diabetes Control and Complications Trial Research GroupThe effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. N Engl J Med. 1993;329:977–986.
3. Nathan DM, Cleary PA, Backlund JY, et alDiabetes Control and Complications Trial/Epidemiology of Diabetes Interventions and Complications (DCCT/EDIC) Study Research Group. Intensive diabetes treatment and cardiovascular disease in patients with type 1 diabetes. N Engl J Med. 2005;353:2643–2653.
4. Cryer PEThe barrier of hypoglycemia in diabetes. Diabetes. 2008;57:3169–3176.
5. Ruan Y, Thabit H, Leelarathna L, et alAP@home Consortium. Variability of insulin requirements over 12 weeks of closed-loop insulin delivery in adults with type 1 diabetes. Diabetes Care. 2016;39:830–832.
8. Thabit H, Hovorka RComing of age: the artificial pancreas for type 1 diabetes. Diabetologia. 2016;59:1795–1805.
9. Garg SK, Weinzimer SA, Tamborlane WVGlucose outcomes with the in-home use of a hybrid closed-loop insulin delivery system in adolescents and adults with type 1 diabetes. Diabetes Technol Ther. 2017;19:155–163.
10. Tauschmann M, Allen JM, Wilinska MEHome use of day-and-night hybrid closed-loop insulin delivery in suboptimally controlled adolescents with type 1 diabetes: a 3-week, free-living, randomized crossover trial. Diabetes Care. 2016;39:2019–2025.
11. Chen C, Zhao XL, Li ZH, Zhu ZG, Qian SH, Flewitt AJCurrent and emerging technology for continuous glucose monitoring. Sensors (Basel). 2017;17:e182.
12. Boyne MS, Silver DM, Kaplan J, Saudek CDTiming of changes in interstitial and venous blood glucose measured with a continuous subcutaneous glucose sensor. Diabetes. 2003;52:2790–2794.
13. Keenan DB, Mastrototaro JJ, Voskanyan G, Steil GMDelays in minimally invasive continuous glucose monitoring devices: a review of current technology. J Diabetes Sci Technol. 2009;3:1207–1214.
14. Pleus S, Schoemaker M, Morgenstern KRate-of-change dependence of the performance of two CGM systems during induced glucose swings. J Diabetes Sci Technol. 2015;9:801–807.
15. Desachy A, Vuagnat AC, Ghazali ADAccuracy of bedside glucometry in critically ill patients: influence of clinical characteristics and perfusion index. Mayo Clin Proc. 2008;83:400–405.
16. Blauw H, Keith-Hynes P, Koops R, DeVries JHA review of safety and design requirements of the artificial pancreas. Ann Biomed Eng. 2016;44:3158–3172.
17. Rijkenberg S, van Steen SC, DeVries JH, van der Voort PHJAccuracy and reliability of a subcutaneous continuous glucose monitoring device in critically ill patients. J Clin Monit Comput. 2018;32:953–964.
18. Rodbard DCharacterizing accuracy and precision of glucose sensors and meters. J Diabetes Sci Technol. 2014;8:980–985.
19. Mauras N, Fox L, Englert K, Beck RWContinuous glucose monitoring in type 1 diabetes. Endocrine. 2013;43:41–50.
20. Bequette BWFault detection and safety in closed-loop artificial pancreas systems. J Diabetes Sci Technol. 2014;8:1204–1214.
21. Castle JR, Ward WKAmperometric glucose sensors: sources of error and potential benefit of redundancy. J Diabetes Sci Technol. 2010;4:221–225.
22. Tauschmann M, Allen JM, Wilinska MESensor life and overnight closed loop: a randomized clinical trial. J Diabetes Sci Technol. 2017;11:513–521.
23. Rice MJ, Smith JL, Coursin DBGlucose measurement in the ICU: regulatory intersects reality. Crit Care Med. 2017;45:741–743.
24. Clarke WL, Kovatchev BContinuous glucose sensors: continuing questions about clinical accuracy. J Diabetes Sci Technol. 2007;1:669–675.
25. Klonoff DC, Lias C, Vigersky R, et alError Grid Panel. The surveillance error grid. J Diabetes Sci Technol. 2014;8:658–672.
26. Obermaier K, Schmelzeisen-Redeker G, Schoemaker MPerformance evaluations of continuous glucose monitoring systems: precision absolute relative deviation is part of the assessment. J Diabetes Sci Technol. 2013;7:824–832.
27. Rice MJ, Coursin DBContinuous measurement of glucose: facts and challenges. Anesthesiology. 2012;116:199–204.
28. Christiansen MP, Garg SK, Brazg RAccuracy of a fourth-generation subcutaneous continuous glucose sensor. Diabetes Technol Ther. 2017;19:446–456.
29. Forlenza GP, Nathan BM, Moran AAccuracy of continuous glucose monitoring in patients after total pancreatectomy with islet autotransplantation. Diabetes Technol Ther. 2016;18:455–463.
30. Elder DA, Jiminez-Vega JM, Hornung LN, et alContinuous glucose monitoring following pancreatectomy with islet autotransplantation in children. Pediatr Transplant. 2017;21:e12998.
32. Liang Y, Wanderer J, Nichols JH, Klonoff D, Rice MJBlood gas analyzer accuracy of glucose measurements. Mayo Clin Proc. 2017;92:1030–1041.
33. Lorencio C, Leal Y, Bonet AReal-time continuous glucose monitoring in an intensive care unit: better accuracy in patients with septic shock. Diabetes Technol Ther. 2012;14:568–575.
34. Boom DT, Sechterberger MK, Rijkenberg SInsulin treatment guided by subcutaneous continuous glucose monitoring compared to frequent point-of-care measurement in critically ill patients: a randomized controlled trial. Crit Care. 2014;18:453.
35. Preiser JC, Lheureux O, Thooft A, Brimioulle S, Goldstein J, Vincent JLNear-continuous glucose monitoring makes glycemic control safer in ICU patients. Crit Care Med. 2018;46:1224–1229.
36. Holzinger U, Warszawska J, Kitzberger RReal-time continuous glucose monitoring in critically ill patients: a prospective randomized trial. Diabetes Care. 2010;33:467–472.
37. Krinsley JS, Chase JG, Gunst JContinuous glucose monitoring in the ICU: clinical considerations and consensus. Crit Care. 2017;21:197–203.
38. Bequette BWChallenges and recent progress in the development of a closed-loop artificial pancreas. Annu Rev Control. 2012;36:255–266.
39. Breton MD, Brown SA, Karvetski CHAdding heart rate signal to a control-to-range artificial pancreas system improves the protection against hypoglycemia during exercise in type 1 diabetes. Diabetes Technol Ther. 2014;16:506–511.
40. Trevitt S, Simpson S, Wood AArtificial pancreas device systems for the closed-loop control of type 1 diabetes: what systems are in development? J Diabetes Sci Technol. 2016;10:714–723.
42. Hirose K, Kawahito S, Mita NUsefulness of artificial endocrine pancreas during resection of insulinoma. J Med Invest. 2014;61:421–425.
43. Mita N, Kawahito S, Soga TStrict blood glucose control by an artificial endocrine pancreas during hepatectomy may prevent postoperative acute kidney injury. J Artif Organs. 2017;20:76–83.
45. Piper HG, Alexander JL, Shukla A, et alReal-time continuous glucose monitoring in pediatric patients during and after cardiac surgery. Pediatrics. 2006;118:1176–1184.
46. Weinstein RL, Schwartz SL, Brazg RL, Bugler JR, Peyser TA, McGarraugh GVAccuracy of the 5-day FreeStyle Navigator Continuous Glucose Monitoring System: comparison with frequent laboratory reference measurements. Diabetes Care. 2007;30:1125–1130.
47. Zijlstra E, Demissie M, Graungaard T, Heise T, Nosek L, Bode BInvestigation of pump compatibility of fast-acting insulin aspart in subjects with type 1 diabetes. J Diabetes Sci Technol. 2018;12:145–151.