The First Hybrid Closed-Loop Insulin Pump: Will It Meet Its Potential?

Soon after the introduction of continuous subcutaneous insulin infusion (CSII), the vision has always been for the development of an “artificial pancreas”. ¹ The progress has been elusive because of the lack of an appropriate glucose monitor. Eventually, continuous glucose monitoring (CGM) became a reality, ² although even today, in a registry of almost 18,000 patients with type 1 diabetes and a decade after its introduction, penetration has been sparse for patients using both CSII (13%) and multiple daily injections (2%). ³ Although there are many reasons for this, ⁴ it is safe to assume that the use of both CSII and CGM is more frequent in the academic diabetes clinics and in registries such as the T1D Exchange. For example, ∼60% of patients in the T1D Exchange (N = 16,061) use CSII, whereas in the United States the use of insulin pump therapy is estimated at 30%–35%. ⁵

Despite the initial low usage of CGM, integration of CSII and CGM has become a reality. The evolution of a truly closed-loop system first requires the use of a hybrid closed-loop pump and sensor system. Diabetes technophiles have been awaiting this merger of these components for many years, and it appears that these futuristic visions are in sight.

In this issue of Diabetes Technology and Therapeutics, ⁶ Garg and colleagues have expanded on their originally published data, ⁷ emphasizing the potential benefit of this technology. After 3 months, significant improvements in HbA1c levels occurred in both adolescents and adults (7.8% ± 0.8% to 7.1% ± 0.6%, P < 0.001, and 7.3% ± 0.9% to 6.8% ± 0.6%, respectively). For both groups, those with HbA1c levels at baseline >7.5% experienced a mean HbA1c reduction of ∼1%. Hypoglycemic exposure was also reduced. Expressed as minutes instead of percentage time, in adolescents, daily time spent below 70 mg/dL was reduced from 62 to 40 min (P = 0.009), whereas in adults, it was reduced from 92 to 40 min (P < 0.001). This technology could only be successful with an accurate CGM system, which has been a criticism of past systems. ⁴ The overall mean absolute relative difference was 10.3% ± 9.0%, similar to the other currently used CGM devices on the market. ⁸ Although no severe hypoglycemia was reported and is encouraging, this should not be over-interpreted for a 3-month study. Longer experiences will provide us with more realistic severe hypoglycemia rates.

There are several important limitations and caveats that require consideration. The authors appropriately note that this was a single-arm nonrandomized study with a primary end point being “safety” of the system as opposed to more traditional glycemic parameters. And although the glycemic end points indeed are impressive, the study duration is for only 3 months. How would these patients do after 6, 12, or 24 months? In a Cochrane meta-analysis of CSII, HbA1c reduction compared with that of multiple injections was small and differences between the two therapies were less with longer duration studies (more than 6 months). ⁹ Theoretically, over longer periods of time, if the patient is using the system correctly and the system is in the “auto mode” (closed loop), patient behaviors would not be as relevant as it is now with completely open loop systems. This hypothesis remains to be proven, however.

One also has to wonder about how frequently alarms were alerting patients, and how “alarm fatigue” may have impacted patient's (and family member's) quality of life (QOL). Anecdotally, this has been a major problem with sensor-augmented pumps, but with all CGM, this has been found to be a barrier in one-third of patients. ⁴ Hopefully, overall QOL measures will be reported in a subsequent publication. The fact that 80% of subjects agreed to participate in a continuation phase suggests a positive improvement in QOL scores.

This hybrid closed-loop system will only function in auto mode closed loop when carbohydrate intake is announced. The investigators should be congratulated in that for the adolescents and adults, the system was in auto mode 75.8% and 88.0%, respectively. Use of pump bolus calculators is quite variable, and increased use is associated with lower HbA1c levels. ¹⁰ The concern of course is for those many patients who do not use the bolus calculators (despite our encouragement) and their continued reluctance once they initiate use of the hybrid closed-loop pump. This is one of several reasons why we should not be surprised if “real-world” results do not match the results of this pivotal trial, particularly with regard to time in auto mode.

Still, the use (or not) of bolus calculators is not the only concern about how this technology will fare outside of this pivotal trial. Like most studies on diabetes technology, the site PIs and research coordinators were experts in CSII and CGM use. The majority of subjects were already using CGM. Will non-CGM-using patients in a real world setting fare as well as this study population, especially those with higher HbA1c levels?

I am particularly concerned about non-CGM-using patients starting this therapy outside of a clinical trial, although those using CGM may have more challenges than we can anticipate. Not reported by the authors were the number of nonscheduled calls, e-mails, texts, and fax messages. It is acknowledged that the frequent contact of the subjects with study personnel may limit the generalizability of the study. Still, one has to wonder in our current era of relative value units counting and productivity-centric systems that lack an infrastructure for this type of therapy, how the system will be able to absorb a potential mass of new communications? And the clinical needs from nonphysician educators are not the only potential barriers, because more paperwork, reimbursement fights, and peer-to-peer discussions with nonexperts will become common. For the endocrinologist in a large multispecialty practice with (hopefully) a part-time registered nurse/certified diabetes educator, will the time intensity to perform this therapy correctly, or even close to what was done in the pivotal trial, be possible?

In my view, any new diabetes therapy cannot be adequately analyzed without a proper download of the data. Although patients have the opportunity to do that on their own, that does not occur as the majority of patients never download at home. ¹¹ And although there are no data about the frequency of meter, pump, or sensor downloads in physician offices, especially outside of large academic centers, my anecdotal observation is the number is small.

For all healthcare professionals, the management of these patients is time intensive, and that is our greatest problem. In a nonscientific survey at a recent endocrinology meeting, out of 200 nonpediatric physicians, approximately one-third had slots for these patients totaling 15 min, another one-third were allotted 20 min, whereas the rest could spend 30 min for a return patient with type 1 diabetes. My prediction: for those physicians with little nonphysician support who are not able to download who spend no more than 15–20 min with each patient, outcomes using the hybrid closed-loop system will be disappointing.

It is my belief that physicians who do not have access to educators (especially knowledgeable with this therapy) need to do a better job of working with industry for both patient education and once-and-for-all develop a system in their offices for downloading. Ideally, the patient should be able to download at home. We have certainly tried this in our clinic and some patients do well with this but most do not. Perhaps that can change.

My overall conclusion is that the hybrid closed-loop pump soon to be available to our patients is a potential major advance in type 1 diabetes therapy. And we certainly are all excited about the potential for a fully closed-loop system, perhaps with a faster insulin. ¹² Watching the technology improve is quite exciting. Still, without appropriate infrastructure, time to spend with the patients (both in person and nonface to face), and financial resources for the healthcare system (yes, the physicians need to be appropriately compensated), my concern is we will underperform the potential for the hybrid closed loop.

I hope my concerns are proven wrong.