Analysis of “Laboratory and Benchtop Performance of a Mealtime Insulin Delivery System”

Abstract

In the current issue of Journal of Diabetes Science and Technology, Dreon et al give a comprehensive overview of the technical performance of a wearable patch for bolus insulin delivery. The test results generated by the manufacturer of the bolus-patch provide the technical prerequisites for clinical application. As the device received FDA clearance already in 2010, positive results from the nonclinical performance testing were to be expected, but present nevertheless interesting insights into the device development. The single-dose accuracy verification results seem especially promising, but tighter accuracy criteria could have been specified and information on outliers is missing from the analysis. The clinical application of the bolus-patch is currently under investigation in a large-scale 44-week intervention trial.

Insulin therapy is a necessity for patients with type 1 diabetes as well as for many patients with type 2 diabetes with progression of the disease. Despite numerous advances in recent years with therapeutic options for patients with type 2 diabetes, many patients still fail to achieve glycemic targets. Major reasons for suboptimal glycemic control are a delay in treatment intensification and a poor adherence to the appropriate therapeutic regimen.^1,2 Therapeutic inertia seems to be more pronounced for starting insulin therapy.³ A recent review identifying barriers to insulin initiation and intensification lists fear of hypoglycemia, burdensome regimens, and weight gain as major barriers alongside fear of injections.⁴ Improvements in injection devices can help to tackle some of the aforementioned barriers, especially by reducing the burden of multiple daily injections (MDI), for example, less interference with daily activities, more discrete insulin administration and reduced injection pain, and overcoming the need for needle injections. Compared to the traditional vial/syringe method of administering insulin, prefilled pen injectors are preferred by patients and are associated with enhanced treatment compliance and health outcomes.^5-8 New injection mechanisms for prefilled pens are being developed to simplify insulin delivery and improve user experience further.⁹ And in a recent editorial in this journal, a strong case was made that new smart pens are likely to be beneficial in improving effective use of insulin.¹⁰ Switching poorly controlled patients with type 2 diabetes from MDI to continuous subcutaneous insulin infusion (CSII) has shown potential to provide benefits in terms of HbA1c reduction and a reduction of total daily insulin dose, but a widespread adoption of insulin pump therapy in this patient collective is hampered by the complexity of the devices with extensive training requirements and a lack of reimbursement.^11-13

Various patch pumps, such as the Omnipod Insulin Management System (Insulet Corp, Bedford, MA, USA), V-Go (Valeritas, Inc, Bridgewater, NJ, USA), PAQ (CeQur, Montreux, Switzerland), or JewelPUMP (Debiotech SA, Lausanne, Switzerland), were designed to provide a simpler, smaller and cheaper technical solution for CSII compared to conventional pumps. In the current article by Dreon et al¹⁴ an even simpler device, a cross-over technology between pen injector and patch pump, was evaluated. Designed by Calibra Medical (Wayne, PA, USA), the device is referred to as a wearable, bolus-only, mealtime insulin delivery system or “bolus-patch.” Although not mentioned in the article, this bolus-patch is (or was?) also known under the names Calibra Finesse and OneTouch Via and received original FDA clearance in 2010, but has not yet been made available commercially.

In the article by Dreon et al¹⁴ a comprehensive overview of the benchtop or technical performance of this bolus-patch was presented and the authors conclude that the patch is a safe, accurate and reliable device for mealtime insulin delivery. The article is of interest because it provides data on the technical prerequisites for use of this patch in patients, something that is more often than not disclosed for new medical devices. For those who are not too familiar with medical device development it may be of particular interest to see how thoroughly regulated this technical testing is, highlighted by the 28 (out of 52) references to standard requirements, guidelines, practice, and quality documents. However, the timing and relevance of the publication, years after receiving FDA clearance for the investigational device, is questionable. The presented data are likely to be a selection of the nonclinical performance data evaluated by the FDA for conformance with the consensus standards and FDA guidance at the time (https://www.accessdata.fda.gov/cdrh_docs/pdf9/k093065.pdf).

For a medical device to receive FDA clearance or CE certification, it has to pass a biocompatibility evaluation and if required additional chemical or toxicological assessments. It is thus not surprising that this FDA cleared patch passed all those standard safety assessments.

From a performance perspective, the most interesting data are presented by the dose accuracy verification testing. For a 2-unit dose to be considered accurate, the authors allowed for a 10% under- or overdose margin, that is, a dose range of 1.8 to 2.2 units. Theoretically the first and last dose administration should be approximately 0.3 units short due to the dead space of the device’s cannula. Excluding the first and last dose, the mean (95%CI) dose in units was 1.98 (1.94-2.03). With more than 12,000 dose administrations performed under different environmental conditions this accuracy testing therefore provides robust data that the delivered dose is accurate in at least 95% of all dosings. Although these results are very promising, especially compared to previously published accuracy evaluations of (patch) pump technologies,¹⁵ unfortunately no data in the article is given on the number and magnitude of outliers. Another minor point of criticism is that the 10% error margin applied in the testing does not meet the accuracy specification of the current ISO guideline 11608-1 on needle-based injection systems for medical use (such as pen injectors). This guideline specifies a 1 unit deviation for doses below 20 units and a tighter 5% error margin for doses of 20 units or more. This also illustrates another minor issue with the dose accuracy testing: the definition of dose and its relevance for clinical practice. The mealtime patch delivers discrete 2-unit doses with each squeeze, but patients will require a multiple of those doses to cover a meal. The technical accuracy of larger doses was formally not tested, but seems to be a given based on the 2-unit dose accuracy data. In clinical practice, dose accuracy will however be influenced by the patient’s ability to count the number of 2-unit doses administered, as the patch does not contain a dose counter. Patient usability studies performed by the patch manufacturer demonstrated that for doses up to 18 units, the correct amount of 2-unit doses was given 100% of the time (unpublished data).

With the technological prerequisites for the device fulfilled, Dreon et al briefly turn to discuss the clinical safety and efficacy data gathered with the device in patients with type 1 and type 2 diabetes. Findings of the published clinical work mainly show that most patients would prefer to use the bolus-patch over continuing using either vial/syringe or pen injector.^16,17 A recently completed 44-week intervention trial in 280 patients with type 2 diabetes who are suboptimally controlled on basal insulin (clinicaltrials.gov identifier NCT02542631) should provide further information on the clinical efficacy of the device.