Pump Users Clamor for Faster Insulin: Is Fast-Acting Insulin Aspart Ready for Them?

Abstract

Recently approved in Europe, Canada, and the United States, fast-acting insulin aspart (FIASP®) is a new rapid acting insulin. Approved for subcutaneous or IV injection use, there is little data available regarding the clinical utility of FIASP in insulin pumps. The article by Zijlstra and colleagues in this issue begins to close this gap by testing pump compatibility of FIASP in the clinic. Reporting on a small (37 subjects) and short (6 weeks) study looking at aspects of infusion set propensity for clogging and malfunction, no cases of infusion set plugging in either FIASP (25 subjects) or insulin aspart (12 subjects) were seen. Unexplained hyperglycemia and premature infusion set changes were more common with FIASP than with insulin aspart. This study demonstrated sufficient safety and efficacy of FIASP in the pump setting to pave the way for longer, larger and more definitive clinical trials.

In this issue of JDST, Zijlstra and colleagues report on the insulin pump compatibility of a newly approved fast-acting insulin analog formulation.¹ It has been recognized for decades that traditional “short” acting insulin products (ie, regular insulin formulations) act both too slowly and for too long a duration to enable a satisfactory matchup between insulin needs and nutrient flux following meals. A major step forward was the introduction of insulin lispro some 20 years ago. More recently, a number of approaches to further accelerate the onset and shorten the duration of mealtime insulin have been undertaken.² The first such subcutaneously delivered product to gain regulatory approval (in Europe, Canada, and more recently, in the United States) is a novel formulation of insulin aspart, containing both niacinamide and l-arginine to enhance local subcutaneous blood flow with resultant acceleration of insulin absorption. So-called FIASP® (Novo Nordisk, Inc, USA, Plainsboro, NJ)³ has demonstrated efficacy and safety in type 1 diabetes subjects studied during the 6-month onset-1 trial using a basal-bolus treatment regimen administered by conventional subcutaneous means.⁴ FIASP was administered either immediately before meal time (N = 381) or 20 minutes after start of meal (N = 382); comparator treatment was insulin aspart administered immediately before meals (N = 380).

The statistical analysis plan for the primary and confirmatory endpoints of the onset-1 trial involved a cascade of hierarchical testing,⁵ beginning with a test for A1C noninferiority comparing mealtime FIASP to insulin aspart. This endpoint was comfortably met, with an upper 95% confidence interval (CI) for the difference between treatments in change from baseline A1C required to be less than 0.4% for statistical success. In point of fact, the measured difference between the means was -0.15% (favoring mealtime FIASP over mealtime insulin aspart), and the 95% CI range (−0.23 to −0.07) demonstrated statistical superiority of FIASP over insulin aspart (although testing for A1C superiority of FIASP over insulin aspart was not part of the protocol-specified statistical hierarchy of endpoints to be tested). Postmeal FIASP also met the noninferiority test when compared to mealtime insulin aspart.

As with any diabetes treatment, A1C results cannot be evaluated in a vacuum, and a careful look at hypoglycemia outcomes is necessary as well. In the onset-1 trial, rates of severe hypoglycemia and a composite of “severe + confirmed hypoglycemia” (blood glucose documented to be below 56 mg/dl) were comparable for all three treatments. Notably, however, there was a clear difference in hypoglycemia rates in the immediate postmeal time frame. Comparing mealtime FIASP to insulin aspart, there was an increase in “severe + confirmed hypoglycemia” in the 0-1 hour and 0-2 hour post mealtime windows.

One might have predicted that the earlier onset of action of FIASP as compared to insulin aspart could result in increased early postmeal hypoglycemia risk with an attendant reduction in late postmeal hypoglycemia events. Overall, it is likely that this is a favorable tradeoff since late postmeal hypoglycemia is often overtreated, resulting in the individual glucose elevations that may trigger compensatory insulin bolus dosing that “stacks” with insulin that is already on board from the meal-related dosing bolus. Chasing this “peaks and valleys” pattern is both frustrating and potentially dangerous. Nonetheless, it would be prudent to counsel persons with diabetes who are starting FIASP that its accelerated action may cause early postmeal hypoglycemia.

Although insulin aspart is specifically approved for use in insulin pumps,⁶ there is no mention whatsoever of pumps in the product labeling for FIASP. Published information supporting the use of FIASP in pumps is very sparse, and includes a euglycemic glucose clamp study⁷ comparing pharmacokinetic and glucodynamic properties of FIASP and insulin aspart. The results demonstrated clear although relatively modest accelerations of both insulin exposure and insulin action along with reduced duration of exposure and action. These characteristics of FIASP might be expected to result in reduced postmeal glucose excursion and reduced A1C along with a shift in the timing of hypoglycemic events as observed in the onset-1 trial discussed above. Confirmation of such results in the insulin pump population must await the results of larger, longer FIASP treatment trials.

Prior to embarking on such treatment trials it was incumbent on the manufacturer to determine the basic compatibility of FIASP with pump hardware (reservoirs, tubing, infusion sets) in the clinical setting. In this issue of JDST the results of such a clinical compatibility comparison of FIASP and insulin aspart are presented.¹ The study was a small (37 subjects, randomized 2:1 FIASP to insulin aspart) and short (6 weeks) trial with a primary endpoint of the number of microscopically confirmed infusion set occlusions during the 6 weeks of treatment. Although this may seem to be a relatively trivial and arcane endpoint, it represents an important step in the effort to validate the efficacy and safety of FIASP in the pump setting.

Whenever insulin formulations and/or means of insulin delivery are changed there is the potential for unintended adverse consequences. Insulin protein misfolding is exacerbated by heat, agitation and other physico-chemical environmental influences, and this may result in development of micro- or macroscopic aggregates (“insulin fibrillation”) which have altered biological activity and as well as undesirable rheologic properties. Efforts to combat this have been undertaken by altering the primary structure of insulin to minimize fibrillation risk.⁸ In any event, the final clinical utility of an insulin product for pump use will depend in part on its propensity to cause infusion set occlusion. A number of studies comparing occlusion risk of the three original rapid acting analog insulins (insulin lispro, insulin aspart, and insulin glulisine) have been undertaken, as reviewed by Klonoff and colleagues,⁹ one can conclude that cannula occlusion events (generally identified clinically by unexplained hyperglycemia events and microscopically by finding of cannula plugging) are relatively common and that mechanisms for pump users to identify impending or established occlusions need to be improved.

The Zijlstra study¹ showed that in 219 infusion sets analyzed in the laboratory there was complete absence of microscopically confirmed infusion set occlusion in either treatment group. Most of these infusion sets were obtained during routine infusion set change during clinic visits; only three sets were from cases of clinically suspected cannula occlusion (two cases of unexplained hyperglycemia and one case of site leakage). All three of the suspect cases were in the FIASP group. A total of seven clinically suspected occlusion events occurred (the aforementioned case of leakage along with six cases of unexplained hyperglycemia); all such cases occurred during FIASP treatment.

In the absence of a readily available and definitive method for detecting cannula blockage the appearance of unexplained hyperglycemia per se has been used as a clinical surrogate for possible infusion set occlusion.¹⁰ In the Zijlstra study¹ the authors report that that of the 25 FIASP-treated subjects, 10 (40%) reported episodes of unexplained hyperglycemia, although only four of these subjects experienced a total of six episodes that were potentially attributed to infusion set occlusion, resulting in premature infusion set change. Among the 12 insulin aspart-treated subjects, 3 (25%) reported unexplained hyperglycemia, but none of these reported this as being related to a possible infusions set occlusion.

Another endpoint in the study was frequency of premature (ie, <72 hours) infusion set changes. There were 21 premature infusion set changes occurring in 11 (44%) of FIASP-treated subjects and four premature changes in two (17%) of insulin aspart-treated subjects. One-third (seven) of the premature FIASP infusion set changes and three-quarters (three) of the premature insulin aspart infusion set changes were attributed to technical issues such as empty reservoir, kinked tubing or dislodged infusion set. In the laboratory analyses of the infusion sets, two cases of FIASP were judged to have minimal particle or crystal formation although in neither case was insulin fibrillation observed.

Efficacy results from this small study were inconclusive but generally supportive of the types of findings seen in the larger onset-1 trial with non–statistically significant improvements in postprandial glucose increment, A1C, fructosamine, and 1,5-anhydrogoglucitol with FIASP treatment as compared to insulin aspart. Hypoglycemia events were more common during the two week run-in period with FIASP but there was no treatment group difference in hypoglycemia during the six week treatment period. This latter finding reinforces the need for appropriate patient education whenever changing from one insulin type to another.

What can be concluded from all of this? As discussed above, cannula occlusions are troublesome and not uncommon. Because of the physico-chemical foibles of insulin in solution, it is incumbent on manufacturers to provide both storage stability data and in-use data such as that reported from the Zijlstra study. From an efficacy and safety perspective, this pump study recapitulates the types of findings seen in the earlier, larger onset-1 trial using standard subcutaneous basal bolus treatment. The pump treatment compatibility results from this study are less convincing. The study clearly met its primary endpoint comparing the rates (zero for both treatments) of documented microscopic cannula occlusion. The problems come in interpreting the softer clinical endpoints such as early infusion set change rates (clearly higher for FIASP as compared to insulin aspart, especially if one discards the events that were attributed to technical issues), and incidence of unexplained hyperglycemia (regardless of whether or not the subject reported a suspected cannula occlusion). On the plus side, the results of the study do not suggest any fundamental safety issues that might reduce enthusiasm for further study of FIASP in pumps. Absent any guidance from the US Package Insert for FIASP for pump users, it would be prudent to advise pump users that FIASP has not yet been adequately studied in large-scale pump trials to establish a full efficacy and safety profile using this delivery modality. Pump users should be advised that FIASP may result in the need for additional premature infusion set changes and that more episodes of unexplained hyperglycemia may be expected and should be scrupulously evaluated. As in everything else in life, risks and benefits need to be balanced, and the likely treatment benefits of FIASP should be weighed against these potential downsides for pump users. When definitive large-scale pump study results are available then these recommendations can be revisited.