Influence of Local Skin Temperature and Choice of Insulin Analog on Catheter Occlusion Rates During Continuous Insulin Infusion: An Exploratory Study

Introduction

T he use of continuous subcutaneous insulin infusion (CSII) therapy in type 1 diabetes has increased significantly over the last decade, partly as a consequence of the results of the Diabetes Control and Complications Trial. ^1,2 However, there remain technical challenges with this mode of insulin delivery such as the potential for occlusions within the infusion system disrupting insulin flow. ³ Interruption of flow can occur anywhere from the insulin reservoir to the infusion site, although the latter appears to be the most vulnerable point resulting in variation in insulin absorption. ⁴

There are several factors that may contribute to occlusions in either the tubing set or indwelling catheter. Air bubbles, which may occur with inadequate priming of the catheter tubing, can occlude the system. Precipitation of insulin may also occur and contribute to occlusions. Different flow rates and different types of insulin are thought to play a role in insulin precipitation. Hirsch et al. ⁵ examined the effect of using two different concentrations of insulin in subjects with type 1 diabetes using CSII. Subjects using U-40 insulin had a significant reduction in occlusion rates. However, it was not clear whether this was related to increased flow rates or decreased time within the catheter.

Precipitation may also be influenced by the type of insulin used. Previous case reports suggested that more occlusions secondary to precipitation were seen in insulin lispro compared with insulin aspart or Velosulin^® BR (human insulin) (Novo Nordisk, Bagsvaerd, Denmark). ⁶ The differences in each of the rapid-acting insulin analogs are created by amino acid substitutions. ⁷ These substitutions confer distinctive properties and may affect decomposition of the analog insulin within the pump. Varying rates of decomposition may also affect the rates of precipitation and occlusion. In some studies insulin aspart has been shown to have the greatest chemical stability and lowest rates of occlusion compared with insulin lispro and insulin glulisine. However, it is unclear which analog insulin is more suited to use in CSII. In a previous laboratory-based study, we examined rates of catheter occlusion study comparing the three commonly used rapid-acting insulins, insulin aspart, insulin lispro, and insulin glulisine. ⁸ Insulin was infused through catheter sets and pumps and kept in an incubator maintained at 32–36°C. In that study early (i.e., within 72 h) catheter occlusions were rare, but late occlusions appeared to be more common with insulin glulisine. Although higher skin temperatures have been shown to enhance subcutaneous insulin absorption, ⁹ there is little evidence on the effect of local skin temperature on insulin precipitation and catheter occlusion rates. This may be relevant as under normal clinical circumstances, there is the potential for variations in skin temperature, in patients using CSII, to influence local insulin flow rates due to the proximity of the infusion catheter to the skin surface.

The aim of this study was to determine whether changes in local skin temperature over which the infusion catheters pass influence the risk of occlusion. Given our previous study showing differences between rapid-acting insulin analogs on the risk of late catheter occlusions at a higher temperature, we also compared the two insulin preparations with the lowest and highest rate of late catheter occlusions from that study.

Subjects and Methods

Healthy volunteers were invited to participate in the trial, which was designed as a randomized crossover study. Approval was obtained from the local ethics committee. Each participant wore an Accu-Chek^® Spirit CSII device with Accu-Chek Rapid-D Link 100-cm infusion sets (Roche, Burgdorf, Switzerland) for 10 days in total. Subjects were instructed to wear the pumps while continuing with all normal daily activities, except for bathing. The insulin was discharged into a small plastic bag containing sponges to absorb the insulin. The infusion set and plastic bag was taped to the skin with micropore in order to reproduce typical pump conditions. Skin temperature was measured using LogTag TRIX-8 temperature data loggers with external bare probes of 1.5-m cable length (LS Technology, Bournemouth, UK). The external probe was held against the skin with a bio-occlusive dressing close to the site of the needle set and infusion set. Each temperature logger (measuring device) was preset to record the skin temperature every 20 min. The insulin pump and temperature logger were kept on for all normal activities and overnight. Each subject was randomly assigned to begin with either insulin aspart (Novorapid^®; NovoNordisk, Crawley, UK) or insulin glulisine (Apidra^®; sanofi-aventis, Guildford, UK). The starting insulin analog was continued for 5 days, and subjects were advised not to change their infusion sets until the end of the first 5-day period unless a catheter occlusion had occurred. After 5 days the insulins were swapped over for a subsequent 5-day period. A washout period was not considered necessary as a new infusion set was used for each insulin. The basal rate for each pump was preset at 0.2 U/h, and participants were asked to give 2-U boluses with meals three times a day. Information on the frequency of boluses was downloaded from the insulin pumps at the end of the study in order to determine the number of boluses given.

Results

After signed consent was obtained, 20 healthy subjects without diabetes were included in the study. There were two men and 18 women, with ages ranging from 21 to 56 years with a mean of 43.8 years. Eight subjects received insulin aspart followed by insulin glulisine, and 12 subjects received insulin glulisine followed by insulin aspart. Seventeen subjects wore the insulin pumps for the full 10-day period. Three participants stopped wearing the pumps after 9 days for different personal reasons. All subjects were asked to give 2-U boluses with meals. In the group starting with insulin glulisine, nine subjects gave the complete 15 boluses, eight subjects omitted one to three boluses, and three subjects omitted four or five boluses. In the insulin aspart group, four subjects gave 15 boluses, 11 subjects omitted one to three bolus doses, and three subjects omitted four to six bolus doses. The median total doses of insulin infused were very similar: 52 units (range, 24–54 units) for insulin glulisine and 50 units (range, 37.2–54 units) for insulin aspart (P=not significant).

Skin temperature

The temperature loggers were programmed to measure the temperature every 20 min. Data were considered to be incomplete if no temperature measurements were recorded in an hour on three different occasions. Overall data monitoring was incomplete for three subjects while using insulin glulisine and five subjects while using insulin aspart. Incomplete data were due to loose wiring in four subjects and for personal reasons in one subject. However, complete temperature data were available for four of five subjects with occlusions for insulin glulisine and three of four subjects with occlusions with insulin aspart. The skin temperature ranged between 33.5°C to 36.68°C for insulin glulisine and 32.35°C to 35.28°C for insulin aspart, although lower values were recorded when the data logger was removed. An example of the variation in skin temperature can be found in Figure 1. Values below 30°C were associated with removal of the temperature logger in keeping with the ambient temperatures at the time of the study. There was no evidence of any difference in mean skin temperature either between insulin aspart and insulin glulisine (estimated difference, 0.53 [95% confidence interval, −0.31 to 1.37]) or between the first 5 days or second 5 days of the overall study period.

OPEN IN VIEWER

FIG. 1.

An example of temperature fluctuations in one subject.

Fluctuations in temperature were generally very small except when the catheter and temperature probe were removed from the skin. There was no difference in the mean fluctuation in temperature between subjects with occlusions and subjects without occlusions for both insulin aspart and insulin glulisine (Fig. 2). Average temperature fluctuations for each of the three preceding hours prior to occlusion were reviewed in three of four occlusions in the insulin glulisine group and four of five occlusions in the insulin aspart group. There was a lower temperature 2 h prior to occlusion in one of three subjects in the insulin glulisine group. However, two subjects in the insulin aspart group had a significantly lower temperature prior to occlusion, and one subject had a significantly higher temperature 2 h before occlusion (Figs. 3 and 4).

OPEN IN VIEWER

FIG. 2.

Mean temperature fluctuations for occlusions (solid columns) and nonocclusions (open columns) for insulin glulisine and insulin aspart.

OPEN IN VIEWER

FIG. 3.

Mean temperature fluctuations in the preceding 3 h prior to occlusion in three subjects (n=3 of 4) on insulin glulisine. *P<0.01.

OPEN IN VIEWER

FIG. 4.

Mean temperature fluctuations in the preceding 3 h prior to occlusion in four subjects (n=4 of 5) on insulin aspart. *P<0.05, **P<0.001.

Discussion

The results of this study suggest that the overall risk of occlusion was similar for the two rapid-acting insulin analogs studied and unrelated to fluctuations in ambient skin temperature over which the infusion catheters passed. Previous in vitro studies have also examined the stability of insulin aspart and insulin lispro in infusion pumps exposed to a higher temperature (37°C) and reported that the potency and stability on the insulins did not change over a 7-day period. In those studies mechanical agitation of the infusion systems was continued throughout. ^10,11 In contrast, in our previous laboratory-based study we found an increased risk of late (>72 h) occlusions with insulin glulisine, but in that study the infusion systems were stationary for the duration of the study, which may be relevant. ⁸ The findings from this current study concur with a recent clinical study comparing insulins glulisine, aspart, and lispro in subjects with type 1 diabetes using CSII and found no overall difference in occlusion rates or hyperglycemia among the three insulin analogs. ¹²

Here, more occlusions occurred during the second period of the study. It is possible that the subjects had adjusted to pump use after the first period and may have been less meticulous with care of the pump tubing. There were only two early occlusions, and both occurred with insulin aspart. One of the early occlusions occurred 6 h after pump setup. This occlusion may have been related to pump setup and air in the system rather than the type of insulin used. Average temperature fluctuations between subjects with occlusions and subjects without were also not different for either insulin. On an individual basis, some fluctuations in temperature were seen in the hours prior to occlusion, although this was not seen in all individuals before occlusion occurred. Because of the small numbers it is not possible to draw any firm conclusions about the influence of fluctuations in skin temperature changes and the risk of subsequent catheter occlusions from these data. The doses of insulin used in this study may be considered small; however, the basal infusion rates of 0.025 U/h are within the pump manufacturer's working range. The levels of insulin used here are also more likely to be found in the pediatric population. ¹³ The use of the skin temperature probe also limited the results of the study as the temperature loggers are designed to be used in static circumstances, and subject movement resulted in loose wires on several occasions. The loggers were also not waterproof, and it was necessary to remove them prior to bathing.

In this study, the cause of occlusion was not evaluated. The early occlusion was more likely to be related to air bubbles following a change in tubing, and later occlusions were thought to be associated with insulin precipitation, although this was unproven. Radioimmunoassay techniques have previously been used to show insulin precipitation ⁶ and should be considered for future studies to aid evaluation of occlusion causes. Also, occlusions in this study would not fully represent occlusions in a clinical setting as there was no insertion into subcutaneous tissue.

In clinical practice our current recommendations are to change insulin pump infusion sets regularly within 72 h, although the changing interval between infusion sets can vary considerably between 2 and 10 days for some patients. ¹⁴ This exploratory study suggests that early occlusions with either insulin analog used here are uncommon and not influenced by fluctuations in ambient skin temperature over which the catheters pass. More extreme temperatures, however, remain an important issue and focus for patient education. ¹⁵ The results of this study do, however, reinforce the need for advice to patients about regular catheter changes, particularly at 72 h.