Macrophage Activation in Type 1 Diabetic Patients With Catheter Obstruction During Peritoneal Insulin Delivery With an Implantable Pump

We included 10 type 1 diabetic patients (C-peptide-negative), after informed consent, and treated them for at least 6 months through IPII with an implantable pump (Minimed Implantable Pump 2001, Minimed Technologies, Sylmar, CA). Pump catheters were made out of silicone-coated polysulfone. The infused insulin was 21 PH neutral semisynthetic human insulin (Hoechst, Frankfurt, Germany) at a concentration of 400 U/ml, stabilized by a glycol-polyethylene-polypropylene surface-active agent (Genapol). Refilling was performed under aseptic conditions every 6 weeks. The pumps were emptied, and the residual insulin was collected for chemotaxis and cytotoxicity analysis.

Group 1 was composed of three diabetic patients (two men and one woman)aged 37-53 years (mean 45.6) who had been suffering from diabetes for 20.6 years on average (range 14-28). During the year after pump implantation, they presented with at least one episode of irreversible catheter obstruction,which led to the removal of the catheter. Catheter obstruction was clinically suspected when the patients presented with hyperglycemia and/or required increasing doses of insulin for several days. This was confirmed by celioscopy, which was used to identify material surrounding the catheter tip. Histology and immunocytochemistry revealed the presence of a fibrous capsule made of several collagen layers with fibroblasts, macrophages, and a few lymphocytes. Amorphous deposits reactive to anti-insulin antibodies were surrounded with giant epithelioid cells and macrophages(4).

Group 2 included seven diabetic patients who did not have catheter obstruction, but who were comparable with group 1 patients in terms of age,diabetic history, body weight, and insulin requirement(Table 1).

Blood samples (150 ml) from diabetic patients were obtained from venous puncture during routine clinical examination. Peripheral mononuclear cells were separated from heparinized blood using Ficoll gradient centrifugation (d= 1.077) and purified with immunomagnetic beads coated with anti-CD14 monocyte antibody.

Mononuclear cells were collected by leukopheresis from five anonymous healthy donors (group 3) and purified through countercurrent centrifugal elutriation (J6M-E; Beckman Coulter, Fullerton, CA). Monocyte purity was checked by morphological examination after Wright-Giemsa staining and by flow cytometry analysis with fluorescein isothiocyanate (FITC)-labeled antibodies directed against CD14. We found 85-90% of the cells were CD14⁺ by using the forward versus sideward light scattering pattern in combination with anti-CD14-FITC antibody. Monocyte viability was checked using the trypan exclusion method.

Purified monocytes were suspended in an RPMI-1640 medium containing glucose(5.5 mmol/l), L-glutamine (2 mmol/l), and 25 mmol/l HEPES. It was supplemented with 100 U/ml penicillin, 100 μg/ml streptomycin (Life Technologies,Paisley, U.K.), and 10% sterile-filtered heat-inactivated (30 min at 56°C)pooled human AB serum (Etablissement de Transfusion Sanguine, Strasbourg,France). To induce macrophage differentiation, 50 U/ml recombinant human granulocyte-macrophage colony-stimulating factor (Life Technologies) was added to the culture medium. The cells were cultured for 7 days at a concentration of 2 × 10⁶ cells/ml in gas-permeable hydrophobic Teflon bags(Polylabo, Strasbourg, France) at 37°C under a 5% CO₂humidified atmosphere. After a 7-day culture, monocytes were differentiated into monocyte-derived macrophages (MDMs) and counted after trypan blue dye staining. MDMs were plated at a concentration of 10⁶ MDMs per well in a 12-well cell culture cluster for 90 min at 37°C (Costar, Cambridge,MA) before chemotaxis and cytotoxicity analyses.

All cell preparations and culture media were endotoxin-free. Endotoxin in the media, additives, and all disposable materials were quantitated using a limulus amebocyte lysate assay (LAL) from Sigma (St Louis, MO). The sensitivity of LAL was 0.015 endotoxin unit/ml using Escherichia colistandard endotoxin. Disposable materials were endotoxin-free, and all glassware was dry-heated at 220°C for 6 h.

We evaluated macrophage chemotaxis toward human insulin using modified Boyden chambers composed of two compartments, one placed on top of the other(17). The upper compartment consisted of a culture insert (Transwell Insert; Costar, Cambridge, MA)equipped with an 8-μm polycarbonate filter. The lower one was filled with 900 μl insulin solution (0.5 U/ml) containing either insulin that was collected from the pump reservoir of the diabetic patients in both groups or insulin that came directly from the bottle (vial insulin). The upper chamber was filled with a cell suspension containing 3 × 10⁵ MDMs from diabetic or healthy subjects. After incubation for 90 min under a humidified atmosphere at 37°C, the upper compartment was removed, and the MDMs that did not cross the membrane were scraped off the upper surface of the membrane. The other side of the membrane, covered with adhering cells, was stained with eosin-hematoxylin (Diff-Quick; Dade, Maurepas, France) for counting the cells that had migrated using direct microscopy (×100) in 10 random fields. The migration response was defined as the mean number of MDMs per field (n = 3). The chemotactic effect of both kinds of insulin was determined by the migration index expressed as the following ratio: the number of MDMs attracted to either native insulin or insulin collected from the pump reservoir divided by the number of MDMs attracted to the control medium.

In each experiment, the chemotactic peptide formyl-methionyl-leucyl-phenylalanine (fMLP) from Sigma was used as a reference. It was shown that 10^-8 mol/l was the concentration at which the chemotactic response of MDM was optimal. The migration of MDMs toward the culture medium was considered as a random migration and was used as negative control.

The MDMs of diabetic and healthy subjects were cultured for 16 h in the culture medium described above containing vial insulin or insulin collected from the reservoir at a concentration of 0.5 U/ml. Under the same conditions,the stimulation of MDMs with 50 ng/ml E. coli lipopolysaccharide(LPS) from Sigma served as the control. After centrifugation at 10,000g for 5 min at 4°C, the supernatants were harvested for tumor necrosis factor-α (TNF-α) determination using the enzyme-linked immunosorbent assay method (Immunotech kit, Beckman Coulter)(n = 3) with a detection limit of 5-10 pg/ml. The working range was 10-1,000 pg/ml with inter- and intra-assay variations of 5.4 and 1.6%,respectively.

All data are expressed as means ± SD. A statistical analysis of macrophage migration indexes and cytokine release in diabetic patients of both groups and healthy subjects was performed using one-way analysis of variance. Significance is shown in Figs. 1 and 2.

Neither insulin recovered from the pump reservoir nor vial insulin stimulated the migration of macrophages from diabetic patients without catheter obstruction. In response to both kinds of insulin, the chemotactic indexes for these macrophages were 0.86 ± 0.15 and 0.81 ± 0.07,respectively. These values were similar to those obtained for macrophages from diabetic patients with catheter obstruction and healthy subjects(Fig. 1).

The migration toward fMLP of macrophages from diabetic patients exhibiting catheter obstruction during IPII was significantly higher compared with that obtained in the presence of either native or reservoir-recovered insulin (3.81± 0.36 vs. 1.07 ± 0.03 and 1.09 ± 0.05, P <0.001). Similar data were found with macrophages from diabetic patients having no catheter obstruction and from healthy subjects(Fig. 1).

LPS significantly stimulated the secretion of TNF-α by macrophages from diabetic patients with catheter obstruction, whereas neither vial nor reservoir-recovered insulin had any effect on the release of this substance(144.83 ± 67.25 vs. 5.15 ± 2.93 and 5.27 ± 2.43 pg/ml[P < 0.001]). Similar results were found with macrophages from diabetic patients without catheter obstruction and healthy subjects(Table 2). When macrophages from diabetic patients or healthy subjects were incubated in the presence of either kind of insulin, the release of TNF-α was similar to that obtained with culture medium alone.

Only the release of TNF-α by macrophages from healthy subjects and the chemotaxis of macrophages toward insulin collected from the pump reservoir of diabetic patients with catheter obstruction have been represented(Fig. 1C and Table 2). Similar data have been obtained with insulin collected from the pump reservoir of diabetic patients without obstruction.

When monocytes from diabetic patients exhibiting catheter obstruction were differentiated into macrophages, their chemotactic index toward both kinds of insulin was significantly higher than that of macrophages from diabetic patients without catheter obstruction (1.09 ± 0.03 vs. 0.86 ±0.15 [P < 0.05] for reservoir-recovered insulin and 1.07 ±0.03 vs. 0.081 ± 0.07 for vial insulin). In response to native insulin,the chemotactic indexes of macrophages from diabetic patients without obstruction and healthy subjects were similar(Fig. 2A and B).

In the same way, the migration toward fMLP of macrophages from patients with catheter obstruction was significantly higher than that observed in macrophages from diabetic patients without obstruction and healthy subjects(3.81 ± 0.36 vs. 2.30 ± 0.89 and 2.60 ± 0.80, P< 0.05) (Fig. 2C).

In response to LPS, the secretion of TNF-α by macrophages from diabetic subjects with or without catheter obstruction reached 144.83 ±67.25 and 130.77 ± 54.32 pg/ml, respectively. These values were not significantly different from those observed in macrophages from healthy subjects (114.81 ± 32.60 pg/ml). However, neither insulin collected from the pump reservoir nor vial insulin stimulated the release of TNF-αby macrophages from the diabetic patients of both groups and the healthy subjects.

Only the migration of macrophages from healthy subjects toward insulin collected from the pump reservoir of diabetic patients with catheter obstruction has been represented (Fig. 2A). Similar data were observed with insulin collected from the pump reservoir of diabetic subjects without catheter obstruction.

This study reports that neither human insulin used in implantable pumps nor its exposure to the pump reservoir surfaces induces macrophage activation in diabetic patients with or without catheter obstruction during IPII. When fMLP, a strong chemotactic factor, was used for a migration study, it was observed that macrophages from diabetic patients with catheter obstruction exhibited a higher level of chemotaxis compared with that of macrophages from either patients without obstruction or healthy subjects. A similar response profile was observed toward both vial and reservoir-recovered insulin.

Highly purified semisynthetic Genapol-stabilized insulin proved to be nonchemotactic toward the macrophages of both diabetic and healthy subjects. Using a modified chemotactic chamber, Josefsen et al.(18) also demonstrated that human insulin exhibited no chemotactic effect on human monocytes. Our results also show that Genapol, which is used as a surfactant to avoid insulin denaturation and aggregation, was devoid of any effect on macrophage migration.

A previous in vitro study demonstrated that a temperature of 37°C and constant shaking led to the formation of a few insulin secondary products(e.g., desamido insulin and derived high-molecular weight molecules)(19). After 4 weeks, 95% of the biological potency remained. However, the antigenicity of these products was not documented. Several studies have reported on the formation of amyloid deposits reacting with anti-insulin antibodies at the catheter tip(4,9,10). Moreover, these aggregates were colonized by fibroblasts and macrophages. The authors speculated that the amyloid deposits could be composed of insulin products that had undergone major modifications Because macrophage chemotaxis is a phenomenon observed at a low concentration of chemotactic factor, it seemed important to evaluate in vivo the potential role of insulin on macrophage migration. Human insulin that was collected from the reservoir of an insulin pump 6 weeks after it was refilled failed to activate macrophages from diabetic patients. Even when insulin was exposed for long periods of time to the environment of the in vivo pump, it did not appear to trigger macrophage activation.

The stimulation of macrophage migration in patients having developed catheter obstruction indicated the high level of macrophage activation during IPII. The causal relationship between increased chemotaxis in vitro and catheter obstruction has not been established. However, this observation constitutes strong evidence, possibly of clinical relevance, of a major complication occurring during IPII. Nevertheless, these results can explain the reduced operating life of peritoneal catheters in patients with previous incidents of catheter obstruction, which is observed in clinical series by the stimulation of macrophage chemotaxis probably induced by IPII(9,10). According to Renard et al. (9),both a longer history of diabetes and peritoneal insulin infusion, either by external or implantable pump, represented the only factors correlating with catheter obstruction. In our study, the patients of both groups (with and without catheter obstruction) had been suffering from diabetes for ∼20 years. No patient had previous peritoneal insulin infusion with an external pump. During the study, the duration of the treatment through an implantable pump was ∼4 years for patients of both groups. Irreversible catheter obstruction occurred in the year after the beginning of IPII and led in all cases to catheter replacement. A previous study reported on an alteration in macrophage functions in a hyperglycemic environment, such as decreased phagocytic ability and stimulated proliferation(20). As shown by the HbA_lc levels, there was no difference in the glycemic control in both groups, thus ruling out this factor as a cause for the activation of macrophages from patients with catheter obstruction.

The high immunogenic response toward insulin reported by several authors in about half of the patients during IPII, together with the presence of amyloid deposits reacting with anti-insulin antibodies observed at the catheter tip,suggested that an immunological reactivity toward insulin could facilitate catheter obstruction(21,22,23). In our study, only one patient who had developed catheter obstruction exhibited a high plasmatic level of anti-insulin antibody. Unexpectedly, a similar concentration was found in two patients without the complication. These data seem to indicate that a high immunogenic response toward insulin was not involved in the stimulation of macrophage migration observed in the patients with catheter obstruction. However, with these results, one must take into account the low number of patients in each group. In our study, the stimulation of macrophage activation concerned only macrophage chemotaxis and not cytokine release. Indeed, regardless of the stimulation (i.e., LPS or reservoir-collected or vial insulin), the level of TNF-α released by macrophages was comparable in diabetic patients (with and without catheter obstruction) and healthy subjects.

In conclusion, human insulin used for intraperitoneal insulin delivery with implantable pumps is not involved in catheter obstruction through the mechanism of macrophage activation. However, the high level of macrophage chemotaxis observed in diabetic patients having developed catheter obstruction suggests an individual sensitivity to peritoneal insulin delivery and might explain the recurrence of this complication that is observed in some of these patients.