Insulin antibody responses after long-term intraperitoneal insulin administration via implantable programmable insulin delivery systems

OBJECTIVE

To determine whether insulin antibodies are generated in diabetic patients after short- and long-term intraperitoneal insulin use and, if so, whether they are of potential clinical interest. Insulin antibodies commonly develop in diabetic patients who use subcutaneous human insulin, although their clinical significance remains controversial. Few data are available regarding insulin antibody responses to intraperitoneal insulin.

RESEARCH DESIGN AND METHODS

We studied insulin antibody levels and clinical diabetes control in 25 type 1 diabetic patients treated for 3-6 years with intraperitoneal surfactant-stabilized porcine modified human insulin delivered by implantable programmable insulin delivery systems.

RESULTS

All patients had preimplantation insulin antibody levels < 20 microU/ml, with a mean value of 2 +/- 2 microU/ml (1 SD). Mean antibody levels increased throughout the study period to a mean maximum of 197 +/- 326 microU/ml (P < 0.02) with 11 of 25 (44%) patients' levels exceeding 20 microU/ml (insulin responders). The mean time to significant antibody development was 21.8 +/- 4.4 months. Of the 11 responder patients, 4 had clinical syndromes that consisted of increasing daily insulin requirements and/or nocturnal hypoglycemia despite minimal nighttime basal insulin infusion rates associated with peak antibody levels > 200 microU/ml. None of the nonresponder patients (antibody levels < 20 microU/ml) had these clinical findings.

CONCLUSIONS

Our results indicate that insulin antibody levels observed during intraperitoneal administration of human insulin are 1) similar to those reported during subcutaneous administration; although the rise in antibody level may be delayed compared with subcutaneous human insulin, 2) associated with a patient subset who are insulin antibody responders after switching from subcutaneous to intraperitoneal human insulin, 3) associated with a decrease in levels among responder patients regardless of whether they discontinue or continue pump use, and 4) associated with increased insulin needs and/or nocturnal hypoglycemia despite minimal basal rate insulin infusion at nighttime when antibody levels exceed 200 microU/ml.

Induction of islet cell surface and insulin antibodies in Balb/c mice by application of porcine insulin and Freund's adjuvant is not associated with insulitis

To investigate whether insulin antibody (IAB) formation is associated with the appearance of islet cell cytoplasmic antibodies (ICA), islet cell surface antibodies (ICSA) and insulitis Balb/c Bln mice were immunized with porcine insulin in combination with or without Freund's adjuvant. The animals received 8 i.p. injections and were followed up to 150 days for the development of antibodies and insulitis. Mice immunized with insulin in CFA developed IAB as well as ICSA. Mice only receiving Freund's adjuvant emulsified in saline also developed ICSA. ICA were not detectable. Inflammatory infiltrates were found in the exocrine pancreatic parenchyma but not in islets. The results show that nonspecific stimulation of the immune system and the application of insulin as antigen leads to both the formation of ICSA and IAB, while insulitis was not detectable.

Immunological resistance to human biosynthetic insulin--effects of immunosuppression and plasmapheresis

A 55-year-old gentleman, after being treated for a short time with a diet and with Chlorpropamide, was switched to purified porcine insulin due to ketonuria and ketoacidosis. After a year the patient developed immunological insulin resistance (mean daily insulin dose: 3.72 U/kg body weight; anti-insulin antibodies 78%). In order to lower anti-insulin antibodies human recombinant DNA insulin was introduced into further therapy. Contrary to expectations, the patient did not reduce whatsoever his anti-insulin antibodies and his daily insulin dose increased up to 5.63 U/kg body weight. Introduction of combined immunosuppressive therapy (prednisone plus azathioprine) together with plasmapheresis resulted in rapid lowering of daily insulin requirement and reduction in anti-insulin antibodies. Immunosuppressive therapy was continued with 10 mg of prednisone and a year later the patients insulin daily requirement was 0.66 U/kg BW while his antibodies were 18%. The possible causes of insulin resistance to human recombinant DNA insulin are discussed as well as the advantage of combined immunosuppressive therapy together with plasmapheresis that was used for rapid lowering of insulin daily requirement and anti-insulin antibodies titer.

[Hypoglycemia and multiple myeloma]

We report the case of a 78-year-old patient with recurrent attacks of severe fasting and late postprandial hypoglycemia, whose plasma showed highly elevated concentrations of immunoreactive insulin evidenced by high titers of spontaneous insulin and proinsulin-binding antibodies. Insulin autoimmune syndrome was diagnosed. Further investigations revealed a multiple myeloma of the kappa-light chain type. The monoclonal insulin-binding antibodies were characterized as IgG2-subclass and were identical with the paraprotein, thereby confirming that the insulin-binding antibodies were in fact produced by the myeloma. Together with initial symptomatic treatment, plasmapheresis was performed repeatedly to reduce the antibody pool. Subsequently octreotide therapy proved successful. The underlying myeloma was treated by chemotherapy.

Effects of human insulin on insulin binding antibody production in nondiabetic subjects

OBJECTIVE

To test the hypothesis that human insulin may have a low immunogenicity and that short-term exposure may not cause endogenous insulin antibody production.

RESEARCH DESIGN AND METHODS

Randomized double-blind prospective study. Serum samples collected for insulin binding antibodies and measured by a sensitive immunochemical assay. Subjects were seven healthy nondiabetic patients who had never received exogenous insulin. Each subject received 6 separate monthly injections of human insulin. On four occasions, both regular and NPH insulin were administered. On the other two occasions, either NPH or regular insulin was administered alone.

RESULTS

Mean +/- SE basal insulin antibody levels (1.2 +/- 0.2 micrograms/L) increased to a maximal level of 4.5 +/- 0.8 micrograms/L after four injections. Thereafter, antibody levels declined to an end-of-study value of 2.5 +/- 0.3 micrograms/L. This represented a highly significant overall increase (P less than 0.001). A control group of six insulin-dependent diabetic subjects treated with human insulin over the same period as the test subjects demonstrated no change in insulin antibody concentrations (2.8 +/- 0.7-2.7 +/- 0.6 micrograms/L).

CONCLUSIONS

These results suggest that human insulin preparations, when administered subcutaneously, may be more immunogenic than previously considered. The antigenic response was rapid, because only four subcutaneous injections were sufficient to produce insulin antibody levels in nondiabetic patients similar to those observed in insulin-dependent diabetic patients receiving chronic insulin replacement therapy.

[Advantages and drawbacks of human insulin]

Insulin (I) preparations used formerly contained a large number of protein contaminants which are thought to be immunogenic and, hence, caused lipodystrophy, I-allergy and sometimes antibody-mediated I-resistance in many patients. Monocomponent (MC)-I and human I (HI) are virtually free of these peptides and are, therefore, very rarely accompanied by the above mentioned immunologic side effects. In this respect, however, HI offers only little advantage over MC-I although HI is the least immunogenic I. On the other hand, the formation of antibody to I in the diabetic mother is an important determinant of fetal outcome. And since children from diabetic mothers treated with HI are less frequently macrosomic, the use of HI is strongly recommended in women with diabetes before and during their childbearing years. Neutral HI action is somewhat shorter, although clinically not to a relevant extent and, furthermore, metabolic control is not improved by using HI compared wtih MC-I. These findings have been regarded as disadvantages of HI, together with the fact that about 20% of HI-treated patients experience a change of hypoglycemia symptoms during the course of their illness. While autonomic symptoms become weaker or disappear, patients have to react to neuroglycogenic symptoms which normally remain constant. However, the incidence of hypoglycemic events does not change during treatment with HI. Several reasons for this change of symptoms are discussed, such as long duration of diabetes, intensified therapy with near-normoglycemia, development of autonomic neuropathy, alcoholic beverages, and often insufficient instruction of patients.(ABSTRACT TRUNCATED AT 250 WORDS)

Continuous subcutaneous insulin infusion (CSII) and insulin antibodies in rabbits

Using prepubertal male New Zealand White rabbits, continuous subcutaneous insulin infusion (CSII), delivered by either an external or an implantable infusion device, resulted in significantly higher insulin antibody (I-Ab) production than bolus injection (BII). We tested the influence during CSII of (1) the insulin species, (2) the insulin diluent, (3) the materials of which the infusion devices were made and (4) the incubation of insulin in a syringe on the backs of rabbits ('sham-infusion'), with the following results: (1) beef and sulphated beef insulins produced high levels of I-Abs, while porcine and human insulins produced moderate levels; (2) with all insulins used, 0.9% NaCl and 0.9% NaCl with 24-26 mmol NaHCO3 added, produced high levels of I-Ab. A buffer containing 0.7% NaCl, 0.136% sodium acetate trihydrate and 0.1% methyl-p-hydroxybenzoate and a buffer containing 16 mg/ml glycerol and 2 mg/ml phenol, produced highly significantly lower I-Abs (P less than 0.001); (3) insulin glass syringes produced much lower I-Ab levels than in standard polypropylene syringes and (4) polypropylene syringes in a 'sham-infusion' technique, resulted in intermediate levels of insulin antibodies [(P less than 0.02) vs CSII; (P less than 0.005) vs BII]. Our data suggest that insulin immunogenicity is influenced by all four factors tested. We suggest that benefits of CSII therapy may be attenuated unless a best possible control of these factors is achieved.

Longitudinal study of islet cell antibodies and insulin autoantibodies and development of diabetes in non-obese diabetic (NOD) mice

We have previously shown the presence of circulating islet cell cytoplasmic antibodies (ICA) and insulin autoantibodies (IAA) in the NOD mouse before onset of insulin-dependent diabetes mellitus (IDDM). Here we have determined the levels of the two autoantibodies in 28 female NOD mice longitudinally from approximately day 40 to day 250, to examine their ontogeny, association and predictive value for diabetes. All animals (11 diabetic, 17 non-diabetic) showed varying levels of ICA at some stage, while IAA activity was found in 21 out of 28 mice. Expression of both the markers was seen in more than half of the animals by day 60, with higher levels and rates occurring subsequently in both diabetic and non-diabetic groups. The expression of ICA did not always correlate with that of IAA. There was no apparent difference in the ontogeny of ICA and IAA between the two groups. During the study period the number of animals with ICA was similar in the two groups, while the number of those with IAA was higher in the diabetic animals. In this group declining and rising levels of ICA were seen just before clinical diabetes with frequent peaks of IAA. In the same animals, eight out of 11 mice showed co-expression of high levels of both markers either intermittently or persistently prior to onset, whereas only one non-diabetic animal showed this. We conclude that the ontogeny and serum level of ICA or IAA alone could not be used to predict the clinical onset of diabetes in these animals. However, co-expression of high levels of both markers prior to onset may suggest a strong predisposition to clinical diabetes. This may have relevance to attempts to predict the onset of IDDM in humans who have one or both of these immunological markers.

Insulin autoimmune syndrome as a cause of hypoglycemia--report of four cases

Insulin autoimmune syndrome is a syndrome consisting of fasting hypoglycemia, hyperinsulinemia and detectable insulin-binding antibodies in patients who have never been exposed to exogenous insulin. Four cases who developed symptoms of hypoglycemic attack with self-limited duration and spontaneous remission were collected in our hospital from 1984 to 1988. The elevated serum total and free insulin and C-peptide levels, as well as the titer of insulin autoantibodies, decreased gradually; but insulin autoantibodies were still present in the serum for more than six months after the initial episodes of hypoglycemia. Three of four patients had Graves' disease and developed the syndrome after methimazole treatment. The fourth one had a history of hemorrhagic cystitis and denied history of specific drug exposure. The cause or stimulus for insulin autoantibody formation is still unknown, but drugs containing a sulfhydryl group like methimazole may play a role in the development of the syndrome. Extremely high insulin antibodies in patients with fasting hypoglycemia along with elevated serum levels of insulin and C-peptide suggest a diagnosis of insulin autoimmune syndrome and usually exclude the possibility of insulinoma or factitious hypoglycemia.

Intractable diabetic ketoacidosis due to insulin antibody--response to steroid therapy

To examine the effect of steroid therapy on insulin antibody titer in insulin-dependent diabetes mellitus, we studied a 58 year-old gentleman with recurrent diabetic ketoacidosis. No any overt precipitating factors could be accounted for, except limited pancreatic beta cell reserve and high titers of anti insulin antibodies. Despite the persistence of high titers of plasma antiinsulin antibodies, the clinical manifestations of diabetic ketoacidosis improved greatly by the administration of steroid. Nevertheless, the patient still showed the great excursion of plasma glucose concentration.

Genetic origin of human anti-insulin antibodies

Numerous studies have characterized the specificity of human anti-insulin antibodies; however, little is known about their genetic origin. To initiate molecular studies, B cells that bind human insulin were selected from the peripheral blood of diabetic donors and transformed with Epstein-Barr virus. The resulting anti-insulin B-cell lines were cultured at limiting dilution and examined for immunoglobulin heavy-chain gene rearrangements on Southern gels. These studies demonstrated the clonality of the B-cell lines and showed that multiple immunoglobulin heavy chain rearrangements are present. When the heavy-chain variable region (VH) gene from one of these Epstein-Barr virus cell lines was cloned, it was found to belong to the recently identified human VHV gene family that represents less than 1% of known human VH genes. Using the polymerase chain reaction, the germline VHV gene of the donor was amplified and sequenced. The sequences showed a high level of homology (98%) between the expressed and germline VHV gene of the donor. While antibodies reactive with autologous insulin (like other autoantibodies) are not extensively mutated from their germline configuration, two replacement substitutions are present in this IgM antibody.

Production of insulin antibodies by mice rejecting insulin transfected cells

Antibodies to insulin appear prior to the development of Type I diabetes and their concentration may correlate with the rate of autoimmune beta cell destruction. In order to study potential mechanisms involved in the production of antibodies to insulin, we transplanted different strains of mice with histoincompatible non-islet cells (AtT20-Ins and NIH-3T3-Ins) synthesizing homologous insulin, in contrast to immunization with non-transfected cells and insulin in Freund's adjuvant. The pituitary cell line (AtT20) and the fibroblast cell line (NIH-3T3) were transfected with the rat insulin-II gene (which encodes an insulin molecule identical to that of mouse insulin-II). No antibodies to insulin were found after subcutaneous injection of AtT20-control cells (without the integrated rat insulin gene) or after injection of rat insulin complete Freund's adjuvant. After subcutaneous injections of living AtT20-Ins or NIH-3T3-Ins cells producing insulin (40 to 60 ng insulin/10(6) cells per injection) in two strains (BALB/cJ, C3H/HeJ) but not in a third (SJL/J), antibodies to insulin rapidly appeared. In addition, when AtT20-Ins cells were transplanted into Wistar-Furth rats, insulin antibodies appeared in three out of four animals. The level of antibodies induced was similar to the concentrations of insulin antibodies of prediabetic NOD mice. This finding suggests that during the immune destruction of a cell synthesizing insulin, humoral 'tolerance' to insulin can be rapidly abrogated. Genetic control of this response is suggested by the difference between response of BALB/cJ and C3H/He vs SJL/J.

[Hypoglycemia induced by insulin antibody during postoperative management with intravenous hyperalimentation--a case report and qualitative analysis of insulin antibody]

A 42-year-old male began to show frequent hypoglycemic attacks, 25 days after total gastrectomy. By that time, he had received intravenous hyperalimentation therapy with bovine insulin. Incidence of these attacks increased despite the dose of glucose was escalated and insulin administration was interrupted. Serum C-peptide level was 11.4 ng/ml and total immunoreactive insulin (IRI) level was 1170 mu u/ml with 1120 mu u/ml (96%) of gamma-globulin-binding IRI. Since insulin antibody formation was suspected, we decreased the dose of glucose to reduce the endogenous insulin production. Consequently total IRI, binding IRI and C-peptide levels decreased, and hypoglycemic attacks disappeared. These results imply that insulin antibody, once induced by bovine insulin, binds with endogenous insulin. Therefore, it is concluded that heterogenous insulin should not be given during hyperalimentation, especially for patients with good glucose tolerance. When insulin antibody developed, it is effective to reduce the dose of glucose in order to decrease endogenous insulin production.

Human insulin autoantibody fine specificity and H and L chain use

Fine specificity and H and L chain isotypes of insulin autoantibodies in sera from 11 subjects were examined. None of these 11 subjects was treated with exogenous insulin. Two patterns of fine specificity were found. In one, the autoantibodies were specific for human insulin, with a requirement for threonine at B30. The conservative substitution in pork insulin (threonine to alanine) abrogated IgG binding by these sera. Insulin autoantibodies in other sera cross-reacted with beef, pork, and human insulin; not requiring threonine at B30. Reciprocal competitive inhibition experiments showed that epitopes recognized by the human specific insulin autoantibodies were exclusively on the B chain, whereas the cross-reactive sera contain autoantibodies that recognize both the B chain and combinatorial (A and B chain) epitopes. The fine specificity of cross-reactive insulin autoantibodies are thus similar to insulin antibodies from insulin-treated subjects. When IgG subclasses and L chains of insulin autoantibodies were examined, however, restricted C region usage was found. The hierarchy was IgG3 greater than G1 greater than G2 greater than G4; with one subclass dominant in each serum, although others were used. L chain use was similarly restricted. There was no correlation between isotype and fine specificity or between H and L chain type. It is concluded that heterogeneity of insulin autoantibodies is restricted. The response is probably more oligo- or pauciclonal than insulin antibody from insulin-treated subjects.

An immunisation protocol which enhances the frequency of antigen-specific monoclonal antibody production

An immunisation protocol has been developed for small molecular weight antigens which results in a high percentage of specific hybridomas being produced after cell fusion. An enzyme-linked immunosorbent assay (ELISA) was developed for screening the desired antibodies in the culture supernatants. A conventional immunisation regimen was followed by doses of antigen in sterile water on each of the last 4 days before fusion. A range of antigen doses was used and the specific efficiency of fusion was increased by selection of the optimum amount. The antigen used as a model antigen in these experiments was biosynthetic human insulin.

The effects of human insulin on antibody formation in pregnant diabetics and their newborns

We studied the immunogenicity of human insulin in 11 diabetic mothers and their newborns. Serum antibody formation was assayed by two different methods. Upon switching four patients from beef/pork insulin to human insulin, we found that elevated baseline antibody levels in three women decreased, in two to undetectable levels at term. The fourth patient had undetectable antibody levels at baseline and borderline levels at term. Only one of their four newborns had antibodies. Upon initiation of insulin treatment in another five diabetics without detectable antibodies at baseline, only two developed antibodies, and only one of their newborns developed antibodies. Two other patients, initially not on insulin, had baseline elevations of antibody that decreased with administration of human insulin; both of their newborns had antibodies. Overt diabetes evolved subsequently in both mothers after pregnancy. We conclude the following: 1) Upon transfer from beef/pork insulin to human insulin, mothers and their newborns show a decrease in insulin antibodies; 2) new patients initiated on insulin develop low levels of antibodies, if any, and their newborns also have low levels of antibodies if any; and 3) the decreased or absent immunogenicity of human insulin supports its use in pregnant diabetics.

Cyclosporin A suppresses insulin autoantibodies and heterologous insulin antibodies in type I diabetic children

We report that cyclosporin A (CsA) suppresses the insulin autoantibodies that are present before insulin therapy in the sera of one-third of studied type I (insulin-dependent) diabetic children. CsA also reversibly blocks the production of antibodies after exogenous insulin injection, whereas high titers of heterologous insulin antibody are observed in all type I patients not receiving CsA.

A comparison of the effects of plasma exchange and immunoadsorption on anti-insulin antibody synthesis in rabbits

Plasma exchange (PE) and ex vivo immunoadsorption (IA) may be applicable to the removal of anti-insulin antibodies (AI-Ab) from diabetic patients. However, the removal of antibodies may prompt an increase in their rate of synthesis and an overshoot of antibody levels which may be deleterious to the patient. The effects of both PE and IA on AI-Ab synthesis were studied in a rabbit model. Rabbits were immunized with insulin and the resulting AI-Abs removed by both plasma exchange and specific immunoadsorption. Following AI-Ab removal by PE no increase in AI-Ab synthesis or antibody overshoot occurred. However a large increase in AI-Ab synthesis and overshoot occurred following specific AI-Ab removal by immunoadsorption. Despite similar reductions in AI-Ab levels by PE and IA, no increase in antibody synthesis occurred due solely to antibody removal. It is likely that antigen released from the immunoadsorbent stimulated the increase in antibody synthesis following immunoadsorption. These findings are relevant to the clinical application of both PE and IA.

Platelet associated IgG (PA IgG) in patients with insulin dependent (type 1) diabetes mellitus

Increased levels of Platelet Associated IgG have been found in 30% of patients with insulin dependent (type 1) diabetes mellitus, closely associated with the simultaneous presence in the serum of insulin antibodies and immune complexes. It is suggested that the increase in PA IgG levels may be of importance in the occurrence of the platelet abnormalities observed in diabetes mellitus.

Characteristics of blood cells responding to insulin in children with insulin-dependent diabetes

Cells responding to insulin in peripheral blood of children with insulin-dependent diabetes mellitus were characterized using monoclonal antibodies and autoradiography. The results indicate that the cells responding to insulin were T cells mainly of the helper (CD4+) phenotype and that they required adherent cells of the monocyte/macrophage lineage for proliferation to occur.