| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Original Studies |
Departments of Medicine I (N.D., M.S., A.B., E.S., S.S., C.B., G.P.) and Nuclear Medicine (F.D., A.S., F.F.), H. San Raffaele Scientific Institute, University of Milan, Milan, Italy; and the Department of Medicine, University of Liége (J.C.S.), Liége, Belgium
Address all correspondence and requests for reprints to: N. Dozio, M.D., Department of Medicine I, H. San Raffaele Scientific Institute, Via Olgettina 60, 20132 Milan, Italy.
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
,
which had a single site dissociation constant (Kd =
10-7 mol/L, by Scatchard analysis), a very fast
dissociation rate of immune complexes, and a very rapid association of
[125I]insulin. Insulin receptors on red blood cells were
down-regulated. The [123I]insulin scintigraphic study
imaged the buffering effect of antibodies on insulin bioavailability.
[123I]Insulin was not removed from the blood, and no
liver or kidney uptake of the hormone occurred. The frequency and
severity of hypoglycemic episodes required treatment. Insulin antibody
levels decreased and [123I]insulin biodistribution
improved after treatment with plasmapheresis and prednisone. Improved
hormone bioavailability was further evidenced by the reduction in the
hypoglycemic delay after iv insulin from 90 min before any treatment to
60 min after plasmapheresis and 30 min after steroid administration.
Glucose tolerance was normal after treatment. Plasmapheresis followed
by steroid treatment can lower the insulin antibody concentration,
abolish severe hypoglycemia, and improve insulin biodistribution and
glucose tolerance in insulin autoimmune hypoglycemia. |
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
-mercaptopropionylglycine, penicillamine, and pyritinol) and with
the clinical or serological evidence of autoimmune disorders (13, 14, 15).
A few idiopathic cases have also been reported (16). In the majority of
the Japanese patients no treatment was required, and spontaneous
remission occurred within 6 months of onset (17). Differences in
associated HLA and clonality of insulin antibodies have been reported
that suggest that the syndrome may vary in ethnically distinct
populations (18). The hypoglycemia and the impairment of glucose metabolism in this syndrome are hypothesized to occur as a result of a buffering effect of high levels of antibody on insulin bioavailability to target tissues. According to this mechanism, the availability of the secreted hormone to receptors in the liver and peripheral tissues would be decreased due to insulin binding to circulating antibodies. This could account for the lack of prompt hypoglycemic response after acute stimulation of insulin secretion or exogenous insulin administration (19) and for the impaired glucose tolerance or overt diabetes reported in some of these patients (11, 14, 20). The half-life of injected insulin in these patients is also significantly prolonged (19), and the release of insulin from the circulating autoantibody pool is expected to be a function of the laws of the equilibrium of mass action and not in response to changes in blood glucose levels, therefore resulting in hypoglycemia. Patients consistently have late hypoglycemia after carbohydrate ingestion or exogenous insulin administration (14, 20), but not after prolonged fasting (11). Although the hypothesis is attractive, it has not been tested using insulin biodistribution studies in vivo. The aim of this study was to demonstrate the buffering effect of antibodies in the autoimmune hypoglycemic syndrome by characterizing insulin autoantibodies in vitro and studying their effects on insulin bioavailability.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The patient was a 58-yr-old white female, slightly overweight
(body mass index, 28.9 kg/m2); presenting no personal or
family medical history for diabetes mellitus, hypoglycemic agents, or
insulin usage or employed in a health-related profession. She reported
a 7-kg weight increase in the year preceding her first documented
hypoglycemic episode. Suddenly, while undergoing traction for
arthrosis, the patient became unconscious. In the Emergency Care
Department her blood sugar was 1.1 mmol/L (20 mg/dL), and she promptly
awakened after iv administration of dextrose. A similar episode
occurred 4 days later. A 75-g oral glucose tolerance test (OGTT) was
indicative of diabetes according to WHO definition and was followed by
late reactive hypoglycemia (Fig. 1
),
whereas a monitored 72-h fast failed to produce a diagnostic fall in
plasma glucose (6 h, 4.6 mmol/L; 12 h, 4.8 mmol/L; 18 h, 4.5
mmol/L; 24 h, 3.9 mmol/L; 30 h, 3.9 mmol/L; 36 h, 3.7
mmol/L; 42 h, 3.6 mmol/L; 48 h, 3.4 mmol/L; 54 h, 3.5
mmol/L; 60 h, 2.3 mmol/L; 66 h, 3.4 mmol/L; 72 h, 2.7
mmol/L). Extremely high levels of total plasma insulin (between
1680–2958 pmol/L) were measured and eventually led to the detection of
insulin antibodies (62% [125I]insulin binding at a 1:2
dilution) and to the diagnosis of insulin autoimmune hypoglycemia. The
patient did not receive any treatment known to be associated with the
insulin autoimmune syndrome. Insulin receptor-, organ-, and
nonorgan-specific autoantibodies were not detectable, nor was other
evidence of autoimmune disorders found. Myeloma was excluded based on
serum and urinary electrophoretic profiles and bone marrow
aspiration.
|
Treatment
The patient was trained to monitor her blood glucose level using a reflectance meter and was discharged on a diet low in simple carbohydrates, consisting of three meals and three snacks (1600 Cal; 45% carbohydrate, 36% fat, and 19% protein). In the next 8 months no comas occurred, but capillary blood glucose levels below 3.3 mmol/L, not associated with specific symptoms, were often recorded. Despite dietary counselling and monthly visits to the out-patient clinic, the patient gained 5 kg. Three episodes of nocturnal hypoglycemic coma then occurred within 10 days and were resolved by glucagon injection and feeding. The patient was hospitalized, and overnight iv infusion of dextrose was required to prevent hypoglycemia.
Four plasmaphereses were performed over 8 days (exchanges of 3 L, with plasma volume replaced by isotonic saline containing 50 g/L albumin). After the second plasmapheresis, no more dextrose infusions were necessary. Immunosuppression using a daily dose of 75 mg prednisone was started and was progressively tapered over 7 months to a maintenance daily dose of 12.5 mg. No further hypoglycemic comas or symptomatic/asymptomatic hypoglycemia occurred during 1 yr of steroid treatment.
Determination of insulin antibodies
Determination of antiinsulin IgG heavy and light chains was performed as previously described by enzymatic immunoassay (6).
Acid-charcoal treatment of serum samples was performed to dissociate insulin from immune complexes and to remove free insulin. Serum aliquots (400 µL) were acidified by adding 120 µL 1 N HCl, followed by 400 µL dextran charcoal and shaking for 5 min at room temperature. After the addition of 680 µL 0.015 mol/L phosphate-buffered saline containing 0.35% BSA and further shaking for 20 min, the samples were centrifuged at 2000 x g for 10 min, and the supernatant was decanted and neutralized with 1 N NaOH.
Insulin antibodies were determined on both untreated and extracted
serum samples. One hundred microliters of serial 2-fold dilutions in
normal serum of either serum or extracted samples were incubated
overnight at 4 C in the presence of a constant amount of
[125I]Tyr-A14 human insulin (39–65 kiloBecquerel/pmol;
20,000 cpm/tube). Immune complexes were precipitated with 18%
Polyethylene Glycol 6000 (PEG 6000, Fluka Chemie, AG, Bucks, CH) in
phosphate buffer, and pellets were counted in a 
-counter. Results
were expressed as the percentage of total added radioactivity. Inter-
and intraassay coefficients of variation were 9.7% (n = 32 in a
2-yr period) and 5.2% (n = 10), respectively (normal values
determined on 102 blood donors samples were 6.9 ± 1.4%;
mean ± SD).
For determination of the affinity of the antibodies by the method of Scatchard, 50 µL extracted serum samples were incubated both overnight at 4 C and at room temperature for 10 min with a constant amount of tracer (as described above) in the presence of increasing amounts of unlabeled insulin, ranging from 6 x 10-11 to 10-6 mol/L. Data were analyzed by the software Ligand produced by Peter Munson.
Dissociation of preformed labeled immune complexes
Undiluted extracted serum was incubated overnight at 4 C in the presence of [125I]human insulin. An excess of cold human insulin (10 IU/mL; Actrapid HM U100, Novo-Nordisk, Bagsvaerd, Denmark) was added, and the mixture was transferred to room temperature under stirring. At selected time points, duplicate aliquots were drawn, immune complexes were precipitated with ice-cold 18% PEG 6000, and both pellet and supernatants were counted.
Association of [125I]insulin to insulin antibodies
[125I]Insulin (100 µL; 20,000 cpm) was added to 1:2 extracted serum dilutions in normal serum at 4 C. Tubes were then transferred at 37 C, and at selected time points, duplicates were precipitated and counted as previously described.
Insulin binding on red blood cells (RBC)
Insulin binding was determined as previously reported (21, 22) on freshly isolated RBC drawn under fasting conditions and purified on cellulose columns. Results are expressed as the percent specific binding for a RBC suspension containing 4 x 106 RBC/µL.
Insulin tolerance test
After an overnight fast, 0.1 IU/kg BW human insulin (Actrapid HM, Novo-Nordisk) were injected iv, and venous plasma glucose was measured at bedside using a glucose oxidase method with a Beckman analyzer (Beckman Instruments, Fullerton, CA) throughout the study. The tests were performed before and immediately after the plasma exchanges and on two occasions during chronic prednisone treatment, 5 and 12 months after plasmapheresis.
Scintigraphic studies with [123I]insulin
Scintigraphic studies using [123I]Tyr-A14
monoiodinated human insulin were performed as previously described (22, 23). The tracer was described to behave as the native hormone (24).
Briefly, the patient was positioned supine under a circular -camera
equipped with medium energy collimator and connected to a dedicated
computer. The field of view included heart, liver, and kidneys. Thyroid
uptake of free iodide was blocked by administering potassium
perchlorate 30 min before the study. One microcurie of
[123I]insulin was injected through an antecubital vein,
and dynamic acquisitions were performed at 30 s/frame for 60 min, both
before and after three plasmaphereses. Regions of interest were drawn
on heart, liver, and kidney where these organs do not overlap. A
background region was drawn in an area without specific uptake, as over
the lung. Time-activity curves for heart, liver, and kidneys were
generated from counts relative to the corresponding regions of interest
after background subtraction. The heart activity was assumed to
represent the blood activity because no uptake of iodinated insulin by
the heart muscle is visualized by this technique. Liver uptake was
quantified as the percentage of the injected dose by calibration with a
liver phantom. To determine the calibration factor between counts per
min and activity in microcuries, images of a human liver phantom
containing known amounts of activities of 123I were
acquired by the same -camera. In each subject a region of interest
including the liver and the right kidney was drawn, and from this, the
equivalent of a region of interest of the left kidney was subtracted.
The liver time activity curve was normalized and expressed as a
percentage of the injected dose using the calibration factor. Ten
healthy volunteers served as controls for these studies.
Analysis of blood radioactivity
At selected times, blood was drawn from an antecubital vein, samples were centrifuged, and 100 µL plasma were treated with 1 mL ice-cold 25% PEG 6000 in water to precipitate the antibody-bound radioactivity. After centrifugation, the supernatant was decanted, and 1 mL 25% trichloroacetic acid (TCA) in water was added to precipitate the remaining protein-associated radioactivity. PEG-precipitated pellet, TCA-precipitated pellet, and the TCA supernatant (containing soluble iodide and iodinated fragments) were counted, and the percentage of radioactivity in each fraction was calculated. The analysis was performed on five normal subjects with the adjunction of a preincubation of plasma with guinea pig antiinsulin antibody (GPAIS 469, provided by Dr. P. H. Wright) to identify immunoreactive insulin.
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Titration of patient serum is shown in Fig. 2, upper panel. Antibody
levels reached 60% in serum diluted 1:5; acid dissociation of
preformed immune complexes did not greatly increase the binding level.
Insulin antibodies were IgG3 with light chain, and no
other subclasses were detected.
|
, lower panel).
Dissociation of the insulin tracer from preformed immune complexes was
very rapid, and most of the bound radioactivity was released from the
antibodies within 5 min (Fig. 2, middle panel).
Insulin binding on RBC was markedly reduced at the time of diagnosis
(Fig. 3), as expected by the sustained
hyperinsulinemia.
|
Figure 1 shows that plasma glucose levels after a 75-g OGTT were
diagnostic for diabetes, and hypoglycemia was observed 300 min after
glucose ingestion.
Intravenous injection of insulin failed to produce a rapid fall in
plasma glucose, and a slow decrease was observed, reaching half of the
initial value 120 min after injection (Fig. 4).
|
shows scintigraphic images 1, 6,
15, and 30 min after tracer injection in a healthy volunteer (Fig. 5, upper panel) and in the patient (Fig. 5, lower
panel). In the normal subject, radioactivity was rapidly removed
from the plasma compartment and taken up by the liver, where it reached
a maximum of 23% of the injected dose at 6 min (23), followed by a
rapid decrease. The uptake of radioactivity in the kidneys paralleled
that in the liver, and 10 min after tracer injection, kidney uptake
became more prominent than liver uptake (Fig. 6, upper panel). In the
patient, insulin biodistribution was significantly altered, with
radioactivity persisting in the blood throughout the study period, as
shown by evident heart and blood vessel images at all times. No
specific uptake of radioactivity over that in the vascular bed was
observed in either liver or kidneys. Time-activity curves showed the
persistence of radioactivity in the blood and the absence of the
rebound observed in normal controls after 6 min (Fig. 6, middle
panel). The curve for the liver reached a maximum corresponding to
4% of the injected dose at 2 min, whereas there was no rise in the
curve for the kidneys, and neither liver nor kidneys had a
radioactivity higher than that in the blood at any time, suggesting
that imaging was due to the blood content in each organ.
|
|
Effect of treatment
Insulin binding to antibody was 62% at the diagnosis and fluctuated between 60–72% during the following year. No changes were observed when repeated hypoglycemic comas occurred. Plasmapheresis resulted in a slight decrease in antibody levels from 72% to 58%, whereas a further progressive decrease was observed during chronic prednisone treatment to 20%.
Insulin bioavailability and glucose tolerance were improved during
prednisone treatment; normal glucose tolerance, determined by a 75-g
OGTT, was restored, and reactive hypoglycemia occurred 150 min after
glucose administration (Fig. 1). After plasmapheresis, the iv injection
of insulin produced a 50% decrease in the plasma glucose level within
60 min, whereas during chronic prednisone treatment, a decrease similar
to that in controls was observed within 30 min (Fig. 4). Moreover, a
decrease in the frequency of capillary blood glucose determinations in
both the low and high ranges was observed after treatment; blood
glucose determinations below 3.3 mmol/L decreased from 14% to 7%, and
88% of capillary blood glucose determinations were in the 3.3–7.8
mmol/L range compared with 71% during treatment with diet alone
(P < 0.005, by 
2 test).
The scintigraphic study performed after plasmapheresis showed a faster
removal of insulin from the blood pool and a slightly higher liver
uptake (Fig. 5, lower panel). Moreover, PEG-precipitable
radioactivity decreased to 81.9% at 1 min and to 39% at 60 min;
conversely, an increase in the level of TCA-soluble radioactivity to
6.5% at 60 min was observed.
Insulin binding on RBC returned to the normal range after a decrease in
antibody level during prednisone treatment (Fig. 3).
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
) was detected with a Scatchard analysis compatible
with a single binding site. The affinity of these insulin antibodies is
low, as Scatchard analysis of the competition experiments showed a
Kd in the micromoles per L range (36), and the
dissociation experiments showed an extremely rapid dissociation of
immune complexes. In view of this low affinity, it is surprising that
insulin bioavailability is so affected (37). However, despite the short
half-life of preformed immune complexes, the rate of association of the
antibodies was very fast. It is likely, therefore, that this high
association constant favors the equilibrium reaction toward the bound
fraction (38). Indeed, as soon as free insulin is in the presence of
the antibodies, it is bound, but it is released very rapidly.
Therefore, a continuous on and off binding process occurs that will
depend upon removal of insulin through a higher affinity interaction
like that with its receptors. In the patient described here, in vivo evidence of a predominant bound insulin fraction is provided by the persistence of [123I]insulin in the blood after iv injection. Insulin is bound to antibodies (94% of labeled insulin is PEG precipitable), and therefore, it is not filtered through kidneys and is prevented from interacting with its own receptors. The resulting images, obtained for the first time in humans, are those of a vascular pool without receptor-mediated uptake in the liver and with nonreceptor-mediated uptake in the kidneys. This pattern of biodistribution is similar to that observed in rats injected with [123I]insulin immune complexes made of guinea pig IgG1 (25).
Unlike the majority of reported cases (11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 30, 33), in this patient treatment became necessary because of life-threatening recurrent hypoglycemic coma. Plasmapheresis resulted in the prompt disappearance of hypoglycemic episodes, partially corrected insulin and glucose tolerance, and improved biodistribution of [123I]insulin despite a minimal decrease in the antibody titer. Normalization of glucose and insulin tolerance was observed with the further decrease in insulin autoantibody levels achieved during long term prednisone treatment. Insulin binding on RBC gradually normalized, because hyperinsulinemia, chronically sustained by release from the antibody-bound pool, decreased with the reduction in insulin antibody levels.
The in vitro characteristics of the antibodies and the pattern of insulin biodistribution observed in the patient described here provide evidence of a buffering effect of insulin antibodies, accounting for the clinical findings of glucose intolerance after an oral glucose load and of delayed hypoglycemia after iv insulin injection observed in patients with insulin autoimmune syndrome. Indeed, in this patient, antibody-bound insulin represents a large and unstable plasma insulin reservoir, and the hormone is delivered to the receptor compartment not according to blood glucose levels, but according to affinity constants and possibly to total insulin concentration. Of note, the impairment of [123I]insulin biodistribution observed in our patient and the associated clinical findings may not be predictable only on the basis of the antibody dissociation constant in the micromolar range, but confirm the importance of the rate of interaction of the antibody with its ligand to determine the overall behavior of immune complexes (37, 38), resulting in persistence of insulin in the plasma compartment.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Received May 13, 1997.
Revised September 3, 1997.
Revised October 28, 1997.
Accepted November 11, 1997.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  S. E. Fineberg, T. T. Kawabata, D. Finco-Kent, R. J. Fountaine, G. L. Finch, and A. S. Krasner Immunological Responses to Exogenous Insulin Endocr. Rev., October 1, 2007; 28(6): 625 - 652. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. H. Kim, J. H. Park, T. S. Park, and H. S. Baek Autoimmune Hypoglycemia in a Type 2 Diabetic Patient With Anti-Insulin and Insulin Receptor Antibodies Diabetes Care, January 1, 2004; 27(1): 288 - 289. [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
) during
prednisone treatment.
) and after (•) charcoal treatment. Middle
panel, Dissociation of [125I]insulin from
preformed immune complexes using the patient’s serum. At time zero, an
excess of cold insulin was added, and aliquots were PEG precipitated at
selected time points. Lower panel, Association of
[125I]insulin to the patient’s serum. At time zero,
tracer was added to serum, and immune complexes were PEG precipitated
at selected time points.
) after
plasmapheresis during prednisone treatment.
