241 HYPOGLYCAEMIA DUE TO AN INSULIN-RECEPTOR ANTIBODY IN HODGKIN’S DISEASE EWAN G. WALTERS1 RICHARD M. DENTON2 JEREMY M. TAVARÉ2 GLYN WALTERS1 Department of Chemical Pathology, Bristol Royal Infirmary,1 and Department of Biochemistry, University of Bristol Medical School2 Summary Severe fasting hypoglycaemia developed in a patient with Hodgkin’s disease after many courses of chemotherapy. Her serum contained a factor which stimulated glucose uptake by rat adipocytes, and this factor was found in the immunoglobulin fraction. The serum also displaced insulin bound to human erythrocytes and both precipitated and phosphorylated insulin receptors extracted from human placenta. The insulin-like substance is probably an antibody to the insulin receptor. Introduction SYMPTOMATIC hypoglycaemia has been reported in association with a wide variety of non-pancreatic tumours,l very few of which have been lymphomas.2.3 Several explanations for the occurrence of such hypoglycaemia have been proposed,4 but the disorder is still not understood and it seems likely that more than one mechanism is responsible. Radioimmunoassays for insulin in plasma are usually negative," but in a proportion of cases high plasma levels of insulin-like growth factors have been reported.5 Autoantibodies to insulin receptors occur in various disorders. In most cases there is apparent blocking of the receptor and associated insulin resistance and hyperglycaemia,6-8 but in a few there is stimulation of the receptors, resulting in hypoglycaemia.9,10 Some of the reported cases had diseases known to be associated with autoimmune disturbances, such as acanthosis nigricans6 and lupus nephritis,9 but in others no initiating cause of autoantibody production was apparent. We now report a patient with Hodgkin’s lymphoma in whom severe hypoglycaemia was apparently caused by an autoantibody to the insulin receptor. Case-report A 50-year-old woman was admitted to hospital in December, 1978, with severe spontaneous hypoglycaemia. She had previously undergone radiotherapy and twelve courses of chemotherapy since the diagnosis of Hodgkin’s lymphoma 10 years previously. Autoimmune haemolytic anaemia had developed 5 years after the onset of the lymphoma, but responded to chemotherapy. Severe hypoglycaemia now recurred episodically, necessitating regular carbohydrate meals, without which the blood glucose fell within a few hours to less than 1.5 mmol/1. The patient needed to be woken during the night to take carbohydrate. When the attacks began there were no clinical signs of Hodgkin’s lymphoma but the plasma alkaline phosphatase activity was persistently 4 times higher than normal; electrophoresis showed that only the liver isoenzyme was increased. The serum copper concentration was also high (41 umol/1; normal range 12-24 µmo1/1). Both abnormalities were consistent with a further relapse of Hodgkin’s disease and hepatic infiltration, but ultrasound and computed tomography scans of the liver were negative, and a liver biopsy taken at laparoscopy revealed no evidence of Hodgkin’s lymphoma. Bone-marrow aspirate and abdominal lymphography were likewise negative. Plasma insulin and C-peptide were measured during hypoglycaemia on several occasions. Both were below the limits of detection of the assay (3 mU/1 and 0-02 -nmol/1, respectively). ’ Calcium infusion and diazoxide infusion tests were carried out; neither result supported the diagnosis of insulinoma, which was therefore thought to have been excluded. The serum cortisol level was 536 nmol/1 (normal early morning range 180-700 nmol/1) and serum growth hormone 12-5 mU/1 (normal range at rest < 2 mU/1) when the blood glucose was less than 1 mmol/1.1 1 mg glucagon intravenously promptly raised blood glucose from 1 5 to 6-5 mmol/1. Serum somatomedin, measured by the incorporation of sulphur-35-labelled sodium sulphate into cartilage, was 75% of the standard level in the United Kingdom; this is just at the lower limit of the normal range (75-120%). Since these tests appeared to exclude a deficiency of these counter- regulatory hormones caused by previous treatments" and a deficiency of hepatic glycogen, further studies were undertaken. Hypoglycaemia was controlled by frequent carbohydrate meals for a time, but the patient deteriorated generally and was readmitted in 1980 with splenomegaly and very frequent attacks of hypoglycaemia difficult to control by frequent meals. Chemotherapy was begun but she died 2 weeks later. Necropsy revealed Hodgkin’s lymphoma. The para-aortic lymph-nodes and those around the stomach and pancreas were enlarged, and there were large masses of tumour in the liver (2065 g) and spleen (420 g). No pancreatic tumour was found on thin slicing of the pancreas, and the other endocrine glands were normal. Microscopically, the enlarged lymph-nodes were fibrotic and contained Reed- Stemberg-like cells and many eosinophils and lymphocytes. The appearance of the tumour in liver and spleen was similar. Re-examination of the lymph-node biopsy on which the diagnosis was made showed that the tumour was then of the nodular, lymphocyte-predominant variety with a granulomatous reaction. Gene probes specific for immunoglobulin heavy chain and lambda and kappa light chains were applied to DNA fragments from frozen tumour by the Southern blot technique. There was no evidence of a monoclonal rearrangement. Further Investigations Bioassay of Insulin-like Activity Epididymal fat pad pieces from male Wistar rats (6-8 weeks) were preincubated for 30 min at 37°C in standard Krebs bicarbonate buffer containing 5-6 mmo1/1 glucose, then in fresh medium for a further 90 min with the addition of various quantities of the patient’s serum, collected during hypoglycaemia, or serum from normal fasting subjects whose plasma insulin was less than 5 mU/1. Porcine insulin, 1 mU/ml, was added to a sample of normal fasting serum to provide a positive control. Glucose uptake by the fat pieces was estimated by measuring the disappearance of glucose from the medium. ’2 Insulin Binding to Erythrocytes Insulin was labelled with iodine-125 at the tyrosine residue (A-14) by lactoperoxidase iodination, and purified by gradient reverse-phase high-performance liquid chromatography with acetonitrile/trifluoracetic acid solvents. Binding of I125-1abe11ed insulin to erythrocytes from a fasting subject was studied by the method of Gambir et al.13 IgG Purification IgG was removed from the patient’s serum by passage through a column of staphylococcal protein-A-’Sepharose’. The bound IgG was eluted with glycine 01 mol/1 at pH 30. Both IgG-containing and IgG-free fractions were concentrated by ultrafiltration (’Millipore’) and subjected to rocket immunoelectrophoresis; this process confirmed the presence of IgG in the glycine eluate and its absence from the other fraction. The recovery of IgG was 65%.
Transcript
241
HYPOGLYCAEMIA DUE TO ANINSULIN-RECEPTOR ANTIBODY IN
HODGKIN’S DISEASE
EWAN G. WALTERS1RICHARD M. DENTON2
JEREMY M. TAVARÉ2GLYN WALTERS1
Department of Chemical Pathology, Bristol Royal Infirmary,1 andDepartment of Biochemistry, University of Bristol Medical School2
Summary Severe fasting hypoglycaemia developed ina patient with Hodgkin’s disease after many
courses of chemotherapy. Her serum contained a factorwhich stimulated glucose uptake by rat adipocytes, and thisfactor was found in the immunoglobulin fraction. Theserum also displaced insulin bound to human erythrocytesand both precipitated and phosphorylated insulin receptorsextracted from human placenta. The insulin-like substanceis probably an antibody to the insulin receptor.
Introduction
SYMPTOMATIC hypoglycaemia has been reported inassociation with a wide variety of non-pancreatic tumours,lvery few of which have been lymphomas.2.3 Several
explanations for the occurrence of such hypoglycaemia havebeen proposed,4 but the disorder is still not understood andit seems likely that more than one mechanism is responsible.Radioimmunoassays for insulin in plasma are usuallynegative," but in a proportion of cases high plasma levels ofinsulin-like growth factors have been reported.5Autoantibodies to insulin receptors occur in various
disorders. In most cases there is apparent blocking of thereceptor and associated insulin resistance and
hyperglycaemia,6-8 but in a few there is stimulation of thereceptors, resulting in hypoglycaemia.9,10 Some of thereported cases had diseases known to be associated withautoimmune disturbances, such as acanthosis nigricans6 andlupus nephritis,9 but in others no initiating cause of
autoantibody production was apparent.We now report a patient with Hodgkin’s lymphoma in
whom severe hypoglycaemia was apparently caused by anautoantibody to the insulin receptor.
Case-reportA 50-year-old woman was admitted to hospital in December,
1978, with severe spontaneous hypoglycaemia. She had previouslyundergone radiotherapy and twelve courses of chemotherapy sincethe diagnosis of Hodgkin’s lymphoma 10 years previously.Autoimmune haemolytic anaemia had developed 5 years after theonset of the lymphoma, but responded to chemotherapy. Severehypoglycaemia now recurred episodically, necessitating regularcarbohydrate meals, without which the blood glucose fell within afew hours to less than 1.5 mmol/1. The patient needed to be wokenduring the night to take carbohydrate.When the attacks began there were no clinical signs of Hodgkin’s
lymphoma but the plasma alkaline phosphatase activity waspersistently 4 times higher than normal; electrophoresis showedthat only the liver isoenzyme was increased. The serum copperconcentration was also high (41 umol/1; normal range 12-24µmo1/1). Both abnormalities were consistent with a further relapse ofHodgkin’s disease and hepatic infiltration, but ultrasound andcomputed tomography scans of the liver were negative, and a liverbiopsy taken at laparoscopy revealed no evidence of Hodgkin’slymphoma. Bone-marrow aspirate and abdominal lymphographywere likewise negative.
Plasma insulin and C-peptide were measured duringhypoglycaemia on several occasions. Both were below the limits ofdetection of the assay (3 mU/1 and 0-02 -nmol/1, respectively).
’ Calcium infusion and diazoxide infusion tests were carried out;neither result supported the diagnosis of insulinoma, which wastherefore thought to have been excluded.The serum cortisol level was 536 nmol/1 (normal early morning
range 180-700 nmol/1) and serum growth hormone 12-5 mU/1(normal range at rest < 2 mU/1) when the blood glucose was lessthan 1 mmol/1.1 1 mg glucagon intravenously promptly raised bloodglucose from 1 5 to 6-5 mmol/1. Serum somatomedin, measured bythe incorporation of sulphur-35-labelled sodium sulphate intocartilage, was 75% of the standard level in the United Kingdom;this is just at the lower limit of the normal range (75-120%). Sincethese tests appeared to exclude a deficiency of these counter-regulatory hormones caused by previous treatments" and a
deficiency of hepatic glycogen, further studies were undertaken.Hypoglycaemia was controlled by frequent carbohydrate meals
for a time, but the patient deteriorated generally and was readmittedin 1980 with splenomegaly and very frequent attacks of
hypoglycaemia difficult to control by frequent meals.
Chemotherapy was begun but she died 2 weeks later.Necropsy revealed Hodgkin’s lymphoma. The para-aortic
lymph-nodes and those around the stomach and pancreas wereenlarged, and there were large masses of tumour in the liver (2065 g)and spleen (420 g). No pancreatic tumour was found on thin slicingof the pancreas, and the other endocrine glands were normal.Microscopically, the enlarged lymph-nodes were fibrotic andcontained Reed- Stemberg-like cells and many eosinophils andlymphocytes. The appearance of the tumour in liver and spleen wassimilar. Re-examination of the lymph-node biopsy on which thediagnosis was made showed that the tumour was then of thenodular, lymphocyte-predominant variety with a granulomatousreaction.Gene probes specific for immunoglobulin heavy chain and
lambda and kappa light chains were applied to DNA fragmentsfrom frozen tumour by the Southern blot technique. There was noevidence of a monoclonal rearrangement.
Further Investigations
Bioassay of Insulin-like Activity
Epididymal fat pad pieces from male Wistar rats (6-8 weeks)were preincubated for 30 min at 37°C in standard Krebsbicarbonate buffer containing 5-6 mmo1/1 glucose, then in freshmedium for a further 90 min with the addition of various quantitiesof the patient’s serum, collected during hypoglycaemia, or serumfrom normal fasting subjects whose plasma insulin was less than 5mU/1. Porcine insulin, 1 mU/ml, was added to a sample of normalfasting serum to provide a positive control. Glucose uptake by thefat pieces was estimated by measuring the disappearance of glucosefrom the medium. ’2
Insulin Binding to ErythrocytesInsulin was labelled with iodine-125 at the tyrosine residue
(A-14) by lactoperoxidase iodination, and purified by gradientreverse-phase high-performance liquid chromatography with
acetonitrile/trifluoracetic acid solvents. Binding of I125-1abe11edinsulin to erythrocytes from a fasting subject was studied by themethod of Gambir et al.13
IgG Purification
IgG was removed from the patient’s serum by passage through acolumn of staphylococcal protein-A-’Sepharose’. The bound IgGwas eluted with glycine 01 mol/1 at pH 30. Both IgG-containingand IgG-free fractions were concentrated by ultrafiltration
(’Millipore’) and subjected to rocket immunoelectrophoresis; thisprocess confirmed the presence of IgG in the glycine eluate and itsabsence from the other fraction. The recovery of IgG was 65%.
242
Immunoprecipitation and Phosphorylation of Human InsulinReceptorsA preparation of insulin receptors was obtained by extracting
human placental syncytiotrophoblast cell membranes with ’TritonX100’ (1 % weight/volume);" 60 µ1 of the preparation (10 mg/mlprotein) was incubated at 0°C for 6 h in a total volume of 180 µ1
- containing 0-25 mmol/l dithiothreitol, 0-15 mol/1 sodium chloride,and 5 mmol/1 sodium phosphate in the presence of the appropriateserum. Protein-A-sepharose (14 mg) was added for the final 1 h.The receptor, antibody, and protein-A-sepharose complex wasisolated by centrifugation (10 OOOg for 10 s at 4°C) and washed threetimes with 500 pl 0- 15 mol/1 sodium chloride in 5 mmol/1 sodiumphosphate, pH 7-4. The pellet was resuspended in a total volume of60 µ1, containing 20 mmol/1 MOPS buffer pH 74, 0 25 mmol/1dithiothreitol,. 1 mmol/1 sodium orthovanadate, 12 mmol/1magnesium chloride, and 2 mmol/1 manganese chloride.
Phosphorylation of the immunoprecipitated insulin receptors wasinitiated by the addition of 100 pnol/1 y-P3z-ATP (500 cpm/pmol).After incubation at 0°C for 30 min the receptors were separated bycentrifugation at 10 000 g for 10 s at 4°C, and the pellet resuspendedin 250 µ1 SDS sample buffer (20% sucrose, 10% sodium dodecylsulphate, 125 mmol/1 "tris"-HCI pH 6-8, and 1%
2-mercaptoethanol). The sample was boiled for 5 min, cooled, andthe supernatant subjected to electrophoresis on a 6% SDS
polyacrylamide gel.15 The gel was fixed, stained, destained, dried,and autoradiographed by means of Kodak ’Xomat S’ film incassettes with intensifying screens for 7 days at -80°C.
Results
Bioassay of Insulin-like Activity
Samples of the patient’s serum increased the rate ofglucose uptake by pieces of rat epididymal adipose tissue bymore than 200% even when the serum was diluted 100-fold(see table). By comparison, a concentration of insulin (1 U/1)giving maximum stimulation resulted in a 356% increase inglucose uptake in the absence of added serum. Controlfasting serum (50 µ1m1) had little effect on basal glucoseuptake but appeared to diminish slightly the effect of addedinsulin. After removal of IgG the patient’s serum no longerstimulated glucose uptake. However, the glycine eluatecontaining IgG had about two thirds of the effect of wholeserum.
Displacement of I125-labelled Insulin Bound to HvmdnErythrocytes
Samples of the patient’s serum at a dilution of 30 µ1/m1buffer were able to displace 55% of I12s-insulin specificallybound to human erythrocytes. This displacement wasequivalent to that found with insulin added at 25 µU/m1 ;
EFFECT OF PATIENT’S SERUM ON GLUCOSE UPTAKE BY RAT
EPIDIDYMAL ADIPOSE TISSUE PIECES
*As % of uptake in absence of additions (2. 22 ± 0 46 µmo1/h/g wet wt oftissue). ’
tSamples from three separate hypoglycaemic attacks.Equivalent to 50 µ1 original serum per ml.§Equivalent to 32 (µ1 original serum per ml.
Autoradiographs of p32-labe11ed proteins after separation by SDSpolyacrylamide gel electrophoresis.
Arrow indicates &bgr;-subunit.Lanes 1 and 2: human placental insulin receptors. Lanes 3 and 4: patient’s
serum diluted 1/10. Lanes 5 and 6: patient’s serum diluted 1/50. Lanes 7 and8: serum B9 diluted 1/50. Lanes 9 and 10: serum from control subject diluted1/10 and 1/50, respectively.
thus the potency of the insulin-like agent in the originalserum of the patient was equivalent to about 750 mU/1insulin.
Specific Receptor ImmunoprecipitationThese studies suggested that the serum of the patient
contained antibodies to the insulin receptor. We confirmedthis suggestion by showing that the serum was able toimmunoprecipitate insulin receptors in the presence of
protein-A-sepharose, which binds IgG. The presence ofinsulin receptors in the immunoprecipitates was detected byfollowing the incorporation of phosphorus-32 from added&Ugr;-P32-ATP into the (3-subunit of the insulin receptor. Theinsulin receptor contains an intrinsic insulin-stimulatedprotein kinase activity, which catalyses the phosphorylationof specific tyrosine residues within the &bgr;-subunit.16,17
Appreciable amounts of phosphorylated &bgr;-subunits werepresent in the immunoprecipitates after incubation of thepatient’s serum with &Ugr;— p32_A TP, but virtually none werepresent after incubation of control serum (see figure).However, the patient’s serum was not as potent as that froma patient (B9) with acanthosis nigricans and severe insulinresistance which, at an equivalent dilution, precipitatedmany more insulin receptors (gift of Prof R. Kahn, JoslinDiabetes Center, Boston, USA). The further addition ofinsulin to the immunoprecipitates had only small effects onthe extent of phosphorylation, indicating that the interactionof human placental insulin receptors with antibodies in bothsera could induce near-maximum autophosphorylation.
Discussion
The stimulatory effect of the patient’s diluted serum onglucose uptake by the rat adipocyte was far greater thancould be accounted for by the non-specific effect of IgG on
243
adipocytes.18 The other results establish that the serum ofthe patient contained antibodies to the insulin receptor. Itseems reasonable to conclude that the patient’shypoglycaemia was due to the presence of these antibodies.Most previous cases of insulin-receptor antibodies,although the antibodies had insulin-like effects in vitro, havebeen characterised by hyperglycaemia and severe insulinresistance in vivo. The reason for the difference between
previous cases and our patient remains to be established. Invitro, autoantibodies to the insulin receptor may reduce thenumber of insulin receptors on cultured cells so thatinsulin resistance might be expected to develop as the titre ofantibodies rises. However, there is no evidence that the
reported patients with insulin resistance had passed througha phase of fasting hypoglycaemia.We have been unable to prove conclusively that there is a
direct link between Hodgkin’s disease and the appearance ofthe anti-receptor antibodies in this patient. However, such alink is suggested by the previous development of anautoimmune haemolytic anaemia, which was overcome bytreatment of the Hodgkin’s disease, and by the lymphocyte-predominant nature of the tumour. It has been suggestedthat the latter variety of Hodgkin’s disease is a tumour of Blymphocytes, with the implication that autoantibodies couldbe produced by a single clone of cells. Exploration of thispossibility by means of gene probes did not reveal evidenceof a monoclonal rearrangement of DNA.We can only speculate on the stimulus giving rise to the
formation of the receptor antibody. Since insulin receptorsare very widely distributed, the expression of an abnormalreceptor by the tumour cells is one possibility; another is theexpression of insulin-receptor antibodies as an anti-idiotyperesponse to insulin-antibody complexes.19 The lack ofevidence for insulin antibodies and the normal plasmainsulin levels appear to exclude this possibility.Hypoglycaemia appears rare in Hodgkin’s lymphoma2O
but our patient is not unique (see Braund and colleagues’paper, p 237 of this issue). Certainly, the possibility thathypoglycaemia may be due to insulin-receptor autoanti-bodies should be considered in all cases of fastinghypoglycaemia for which a cause cannot readily be found.When the underlying cause cannot be treated steroids mayalleviate the symptoms through their effect on antibodyfonnation.8
We thank Dr B, T. Hale for permission to include details of his patient,Dr N. Rooney for the DNA studies, and Dr G. S. G. Spencer of the AFRCInstitute of Food Research, Langford, for the somatomedin assay. Thestudies were supported by grants from the Bristol and Weston HealthAuthority and the British Diabetic Association.
Correspondence should be addressed to G. W., Department of ChemicalPathology, Bnstol Royal Infirmary, Bristol BS2 8HW. ,
REFERENCES
1 Marks V. Hypoglycaemia. 2. Other causes Clin Endocrinol Metab 1976; 5: 769-82.2. Wanebo HJ, Schlessinger I, Tashima CK. Severe hypoglycaemia associated with
terminal lymphomas. Report of five cases. Cancer 1966, 19: 1451-58.3 Abbasi A, Power L Insulin and insulin-like activity in extracts of tumours associated
with hypoglycaemia Diabetes 1973; 22: 762-67.4 Unger RH. The riddle of tumour hypoglycemia. Am J Med 1966; 40: 325-30.5 Gorden P, Hendricks CM, Kahn CR, Megyesi K, Roth P Hypoglycemia associated
with non-islet-cell tumour and insulin-like growth factors. N Engl J Med 1981, 305:1452-55
6. Kahn CR, Flier JS, Bar RS, et al. The syndromes of insulin resistance and acanthosisnigricans. N Engl J Med 1976; 294: 739-45.
7 Taylor SI, Marcus-Samuels B. Anti-receptor antibodies mimic the effect of insulin todown-regulate insulin receptors m cultured human lymphoblastoid (IM-9) cells. JClin Endocrinol Metab 1984; 58: 182-86
responsive diabetes mellitus in a patient with myositis Br Med J 1984; 288: 1578.
When the following paper by Dr Anand and his colleaguesappeared in our issue of Jan 10 we failed to include the fïveappeared in our issue of Jan 10 we failed to include the fivefigures referred to in the text. With our renewed apologies to theauthors, we now reproduce the complete paper.
RANDOMISED TRIAL OF FENTANYLANAESTHESIA IN PRETERM BABIES
UNDERGOING SURGERY: EFFECTS ON THESTRESS RESPONSE
K. J. S. ANAND*
Department of Paediatrics, John Radcliffe Hospital, Oxford
Department of Child Health, Royal Victoria Infirmary,Newcastle upon Tyne
Summary In a randomised controlled trial, pretermbabies undergoing ligation of a patent
ductus arteriosus were given nitrous oxide and d-
tubocurarine, with (n = 8) or without (n = 8) the addition offentanyl (10 µg/kg intravenously) to the anaesthetic
regimen. Major hormonal responses to surgery, as indicatedby changes in plasma adrenaline, noradrenaline, glucagon,aldosterone, corticosterone, 11-deoxycorticosterone, and11-deoxycortisol levels, in the insulin/glucagon molar ratio,and in blood glucose, lactate, and pyruvate concentrationswere significantly greater in the non-fentanyl than in thefentanyl group. The urinary 3-methylhistidine/creatinineratios were significantly greater in the non-fentanyl groupon the second and third postoperative days. Compared withthe fentanyl group, the non-fentanyl group had circulatoryand metabolic complications postoperatively. The findingsindicate that preterm babies mount a substantial stress
response to surgery under anaesthesia with nitrous oxide
*Present address: Department of Anesthesia, Harvard Medical School,Children’s Hospital, 300 Longwood Avenue, Boston MA 02115, USA.
9. Tardella L, Rossetti L, De Pirro R, et al. Circulating anti-insulin receptor antibodies ina patient suffering from lupus nephritis and hypoinsulinemic hypoglycaemia JClin Lab Immunol 1983; 12: 159-65
10. Taylor SI, Grunberger G, Marcus-Samuels B, et al. Hypoglycemia associated withantibodies to the insulin receptor. N Engl J Med 1982; 307: 1422-26.
11. Boden G, Reichard GA, Hoeldtke RD, Rezvani I, Owen OE. Severe insulin-inducedhypoglycemia associated with deficiencies to the release of counterregulatoryhormones. N Engl J Med 1981; 305: 1200-05
12. Slein MW. D-glucose determinations with hexokinase and glucose-6-phosphatedehydrogenase. In Bergmeyer HY, ed. Methods of enzymatic analysis. London,New York. Academic Press, 1965: 117-30.
13. Gambir KK, Archer JA, Carter L. Insulin radioreceptor assay for human
