580
are cytotoxic effector cells against trophoblast in early pregnancy. Our data show that changes of NK activities are related
NK cells
outcome after immunotherapy. NK activity before might be important in decisions about immunotherapy for patients with unexplained recurrent spontaneous abortion. In those with low pretreatment NK, activity factors other than NK activity may contribute to abortion, or abortion may be due to different factors in these patients and those with high NK activity. Finally, we are concerned about the patients in whom NK activity was raised by immunotherapy but who had not conceived more than a year after immunotherapy, and believe that further investigation is necessary.
clinical
to
treatment
This work
was
supported by
a
Grant-in-Aid from the Ministry of
Education, Science, and Culture of Japan.
RYUJI MAKIDA Departments of Transfusion Medicine and Immunohaematology, and Obstetrics and Gynaecology, Faculty of Medicine, University of Tokyo, Tokyo 113, Japan
MUTSUHIKO MINAMI MASARU TAKAMIZAWA TAKEO JUJI TOMOYUKI FUJII MASAYUKI MIZUNO
Fujii T, Takamizawa M, Juji T, Kawana T, Mizuno M. Outcome of pregnancy after injection of mononuclear cells. Am J Obstet Gynecol 1988; 158: 1015-16. 2. King A, Loke YW. Uterine large granular lymphocytes: a possible role in embryonic implantation. Am J Obstet Gynecol 1990; 162: 308-10. 3. Bumig JW. Carboxy-fluorescein fluorochromasia assay: I non-radioactively labeled 1.
4. De
cell-mediated lympholysis. J Immunol Methods 1980; 33: 33-44. Fougerolles AR, Baines MG. Modulation of the natural killer cell activity in pregnant mice alters spontaneous abortion rate. J Reprod Immunol 1987; 11: 147-53.
Helix pomatia in breast
cancer
SIR,-Dr Brooks and Dr Leathern (July 13, p 71) report that staining of primary breast cancers for Helix pomatia (HPA) is an "excellent predictor of long-term patient prognosis" and that it is a "possible staging adjunct in patients apparently lymph-node negative". We have done a pilot study on 363 primary breast cancers taken at random from the files of the International (Ludwig) Breast Cancer Study Group Trial V conducted between 1981 and 1985. Details of this trial have been published elsewhere.1,2 All cases were stained immunocytochemically with the lectin and specificity was confirmed by absorption with the appropriate sugar. Scoring, as described by Brooks and Leathem, was done by two pathologists. All samples were coded and the results sent to the Biostatistics Center, Dana Farber Cancer Institute, where the clinical correlations were studied. A detailed analysis has been done but what follows is limited to the questions raised by Brooks and Leathem’s paper. There was a weak correlation between HPA binding and lymph node involvement. In contrast to the findings of Brooks and Leathern only 70% of node-positive cases were positively stained, making HPA binding a poor predictor of lymph node staging in our material:
Disease-free survival (DFS) was the same for HPA non-stainers for HPA stainers (6-year DFS 56% [SE 5] and 56% [3], respectively). We evaluated DFS and overall survival according to lymph node status and HPA binding. While lymph node status was significant, as expected, HPA binding had no predictive value: as
*p< 0 0001 for node+ vs node - (univariate analysis by log-rank test) tNo significant differences for HPA- vs HPA+ by umvanate analysis or by multivarite analyses based on Cox model with lymph node status (N -, 1-3N +, > 3N +), H PA, grade, and tumour size
37% of node-negative patients received no chemotherapy and 63 % received one cycle of perioperative chemotherapy. Among the
node-positive patients, 67% received six or seven cycles of chemotherapy and 33% received one cycle of perioperative chemotherapy. Thus, in contrast to Brooks and Leathem, we can find no evidence that HPA staining is of value as a prognostic marker. In our cases lymph node status was determined pathologically after total mastectomy with complete axillary clearance, and an average of 14 nodes
were
examined per
case.
It is difficult
to
draw direct
comparisons with the Brooks study since no details are provided on how lymph node status was assessed. Treatment may affect outcome and information on this is not provided by Brooks and T eathem Section of Cell Biology and Experimental Pathology, Institute of Cancer Research, Haddow Laboratories, Sutton, Surrey SM2 5NG, UK
International
Breast Cancer
(Ludwig) Study Group (IBCSG),
Boston, Massachusetts, USA and Bern, Switzerland
C. W. TAYLOR R. ANBAZHAGAN H. JAYATILAKE A. ADAMS B. A. GUSTERSON K. PRICE R. D. GELBER A.
GOLDHIRSCH,
behalf of pathologists and participants of the IBCSG
on
Ludwig Breast Cancer Study Group. Combination adjuvant chemotherapy for node-positive breast cancer: inadequacy of a single perioperative cycle. N Engl J Med 1988, 319: 677-83. 2. Ludwig Breast Cancer Study Group. Prolonged disease-free survival after one course of perioperative adjuvant chemotherapy for node negative breast cancer. N Engl J 1.
Med 1989; 320: 491-96.
***This letter has been shown to Dr Brooks and Dr Leathern, whose reply to it and a previous letter follows.-ED. L. SIR,-Mr Galea and colleagues (Aug 10, p 392) and Dr Taylor et al raise questions that seem to relate directly to our methodology. Our earlier observation1 that binding of Helix pomatia lectin (HPA) to primary breast cancers was associated with shorter survival was followed by three publications confirming our results with an indirect antibody, an avidin-biotin, and a fluorescein method.2-4 Of six different methods we tried on cancers with long versus short survival, we found that an indirect antibody method gave staining results that most closely reflected survival. We described earlier how various methods to detect lectin binding to tissue sections could give different results.5,6Last year we tried to develop a simple staining and scoring system to detect HPA binding to breast cancer sections. We chose a direct HPAperoxidase conjugate (Sigma) to stain our series and obtained clean staining that was straightforward to interpret as positive or negative. In parallel we used an indirect antibody method. Comparing results with survival showed a trong correlation with prognosis by the indirect method (p < 0-00001) but only a weak correlation by the direct method (p < 0-03), and we stopped using this direct method. We were surprised by this difference. The earlier HPA results of the Nottingham group in 1987,z with an indirect antibody method, showed a close association between HPA binding and survival (p < 0-001) but the group’s recent results (Aug 10) with a direct conjugate show no significant differences between stainers and non-stainers. This is in accord with the results of our direct conjugate staining method. We have discussed this with them and agree that "the most probable explanation must lie in the difference between the staining method and the source of HPA. Although, as Galea et al mention, many of our lectins and conjugates have been home-made; both the native and conjugated HPA described here have been obtained from Sigma. Our rabbit polyclonal antiserum reacts with all HPA sources we have tested. We are evaluating a monoclonal antibody to HPA that is directed against an epitope seemingly distinct from the binding site and produces good staining, but we do not yet know how the results compare with tumour behaviour. Taylor et al do not show that HPA has a predictive value for node status or prognosis, but the methodology is not explained. To compare results from different centres the basic methodology must be the same, and polyclonal rabbit antiserum is available from us and others to centres wishing to test this out. There are several explanations for the differences noted that relate to the binding of a lectin. First, the binding of lectins, just as that of antibodies and other
581
receptor molecules, is an equilibrium based on the spatial arrangement of receptors and ligands, such that the closer theyare, the tighter or more effective the binding (eg, the polyvalent IgM or the binding of Fc receptors). The differences we describe on tissue sections may be an artifact based on spatial arrangement of GaINAc; there is some evidence for this in the complex bands seen by PAGE/lectin blotting of both Helix-positive and Helix-negative tumours, which can be completely inhibited by 10 mmol/1 GaINAc. Second, if two lectins with the same nominal simple monosaccharide specificity are incubated on the same tissue, however, then quite different histological structures may be stained, although both lectins may be totally inhibited by the same simple sugar. Most lectins have more than one binding site and some are known to be bifunctional (eg, discoidin 1); GaINAc binding lectins such as DBA and SBA can also bind to non-carbohydrate ligands, such as adenine. Third, what lectin binding detects is how many sugars are exposed and how close they are. But binding is determined not only by the sugars but also by their 0 or N linkages to peptides, topology, adjacent structures, charge, hydrophobicity, and van de Waals forces. If the binding site is altered (as by conjugation) simple sugars may continue to inhibit binding to a section but the specificity to complex sugars is altered. If the hydrogen bond energy is donated by peroxidase, for example, then the binding tightness with the glycoconjugate at that point is weakened. Such derivation of lectins can alter the binding characteristics of lectins-for example, succinylation of WGA reduces the affinity for sialic acid although it does not affect binding to G1cNAc, and this property has been applied to tissue sections. Peroxidase is highly glycosylated and avidly reacts with the binding sites of concanavalin A which at the same time can react with tissue glycoconjugates. We do not know how peroxidase reacts with HPA. Concanavalin A conjugated to ’Sepharose’ shows a different spectrum of glycoprotein binding to the native lectin. But this is an unexplored field. With HPA-HRP the staining pattern is very clear and presumably reflects some specific structures, which can be blocked by Ga1NAc, but not the same structure as those detected by native HPA. The use of direct peroxidase-lectin seems to give different results from the indirect methods. Until the HPA binding glycoconjugates are characterised and the methodology is better understood, we suggest that the direct peroxidase-HPA method may be less useful in this system than an indirect method. Department of Histopathology, University College and Middlesex School of Medicine, Middlesex Hospital, London W1P 7PN, UK
SUSAN BROOKS ANTHONY LEATHEM
1 Leathem A, Brooks S. Predictive value of lectin binding on breast-cancer recurrence and survival. Lancet 1987; i 1054-56. 2. Fenlon S, Ellis IO, Bell J, Todd JH, Elston CW, Blamey RW. Helix pomatia and Ulex europeus lectin binding in human breast carcinoma. J Pathol 1987; 152: 169-76. 3. Alam SM, Whitford P, Cushley W, George WD, Campbell AM. Flow cytometric analysis of cell surface carbohydrates in metastatic human breast cancer. Br J Cancer 1990; 62: 238-42. 4. Fukutomi T, Itabashi M, Tsuane S, Yamamoto H, Nanasawa T, Hiroto T. Prognostic contributions of Helix pomatia and carcinoembryonic antigen using histochemical techniques in breast carcinomas. Jpn J Clin Oncol 1989; 19: 127-34. 5. Leathem A, Atkins N. Lectin binding to formalin-fixed paraffin sections. J Clin Pathol 1983; 36: 747-50. 6. Leathem A. Lectin histochemistry. In: Polak J, Van Noorden S, eds. 2nd ed. Bristol: Wnght, 1986: 167-87.
Beclobrate and fatal acute
hepatitis
SIR,-We report the death of a 64-year-old man from acute hepatic failure attributable to beclobrate, a lipid-lowering agent. The patient’s hypercholesterolaemia, first diagnosed at the age of 56, was treated by diet and then with a succession of hypolipidaemic agents. In addition, he received chondroitin sulphate 500 mg thrice daily for osteoarthrosis of the hip. There was no history of self-medication or alcohol abuse. 6½ months after he had started treatment with beclobrate 100 mg daily he became anorexic and jaundiced. His total bilirubin was 304 umol/1 (normal < 17), and transaminases were very much raised. He was admitted to hospital 3 weeks later with a diagnosis of acute hepatitis. He died 7 days later. Investigations revealed no evidence for an infectious (viral or toxoplasmosis) or obstructive cause.
At necropsy there was liver panlobular hepatic necrosis with bridging, consistent with an adverse drug reaction. The histological appearances were identical to those found by the same histopathologist (L. B.) in a previously reported fatal case of acute hepatitis attributed to beclobrate in a 63-year-old woman.’
Beclobrate has been licensed in Switzerland since 1988 for the of hyperlipidaemia. Increased serum levels of liver enzymes and, less frequently, hepatitis have been reported with other lipid-lowering agents of the fibrate type2-5 but the two patients discussed here are the first with fatal acute liver failure during 35 000 patient-years of treatment with beclobrate. Since there are no unequivocal histological features of drug-related hepatic damageit is not possible to prove that the hepatitis was due to beclobrate. However, the histological features, the unpredictable and rare nature of the reaction in the absence of any evidence of a dose-relationship, hypersensitivity, or similar response in animal models suggests that this is an idiosyncratic drug reaction.7 Despite the wide use of hypolipidaemic drugs, little is known about their treatment
hepatotoxicity. Beclobrate has now been withdrawn from the market since the risk of hepatic damage is believed to outweigh the potential benefits of the drug. CHU Vaudois, 1011 Lausanne, Switzerland
D. VOUILLAMOZ M.-D. SCHALLER
Institute of
Pathology, Hôpital Universitaire, Basle
L. BIANCHI
University Institute of Pathology, Lausanne
P. CHAUBERT
Department of Internal Medicine, Inselspital, Berne
W. REINHART
CHU Vaudois, Lausanne
J. THORENS
D. ARMSTRONG A. L. BLUM
1. Aebi U, Reinhart WH. Beclobrate and liver disease. Ann Intern Med 1990; 114: 483. 2. Illingworth DR. Lipid-lowering drugs: an overview of indications and optimum therapeutic use. Drugs 1987; 33: 259-79. 3. Gendreau-Tranquart C, Barbieux JP, Gillion JM, et al. Un cas d’hépatitie à l’exifone et au bézafibrate. Gastroenterol Clin Biol 1989; 13: 427. 4. Valmalle R, Bacq Y, Furet Y, Dorval E, Barbieux JP, Metman EH. Hépatite aigue mortelle lors d’un traitement par maleate de perhexiline et bézafibrate. Gastroenterol Clin Biol 1989; 13: 530. 5. Uglesic A, Ljachnicky N, Langzzeittherapie der typ-II-und der typ-IVhyperlipoproteinämie mit beclobrat. Schweiz Rundsch Med Prax 1988; 77: 755-61. 6. Bianchi L, De Groote J, Desmet V, et al. Guidelines for diagnosis of therapeutic drug-induced liver injury in liver biopsies. Lancet 1974; i: 854-57. 7. Kaplowitz N, Yee Aw T, Simon FR, Stolz A. UCLA conference: drug-induced hepatotoxicity. Ann Intern Med 1986; 104: 826-39.
Oral naloxone in
opioid-associated constipation
SIR,-In Dr Sykes’ letter (June 15, p 1475) it is not clear how the naloxone dose was calculated-was x% of the total 24 h opioid dose given every 4 h or was the 24 h naloxone dose x% of the 24 h opioid dose? We undertook a similar study to ascertain whether oral naloxone would reverse morphine-associated constipation in patients with cancer. The study was approved by the Canterbury Area Health Board ethics committee and patients gave written informed consent. Twelve patients completed the study. The daily dose of morphine ranged between 10 and 380 mg (median 120). The dose of naloxone was escalated from 0-4 mg up to 4 mg, with 4 patients treated at two dose levels and 4 at three or four dose levels. The maximum single dose of naloxone used was 0-5% of daily morphine dose in 3 patients, 0.5-1.0% in 2,1-2% in 5,4% in 1, and 10% in 1. None had a laxative effect at any of these dose levels; none experienced colic although one noticed increased bowel activity and none recorded any increase in pain either on linear analogue scale or as increased analgesia requirement. Doses in excess of 12 mg were not used because of a previous report of painful abdominal cramps and antagonism of analgesia.’ Sykes makes no mention of cost. In New Zealand, naloxone costs$13.15 for 0-4 mg or £11.48/mg. In view of the lack of efficacy at doses up to 4 mg, we have suspended further investigation since naloxone is currently not a cost-effective laxative
are cytotoxic effector cells against trophoblast in early pregnancy. Our data show that changes of NK activities are related
NK cells
outcome after immunotherapy. NK activity before might be important in decisions about immunotherapy for patients with unexplained recurrent spontaneous abortion. In those with low pretreatment NK, activity factors other than NK activity may contribute to abortion, or abortion may be due to different factors in these patients and those with high NK activity. Finally, we are concerned about the patients in whom NK activity was raised by immunotherapy but who had not conceived more than a year after immunotherapy, and believe that further investigation is necessary.
clinical
to
treatment
This work
was
supported by
a
Grant-in-Aid from the Ministry of
Education, Science, and Culture of Japan.
RYUJI MAKIDA Departments of Transfusion Medicine and Immunohaematology, and Obstetrics and Gynaecology, Faculty of Medicine, University of Tokyo, Tokyo 113, Japan
MUTSUHIKO MINAMI MASARU TAKAMIZAWA TAKEO JUJI TOMOYUKI FUJII MASAYUKI MIZUNO
Fujii T, Takamizawa M, Juji T, Kawana T, Mizuno M. Outcome of pregnancy after injection of mononuclear cells. Am J Obstet Gynecol 1988; 158: 1015-16. 2. King A, Loke YW. Uterine large granular lymphocytes: a possible role in embryonic implantation. Am J Obstet Gynecol 1990; 162: 308-10. 3. Bumig JW. Carboxy-fluorescein fluorochromasia assay: I non-radioactively labeled 1.
4. De
cell-mediated lympholysis. J Immunol Methods 1980; 33: 33-44. Fougerolles AR, Baines MG. Modulation of the natural killer cell activity in pregnant mice alters spontaneous abortion rate. J Reprod Immunol 1987; 11: 147-53.
Helix pomatia in breast
cancer
SIR,-Dr Brooks and Dr Leathern (July 13, p 71) report that staining of primary breast cancers for Helix pomatia (HPA) is an "excellent predictor of long-term patient prognosis" and that it is a "possible staging adjunct in patients apparently lymph-node negative". We have done a pilot study on 363 primary breast cancers taken at random from the files of the International (Ludwig) Breast Cancer Study Group Trial V conducted between 1981 and 1985. Details of this trial have been published elsewhere.1,2 All cases were stained immunocytochemically with the lectin and specificity was confirmed by absorption with the appropriate sugar. Scoring, as described by Brooks and Leathem, was done by two pathologists. All samples were coded and the results sent to the Biostatistics Center, Dana Farber Cancer Institute, where the clinical correlations were studied. A detailed analysis has been done but what follows is limited to the questions raised by Brooks and Leathem’s paper. There was a weak correlation between HPA binding and lymph node involvement. In contrast to the findings of Brooks and Leathern only 70% of node-positive cases were positively stained, making HPA binding a poor predictor of lymph node staging in our material:
Disease-free survival (DFS) was the same for HPA non-stainers for HPA stainers (6-year DFS 56% [SE 5] and 56% [3], respectively). We evaluated DFS and overall survival according to lymph node status and HPA binding. While lymph node status was significant, as expected, HPA binding had no predictive value: as
*p< 0 0001 for node+ vs node - (univariate analysis by log-rank test) tNo significant differences for HPA- vs HPA+ by umvanate analysis or by multivarite analyses based on Cox model with lymph node status (N -, 1-3N +, > 3N +), H PA, grade, and tumour size
37% of node-negative patients received no chemotherapy and 63 % received one cycle of perioperative chemotherapy. Among the
node-positive patients, 67% received six or seven cycles of chemotherapy and 33% received one cycle of perioperative chemotherapy. Thus, in contrast to Brooks and Leathem, we can find no evidence that HPA staining is of value as a prognostic marker. In our cases lymph node status was determined pathologically after total mastectomy with complete axillary clearance, and an average of 14 nodes
were
examined per
case.
It is difficult
to
draw direct
comparisons with the Brooks study since no details are provided on how lymph node status was assessed. Treatment may affect outcome and information on this is not provided by Brooks and T eathem Section of Cell Biology and Experimental Pathology, Institute of Cancer Research, Haddow Laboratories, Sutton, Surrey SM2 5NG, UK
International
Breast Cancer
(Ludwig) Study Group (IBCSG),
Boston, Massachusetts, USA and Bern, Switzerland
C. W. TAYLOR R. ANBAZHAGAN H. JAYATILAKE A. ADAMS B. A. GUSTERSON K. PRICE R. D. GELBER A.
GOLDHIRSCH,
behalf of pathologists and participants of the IBCSG
on
Ludwig Breast Cancer Study Group. Combination adjuvant chemotherapy for node-positive breast cancer: inadequacy of a single perioperative cycle. N Engl J Med 1988, 319: 677-83. 2. Ludwig Breast Cancer Study Group. Prolonged disease-free survival after one course of perioperative adjuvant chemotherapy for node negative breast cancer. N Engl J 1.
Med 1989; 320: 491-96.
***This letter has been shown to Dr Brooks and Dr Leathern, whose reply to it and a previous letter follows.-ED. L. SIR,-Mr Galea and colleagues (Aug 10, p 392) and Dr Taylor et al raise questions that seem to relate directly to our methodology. Our earlier observation1 that binding of Helix pomatia lectin (HPA) to primary breast cancers was associated with shorter survival was followed by three publications confirming our results with an indirect antibody, an avidin-biotin, and a fluorescein method.2-4 Of six different methods we tried on cancers with long versus short survival, we found that an indirect antibody method gave staining results that most closely reflected survival. We described earlier how various methods to detect lectin binding to tissue sections could give different results.5,6Last year we tried to develop a simple staining and scoring system to detect HPA binding to breast cancer sections. We chose a direct HPAperoxidase conjugate (Sigma) to stain our series and obtained clean staining that was straightforward to interpret as positive or negative. In parallel we used an indirect antibody method. Comparing results with survival showed a trong correlation with prognosis by the indirect method (p < 0-00001) but only a weak correlation by the direct method (p < 0-03), and we stopped using this direct method. We were surprised by this difference. The earlier HPA results of the Nottingham group in 1987,z with an indirect antibody method, showed a close association between HPA binding and survival (p < 0-001) but the group’s recent results (Aug 10) with a direct conjugate show no significant differences between stainers and non-stainers. This is in accord with the results of our direct conjugate staining method. We have discussed this with them and agree that "the most probable explanation must lie in the difference between the staining method and the source of HPA. Although, as Galea et al mention, many of our lectins and conjugates have been home-made; both the native and conjugated HPA described here have been obtained from Sigma. Our rabbit polyclonal antiserum reacts with all HPA sources we have tested. We are evaluating a monoclonal antibody to HPA that is directed against an epitope seemingly distinct from the binding site and produces good staining, but we do not yet know how the results compare with tumour behaviour. Taylor et al do not show that HPA has a predictive value for node status or prognosis, but the methodology is not explained. To compare results from different centres the basic methodology must be the same, and polyclonal rabbit antiserum is available from us and others to centres wishing to test this out. There are several explanations for the differences noted that relate to the binding of a lectin. First, the binding of lectins, just as that of antibodies and other
581
receptor molecules, is an equilibrium based on the spatial arrangement of receptors and ligands, such that the closer theyare, the tighter or more effective the binding (eg, the polyvalent IgM or the binding of Fc receptors). The differences we describe on tissue sections may be an artifact based on spatial arrangement of GaINAc; there is some evidence for this in the complex bands seen by PAGE/lectin blotting of both Helix-positive and Helix-negative tumours, which can be completely inhibited by 10 mmol/1 GaINAc. Second, if two lectins with the same nominal simple monosaccharide specificity are incubated on the same tissue, however, then quite different histological structures may be stained, although both lectins may be totally inhibited by the same simple sugar. Most lectins have more than one binding site and some are known to be bifunctional (eg, discoidin 1); GaINAc binding lectins such as DBA and SBA can also bind to non-carbohydrate ligands, such as adenine. Third, what lectin binding detects is how many sugars are exposed and how close they are. But binding is determined not only by the sugars but also by their 0 or N linkages to peptides, topology, adjacent structures, charge, hydrophobicity, and van de Waals forces. If the binding site is altered (as by conjugation) simple sugars may continue to inhibit binding to a section but the specificity to complex sugars is altered. If the hydrogen bond energy is donated by peroxidase, for example, then the binding tightness with the glycoconjugate at that point is weakened. Such derivation of lectins can alter the binding characteristics of lectins-for example, succinylation of WGA reduces the affinity for sialic acid although it does not affect binding to G1cNAc, and this property has been applied to tissue sections. Peroxidase is highly glycosylated and avidly reacts with the binding sites of concanavalin A which at the same time can react with tissue glycoconjugates. We do not know how peroxidase reacts with HPA. Concanavalin A conjugated to ’Sepharose’ shows a different spectrum of glycoprotein binding to the native lectin. But this is an unexplored field. With HPA-HRP the staining pattern is very clear and presumably reflects some specific structures, which can be blocked by Ga1NAc, but not the same structure as those detected by native HPA. The use of direct peroxidase-lectin seems to give different results from the indirect methods. Until the HPA binding glycoconjugates are characterised and the methodology is better understood, we suggest that the direct peroxidase-HPA method may be less useful in this system than an indirect method. Department of Histopathology, University College and Middlesex School of Medicine, Middlesex Hospital, London W1P 7PN, UK
SUSAN BROOKS ANTHONY LEATHEM
1 Leathem A, Brooks S. Predictive value of lectin binding on breast-cancer recurrence and survival. Lancet 1987; i 1054-56. 2. Fenlon S, Ellis IO, Bell J, Todd JH, Elston CW, Blamey RW. Helix pomatia and Ulex europeus lectin binding in human breast carcinoma. J Pathol 1987; 152: 169-76. 3. Alam SM, Whitford P, Cushley W, George WD, Campbell AM. Flow cytometric analysis of cell surface carbohydrates in metastatic human breast cancer. Br J Cancer 1990; 62: 238-42. 4. Fukutomi T, Itabashi M, Tsuane S, Yamamoto H, Nanasawa T, Hiroto T. Prognostic contributions of Helix pomatia and carcinoembryonic antigen using histochemical techniques in breast carcinomas. Jpn J Clin Oncol 1989; 19: 127-34. 5. Leathem A, Atkins N. Lectin binding to formalin-fixed paraffin sections. J Clin Pathol 1983; 36: 747-50. 6. Leathem A. Lectin histochemistry. In: Polak J, Van Noorden S, eds. 2nd ed. Bristol: Wnght, 1986: 167-87.
Beclobrate and fatal acute
hepatitis
SIR,-We report the death of a 64-year-old man from acute hepatic failure attributable to beclobrate, a lipid-lowering agent. The patient’s hypercholesterolaemia, first diagnosed at the age of 56, was treated by diet and then with a succession of hypolipidaemic agents. In addition, he received chondroitin sulphate 500 mg thrice daily for osteoarthrosis of the hip. There was no history of self-medication or alcohol abuse. 6½ months after he had started treatment with beclobrate 100 mg daily he became anorexic and jaundiced. His total bilirubin was 304 umol/1 (normal < 17), and transaminases were very much raised. He was admitted to hospital 3 weeks later with a diagnosis of acute hepatitis. He died 7 days later. Investigations revealed no evidence for an infectious (viral or toxoplasmosis) or obstructive cause.
At necropsy there was liver panlobular hepatic necrosis with bridging, consistent with an adverse drug reaction. The histological appearances were identical to those found by the same histopathologist (L. B.) in a previously reported fatal case of acute hepatitis attributed to beclobrate in a 63-year-old woman.’
Beclobrate has been licensed in Switzerland since 1988 for the of hyperlipidaemia. Increased serum levels of liver enzymes and, less frequently, hepatitis have been reported with other lipid-lowering agents of the fibrate type2-5 but the two patients discussed here are the first with fatal acute liver failure during 35 000 patient-years of treatment with beclobrate. Since there are no unequivocal histological features of drug-related hepatic damageit is not possible to prove that the hepatitis was due to beclobrate. However, the histological features, the unpredictable and rare nature of the reaction in the absence of any evidence of a dose-relationship, hypersensitivity, or similar response in animal models suggests that this is an idiosyncratic drug reaction.7 Despite the wide use of hypolipidaemic drugs, little is known about their treatment
hepatotoxicity. Beclobrate has now been withdrawn from the market since the risk of hepatic damage is believed to outweigh the potential benefits of the drug. CHU Vaudois, 1011 Lausanne, Switzerland
D. VOUILLAMOZ M.-D. SCHALLER
Institute of
Pathology, Hôpital Universitaire, Basle
L. BIANCHI
University Institute of Pathology, Lausanne
P. CHAUBERT
Department of Internal Medicine, Inselspital, Berne
W. REINHART
CHU Vaudois, Lausanne
J. THORENS
D. ARMSTRONG A. L. BLUM
1. Aebi U, Reinhart WH. Beclobrate and liver disease. Ann Intern Med 1990; 114: 483. 2. Illingworth DR. Lipid-lowering drugs: an overview of indications and optimum therapeutic use. Drugs 1987; 33: 259-79. 3. Gendreau-Tranquart C, Barbieux JP, Gillion JM, et al. Un cas d’hépatitie à l’exifone et au bézafibrate. Gastroenterol Clin Biol 1989; 13: 427. 4. Valmalle R, Bacq Y, Furet Y, Dorval E, Barbieux JP, Metman EH. Hépatite aigue mortelle lors d’un traitement par maleate de perhexiline et bézafibrate. Gastroenterol Clin Biol 1989; 13: 530. 5. Uglesic A, Ljachnicky N, Langzzeittherapie der typ-II-und der typ-IVhyperlipoproteinämie mit beclobrat. Schweiz Rundsch Med Prax 1988; 77: 755-61. 6. Bianchi L, De Groote J, Desmet V, et al. Guidelines for diagnosis of therapeutic drug-induced liver injury in liver biopsies. Lancet 1974; i: 854-57. 7. Kaplowitz N, Yee Aw T, Simon FR, Stolz A. UCLA conference: drug-induced hepatotoxicity. Ann Intern Med 1986; 104: 826-39.
Oral naloxone in
opioid-associated constipation
SIR,-In Dr Sykes’ letter (June 15, p 1475) it is not clear how the naloxone dose was calculated-was x% of the total 24 h opioid dose given every 4 h or was the 24 h naloxone dose x% of the 24 h opioid dose? We undertook a similar study to ascertain whether oral naloxone would reverse morphine-associated constipation in patients with cancer. The study was approved by the Canterbury Area Health Board ethics committee and patients gave written informed consent. Twelve patients completed the study. The daily dose of morphine ranged between 10 and 380 mg (median 120). The dose of naloxone was escalated from 0-4 mg up to 4 mg, with 4 patients treated at two dose levels and 4 at three or four dose levels. The maximum single dose of naloxone used was 0-5% of daily morphine dose in 3 patients, 0.5-1.0% in 2,1-2% in 5,4% in 1, and 10% in 1. None had a laxative effect at any of these dose levels; none experienced colic although one noticed increased bowel activity and none recorded any increase in pain either on linear analogue scale or as increased analgesia requirement. Doses in excess of 12 mg were not used because of a previous report of painful abdominal cramps and antagonism of analgesia.’ Sykes makes no mention of cost. In New Zealand, naloxone costs$13.15 for 0-4 mg or £11.48/mg. In view of the lack of efficacy at doses up to 4 mg, we have suspended further investigation since naloxone is currently not a cost-effective laxative
