1504
other causes in these studies and we are dealing with widely used and tested drugs. Our original intention was to select all the studies bearing on the question "Do beta-blockers influence variceal haemorrhage and mortality in patients with portal hypertension and varices?" The breadth of our selection criteria ran the risk of dilution of any treatment effects; we identified them despite any such dilutions. The fact that the effects increased when we restricted ourselves to the methodologically superior studies shows that we were not simply detecting biases from poorer experimental technique. We claim, not that our conclusions apply uniformly across all patients, but that the direction of effects will be right and that any overestimates in some populations will be balanced by underestimates in others. Our stricter criteria for selection did indeed exclude some large trials with interesting results but selection criteria must be chosen independently of the results of trials, and this is what we did. The inclusion of a placebo may be debatable in this area of medicine, but many triallists have decided that it is necessary, and we wanted to be on the side of the purists. The weight given to small trials was the appropriate statistical one related to the number of end-points accrued. We have no evidence of publication bias arising from the inclusion of small studies. No-one has referred us to additional unpublished small studies, and the collected knowledge of our joint contacts probably covers almost all the clinical trials that have been done. We regret that Pagliaro et al had difficulties finding the data from our references. Much of the detail is in an earlier paper’ which we cited. Studies where deaths from bleeding were not given in the published reports were omitted from the analysis of this end-point. Uribe et al reported no deaths in their small study, and so we assumed there were none. Cerbelaud et al reported deaths from bleeding but not total mortality, and we used these deaths in our analysis. Omission of these two studies does not affect our conclusions. Our reference to Mills, Garden, et al should have been the Garden, Mills, et al reference cited by two of the correspondents (the data in this paper are slightly updated). Transposition of the Lebrec references has been corrected (Aug 4, p 324). The proposal of Poynard et al that sensitivity be increased by collecting and analysing individual survival times and incorporating prognostic factors is a good one. It also allows clearer estimation of the effects of treatment on survival through the use of survival curves.2 However, it is much more difficult to collect comprehensive data. We have some criticism of the meta-analysis of Pagliaro et al: in the study of Ideo et al they included amongst their controls a group of patients treated with ranitidine; the study of Columbo et al was included twice, comparing the same control group with different beta-blockers; and in the study of Cerbelaud et al they included the 16 patients who never started treatment. These are not major issues because of the magnitude of the treatment effects. On the other hand the ranges for the number of patients who need to be treated to prevent one episode of bleeding look too wide. They seem to have been calculated from the range of study results rather than from the confidence intervals on pooled relative risk. The latter strategy suggests that the number ranges from about 5 to 11 treated patients to prevent 1 first bleeding and 2 to 3 treated patients to prevent 1 rebleeding. There are few areas of medicine where such response rates can be claimed. Department of Medicine, Royal Infirmary, Edinburgh
PETER C. HAYES
Medical Affairs Department, ICI Pharmaceuticals
JILL M. DAVIS
Free
haemoglobin and pre-eclampsia
SIR,-Microangiopathic haemolysis is a late feature of preeclampsia. Therefore it is unlikely, as Sarrel and colleagues suggest (Oct 27, p 1030), that increased amounts of free haemoglobin might have primary responsibility for raised peripheral vascular resistance in pre-eclampsia. In contrast to Sarrel and colleagues’ experience in their coronary artery study, Vallance et aF found no inhibitory effect of 10mol/1 haemoglobin on the vasodilator response to acetylcholine or bradykinin in the forearmThis finding is with in-vitro observations that whole human erythrocytes inhibit endothelium-derived relaxing factor (EDRF) as effectively as does free human haemoglobin.3 Thus it seems that the red cell membrane does not impede access of EDRF to intracellular haemoglobin and that erythrocytic (luminal) inhibition of EDRF will be greatest in patients with a haemoglobin over 4 g/dlSarrel et al do not say whether the response to a second acetylcholine exposure (ie, the appropriate control for that experiment) was or was not attenuated. Neither do they state whether the coronary arteries infused were normal (ie, intact endothelium); infusion of acetylcholine into angiographically normal coronary arteries of patients with ischaemic heart disease has been reported to cause vasoconstriction.4 Not all women who have hypertension in pregnancy will go on to manifest the other signs that characterise pre-eclampsia. Loss of endothelial integrity is not a feature of mild disease. With a morphologically intact endothelium, free haemoglobin can only inhibit luminally released, locally acting EDRF. The more substantial abluminal component thus remains protected.3 .3 evidence for reduced or of EDRF in release Experimental activity pre-eclampsia is scarce. However, Aalkjaer et al’ have shown lower rate of relaxation in angiotensin II precontracted isolated omental resistance arteries from women with pre-eclampsia than from pregnant normotensive controls. This in-vitro work was done with saline buffer (ie, in the absence of free haemoglobin, which of course causes reversible inhibition of EDRF). There have been many reports describing the association and correlation between pregnancy-induced hypertension and circulating inhibitors of the sodium pump.6 We have demonstrated that low concentrations of the cardiac glycoside ouabain irreversibly inhibit the N"-monomethyl-L-arginine-sensitive component of endothelium-dependent relaxation in human subcutaneous resistance arteries.’ Perhaps by this mechanism release or action of EDRF is impaired in pregnancy-induced hypertension. It is most unlikely that increased free haemoglobin has an important role. consistent
St Philip’s Hospital, London WC2A 2EX, UK
R. G. WOOLFSON D. J. WILLIAMS
1 Vardi J, Fields GA. Microangiopathic hemolytic anemia m severe pre-eclampsia. Am J Obs Gynecol 1974; 119: 617-22. 2. Vallance P, Benjamin N, Collier J EPO, haemoglobin, and hypertensive crises Lancet 1988; i: 1107 3. Evans HG, Ryley HC, Hallett I, Lewis MJ. Human red blood cells inhibit endothelium-denved relaxing factor (EDRF) activity. Eur J Pharmacol 1989; 163: 361-64 4 Werns SW, Walton JA, Hsia HH, Nabel EG, Sanz ML, Pitt B. Evidence of endothelial dysfunction in angiographically normal coronary arteries of patients with coronary artery disease. Circulation 1989; 79: 287-91. 5. Aalkjaer C, Danielsen H, Johannesen P, Pedersen EB, Rasmussen A, Mulvany MJ Abnormal vascular function and morphology in pre-eclampsia: a study of isolated resistance vessels. Clin Sci 1985; 69: 477-82. 6. Poston L. Sodium transport inhibitors in pregnancy-induced hypertension Cardiovasc Drugs Ther 1990; 4: 351-56 7 Woolfson RG, Poston L. Ouabain inhibits acetylcholine-induced relaxation through the EDRF pathway in human resistance vessels. J Hypertens 1990; 8 (suppl 3): 102A.
Medical Affairs Department, ICI Pharmaceuticals,
Alderley Park, Macclesfield SK10 4TG, UK
JOHN A. LEWIS
Department of Medicine, Royal Infirmary, Edinburgh
IAN A. D. BOUCHIER
Chiropractors and low back pain SiR,—Your editorial, based on the findings of Meade et al,’ on the
relationship between the medical profession and chiropractors (July 28, p 220) suggested that doctors cooperate with chiropractors and
D, Cameron HA Beta-blockers m portal hypertension, an overview. Drugs 1989, 37 (suppl 2). 62-69 Early Breast Cancer Trialists’ Collaborative Group. Effects of adjuvant tamoxifen and of cytotoxic therapy on mortality in early breast cancer. N Engl J Med 1988; 319:
1. Lewis JA, Davis JM, Allsopp 2
1681-91.
and adopt their manipulative techniques. However sensible these conclusions might seem, the acceptance of chiropractic techniques as the therapy for back diseases must be
physiotherapists, disputed.
1505
In patients with acute low back pain (LBP), Meade et al showed superiority for chiropractic intervention, a finding consistent with spontaneous regression of acute LBP in more than 80% of patients.2 Patients with acute LBP are frequently offered year-long no
prevention by chiropractors, a practice not justified by work or based on experience with conventional treatment. On the contrary, these manipulations may turn the "person into a patient". We believe that strengthening the back is the correct means of preventing recurrent LBP, and prefer active to passive therapy.3 In patients with chronic LBP, Meade et al observed an effect with prolonged therapy, in accordance with the present interpretation of this disease as a mixture of physical and psychological problems.’ However, chiropractic intervention lasted longer than control, and failure to equalise the number and duration recurrence
published
of treatments in treatment and control groups is a drawback to studies in this field.’ We interpret the study by Meade et al as showing that chiropractic manipulation may be superior to the mixture of treatments that the medical profession is in the process of doing away with. It is increasingly evident that the answer to chronic LBP is usually active training. The emphasis of modem physical therapy is on exercise with behavioural support.6-9 The background to this development is knowledge of the complicated pathogenesis of LBP, with pain-coping problems involved in most cases. We now have means of treating the spine with cautious techniques in order to avoid the complications of chiropractic manipulations, which may, albeit rarely, lead to severe disablement.9 Thus, we do not agree that chiropractic manipulation be given higher priority. The trend toward a wide perspective on the issues of therapy of LBP should continue.
the absence of binding to zwitterion phospholipids. This has been shown by the Ouchterlony procedure,z inhibition studies,3.4 and ELISA tests.4,s The preferential binding of aPL to negatively charged phospholipids rather than to zwitterion phospholipids or to other antigens containing phosphodiester groups (eg, DNA 3’) has persuaded most investigators that anionic phospholipids are the true antigen for aPL antibodies. These antigens may be presented alone or possibly as a protein-phospholipid complex (c). Division of
Department of Rheumatology, State University Hospital, Copenhagen 2100 Ø, Denmark
HENNING BLIDDAL TOM BENDIX
1. Meade TW, Dyer S, Browne W, Townsend J, Frank AO. Low back pain of mechanical origin, randomised comparison of chiropractic and hospital outpatient treatment. Br Med 1990; 300: 1431-37. J 2. Quebec Task Force on Spinal Disorders Scientific approach to the assessment of activity-related spinal disorders Spine 1987; 12 (suppl 1). 3. Biering-Sørensen F. Physical measurements as risk indicators for low-back trouble over a one-year period Spine 1984; 9: 106-19. 4. Waddell G. A new clinical model for the treatment for low-back pain Spine 1987; 12: 632-44. 5 Coxhead CE, Inskip H, Meade TW, North WRS, Troup JDG Multicentre trial of physiotherapy in the management of sciatic symptoms Lancet 1981; i: 1065-68. 6 Manniche C, Hesselsøe G, Bentzen L, Christensen I, Lundberg E. Clinical trial of intensive muscle training for chronic low back pain. Lancet 1988; ii: 1473-76 7 Mayer TG, Gatchel RJ, et al. Objective assessment of spine function following industrial injury. Spine 1985; 10: 482-93. 8 Hazard RG, Fenwick JW, et al Functional restoration with behavioural support. Spine 1989, 14: 157-61. 9. Povlsen UJ, Kjaer L, Arlien-Søborg P Locked-in syndrome following cervical manipulation Acta Neurol Scand 1987, 76: 486-88.
What is the "true"
antigen for
antiphospholipid antibodies? SIR,-We wish to clarify the issues in the debate about the "true" antigen of antiphospholipid antibodies (aPL) (Aug 25, p 505; Oct 13, p 952). The three possible antigens, expressed
diagrammatically, are: Dr Galli and co-workers (June 30, p 1544) state that aPL bind &bgr;2-glycoprotein I only (a). McNeil et all claim that the antibodies bind anionic phospholipids (such as cardiolipin) only when they are linked to &bgr;z-glycoprotein I (c). We have stated, on the basis of our own and other data, that aPL bind anionic phospholipids (b) with or without &bgr;2-glycoprotein 1. It may be that &bgr;2-glycoprotein I alters the conformation of anionic phospholipids (c) so that they are more avidly bound by aPL, but we find little experimental basis for Galli and colleagues’ view that the protein alone is the antigen. Bevers and Galli (Oct 13 letter) now seem to argue more in favour of structure (c) being the antigen. Dr Matsuura and colleagues (July 21, p 177) also seem to suggest that autoimmune aPL antibodies, in contrast to syphilis antibodies, bind the phospholipid-glycoprotein complex. Bevers and Galli state correctly that any experiment which shows that aPL bind negatively charged phospholipids should also show
Rheumatology, Department of Medicine, University of Louisville, Louisville, Kentucky 40292,
NIGEL E. HARRIS SILVIA PIERANGELI
USA
1. McNeil HP, Simpson RJ, Chesterman CN, Krillis SA. Antiphospholipid antibodies are directed against a complex antigen that includes a lipid-binding inhibitor of coagulation &bgr;2 glycoprotein I(apo H). Proc Natl Acad Sci (USA)1990; 87: 4120-24. 2. Thiagarajan P, Shapiro SS, DeMarco L A monoclonal immunoglobulin M coagulation inhibitor with phospholipid specificity mechanism of a lupus anticoagulant. J Clin Invest 1980; 66: 397-405. 3. Harris EN, Gharavi AE, Loizou S, Derue G, Chan JK. Crossreactivity of anti-phospholipid antibodies. J Clin Lab Invest 1985; 66: 397-405. 4 Pengo V, Thiagarajan P, Shapiro SS, Haine MJ. Immunological specificity and mechanism of action of IgG lupus anticoagulants. Blood 1987; 70: 69-76. 5. Harris EN, Gharavi AE, Tincani A, et al. Affinity purified anticardiolipin antibodies and anti-DNA antibodies. J Clin Lab Immunol 1985, 17: 155-62.
Cholecystectomy in Saudi Arabia SiR,—Iread with dismay the first sentence of Professor Tamimi and colleagues’ summary of their report (Nov 17, p 1235), which stated: "Gallstones have become increasingly prevalent in Saudi Arabia ...". Could I possibly have missed the epidemiological data despite close attention to published work? The answer is no, because there is no such published information and nor do Tamimi and colleagues’ article provide any. The increased surgical activity they report probably reflects investment of oil revenues in health care, so that diagnostic ultrasonography and operations are available at a much higher level than before. It would indeed be very interesting, and compatible with evidence from the UK, Japan, and Norway, if gallstone prevalence had increased in Saudi Arabia lately, but this would depend on proper population-based surveys or necropsy reviews. Cholecystectomy rates might give quite false impressions about changes in gallstone prevalence. Rates have fallen strikingly in Sweden and Scotland at a time when gallstone disease is known to remain very common. General
Hospital, Bishop Auckland, County Durham DL14 6AD, UK 1 Bateson MC Gallstone variable. Lancet 1984;
prevalence ii:
521-25
M. C. BATESON and
cholecystectomy
rate are
independently
other causes in these studies and we are dealing with widely used and tested drugs. Our original intention was to select all the studies bearing on the question "Do beta-blockers influence variceal haemorrhage and mortality in patients with portal hypertension and varices?" The breadth of our selection criteria ran the risk of dilution of any treatment effects; we identified them despite any such dilutions. The fact that the effects increased when we restricted ourselves to the methodologically superior studies shows that we were not simply detecting biases from poorer experimental technique. We claim, not that our conclusions apply uniformly across all patients, but that the direction of effects will be right and that any overestimates in some populations will be balanced by underestimates in others. Our stricter criteria for selection did indeed exclude some large trials with interesting results but selection criteria must be chosen independently of the results of trials, and this is what we did. The inclusion of a placebo may be debatable in this area of medicine, but many triallists have decided that it is necessary, and we wanted to be on the side of the purists. The weight given to small trials was the appropriate statistical one related to the number of end-points accrued. We have no evidence of publication bias arising from the inclusion of small studies. No-one has referred us to additional unpublished small studies, and the collected knowledge of our joint contacts probably covers almost all the clinical trials that have been done. We regret that Pagliaro et al had difficulties finding the data from our references. Much of the detail is in an earlier paper’ which we cited. Studies where deaths from bleeding were not given in the published reports were omitted from the analysis of this end-point. Uribe et al reported no deaths in their small study, and so we assumed there were none. Cerbelaud et al reported deaths from bleeding but not total mortality, and we used these deaths in our analysis. Omission of these two studies does not affect our conclusions. Our reference to Mills, Garden, et al should have been the Garden, Mills, et al reference cited by two of the correspondents (the data in this paper are slightly updated). Transposition of the Lebrec references has been corrected (Aug 4, p 324). The proposal of Poynard et al that sensitivity be increased by collecting and analysing individual survival times and incorporating prognostic factors is a good one. It also allows clearer estimation of the effects of treatment on survival through the use of survival curves.2 However, it is much more difficult to collect comprehensive data. We have some criticism of the meta-analysis of Pagliaro et al: in the study of Ideo et al they included amongst their controls a group of patients treated with ranitidine; the study of Columbo et al was included twice, comparing the same control group with different beta-blockers; and in the study of Cerbelaud et al they included the 16 patients who never started treatment. These are not major issues because of the magnitude of the treatment effects. On the other hand the ranges for the number of patients who need to be treated to prevent one episode of bleeding look too wide. They seem to have been calculated from the range of study results rather than from the confidence intervals on pooled relative risk. The latter strategy suggests that the number ranges from about 5 to 11 treated patients to prevent 1 first bleeding and 2 to 3 treated patients to prevent 1 rebleeding. There are few areas of medicine where such response rates can be claimed. Department of Medicine, Royal Infirmary, Edinburgh
PETER C. HAYES
Medical Affairs Department, ICI Pharmaceuticals
JILL M. DAVIS
Free
haemoglobin and pre-eclampsia
SIR,-Microangiopathic haemolysis is a late feature of preeclampsia. Therefore it is unlikely, as Sarrel and colleagues suggest (Oct 27, p 1030), that increased amounts of free haemoglobin might have primary responsibility for raised peripheral vascular resistance in pre-eclampsia. In contrast to Sarrel and colleagues’ experience in their coronary artery study, Vallance et aF found no inhibitory effect of 10mol/1 haemoglobin on the vasodilator response to acetylcholine or bradykinin in the forearmThis finding is with in-vitro observations that whole human erythrocytes inhibit endothelium-derived relaxing factor (EDRF) as effectively as does free human haemoglobin.3 Thus it seems that the red cell membrane does not impede access of EDRF to intracellular haemoglobin and that erythrocytic (luminal) inhibition of EDRF will be greatest in patients with a haemoglobin over 4 g/dlSarrel et al do not say whether the response to a second acetylcholine exposure (ie, the appropriate control for that experiment) was or was not attenuated. Neither do they state whether the coronary arteries infused were normal (ie, intact endothelium); infusion of acetylcholine into angiographically normal coronary arteries of patients with ischaemic heart disease has been reported to cause vasoconstriction.4 Not all women who have hypertension in pregnancy will go on to manifest the other signs that characterise pre-eclampsia. Loss of endothelial integrity is not a feature of mild disease. With a morphologically intact endothelium, free haemoglobin can only inhibit luminally released, locally acting EDRF. The more substantial abluminal component thus remains protected.3 .3 evidence for reduced or of EDRF in release Experimental activity pre-eclampsia is scarce. However, Aalkjaer et al’ have shown lower rate of relaxation in angiotensin II precontracted isolated omental resistance arteries from women with pre-eclampsia than from pregnant normotensive controls. This in-vitro work was done with saline buffer (ie, in the absence of free haemoglobin, which of course causes reversible inhibition of EDRF). There have been many reports describing the association and correlation between pregnancy-induced hypertension and circulating inhibitors of the sodium pump.6 We have demonstrated that low concentrations of the cardiac glycoside ouabain irreversibly inhibit the N"-monomethyl-L-arginine-sensitive component of endothelium-dependent relaxation in human subcutaneous resistance arteries.’ Perhaps by this mechanism release or action of EDRF is impaired in pregnancy-induced hypertension. It is most unlikely that increased free haemoglobin has an important role. consistent
St Philip’s Hospital, London WC2A 2EX, UK
R. G. WOOLFSON D. J. WILLIAMS
1 Vardi J, Fields GA. Microangiopathic hemolytic anemia m severe pre-eclampsia. Am J Obs Gynecol 1974; 119: 617-22. 2. Vallance P, Benjamin N, Collier J EPO, haemoglobin, and hypertensive crises Lancet 1988; i: 1107 3. Evans HG, Ryley HC, Hallett I, Lewis MJ. Human red blood cells inhibit endothelium-denved relaxing factor (EDRF) activity. Eur J Pharmacol 1989; 163: 361-64 4 Werns SW, Walton JA, Hsia HH, Nabel EG, Sanz ML, Pitt B. Evidence of endothelial dysfunction in angiographically normal coronary arteries of patients with coronary artery disease. Circulation 1989; 79: 287-91. 5. Aalkjaer C, Danielsen H, Johannesen P, Pedersen EB, Rasmussen A, Mulvany MJ Abnormal vascular function and morphology in pre-eclampsia: a study of isolated resistance vessels. Clin Sci 1985; 69: 477-82. 6. Poston L. Sodium transport inhibitors in pregnancy-induced hypertension Cardiovasc Drugs Ther 1990; 4: 351-56 7 Woolfson RG, Poston L. Ouabain inhibits acetylcholine-induced relaxation through the EDRF pathway in human resistance vessels. J Hypertens 1990; 8 (suppl 3): 102A.
Medical Affairs Department, ICI Pharmaceuticals,
Alderley Park, Macclesfield SK10 4TG, UK
JOHN A. LEWIS
Department of Medicine, Royal Infirmary, Edinburgh
IAN A. D. BOUCHIER
Chiropractors and low back pain SiR,—Your editorial, based on the findings of Meade et al,’ on the
relationship between the medical profession and chiropractors (July 28, p 220) suggested that doctors cooperate with chiropractors and
D, Cameron HA Beta-blockers m portal hypertension, an overview. Drugs 1989, 37 (suppl 2). 62-69 Early Breast Cancer Trialists’ Collaborative Group. Effects of adjuvant tamoxifen and of cytotoxic therapy on mortality in early breast cancer. N Engl J Med 1988; 319:
1. Lewis JA, Davis JM, Allsopp 2
1681-91.
and adopt their manipulative techniques. However sensible these conclusions might seem, the acceptance of chiropractic techniques as the therapy for back diseases must be
physiotherapists, disputed.
1505
In patients with acute low back pain (LBP), Meade et al showed superiority for chiropractic intervention, a finding consistent with spontaneous regression of acute LBP in more than 80% of patients.2 Patients with acute LBP are frequently offered year-long no
prevention by chiropractors, a practice not justified by work or based on experience with conventional treatment. On the contrary, these manipulations may turn the "person into a patient". We believe that strengthening the back is the correct means of preventing recurrent LBP, and prefer active to passive therapy.3 In patients with chronic LBP, Meade et al observed an effect with prolonged therapy, in accordance with the present interpretation of this disease as a mixture of physical and psychological problems.’ However, chiropractic intervention lasted longer than control, and failure to equalise the number and duration recurrence
published
of treatments in treatment and control groups is a drawback to studies in this field.’ We interpret the study by Meade et al as showing that chiropractic manipulation may be superior to the mixture of treatments that the medical profession is in the process of doing away with. It is increasingly evident that the answer to chronic LBP is usually active training. The emphasis of modem physical therapy is on exercise with behavioural support.6-9 The background to this development is knowledge of the complicated pathogenesis of LBP, with pain-coping problems involved in most cases. We now have means of treating the spine with cautious techniques in order to avoid the complications of chiropractic manipulations, which may, albeit rarely, lead to severe disablement.9 Thus, we do not agree that chiropractic manipulation be given higher priority. The trend toward a wide perspective on the issues of therapy of LBP should continue.
the absence of binding to zwitterion phospholipids. This has been shown by the Ouchterlony procedure,z inhibition studies,3.4 and ELISA tests.4,s The preferential binding of aPL to negatively charged phospholipids rather than to zwitterion phospholipids or to other antigens containing phosphodiester groups (eg, DNA 3’) has persuaded most investigators that anionic phospholipids are the true antigen for aPL antibodies. These antigens may be presented alone or possibly as a protein-phospholipid complex (c). Division of
Department of Rheumatology, State University Hospital, Copenhagen 2100 Ø, Denmark
HENNING BLIDDAL TOM BENDIX
1. Meade TW, Dyer S, Browne W, Townsend J, Frank AO. Low back pain of mechanical origin, randomised comparison of chiropractic and hospital outpatient treatment. Br Med 1990; 300: 1431-37. J 2. Quebec Task Force on Spinal Disorders Scientific approach to the assessment of activity-related spinal disorders Spine 1987; 12 (suppl 1). 3. Biering-Sørensen F. Physical measurements as risk indicators for low-back trouble over a one-year period Spine 1984; 9: 106-19. 4. Waddell G. A new clinical model for the treatment for low-back pain Spine 1987; 12: 632-44. 5 Coxhead CE, Inskip H, Meade TW, North WRS, Troup JDG Multicentre trial of physiotherapy in the management of sciatic symptoms Lancet 1981; i: 1065-68. 6 Manniche C, Hesselsøe G, Bentzen L, Christensen I, Lundberg E. Clinical trial of intensive muscle training for chronic low back pain. Lancet 1988; ii: 1473-76 7 Mayer TG, Gatchel RJ, et al. Objective assessment of spine function following industrial injury. Spine 1985; 10: 482-93. 8 Hazard RG, Fenwick JW, et al Functional restoration with behavioural support. Spine 1989, 14: 157-61. 9. Povlsen UJ, Kjaer L, Arlien-Søborg P Locked-in syndrome following cervical manipulation Acta Neurol Scand 1987, 76: 486-88.
What is the "true"
antigen for
antiphospholipid antibodies? SIR,-We wish to clarify the issues in the debate about the "true" antigen of antiphospholipid antibodies (aPL) (Aug 25, p 505; Oct 13, p 952). The three possible antigens, expressed
diagrammatically, are: Dr Galli and co-workers (June 30, p 1544) state that aPL bind &bgr;2-glycoprotein I only (a). McNeil et all claim that the antibodies bind anionic phospholipids (such as cardiolipin) only when they are linked to &bgr;z-glycoprotein I (c). We have stated, on the basis of our own and other data, that aPL bind anionic phospholipids (b) with or without &bgr;2-glycoprotein 1. It may be that &bgr;2-glycoprotein I alters the conformation of anionic phospholipids (c) so that they are more avidly bound by aPL, but we find little experimental basis for Galli and colleagues’ view that the protein alone is the antigen. Bevers and Galli (Oct 13 letter) now seem to argue more in favour of structure (c) being the antigen. Dr Matsuura and colleagues (July 21, p 177) also seem to suggest that autoimmune aPL antibodies, in contrast to syphilis antibodies, bind the phospholipid-glycoprotein complex. Bevers and Galli state correctly that any experiment which shows that aPL bind negatively charged phospholipids should also show
Rheumatology, Department of Medicine, University of Louisville, Louisville, Kentucky 40292,
NIGEL E. HARRIS SILVIA PIERANGELI
USA
1. McNeil HP, Simpson RJ, Chesterman CN, Krillis SA. Antiphospholipid antibodies are directed against a complex antigen that includes a lipid-binding inhibitor of coagulation &bgr;2 glycoprotein I(apo H). Proc Natl Acad Sci (USA)1990; 87: 4120-24. 2. Thiagarajan P, Shapiro SS, DeMarco L A monoclonal immunoglobulin M coagulation inhibitor with phospholipid specificity mechanism of a lupus anticoagulant. J Clin Invest 1980; 66: 397-405. 3. Harris EN, Gharavi AE, Loizou S, Derue G, Chan JK. Crossreactivity of anti-phospholipid antibodies. J Clin Lab Invest 1985; 66: 397-405. 4 Pengo V, Thiagarajan P, Shapiro SS, Haine MJ. Immunological specificity and mechanism of action of IgG lupus anticoagulants. Blood 1987; 70: 69-76. 5. Harris EN, Gharavi AE, Tincani A, et al. Affinity purified anticardiolipin antibodies and anti-DNA antibodies. J Clin Lab Immunol 1985, 17: 155-62.
Cholecystectomy in Saudi Arabia SiR,—Iread with dismay the first sentence of Professor Tamimi and colleagues’ summary of their report (Nov 17, p 1235), which stated: "Gallstones have become increasingly prevalent in Saudi Arabia ...". Could I possibly have missed the epidemiological data despite close attention to published work? The answer is no, because there is no such published information and nor do Tamimi and colleagues’ article provide any. The increased surgical activity they report probably reflects investment of oil revenues in health care, so that diagnostic ultrasonography and operations are available at a much higher level than before. It would indeed be very interesting, and compatible with evidence from the UK, Japan, and Norway, if gallstone prevalence had increased in Saudi Arabia lately, but this would depend on proper population-based surveys or necropsy reviews. Cholecystectomy rates might give quite false impressions about changes in gallstone prevalence. Rates have fallen strikingly in Sweden and Scotland at a time when gallstone disease is known to remain very common. General
Hospital, Bishop Auckland, County Durham DL14 6AD, UK 1 Bateson MC Gallstone variable. Lancet 1984;
prevalence ii:
521-25
M. C. BATESON and
cholecystectomy
rate are
independently
