1001

LETTERS to the EDITOR Beta-blockers for preventing variceal

bleeding SIR,-We would take issue with some aspects of the meta-analysis by Dr Hayes and colleagues (July 21, p 153). The interpretation of a meta-analysis of clinical trials is intrinsically linked with the way trials were chosen for evaluation.1 The significant (or borderline significant) heterogeneity in the therapeutic effect of beta-blockers on bleeding and mortality in some subsets of the trials evaluated (table I in their paper) may be accounted for by their use of a "mixing apples and oranges" design.They pooled randomised and some non-randomised trials; trials in cirrhotic and in non-cirrhotic portal hypertension, in patients with and without varices; and trials of propranolol alone versus no active treatment, and of propranolol combined with sclerotherapy versus sclerotherapy alone. Their criteria for selecting eleven trials for a more precise evaluation were questionable. They excluded studies which did not use a true placebo, including two of the six largest trials-the Italian Multicentre Project (174 patients) which showed more deaths in the propranolol group and the trial of Queuniet et al (99 patients) which showed no difference in rebleeding or mortality. However, they included trials of non-cirrhotic hypertension and small trials showing therapeutic benefit (Lopez Fuerte, 19 patients; Gatta, 24 patients). The heterogeneity test did not achieve significance (p=0-073) for the effect on bleeding. However, this test has notoriously low power and no statistically significant difference does not mean that important sources of heterogeneity do not exist. Moreover in the context of these beta-blocker trials, in which only that by Bosch et al is double-blind, the absence of a true placebo does not invalidate the evaluation of outcomes (death and bleeding) since these are sufficiently hard end-points.

Furthermore given the well-known publication bias for small trials showing therapeutic benefit of a new treatment over small trials that show no therapeutic advantage (which thus may remain unknown) the weight given to these trials by their inclusion in the "selected group" is surprising. Hayes and colleagues’ paper neither reports the raw data from preliminary studies nor gives the source of several unpublished data (eg, death from bleeding), which is missing from most of the abstracts they cite, and overall mortality, not reported by Cerbelaud et al and Uribe et al. Readers cannot repeat the analysis for themselves. We wanted to do so to compare the interpretations by Hayes et al with our more conservative ones, published as part of an extensive meta-analysis of all treatments for the prevention of variceal bleeding.2 We found that beta-blockers significantly reduced the risk of bleeding and rebleeding but without significant decrease in mortality. No heterogeneity was detected. We only included randomised trials, in which there was no active treatment in controls and in which the whole population was cirrhotic. We separated primary from secondary prevention studies since they target different populations with a very different baseline average bleeding risk (about 30% for first bleeding and 70% for

rebleeding). The accompanying table shows an updated meta-analysis including the latest published trials (not analysed by Hayes et al and those analysed by them only as interim reports3-6). This analysis confirms our previous results. The number of patients that need to be treated to prevent 1 episode of first bleeding or rebleeding ranges between 4 and 20 and 2 and 20, respectively.7 Thus, many patients undergo treatment without benefit. This suggests the need for studies which are sufficiently large and/or have particular inclusion criteria (perhaps good liver function or an early

UPDATE OF META-ANALYSIS OF RANDOMISED TRIALS OF BETA-BLOCKER IN VARICEAL HAEMORRHAGE I

RR=relative risk

r

I

I

(Peto/Mantel-Haenszel) plus 95% confidence interval, NTT-number HET=heterogeneity; NR =not reported; NE- not evaluable *See references in paper by Hayes et al

to treat

for preventing

one

bleeding episode,7 PRR=pooled

relative risk;

1002

of splanchnic haemodynamic response) to detect responders to treatment. This should allow better selection of cirrhotic patients with varices for beta-blocker therapy in the future.

assessment

Medical Clinic R, University of Palermo,

Palermo, Italy

L. PAGLIARO

Hepatobiliary and Liver Transplantation Unit, Royal Free Hospital and School of Medicine, London NW3 2QG, UK

A. K. BURROUGHS

Department of Medicine, Hvidovre University Hospital, Copenhagen, Denmark

T. I. A. SORENSEN

Liver

J. C. PETITHORY Centre Hospitalier, 95500 Gonesse, France, and Centre Hospitalier,

B. PAUGAM P. BUYET-ROUSSET A. PAUGAM

Aulnay-sous-Bois

Nagayo T. Histogenesis and precursors of human gastric cancer. Berlin: SpringerVerlag, 1986 2. Asaishi K, Nishino C, Hayasaka H. Geographical distribution and epidemiology. In. Ishikura H, Namiki M, eds. Gastric anisakiasis in Japan. Berlin: Springer-Verlag, 1.

Physiology Research Unit,

1989: 31-36.

INSERM U-24,

Hirayama T. A study of epidemiology of stomach cancer, with special reference to the effect of the diet factor. Bull Inst Publ Health 1963, 12: 85-96. 4. Desowitz RS. Human and experimental anisakiasis in the United States. Hokkaïdo Igaku Zasshi 1986; 61: 358-71 5. Petithory JC, Lapierre J, Rousseau M, Clique MT Diagnostic sérologique de l’anisakiase par précipitation en milieu gélifié. Méd Mal Infect 1986; 3: 157-62 6 Petithory JC, Moisan F, Vigier G. Serologic diagnosis of subacute and chronic anisakiasis: report of 23 cases. Bull Soc Fr Parasitol 1990; 8 (suppl 2): 1018

3.

Hôpital Beaujon, Clichy, France

D. LEBREC

Institute of Biometrics, University of Milan, Milan, Italy

A. MORABITO

Division of Medicine, Ospedale V Cervello, Palermo

G. D. AMICO

Medical Clinic R, University of Palermo

F. TINE

Furberg CD, Morgan TM. Lessons from overviews of cardiovascular trials. Stat Med 1987; 6: 295-303. 2. Pagliaro L, Burroughs AK, Sorensen TIA, et al. Therapeutic controversies and randomised controlled trials (RCTs): prevention of bleeding and rebleeding in cirrhosis. Gastroenterol Int 1989; 2: 71-84. 3. Andreani T, Poupon RE, Balkau B, et al. Prophylaxie de la premier hemorragie digestive chez le cirrhotique: etude prospective randomosee comparant sclerose endoscopique, propranol et placebo. Gastroenterol Clin Biol 1990; 14 (S2). A5. 4. The PROVA Study Group. Randomized multicenter trial of sclerotherapy and/or propranolol for prophlaxis of first hemorrhage from oesophageal vances m cirrhosis. In: Abstracts of World Congress of Gastroenterology (Sydney, 1990). Abingdon: Medicine Group (UK), 1990: 1192. 5. Italian Multicentre Project for Propranolol in Prevention of Bleeding. Propranolol prevents first gastrointestinal bleeding in non-ascitic cirrhotic patients. J Hepatol 1989; 9: 75-83. 6. Garden OJ, Mills PR, Bimie GG, Murray GD, Carter DC. Propranolol in the prevention of recurrent variceal hemorrhage in cirrhotic patients: a controlled trial Gastroenterology 1990; 98: 185-90. 7. Laupacis A, Sackett DL, Roberts RS. An assessment of clinically useful measures of 1.

the consequences of treatment.

These epidemiological, experimental, and histopathological data suggest that A simplex may be a co-factor for certain forms of gastric cancer-a hypothesis that can be tested by longitudinal studies, which take into account the long incubation period of cancers.

N Engl J Med 1988; 378: 1728-33.

Anisakis simplex, a co-factor of gastric cancer? SIR,-Dr Davis and colleagues (Aug 25, p 474) note that mortality from stomach cancer is still falling in nearly all the countries studied "although the oldest groups in Italy and Japan show some continuing increases". Japan has the highest national mortality rate from gastric cancer (52-3 per 100 000 in men in 1977), though the rate has been falling.l Gastric anisakiasis is also very common in Japan, with 11 232 cases in 1988.zA limited relation has been found between gastric cancer mortality3 and the consumption of salted fish that could be the origin of contamination by Anisakis simplex. Desowitz,4 in work on anisakiasis, found a low molecular weight fraction of extract with "tumour-promoter-like activity". By serological diagnostic methods’ we have detected ten subacute and chronic cases of anisakiasis. These patients had gastrointestinal symptoms, they ate raw or undercooked fish, and they had moderate hypereosinophilia. One of these cases is of special interest. A 32-year-old woman was referred to us with eosinophilia (710/p.I) and persistent digestive troubles over several years. She told us that she ate raw fish at least once a month in Japanese restaurants. A serological test with a crude third-stage larval extract ofA simplex gave an arc with the Ouchterlony technique, an arc with electrosyneresis, and five bands on western blotting. She stopped eating raw fish and her symptoms disappeared, serological tests became negative, and the eosinophil count fell to normal after 6 months. 312 years later, digestive troubles reappeared and gastroscopy revealed three polyps. Histopathological study indicated that these were polyadenomas infiltrated with eosinophils (15 per microscopic field x 400). In Japan such polyadenomas are seen as precancerous lesions.’

Living non-related kidney transplantation in Bombay SiR,—In 1986

we reported our early experience in non-related kidney living transplantation done in Bombay and followed up by us.1 All the patients then (as now) went to Bombay for renal transplantation against our advice. The results reported then related to the first 20 such patients and showed, as found by Dr Salahudeen and colleagues (Sept 22, p 725), a high incidence of complications and graft failure. Later experience in a further 90 patients coming back from Bombay since that time revealed a worrying incidence of

HIV infection :2 so far there have been 6 cases, of whom 4 have died. This complication was also reported by Salahudeen et al. However, unlike Salahudeen et al, we have noticed an improvement in one-year patient and graft survival (95 % and 85 %, respectively) compared with our early experience.’ This is perhaps not surprising: after many transplants had been done in Bombay the surgical technique had largely been mastered and postoperative and medical care is largely given by us because the patients return to Saudi Arabia 13 days postoperatively on average. It is therefore crucial that opposition to commercial transplantation be based not on the poor results (which have already shown improvement) but on total abhorrence of commercialism in organ donation, whatever the result. Department of Nephrology, Riyadh Armed Forces Hospital,

A. A. AL-KHADER M. AL-SULAIMAN J. M. DHAR

PO Box 7897,

Riyadh 11159, Saudi Arabia

1. Al-Khader AA, Abomelha, Saltissi D, Chang R, Jawdat M, Etaibi K. Our experience in live non-related renal transplantation performed in Bombay and followed up in Saudi Arabia. Proceedings of XIth International Congress of the Transplantation

Society, 1986; S11.4 A, Al-Khader AA, Al-Hasani MK, Dhar JM. The impact of HIV infection on dialysis and renal transplantation. Transplant Proc 1989; 21: 1970-71.

2 Al Sulaiman

Compliance and clinical trials in heart disease SIR,-Professor Horwitz and colleagues (Sept 1, p 542) have re-examined the relation between compliance and mortality, using data from the Beta-Blocker Heart Attack Trial (BHAT).1 It would be interesting to know if an analysis over the whole duration of the study also shows a relation. In the Lipid Research Clinics Coronary Primary Prevention Trial (LRC-CPPT)2 adherence was not found to be related to event rate in the placebo group. Horwitz et al argue that this may be because the LRC-CPPT was a primary prevention trial, whereas BHAT and the Coronary Drug Project (CDP)’ were secondary prevention trials. LRC-CPPT differs from BHAT and CDP in other ways that seem more likely to explain the divergent findings. The patients in LRC-CPPT were younger and better

Beta-blockers for preventing variceal bleeding.

1001 LETTERS to the EDITOR Beta-blockers for preventing variceal bleeding SIR,-We would take issue with some aspects of the meta-analysis by Dr Haye...
274KB Sizes 0 Downloads 0 Views