399
PYRAZINAMIDE IS A MAJOR ANTITUBERCULOSIS DRUG
colleagues (Nov. 7, p. 1056), writing under SIR,-Dr the title "Pyrazinamide fulminant hepatitis: an old hepatotoxin strikes again", ascribe the death of a tuberculosis patient to Danan and
pyrazinamide. Any observation on the side-effects of antituberculosis drugs, above all if it is dramatic, should be published so that other clinicians can be alerted. However, having some experience with complex antituberculosis treatments including several hepatotoxic drugs, we think that it is also important to interpret biological data carefully. Several cases of jaundice were reported years ago, when pyrazinamide was prescribed at daily doses of 3 g or more to seriously ill patients whose immune resistance was flagging and who had already taken hepatotoxic drugs for a long time. Among these patients a few deaths due to therapy have been reported, but it is often difficult to determine which drug or combination of drugs was responsible, with the possible exception of Potter and Chang’s report (this and other references in this letter may be found in our 1980 review): pyrazinamide was prescribed as monotherapy to twenty-seven patients and one death from hepatitis was recorded. In other cases (Petty and Mitchell, MacDermott, and Donnemberg et al.) pyrazinamide was prescribed in association with isoniazid at high doses. In Girling’s review and in our own no other death due to pyrazinamide could be found. Danan’s patient had been given several drugs. The phenotype of isoniazid acetylation was not reported, and though the dosage of isoniazid seems to have been low, this drug is hepatotoxic too. Isoniazid was still given when serum transaminase activities were three times higher than normal, when cytolysis would already have been present. It is difficult to ascribe this death to pyrazinamide alone.
Pyrazinamide is hepatotoxic, as is isoniazid, and continuous laboratory checks are needed: repeated rises in transaminase values justify withdrawal of the drug. We have given pyrazinamide, with isoniazid, rifampicin, and streptomycin, to eighty-four patients for 18 weeks, during a clinical trial now in the follow-up stage. Biological tolerance has been satisfactory and the few slight clinical problems have not obliged us to stop treatment. Pyrazinamide, whose sterilising activity is well established, is recognised by World Health Organisation experts as one of the three major antibiotics against the tubercle bacillus, together with rifampicin and isoniazid. Respiratory Diseases Service, C H.U. Cochin, Paris
S. PRETET
INSERM Unit 179, 44 chemin de Ronde, 78110 Le Vesinet, France
R. LIARD S. PERDRIZET
HOW LONG HAS HE GOT, DOCTOR?
SIR,-Can any statistician tell me if it would be wrong to use the "half-life" principle to assist in the interpretation of cumulative survival curves? The usual method of interpreting such a curve is to read off the percentage alive at a given interval after treatment, usually five years. Such a rate suffers from two drawbacks, one theoretical and one practical. Unless many patients have been studied for much longer than five years, calculation of the survival curve beyond about the third year of treatment is based on very few patients. This is certainly true of some cancers where five year survival is poor. Thus, a finding that surgery increased survival at five years by 1007o may be based on only a handful of patients. The practical objection is that relatives of a cancer patient never ask about chances of a five year cure but about how long the patient will live-a question which cannot be answered from median survival times. Since patients are continually being added to a series, and some are still alive, the median survival cannot be calculated. Another way to express survival would be to quote the "half-life"-i.e., the length of time which elapses before the survival rate drops to 50%. The figure shows the survival of patients with fixed nodes in the neck secondary to a head and neck cancer. The three curves represent untreated patients, those treated by radiotherapy, and those treated by surgery. Survival figures at five years were 0%, 0%, and 16%, respectively. This result appears strongly to favour surgery, but in the whole series of 139 patients, only 4 survived beyond three years and only 2 for more than four years. At one year, survival figures for the three groups were 6%, 34%, and 58%, respectively, but even these figures are based on only 22 out of 139 patients-less than 16% of the total. Furthermore, these figures do not help to predict the survival of individual patients. The times elapsing until the survival rate had dropped to 50% were 105, 275, and 413 days, respectively. In simple terms, the untreated patients had about three months to live, those treated by radiotherapy had nine months, and the surgical group had 15 months, with a very small chance of cure. One positive aspect is that there are simple tests available for analysis of the survival times of those who die before the half-life survival time is reached. These are the Wilcoxon rank sum test, if there are only two groups, and the Kruskal-Wallis one way analysis of variance for non-parametric data (Siegel) for 3 or more groups-which, when applied to the data in the figure revealed significant differences between the half-life survival times of these groups (X2= 8 -45, p<0 02). These tests might usefully supplement the standard log rank test described by Peto et awl.2 University Department of Oto-rhino-laryngology, Royal Liverpool Hospital, Liverpool L69 3BX
P. M. STELL
LUNG CANCER IN BUTCHERS
SIR,-The excess mortality from lung cancer among butchers in England and Wales around the 1971 census, to which Professor Fox and his colleagues draw attention (Jan. 16, p. 165) was also observed in the 1959-63 period. Butchers and meat cutters had a standardised mortality ratio (SMR) for lung cancer of 127, based on 436 deaths of men under 65 years.2A similar excess was noted in the occupational mortality report of the 1951census. The SMR for meat and fish curers and smokers was 200 (based on only 10 deaths) and that of slaughterhouse workers was 131 (17 deaths). The earlier report does not identify butchers separately. They would have been included in the occupational group "Proprietors etc of retail businesses for the sale of greengrocery, meat, fish, poultry and other food goods", a group with an SMR for lung cancer of 151 based on
1.
Siegel S. Non parametric statistics. Tokyo: McGraw-Hill, Kogakusha, 1956. R, Pike MC, Armitage P, et al. Design and analysis of randomised clinical trials requiring prolonged observation of each patient. Br J Cancer 1977; 35: 1-39
2. Peto
471 deaths. 18 Vtciona Terrace, Beaumans, Gwynedd LL58 8BU
G. WYNNE GRIFFITH
1 Pretet S, Perdrizet S. La toxicité du pyrazinamide dans les traitements antituberculeux
2. 3
Rev Fr Mal Resp 1980; 8: 307. Registrar General’s decennial supplement England and Wales, mortality, tables. London. HMSO, 1971. Registrar General’s decennial supplement England and Wales, mortality, pan II, vol 2, tables. London: HMSO, 1957
1961:
Occupational
1951:
Occupational
Survival curves of patients with nodes in the neck.
a
head and neck
Reading downwards from top, the curves represent surgery, those who received radiotherapy, and those who
cancer
and fixed
who had untreated.
patients were