318
CLOFIBRATE SiR,-Your editorial’ under the heading Clofibrate: a Final Verdict? was a great disappointment in that a verdict was reached after a superficial review of a very complex investigation. We would have valued an independent appraisal of a trial which raises several statistical and clinical issues, most of which you did not tackle. The interpretation of the results of this trial is not as clear as your editorial implies, and you do not do justice to the investigators themselves who presented more deeply considered interpretations of the analyses.2 The major end-points are morbidity and mortality from ischaemic heart-disease and deaths from all causes subdivided into deaths from causes other than I.H.D. Prof. M. F. Oliver, summarising the results of the trial in which he took part, stated that "what is clear, at last, is that reduction of serum cholesterol concentrations is relevant to the prevention of coronary heart disease".’ This conclusion was reached from a community study in which clofibrate was used in a way that has never been recommended in clinical practice. With regard to mortality, the investigators stated that "There was no significant difference in age standardised mortality from all causes between any of the groups". Quite properly, they attempted by a series of further analyses to define the place of clofibrate in the treatment of hypercholesteroleemia to prevent I.H.D. rather than rest on these broad conculsions. The men selected for the two high-cholesterol groups i and 11 were subsequently found to have an average serum-cholesterol of 248 mg/dl, a value many clinicians would consider as a high "normal" concentration. Half these men were then given clofibrate for more than 5 years without the benefit of ancillary treatment and irrespective of whether their cholesterol levels fell, remained unchanged, or rose. Under these circumstances the incidence of non-fatal myocardial infarctions was reduced by 25% associated with a modest reduction of the serum-cholesterol of 9%. When the men are subgrouped into those most likely to receive clofibrate in clinical practicenamely those with cholesterol levels above the mean showing a fall in levels under treatment-the reduction is greater and may be further enhanced if other risk factors are also present. The failure of clofibrate to reduce the incidence of fatal heartattacks is disappointing but the ratio of fatal to non-fatal attacks is lower than would be expected. The results from this trial clearly show that reduction of serum-cholesterol by the administration of clofibrate reduces the incidence of non-fatal heart-attacks and that the reduction can be further enhanced by the selection of appropriate patients together with a vigorous attempt to reduce the cholesterol levels. ’-Most if not all active drugs have adverse as well as beneficial effects and it is a requirement that the benefit/risk ratio must be favourable. The difference between the number of deaths from all causes in groups I and n was subjected to many further analyses because the difference was statistically significant when expressed numerically (162 v. 127) and as a crude rate although not when age-standardised procedures were used. Furthermore, "A comparison of the age-standardised rates in groups 1,11 and in for all causes other than I.H.D. (4.1, 3-1, 3-9) suggests that part of the difference between groups i and II is partly or largely the result of a low mortality in group ii rather than a high mortality in group i. For cancer the corresponding rates are 2.2, 1-7and 2.5 respectively, with a stronger indication of the same point." Most drug-induced toxicity tends to be targeted to one organ or organ system so it was entirely appropriate to classify the subjects by cause of’ death. Except for deaths after cholecystectomy and, possibly, deaths grouped under the heading of regional pathology, no target-oriented toxicity was demonstrated. The major cause of non-i.H.D. deaths other than vascular diseases and accidents was cancer, accounting for some 80% of the total. Further ex1. Lancet, 1978, ii, 1131. 2. Committee of Principal Investigators Br. Heart J. 1978, 40, 1069. 3. Oliver, M. F. New Engl. J. Med. 1978, 299, 1360.
amination of the deaths from cancer, including a comparison with rates calculated from the official mortality statistics for men from the three centres, led the investigators to write "Thus the data from all cancer do not give rise to special concern." There remains the group of other medical causes for which there is a difference between groups i and II (16 and 5). From the tabulation in table 14 these 21 deaths are distributed between thirteen different diagnoses. Only under two headings (cholecystectomy and pancreatitis) does the number of deaths reach 3, and it is likely that both causes were associated with gallstones. If the deaths from this wide range of diseases were to be attributable to clofibrate it would be reasonable to expect that these same diseases were the cause of morbidity as well as mortality and this would be reflected in the number of withdrawals for medical reasons. Table 15 clearly shows this not to be so. It is also worthy of note that the excess of deaths in group i compared with group II "was most evident in Budapest and virtually absent in Edinburgh". Furthermore, threequarters of the excess deaths from all causes other than I.H.D. occurred in Budapest. Finally, in an attempt to seek a target organ, a number of deaths previously included under other headings were retrospectively assembled under the heading of "regional pathology", yielding an apparent excess of deaths in groupi compared with group n. This difference is reflected in the crude rates but age-standardisation shows the rates for group in not to be dissimilar to those of group i and thus raises again the possibility that the rate in group ii is fortuitously low. If this was indeed so and no organ-specific toxicity was identified, then the overall difference between groups i and ii would be similarly represented in all subsequent subgrouping. Examination of the tables shows this premise to be correct with very few exceptions. I have tried to indicate in this very brief resume of the results and the investigators’ comments that the appraisal of the trial is both difficult and complex and likely to be debated for some time to come. The more immediate problem concerns the use of clofibrate in the treatment of patients with hyperlipidasmia. I.C.I. agrees with the investigation committee and does not recommend the use of clofibrate for community-wide primary prevention of ischaemic heart-disease as it was used in the trial. This conclusion should not be extended to patients individually treated for recognised disorders of lipid metabolism which may put them at increased risk of heart-disease. The trial did not make any formal study of such patients and the information from many other studies must be considered in relation to the treatment of these patients with diet, clofibrate, and other agents available for lowering blood-lipids. The collective evidence, including the findings of this large trial, does show that lowering serum-cholesterol levels can result in a significant reduction in the incidence of ischsemic heart-disease. I.C.I. Ltd., Pharmaceuticals Division
Alderley Park, Macclesfield, Cheshire
*)* It is understandable that
COLIN C. DOWNIE
our editorial (and the results we the manufacturers of clofibrate. However, the results of a randomised trial must be judged in relation to its primary intentions and design. One major end-point (non-fatal infarction) showed a significant benefit, a second (total mortality) was significantly adverse, and the third (nonfatal infarction) showed no effect. With hindsight and special pleading one can then proceed to analyse subgroups, perform extra standardisations, or contrast results with those in other patients outside the randomisation procedure; but this is lightweight evidence, and must not be allowed to obscure the basic findings. We are glad that I.C.I. agrees that the drug should not be used for primary prevention in the community. Even in gross hypercholesterolaemia the effect on survival has yet to be demonstrated.-ED.L.
reviewed) disappointed
CLOFIBRATE SiR,-Your editorial’ under the heading Clofibrate: a Final Verdict? was a great disappointment in that a verdict was reached after a superficial review of a very complex investigation. We would have valued an independent appraisal of a trial which raises several statistical and clinical issues, most of which you did not tackle. The interpretation of the results of this trial is not as clear as your editorial implies, and you do not do justice to the investigators themselves who presented more deeply considered interpretations of the analyses.2 The major end-points are morbidity and mortality from ischaemic heart-disease and deaths from all causes subdivided into deaths from causes other than I.H.D. Prof. M. F. Oliver, summarising the results of the trial in which he took part, stated that "what is clear, at last, is that reduction of serum cholesterol concentrations is relevant to the prevention of coronary heart disease".’ This conclusion was reached from a community study in which clofibrate was used in a way that has never been recommended in clinical practice. With regard to mortality, the investigators stated that "There was no significant difference in age standardised mortality from all causes between any of the groups". Quite properly, they attempted by a series of further analyses to define the place of clofibrate in the treatment of hypercholesteroleemia to prevent I.H.D. rather than rest on these broad conculsions. The men selected for the two high-cholesterol groups i and 11 were subsequently found to have an average serum-cholesterol of 248 mg/dl, a value many clinicians would consider as a high "normal" concentration. Half these men were then given clofibrate for more than 5 years without the benefit of ancillary treatment and irrespective of whether their cholesterol levels fell, remained unchanged, or rose. Under these circumstances the incidence of non-fatal myocardial infarctions was reduced by 25% associated with a modest reduction of the serum-cholesterol of 9%. When the men are subgrouped into those most likely to receive clofibrate in clinical practicenamely those with cholesterol levels above the mean showing a fall in levels under treatment-the reduction is greater and may be further enhanced if other risk factors are also present. The failure of clofibrate to reduce the incidence of fatal heartattacks is disappointing but the ratio of fatal to non-fatal attacks is lower than would be expected. The results from this trial clearly show that reduction of serum-cholesterol by the administration of clofibrate reduces the incidence of non-fatal heart-attacks and that the reduction can be further enhanced by the selection of appropriate patients together with a vigorous attempt to reduce the cholesterol levels. ’-Most if not all active drugs have adverse as well as beneficial effects and it is a requirement that the benefit/risk ratio must be favourable. The difference between the number of deaths from all causes in groups I and n was subjected to many further analyses because the difference was statistically significant when expressed numerically (162 v. 127) and as a crude rate although not when age-standardised procedures were used. Furthermore, "A comparison of the age-standardised rates in groups 1,11 and in for all causes other than I.H.D. (4.1, 3-1, 3-9) suggests that part of the difference between groups i and II is partly or largely the result of a low mortality in group ii rather than a high mortality in group i. For cancer the corresponding rates are 2.2, 1-7and 2.5 respectively, with a stronger indication of the same point." Most drug-induced toxicity tends to be targeted to one organ or organ system so it was entirely appropriate to classify the subjects by cause of’ death. Except for deaths after cholecystectomy and, possibly, deaths grouped under the heading of regional pathology, no target-oriented toxicity was demonstrated. The major cause of non-i.H.D. deaths other than vascular diseases and accidents was cancer, accounting for some 80% of the total. Further ex1. Lancet, 1978, ii, 1131. 2. Committee of Principal Investigators Br. Heart J. 1978, 40, 1069. 3. Oliver, M. F. New Engl. J. Med. 1978, 299, 1360.
amination of the deaths from cancer, including a comparison with rates calculated from the official mortality statistics for men from the three centres, led the investigators to write "Thus the data from all cancer do not give rise to special concern." There remains the group of other medical causes for which there is a difference between groups i and II (16 and 5). From the tabulation in table 14 these 21 deaths are distributed between thirteen different diagnoses. Only under two headings (cholecystectomy and pancreatitis) does the number of deaths reach 3, and it is likely that both causes were associated with gallstones. If the deaths from this wide range of diseases were to be attributable to clofibrate it would be reasonable to expect that these same diseases were the cause of morbidity as well as mortality and this would be reflected in the number of withdrawals for medical reasons. Table 15 clearly shows this not to be so. It is also worthy of note that the excess of deaths in group i compared with group II "was most evident in Budapest and virtually absent in Edinburgh". Furthermore, threequarters of the excess deaths from all causes other than I.H.D. occurred in Budapest. Finally, in an attempt to seek a target organ, a number of deaths previously included under other headings were retrospectively assembled under the heading of "regional pathology", yielding an apparent excess of deaths in groupi compared with group n. This difference is reflected in the crude rates but age-standardisation shows the rates for group in not to be dissimilar to those of group i and thus raises again the possibility that the rate in group ii is fortuitously low. If this was indeed so and no organ-specific toxicity was identified, then the overall difference between groups i and ii would be similarly represented in all subsequent subgrouping. Examination of the tables shows this premise to be correct with very few exceptions. I have tried to indicate in this very brief resume of the results and the investigators’ comments that the appraisal of the trial is both difficult and complex and likely to be debated for some time to come. The more immediate problem concerns the use of clofibrate in the treatment of patients with hyperlipidasmia. I.C.I. agrees with the investigation committee and does not recommend the use of clofibrate for community-wide primary prevention of ischaemic heart-disease as it was used in the trial. This conclusion should not be extended to patients individually treated for recognised disorders of lipid metabolism which may put them at increased risk of heart-disease. The trial did not make any formal study of such patients and the information from many other studies must be considered in relation to the treatment of these patients with diet, clofibrate, and other agents available for lowering blood-lipids. The collective evidence, including the findings of this large trial, does show that lowering serum-cholesterol levels can result in a significant reduction in the incidence of ischsemic heart-disease. I.C.I. Ltd., Pharmaceuticals Division
Alderley Park, Macclesfield, Cheshire
*)* It is understandable that
COLIN C. DOWNIE
our editorial (and the results we the manufacturers of clofibrate. However, the results of a randomised trial must be judged in relation to its primary intentions and design. One major end-point (non-fatal infarction) showed a significant benefit, a second (total mortality) was significantly adverse, and the third (nonfatal infarction) showed no effect. With hindsight and special pleading one can then proceed to analyse subgroups, perform extra standardisations, or contrast results with those in other patients outside the randomisation procedure; but this is lightweight evidence, and must not be allowed to obscure the basic findings. We are glad that I.C.I. agrees that the drug should not be used for primary prevention in the community. Even in gross hypercholesterolaemia the effect on survival has yet to be demonstrated.-ED.L.
reviewed) disappointed
