Calcium Antagonist Controversy, Does the Accusing Finger Also

Commentary: Calcium antagonist controversy, does the accusing ﬁnger also point to beta-blockers and methyldopa?

LH Opie Medical Research Council, Heart Research Unit, University of Cape Town, Cape Town, South Africa

Keywords: calcium antagonists; mortality; case control; hypertension; methyldopa; beta-blockers

‘Stop it at the start, it’s late for medicine to be nature, retrospective and often refer to patterns of prepared when disease has grown strong through drug usage that no longer hold, such as the use of delays’ Ovid, in Remedia Amoris, about 10 bc. short acting CCAs for hypertension, which is no longer common practice. Given all these caveats, it Ever since Ovid, the correct philosophy has been to is no surprise that case control studies do not pro- stop a disease in its early stages, so that antihyper- vide as strong evidence as other methods of study tensives are commonly started on the basis of elev- such as cohort studies (somewhat stronger) or ran- ated blood pressure (BP) readings, using a variety of domized controlled trials (RCTs; much stronger agents, rather than waiting for end-organ damage to evidence). Nonetheless, case control studies cannot develop. Calcium antagonists, also called calcium be ignored at a time when every morsel of available channel blockers (CCBs) or calcium channel antag- evidence needs careful assessment, and when the onists (CCAs), were once the darling of the hyper- number of outcome RCTs with CCAs is strictly lim- tension fraternity, yet are now under scrutiny. The ited. ‘complaint’ is chiefly based on three reports, each The case control study of Psaty et al1 concluded using a different technique: case control,1 cohort,2 that ‘the use of short-acting calcium channel block- or meta-analysis.3 Each has given rise to debate and ers, especially in high doses, was associated with an controversy,4,5 The present study by a British group6 increased risk of myocardial infarction’ (AMI) when is case control in design and contrasts with the pro- compared with the reference group which was the posal of Psaty et al1 that CCAs are harmful when beta-blocker group. The major defects of this study compared with beta-blockers, so that both the include that: (1) in one-third of cases the pretreat- reports warrant very careful examination. ment BP measurements were missing; (2) there were A number of major defects of case control studies confounding differences between the CCA and the are well known and include the following. First, beta-blocker group, such as the fact that more of the there is usually no guarantee that the subject had CCA group were diabetic; and (3) fatality rates were actually taken the drug or had used the right dosage. not emphasized but were equal in the CCA and in It must also be assumed that there was no change of the beta-blocker groups. It follows that the increase medication during the period of the study, unless in AMI associated with CCAs was in non-fatal but there is effective follow-up which is unusual for a not in fatal infarctions. case control study. Second, there is no knowledge about the reason for choosing one drug rather than another, but it is reasonable to suppose that CCAs The present study might have been preferred over diuretics for those with pre-existing ischaemic heart disease. Thirdly, Against that background, the study by Bulpitt et al6 confounding factors are difficult to obviate.7 For from the Royal Postgraduate Medical School in Lon- example, if the perception is that diabetic patients don is of considerable interest. First, the total num- should be given a CCA rather than a beta-blocker, ber of cases given CCAs is much larger (349 given then there will be more diabetics, with their greater CCAs in Britain vs 56 in the American study) also cardiovascular risk, in the CCA than in the beta- with many more given the combination of a CCA blocker group, as happened in the study of Psaty et plus a diuretic (643 vs 24, respectively). Second the al.1 Fourthly, case control studies are by their number given diuretics was also much larger, being 2525 vs 452. Likewise there were many more given beta-blockers with a diuretic (269 vs 34) and ACE Correspondence: Professor LH Opie, Medical Research Council, inhibitors, with or without a diuretic. Thus the stat- Heart Research Unit, University of Cape Town, Medical School, istical power of the British study was greater. The Observatory 7925, Cape Town, South Africa average time that patients were studied was about Calcium antagonist controversy LH Opie 202 11 years in the British study and 7 years in the myself) especially in uncomplicated hypertensive American study. The most important difference is patients. Diuretics are relatively safe and easy to use. that the British study was sufficiently powered to In several studies initial diuretic treatment with low give data on mortality in three categories – dose, potassium-sparing agents has been used with ischaemic heart disease (IHD), cardiovascular mor- outcome benefit and reduction of mortality.8,11 In tality and total mortality. The key results may be particular, potassium-sparing agents seem safer than summarized as follows. others on the evidence of case control studies.9,10 Mortality due to IHD was non-significantly Potassium retainers maintain body potassium and increased in all groups (methyldopa, CCA, beta- magnesium better than non-retainers12 and the blocker and ACE inhibitor) when compared with a consequence may be a decreased risk of serious car- diuretic alone. Cardiovascular mortality was diac arrhythmias. The problem with singing the increased only by methyldopa. Total mortality was praises of diuretics on the basis of this British study increased by all three types of agents when com- is that neither type nor dose of diuretic are known. pared with diuretics, but only significantly so by If the British doctors, in their wisdom, had chosen methyldopa (adjusted for treated BP), or by CCAs predominantly low-dose potassium-sparing agents when adjusted for untreated BP, the latter being then this would explain why the beta-blockers in unavailable in 42% of the cases, whereas treated BP this study had a relative increase in mortality com- levels were known in 73% of patients. Regarding the pared with the diuretic groups. ‘with diuretic groups’, both IHD and cardiovascular Other problems with the present study, are those mortality tended to be higher with CCAs, and were common to all case control studies, including the statistically higher with methyldopa. Total mor- fundamental fact that it will never be clearly known tality, most surprisingly, was increased not why a given drug or combination of drugs was decreased in three groups, specifically methyldopa chosen rather than another. It must also be assumed plus diuretic, CCA plus diuretic and beta-blocker that the drugs were distributed to hypertensive plus diuretic all with relative risks of 1.4 or slightly populations which were roughly comparable, to more. ACE inhibitors were excluded from this analy- make up for the almost irreparable defect of non- sis because of low numbers. The authors conclude randomization. However, it is difficult to exclude that methyldopa, CCAs and beta-blockers were all the possibility that the diuretic group in the British associated with higher mortalities than the reference study were in fact those hypertensives who did not diuretic groups, and that the results in the three non- have associated end-organ damage, because, for diuretic groups were similar. It needs re-emphasis example, the diuretic group had a majority of that the selection bias cannot eliminate the possi- females and the beta-blockers group a majority of bility and indeed the probability (Table 1) that males. Not enough data are available on various patients with higher BP values were more likely to potential confounders such as diabetes. Nonethe- receive initial combination therapy with diuretics less, on the given data, the various groups seem to and another agent. have comparative BP levels at entry, about 180/110 Allowing for these many serious defects, the Brit- mm Hg, except for beta-blockers without diuretics ish study points the accusing finger not only at CCAs (about 170/105 mm Hg). BP levels during treatment but also at beta-blockers and methyldopa. In fact, the were also comparable in all groups except that the results with methyldopa are so bad (Table 2) that values again seemed lowest in the beta-blocker serious consideration might now be given to specific group. One interpretation could be that beta-block- advice not to use this drug either on its own or in ers had the same dangers as other drugs but perhaps combination with a diuretic. An exception would evident despite lower BP levels. remain the short term use of methyldopa in preg- nancy, where this drug has been studied better than Is there a case against beta-blockers in most others. A second exception is in the elderly, when monotherapy by a potassium-retaining hypertension therapy? diuretic has failed.8 There is already evidence that Further evidence against beta-blockers for hyperten- beta-blockers, when compared with potassium-spar- sion is as follows. There have been only two large ing diuretics, tend to cause sudden death9,10 but the placebo controlled studies (RCTs) both conducted present data are the first to directly accuse beta- by the British Medical Research Council (MRC). In blockers when compared with diuretics. Further, the the first, on middle-aged subjects, the beta-blocker detrimental effect was especially noted when com- gave limited benefits in non-smokers and the fall in paring a beta-blocker combined with diuretics alone mortality was slight and not significant.13 In smok- (relative risk, 1.57, CI 1.09–2.26). Regarding the ing men or women, propranolol did not reduce CCAs, the authors have perhaps taken too lightly strokes, whereas the diuretic did. Again, in smokers, their finding of an increased relative risk (RR) for propranolol did not reduce coronary events (nor did total mortality with CCAs when adjusted for the diuretic). In non-smokers propranolol reduced untreated rather than treated BP (Table 3, right hand stroke and coronary events in males but not in column), although in nearly 50% the untreated BP females. These results do not exactly show a dra- was not known which weakens this finding. matic benefit for propranolol, and certainly do not The consoling factor to emerge from this large argue strongly for a mortality benefit, because all- British study is that diuretic treatment seems least cause mortality was low and unaffected by propran- harmful. In general, diuretics are often chosen for olol. In the second trial, on patients aged 65–74, first line treatment (and favoured by many including despite equal BP reduction by a diuretic and the Calcium antagonist controversy LH Opie 203 beta-blocker atenolol, ‘the beta-blocker group sion,18 and with increased mortality in the very eld- showed no significant reductions in these end erly in one cohort study2 and with acute ischaemic points’, which were stroke, coronary event and all syndromes in one meta-analysis.3 In another meta- cardiovascular events.11 Rather, there was an analysis reported to the FDA and yet to be pub- increased incidence of cardiovascular deaths in the lished, it was the short but not long acting form of beta-blocker group (66 in diuretic and 95 in beta- nifedipine that was associated with adverse out- blocker, P = 0.03), and a trend towards a greater comes. number of total deaths in the beta-blocker group (P In summary, the present case control study suffers = 0.07). There was a suggestion of increased cancer from the class defects of such studies, but the num- in the beta-blocker group. Overall, these results from bers are much larger than in previous reports and placebo-controlled trials are hardly impressive for allow specific analysis of the incidence of deaths. the beta-blocker and do not allow the conclusion The therapy seemingly safest was diuretics, in com- that mortality was lessened. In comparative trials parison with which not only CCAs but also beta- with diuretics in middle-aged subjects, mortality blockers and especially methyldopa did badly. The was similar in the HAPPHY study14 and lower in the numbers with ACE inhibitors are too small to allow beta-blocker group in the MAPHY study15 – the lat- comment. The adverse effects of the non-diuretic ter study evolved from the former and was in reality drugs could simply be explained by preferential a follow-up open-label study; it was not a simple choice of such drugs over diuretics for complicated prospective comparison. Of interest in the MAPHY hypertensives. Nonetheless, the disappointing data study,15 the better result with the beta-blocker meto- with the two adrenergic modifying drugs, beta- prolol was confined to smokers, in contradiction to blockers and methyldopa, are difficult to understand the MRC study, only the latter being placebo- but may be related to intermittent adrenergic dis- controlled. Also relevant is that the comparator charge for a short-acting beta-blocker,19 and to the diuretic used in the MAPHY study15 was a non-pot- well-described potentially fatal side effects of assium retaining thiazide, which could have biased methyldopa such as severe haemolytic anaemia and results towards the beta-blocker.9 If the MAPHY liver failure. This report reinforces the view that evi- study15 is the only evidence there is for a benefit of dence-based medicine requires outcome data for the beta-blockers on mortality in hypertensives, then a use of both CCAs (which are underway) and beta- modern view would have to be that the trial design blockers in hypertension. This report gives no sup- was so faulted and the resultant evidence so weak, port to the view that beta-blocking drugs beneficially that it remains possible beta-blockers have no mor- modify mortality in hypertensive patients, a view ality benefit or even possible harm as suggested by often ascribed to the authors of JNC V. Rather, out- the MRC study in the elderly and by the present case come data are seriously lacking. The often-quoted control study. ‘First do no harm’ means that we should now be serious about collecting good prospective data for Comparison with previous case control the real benefit in the treatment of hypertension not only for CCAs and ACE inhibitors but also for studies beta-blockers. How do these results compare with other case control studies? It must be stated that each of the other case report studies has a different end-point. For Conclusion example, Psaty reported on non-fatal MI in patients The present report, although suffering from the seri- many of whom had other risk factors, the present ous defects inherent to case control studies, could study reports on deaths, the Jick study on AMI in suggest that diuretics are safer than CCAs, beta- hypertensives with no other risk factors,16 and the blockers or methyldopa. The slow realization that Aursnes study on patients with AMI who did or did the case for data showing the long-term safety of not have hypertension.17 Although the simplest both CCAs and beta-blockers when given for hyper- would be to group these studies together and to say tension is not immaculate, results from the new that taken together, the charge against CCAs is not critical eye which has been opened by contro- proven and the balance of evidence shows no large versy.3,20 scale adverse effects when compared to beta-block- ade, the mortality evidence from the present study ‘Controversy should be welcomed, it is only through 21 does point against both beta-blockers and CCAs controversy that science advances’. when compared with diuretics. Geographic differences in patterns of drug use References may be important. The negative (for CCAs) case control study of Psaty was conducted in the USA, 1 Psaty BM et al. The risk of myocardial infarction asso- whereas the three studies showing no adverse ciated with antihypertensive drug therapies. JAMA effects were carried out in Europe. In general, slow 1995; 274: 620–625. acting formulations of CCAs were introduced much 2 Pahor M et al. Long term survival and use of antihypertensive medications in older persons. J Am Geriat Soc earlier in Europe. The difference between short and 1995; 43: 1191–1197. long acting preparations is especially marked in the 3 Furberg CD, Psaty BM, Meyer JV. Nifedipine dose- case of nifedipine, where the short acting form is related increase in mortality in patients with coronary clinically known to precipitate angina as a side heart disease. Circulation 1995; 92: 1326–1331. effect, and is associated with risk of excess hypoten- 4 Opie LH, Messerli FH. Nifedipine and mortality. Grave Calcium antagonist controversy LH Opie 204 defects in the dossier. Circulation 1995; 92: 1068– 13 MRC Study. Medical Research Council trial of treat- 1073. ment of mild hypertension: principal results. Br Med 5 Tijssen JGP, Hugenholtz PG. Critical appraisal of J 1985; 291: 97–104. recent studies on nifedipine and other calcium chan- 14 HAPPHY Trial, Wilhelmsen L et al. Beta-blockers ver- nel blockers in coronary artery disease and hyperten- sus diuretics in hypertensive men: main results from sion. Eur Heart J 1996; 17: 1152–1157. the HAPPHY trial. J Hypertens 1987; 5: 561–572. 6 Bulpitt CJ et al. Calcium channel blockers and cardiac 15 MAPHY Study, Wikstrand J et al. Primary prevention mortality in the treatment of hypertension: A report with metroprolol in patients with hypertension. JAMA from the Department of Health Hypertension Care 1988; 259: 1976–1982. Computing Project (DHCP). J Hum Hypertens 1997; 11: 16 Jick H, Derby LE, Gurewich V, Vasilakis K. The risk of (in press). myocardial infarction associated with antihyperten- 7 Mancia G, Zwieten PA van. How safe are calcium sive drug treatment in persons with uncomplicated antagonists in hypertension and coronary heart dis- essential hypertension. Pharmacotherapy 1996; 16: ease? J Hypertens 1996; 14: 13–17. 321–326. 8 Amery A et al. Efficacy of antihypertensive drug treat- 17 Aursnes I, Litleskare I, Froyland H, Abdelnoor M. ment according to age, sex, blood pressure and pre- Association between various drugs used for hyperten- vious cardiovascular disease in patients over the age sion and risk of acute myocardial infarction. Blood of 60. Lancet 1986; ii: 589–592. Press 1995; 4: 157–1163. 9 Siscovick DS, Raghunathun TE, Psaty BM. Diuretic 18 Grossman E, Messerli FH, Grodzicki T, Kowey P. therapy for hypertension and the risk of primary cardiac arrest. New Eng J Med 1994; 330: 1852–1857. Should a moratorium be placed on sublingual nifedip- 10 Hoes AW et al. Diuretics, ␤-blockers, and the risk for ine capsules given for hypertensive emergencies and sudden cardiac death in hypertensive patients. Ann pseudo emergencies. JAMA 1996; 276: 1328–1331. Intern Med 1995; 123: 481–487. 19 Leenen FH. 1.4-Dihydropyridines versus beta-blockers 11 MRC Study. Medical Research Council trial of treat- for hypertension: are either safe for the heart? Cardiov- ment of hypertension in older adults: principal results. asc Drugs Ther 1996; 10: 307–402. Br Med J 1992; 304: 405–412. 20 Messerli F. Whatever happened to the calcium antag- 12 Widmann L, Dyckner T, Wester PO. Effects of triamter- onist controversy? (Editorial). JACC 1996; 28: 12–13. ene on serum and skeletal muscle electrolytes in 21 Frohlich ED, Zanchetti A. Antihypertensive therapy: diuretic-treated patients. Eur J Clin Pharmacol 1988; safety and efficacy of drugs and publications. Hyper- 33: 577–579. tension 1996; 27 (Part 1): 317–318.