Zidovudine for symptomless HIV infection Last month the Food and Drug Administration in the USA extended the licence for the use of zidovudine to symptom-free patients infected with human immunodeficiency virus (HIV) with peripheral blood CD4 counts below 500/p.l. A consensus meeting at the National Institutes of Health, Washington, also in March, strongly recommended that clinicians should follow this policy, although serious doubts were raised about its rationale.2 As a result of these deliberations about 50% of all HIV-infected people could be offered zidovudine-potentially a ten-fold increase in the use of the drug, with enormous implications for cost and resources. The evidence supporting this extension of drug use has been known informally for some time,3but the first report of the US clinical trial (ACTG 019) has only now been published.4 Before clinicians and their patients elsewhere decide whether to follow the US policy they should reflect on what is already known about zidovudine and on what the new trial has and has not shown. ACTG 019 was designed to test the efficacy of zidovudine in symptom-free patients. It was a doubleblind placebo-controlled study of 3200 subjects with three treatment groups-high dose (1500 mg/day), low dose (500 mg/day), and placebo. The study protocol allowed separate analysis of the 1338 patients who at entry had fewer than 500 CD4 cells. It is not clear why the level of 500 was chosen. Unfortunately, patients only had a single CD4 count recorded before entry and no quality control data are provided to assess the size of biological and laboratory variation. Patients were followed for a mean of 55 weeks (range 19-107) and the main end point was progression of disease to AIDS-related complex or AIDS. Last August the trial for patients with CD4 counts below 500 was stopped prematurely because the progression rate had been halved from 7-6 per 100-person-years of observation in the placebo groups to 3-6 (p=001) with low-dose and 4-3 (p=0-10) with high-dose
zidovudine. Further analysis showed that the number of patients progressing to AIDS fell from 33 with placebo to 11 (p = 0002) with low-dose and 14 (p=0-05) with high-dose therapy. High-dose zidovudine led to anaemia in 29 patients (6-3%) vs 5 patients (1-1%) in the low-dose group. Neutropenia was not observed with the 500 mg dose. In view of these results, the data and safety monitoring board recommended that the trial be stopped in the subgroup of patients with CD4 counts less than 500; these patients were then offered open zidovudine at the low dose. There were some practical problems with this trial. Compliance was poor, about 9% of placebo-group patients may have been on active drug, and 7% of patients were lost to follow-up. Blinding was also poor because clinicians had access to mean corpuscular volume results, which are well known to increase with zidovudine. Nevertheless, it is very likely that the drug, taken over the period of the study, can limit the progression of disease in a few symptomfree individuals. What do these results mean for patients? Clearly, zidovudine should be offered to patients with HIV infection but the study does not tell us when to start treatment. Should zidovudine begin early or be delayed until the disease has progressed? The US recommendations have balanced the proven risk of taking low-dose drug (about 1 % anaemia) with the proven short-term benefit of early treatment (4% of patients avoid early progression). The recommendations therefore make some important assumptions on the probable benefits and lack of risks of early treatment. It may be helpful to consider the likely validity of these assumptions. First, it is assumed that the benefit conferred by zidovudine to 4% of the trial patients over the first 12-24 months of treatment can be extrapolated to the remaining 96% of patients if they continue taking the drug for many years. Inspection of the time-to-disease progression curves suggest that the curves are parallel rather than divergent. There are inevitably very large errors at the end of survival curves because at that stage they are based on small numbers, but it is very
possible that zidovudine delays disease progression for only 7 months rather than halving the rate of progression. The limited effect may be explained by the development of drug resistance, which has already been shown in HIV strains isolated from patients with AIDS after about 6-12 months’ treatment.5 Second, it is assumed that progression to symptomatic disease represents an irreversible step in the course of the disease which determines the patient’s subsequent morbidity and survival. However, it is possible that the effectiveness of zidovudine, if first given after disease progression, is sufficient to reverse this process, thereby rendering
symptomatic patients symptom-free again. Third, little information is available about the subsequent clinical course of patients who progressed despite zidovudine. Patients taking the drug early might be assuming that a new effective antiviral agent will emerge fast enough to become generally available before their own disease progresses. This may not be so. Fourth, it is assumed that the low toxicity observed is maintained with time. Lastly, patients who take zidovudine early will deny themselves the opportunity of using the drug in combination with new antiviral agents later in the disease. Although such treatments have not yet been tested, experience with chemotherapy against tuberculosis and malignancies suggests that only combination therapy will allow long-term suppression of HIV infection without the development of resistance. Volberding et al4 conclude their discussion by stating "The overall benefits of the treatment of early HIV disease with zidovudine must be weighed against potential toxicity and the costs associated with therapy as well as the uncertainly that it will confer a long-term benefit in survival". These sentiments admirably express the aims of all clinical trials. Unfortunately, the results are found wanting. The study merely shows that zidovudine is safe and effective in the few symptom-free patients with CD4 counts under 500 who, if left untreated, were destined to show disease progression within 12-24 months. More information, including details on the reliability of CD4 counts, is required before the drug can be used with confidence in all low-CD4 symptom-free individuals. We hope that the MRC-INSERM Concorde trial and Wellcome trials still in progress will continue long enough to provide the answers. Meanwhile, clinicians should change their prescribing habits with caution. The use of zidovudine in symptomless infection represents prophylaxis rather than treatment; many patients may prefer to be free of drugs unless there is a clear indication of overall clinical benefit. Thus it would not be unreasonable to offer zidovudine only to those patients who are keen to take the drug regularly to avoid early progression of disease, despite the long-term uncertainties of the treatment. There is no evidence to suggest that clinicians should be encouraged to give zidovudine to most symptom-free patients in the hope of short-term benefit to a few.
1. Medical News and Perspectives. Recommendations for zidovudine: early infection. JAMA 1990; 263: 1606-09. 2. Medical News and Perspectives. Controversy continues as experts ponder zidovudine’s role in early HIV infection. JAMA 1990; 263: 1605. 3. Editorial. Concorde remains aloft. Lancet 1989; ii: 1017-18. 4. Volberding PA, Lagakos SW, Koch MA, et al. Zidovudine in asymptomatic human immunodeficiency virus infection. N Engl J Med 1990; 322: 941-49. 5. Larder BA, Darby G, Richman DD. HIV reduced sensitivity to zidovudine (AZT) isolated during prolonged therapy. Science 1989; 243: 1731-34.
The aura phase of a classical migraine attack seems to be mediated by Leao’s spreading depression moving across the cerebral cortex.i Changes in regional cerebral blood flow seen in this type of migraine are probably the consequence rather than the cause of the cortical disturbance. However, the process that initiates this response remains elusive, as does its precise relation to the headache phase. In common migraine headache may occur without a preceding aura, or an aura may never proceed to headache. Whilst Wolff’s hypothesis that the aura is the direct result of cerebral vasospasm is no longer tenable,4the evidence that the headache is in some way related to dilated blood vessels remains far more convincing. In many patients headache is relieved by pressure on the superficial temporal artery 56and ergotamine relieves headache simultaneously with a reduction in arterial
pulse-wave amplitude.’ Vasodilators (eg, nifedipine, indomethacin, and nitroglycerine) will often produce headache.8,9 The pallor of some patients during attacks led some workers to suggest that the painful dilatation often affects the meningeal circulation,6and others that arteriovenous shunting might explain the combination of arterial dilatation and cutaneous capillary constriction.10,11 Swelling of a blood vessel wall might be the mechanism whereby increased pulse-wave amplitude can be accompanied by angiographic narrowing.l2 The findings that cerebral blood flow can fall to ischaemic levels13 and that migraineurs occasionally have cerebral infarcts14,15 are further evidence of vascular involvement in classical
it has become increasingly apparent that the vascular hypothesis cannot explain all the features of migraine-in particular, the unilaterality of both the aura and the headache, and the finding that the sides affected are not consistently related.16 The possibility that the headache may be of neurogenic origin has therefore received considerable attention. The scalp and meningeal arteries, as well as those of the circle of Willis, are supplied by branches of the trigeminal nerve,17 and this seems to be the pathway by which headache is perceived.18,19 Lance and his colleagues2O in Sydney have shown that the circulation in the external carotid artery in monkeys may be