Editorial: Antimalarial drug resistance and mortality in falciparum malaria

Nobody knows the true mortality of untreated falciparum malaria. It depends largely on age and previous exposure, but even in nonimmune patients mortality is probably less than 5%, and may be much lower ( Sudre et al. 1992 ). The mortality of treated symptomatic drug-sensitive infections is approximately 0.1% ( Meek et al. 1986 ; Luxemburger et al. 1996 ). Antimalarial treatment therefore reduces the chances of dying from falciparum malaria up to 50 fold. Once the infection has progressed to the stage of severe malaria, the treated mortality rises to between 15% and 20% ( Waller et al. 1995 ; Hien et al. 1996 ; van Hensbroek et al. 1996 ; Newton & Krishna 1998). If no antimalarial treatment is given, most authorities consider severe malaria as a usually fatal condition. Thus in the context of severe malaria, mortality is reduced up to fivefold by effective antimalarial treatment. It is evident that both in terms of morbidity and mortality the greatest benefit is obtained by preventing the progression from an uncomplicated to a severe infection with effective antimalarial drugs. The likelihood of progressing from uncomplicated to severe malaria, and ultimately fatal infection, depends not only on the intervention with antimalarial treatment, but also on the background level of immunity and the age of the patient. What happens as antimalarial drug resistance develops? In areas of low malaria transmission people have little or no background immunity, and symptomatic malaria infections are encountered at all ages. Resistance manifests itself as an increase in the rate of recrudescence. Where there is some background immunity, recrudescence rates increase first in children and in pregnant women who have less immunity ( ter Kuile et al. 1995 ; Price et al. 1997 ). As resistance worsens, the mean time to recrudescence shortens, symptom resolution becomes slower, and eventually patients are encountered whose parasitaemia does not clear and whose symptoms do not resolve. In this context quinine, or an artemisinin derivative, are usually used for the treatment of severe malaria. Both are reliably effective. Although there is some evidence of a slowing in the clinical response to quinine in areas of resistance ( Pukrittayakamee et al. 1994 ), there is no evidence yet for an increase in case-specific mortality in severe malaria. But even if effective parenteral treatments for severe malaria are available at health centres and hospital, there will still be an increase in the number of patients who present with severe malaria if ineffective first-line oral treatments are used. Thus morbidity and mortality rise, not because of ineffective treatment of severe malaria, but because of ineffective first-line oral treatment which causes an increasing proportion of patients to develop severe disease. This situation is reflected well by conventional in vivo antimalarial treatment assessments which show high-level resistance (an increasing proportion of early R2 and R3 treatment failures). Persistent use of ineffective treatments may lead to an increase in the overall incidence of falciparum malaria and a change in epidemiology (Nosten et al. unpublished observations). The transition from low to high-level resistance is slow for the quinoline antimalarials (chloroquine, amodiaquine and mefloquine), but may be more rapid for pyrimethamine-sulphonamide combinations, and is particularly abrupt for atovaquone. As adults are affected almost as much, or in some epidemiological contexts, more so than children, the impact of drug resistance on disease severity is obvious if there is surveillance. But areas labelled as having low or unstable transmission are often located in remote regions, and may conceal considerable heterogeneity of transmission of intensities. Surveillance and thus reliable epidemiological information is the exception, not the rule. Patients often cannot reach treatment facilities, self-medication is usual, and the outcome of the infection depends on the efficacy of available drugs. Deaths frequently go unreported and mortality from malaria is underestimated. This undocumented mortality will rise as the available drugs become ineffective. At higher levels of transmission and frequency of infection, the impact of antimalarial drug resistance on mortality is also underestimated. In this context nearly everybody is infected with malaria all the time, but the majority of infections are controlled at a level below that which causes illness. Severe disease is confined to childhood, and, at very high levels of transmission, to the first three years of life ( Marsh et al. 1996 ). Unfortunately in vivo assessments of drug efficacy are often done in older children or sometimes in asymptomatic adults. This provides misleading information as these age groups are likely to self-cure. Many patients would have apparently good responses to treatment, even if no active antimalarial drug were given! Furthermore, as background immunity provides such a large antiparasitic effect, only a small additional boost from a relatively ineffective antimalarial drug may be sufficient to drive the infection below the detection threshold, whereas the same drug-resistant parasites causing an infection in a nonimmune patient could cause an early and potentially dangerous treatment failure. In a high stable transmission area only young children (mainly in the first 3 years of life) would demonstrate such an unsatisfactory response to ineffective treatment, and even within this narrow age-range there may be considerable variation as immunity is acquired rapidly through continuous infection. Worsening antimalarial drug resistance will go unnoticed unless trials are conducted specifically in very young children. Another point needs to be considered. There is considerable variation in the multiplication of Plasmodium falciparum in vivo. This results from intrinsic differences between parasites in their multiplication rates, and their susceptibilities to density-dependent feedback control mechanisms, antimalarial drugs and host immune responses. There is also considerable variation between human hosts in terms of drug absorption and disposition, red cell susceptibility to parasite invasion, the speed and efficiency of induction of host immune defence mechanisms and the reactions to the infecting parasites. This results in significant differences in the severity of illness and the therapeutic responses. In areas of high transmission where infections are very frequent, even an unusual or aberrant response will occur relatively frequently, whereas it will be a rare event in a low-transmission area. In areas of intense transmission heavy parasitaemias are common in the vulnerable age group (i.e. young children), and the difference between a well-controlled and a lethal infection may be a single multiplication cycle. Lethal disease often develops rapidly and unpredictably. Attributing the death of an infant living in a high transmission area to malaria is also notoriously difficult. As resistance develops, death resulting from an inadequate initial treatment response may represent initially only the very tail end of a wide distribution of responses, but because infection is so common, this is not a rare event. In contrast, in low transmission areas the progression from first symptom to death is usually several days, i.e. more than one asexual cycle, and there is more time to detect an inadequate treatment response (WHO 1990). Thus in areas of high transmission an increase in mortality in babies and young children may be the first sign of significant antimalarial drug resistance. The recent detailed longitudinal epidemiological studies from Senegal reported by Trape et al. (1998) indicate a dramatic increase in childhood mortality attributed directly to chloroquine resistance, despite what, by conventional assessment, would be considered ‘moderate’ levels of chloroquine resistance in the African context. In most of the tropical world, the impact of antimalarial drug resistance on malaria mortality is almost certainly underestimated.