CHAPTER

rek ©2 ah 005 Do / La Co No nde pyri t D s B ght ist i rib osc ute ien

ce

The Epidemiological Consequences of Reducing the Transmission Intensity of P. falciparum Hugh Reyburn* and Chris Drakeley

Abstract

C

ontrol methods which aim to reduce the human host’s exposure to infected mosquitoes (as is the desired outcome of a release of transgenic mosquitoes) will ideally reduce the frequency of malarial infection. This change in transmission intensity will affect the development of immunity to malaria and the age at which infections are acquired. Both of these have important consequences for the overall burden of malarial disease with some suggestion that transmission reduction might not be beneficial in all scenarios. In this chapter we review the methods for measuring malaria transmission and summarise the data on disease epidemiology in these settings.

Introduction

Eu

Transgenic technology applied to malaria control has led to the idea of using genetically modified (GM) mosquitoes able to interrupt malaria transmission. While technological advances are being made in this area1 there remains uncertainty as to how their introduction would affect the human burden of malaria. Such an impact will depend strongly on the efficiency of refractoriness2 and on the preintervention levels of transmission.3 In addition there are fundamental epidemiological questions relating to the level of population immunity and the consequent burden of malaria are common to any intervention designed to reduce exposure to infection and are the focus of this chapter. At high levels of transmission there have been concerns that reducing transmission could result in a rebound in mortality in older age groups as the population level of immunity realigns to new levels of exposure. It has been argued that high rates of infection in early life may stimulate active immunity at a time when other mechanisms such as maternally acquired antibodies and the persistence of foetal haemoglobin reduce the risk of severe disease.4 Other age-dependent effects may also operate, particularly that cerebral manifestations of malaria are relatively more common among older children than infants. These concerns were first raised in the context of the WHO campaign for the eradication of malaria in the 1950s5,6 and have resurfaced with the growing use of insecticide treated nets (ITNs).4 At the other end of the transmission spectrum reductions from low and very low levels may result in a change from ‘stable’ to ‘unstable’ (i.e., epidemic-prone) malaria. There are *Corresponding Author: Hugh Reyburn—Department of Infectious and Tropical Diseases, London School of Hygiene and Tropical Medicine, Keppel Street, London WC1E 7HT, U.K.; Joint Malaria Programme, Moshi, Tanzania. Email: [email protected]

Genetically Modified Mosquitoes for Malaria Control, edited by Christophe Boëte. ©2005 Eurekah.com.

Boete(Reyburn)

1

8/23/05, 12:58 PM

Genetically Modified Mosquitoes for Malaria Control

2

Table 1. Classification of malaria transmission Parasite Prevalence1 (% in children age 2-9 yrs)

EIR2 (infectious bites per person per year)

Endemicity

1-10% 10-50% 50-75% >75%

1-10% 10-50% 50-75% >75%

100

Hypoendemic Mesoendemic Hyperendemic Holoendemic

rek ©2 ah 005 Do / La Co No nde pyri t D s B ght ist i rib osc ute ien

ce

Spleen Rate1 (% palpable spleen in children age 2-9 yrs)

1 Garnham PCC. Malaria parasites and other haemosporidia. Oxford: Blackwell, 1966. 2 Beier JC, Killeen GF, Githure JI. Short report: entomologic inoculation rates and Plasmodium

falciparum malaria prevalence in Africa. Am J Trop Med Hyg 1999; 61(1):109-113.

few studies on which to base conclusions about this scenario but we consider the evidence later in the chapter. The primary problem in resolving these questions, particularly at high transmission, is the difficulty in maintaining a controlled trial of an effective intervention for more than about two years, and this is insufficient time to observe rebound mortality. An alternative approach to prolonged controlled trials is to compare the burden of malaria in populations naturally exposed to different levels of transmission but this method has limitations.7-10 Stable populations living under different levels of challenge have higher frequencies of genes that protect against malaria mortality and parasite diversity and possibly resistance to antimalarials may be greater at high transmission.3 In addition, malaria itself may be the cause and result of economic differences which are likely to affect overall mortality due to its multiple effects on nutrition, access to and quality of care etc.11 Finally, no ideal measure of exposure to Plasmodium falciparum (P. falciparum) currently exists that can reliably characterise the exposure of large populations that are needed if outcomes of severe malaria and mortality are to be compared.12

Measures of Malaria Transmission

Traditional methods of categorising transmission intensity of P. falciparum are summarised in Table 1. Each method has limitations that constrain current knowledge of how the human burden of malaria varies with transmission. However, in the context of GM mosquitoes, a combination of the approaches discussed below could provide improved measures of long-term exposure over large populations.

Entomological

Eu

The gold standard measure for (mosquito-man) transmission intensity is the entomological inoculation rate (EIR). The EIR is classically derived from the density of man-biting anopheline mosquitoes, sporozoite rate and the human blood index and represents the number of infectious bites an individual is likely to be exposed to over a defined period of time usually one year (ib/p/year). The human biting catch (HBC) is the most accurate method for assessing man biting rates though this technique is limited due to ethical and logistical constraints. Light trap catches (LTC) and pyrethrum spray catches (PSC) represent viable alternatives and have been evaluated against HBC.13,14 Obtaining reliable and reproducible estimates of EIR is time consuming (and expensive) and subject to seasonal or meteorological fluctuations. Knowledge of local vectors and their feeding behaviour is clearly vital. A review of EIR in Africa found there to be marked heterogeneity in malaria risk across the continent15 from more than 2 infectious bites per night (884 ib/p/yr) to zero. Most reported

Boete(Reyburn)

2

8/23/05, 12:58 PM

The Epidemiological Consequences of Reducing the Transmission Intensity of P. falciparum

3

studies have yet to consider confidence intervals (CI) around an EIR estimate which in areas of unstable and low malaria transmission are likely to be wide as the chances of catching an infected mosquito are extremely low. Any intervention which aims to reduce the proportion of infected mosquitoes (e.g., genetically modified mosquitoes) must be sufficiently powered to detect a significant reduction in this proportion pre and post intervention.16

Parasitological

Clinical

rek ©2 ah 005 Do / La Co No nde pyri t D s B ght ist i rib osc ute ien

ce

Transmission estimates based on the prevalence of human infection (the parasite rate, PR) are informative as they represent actual (rather than potential) infections and are a direct measure of the disease within the community. In theory, sufficient numbers of people can quickly (and inexpensively) be tested (and retested) to obtain reliable estimates. PR has been used to define malaria endemicity17 and it has been used to classify malaria endemicity and recently has been correlated with EIR.18 However there are similar caveats for PR and the EIR and relationships are not straightforward even in relatively small geographical locations;12,19 in particularly PR is susceptible to micro heterogeneity caused by climatic factors and the socio-economic determinants of health seeking behaviour. A classical correlate of transmission intensity is the measurement of spleen rates in children.20,21 Measurement of anaemia either as direct haemoglobin levels (Hb) or classification into the prevalence of mild or moderate forms has also been shown to correlate with transmission intensity.12 The utility of these measures, Hb in particular, is that they are likely to be more reliable than a single blood film and will reflect longer term trends in exposure to malaria. However, different methods are often used to assess Hb levels and spleen rates are subject to observer variation. There are also potential confounders with other diseases which can cause both splenomegaly and/or anaemia, the latter may also be due to nutrient deficiency.

Serological

Eu

One possible alternative is to combine parasite prevalence with serological measurement of anti-parasite IgM.22 This has been used to demonstrate reduced force of infection in bed net trials and subsequently used to characterise transmission intensity.23 Other studies have shown contrasting results with no change in antibody levels in studies with treated curtains24 and a reduction in antibody recognition with treated nets.25 Similarly, contrasting data have come comparing EIR with responses to merozoite surface protein 1.26,27 Clearly, both the choice of antigen (and the longevity of the antibody response it elicits) and study subjects are important. However the screening of large numbers of samples quickly and the potential for using more immunogenic antigens at lower transmission makes this approach attractive especially when blood taking can be combined with community health surveys.27 One intriguing possibility is the measurement of anti-mosquito saliva antibodies in exposed populations (Mitchell, Drakeley, and Billingsley unpublished data). Total antibody levels may reflect overall exposure and specific isotypes (IgE and IgM) more recent exposure history. In theory, as the proteome of saliva becomes known, recombinant antigens could be used to refine identification to biting of specific species and reductions in vector host contact.

Geographical Information Systems Recent advances in data collection techniques and remote sensing have led to increasingly advanced models for malaria transmission.28-31 These are largely climate driven with additional information used to refine the model for a particular scenario: malaria risk,28 vector distribution,30 epidemic prediction31 and urban malaria.32 However GIS models may lack precision to define transmission at the local scale and a recent analysis suggested that they over-estimate those at no risk.33 Modelling is becoming increasingly important in predicting malaria outbreaks in epidemic prone areas.34

Boete(Reyburn)

3

8/23/05, 12:58 PM

Genetically Modified Mosquitoes for Malaria Control

4

The Effect of Reducing Transmission Intensity in Endemic Areas

ce

Limitations in methodologies for measuring the transmission intensity of P. falciparum mean that transmission is routinely classified into broad bands (Table 1). Existing evidence has used these bands to predict the effects of reducing transmission on uncomplicated and severe malaria and how these are reflected in malaria-specific and all-cause mortality. As will be seen, these effects vary with transmission intensity and may change over time as the population level of immunity adjusts to new levels of exposure; concerns have been expressed that the result may not always be beneficial.4

Nonsevere Malaria

rek ©2 ah 005 Do / La Co No nde pyri t D s B ght ist i rib osc ute ien

Malaria can present with a wide range of nonspecific symptoms, and generally it is not possible to discriminate between malaria and a number of other common illnesses. Non severe malaria is thus defined as a febrile illness with a positive test for malaria parasitaemia and no obvious alternative cause of illness. Since, in endemic areas, asymptomatic parasitaemia is common many cases meeting the above definition are likely to be suffering from a nonmalarial illness, i.e., the definition of nonsevere malaria lacks specificity even when slide results are available. While the inclusion of the density of parasitaemia can provide some refinement to the definition, the problem of specificity remains.35 In controlled trials the use of ITNs has been associated with a reduction in the incidence of nonsevere malaria by almost 50%.36 The sustainability of this reduction can be estimated by comparing stable populations living under different levels of P. falciparum challenge. In children under the age of 18 months naturally exposed to a 10-fold difference in EIR in Tanzania there was a 1.6 fold increase in the incidence of parasitaemia with fever in the high compared to the lower transmission area, suggesting that in infants and young children artificial reductions in transmission intensity are likely to result in a sustained reduction in nonsevere malaria.37 However, this effect may not be sustained among older children and adults. In two closely monitored villages in Senegal (Dielmo and Ndiop) exposed to EIR of 200 and 20 ib/p/yr respectively the age-specific incidence of any clinical episode of malaria was ‘right shifted’ in the low transmission village compared to the high transmission village. The lifetime number of malaria attacks, projected from population and malaria incidence data, was found to be 43 per person by the age of 60 years in the high transmission village and 62 per person by the age 60 in the low transmission village (Fig. 1).38 Similarly in Kenya, Clarke et al found the incidence of such attacks in school children in the Kenyan highlands, an epidemic prone area, to exceed that in schoolchildren in a holoendemic area suggesting that the burden of nonsevere disease may remain comparable at all levels of malaria transmission.39 The problems of accuracy of diagnosis and attributing the cause of febrile episodes make a definitive answer difficult as at high transmission levels the probability that febrile illness in the presence of parasitaemia but which are due to nonmalarial causes increases.35

Severe Malaria

Eu

The definition of severe malaria is based on the presence of P. falciparum parasitaemia, clinical features which have been empirically found to be associated with significant mortality, and the absence of an obvious alternative cause.40 In African children the presence of one or more of 6 ‘syndromes’ capture the large majority of severe and fatal cases of malaria.41,42 These are: 1. Unrousable coma (a lower level of coma called ‘impaired consciousness’ is a supporting criterion) 2. Respiratory distress (abnormally deep breathing associated with in-drawing of the chest wall on inspiration). 3. Severe anaemia (Hb