THE CHANGING EPIDEMIOLOGY OF MALARIA

Over the past decade there has been unprecedented investment into malaria control and elimination. Increased coverage and access to malaria control measures such as bed nets, indoor residual spraying, preventive treatment and the introduction of the highly efficacious artemisinin drugs has had considerable impact on the malaria burden. Between 2000 and 2013, estimated malaria mortality rates decreased by 47% worldwide (to 584 000 in 2013) predominantly in African children under the age of 5 years (WHO, 2014). The prevalence of Plasmodium spp. infection (symptomatic and asymptomatic) has also decreased with recent analysis showing a relative decline of 48% in average infection prevalence in children aged 2–10 years in sub-Saharan Africa and a 26% reduction overall (WHO, 2014). Since 2000, 55 countries have recorded >75% decrease in case incidence, and while the greatest gains have been observed in areas of high stable transmission in sub-Saharan Africa, impressive gains have also been observed in areas of relatively low transmission areas in Asia (>50% reduction in reported malaria incidence rates between 2000 and 2013); it is in low transmission areas where efforts have focussed on achieving the end goal of malaria elimination.

Geographically, malaria transmission is highest in sub-Saharan Africa and parts of the Pacific (up to 1000 infectious bites/year) and Plasmodium falciparum is the dominant species (Gething et al. Reference Gething, Elyazar, Moyes, Smith, Battle, Guerra, Patil, Tatem, Howes, Myers, George, Horby, Wertheim, Price, Mueller, Baird and Hay2012, Reference Gething, Patil, Smith, Guerra, Elyazar, Johnston, Tatem and Hay2011). In higher transmission areas, malaria is holoendemic and symptomatic disease is confined to young children while older children and adults are typically protected from malaria illness and often asymptomatic parasitemias. Conversely, in Asia and South and Central America, transmission is typically low (⩽1 infectious bites/year), and seasonal and P. falciparum and Plasmodium vivax are often both prevalent (Gething et al. Reference Gething, Elyazar, Moyes, Smith, Battle, Guerra, Patil, Tatem, Howes, Myers, George, Horby, Wertheim, Price, Mueller, Baird and Hay2012, Reference Gething, Patil, Smith, Guerra, Elyazar, Johnston, Tatem and Hay2011). In these low transmission areas, transmission is typically unstable and symptomatic disease occurs in all age groups. The differences in the clinical consequences of Plasmodium spp. infection according to transmission is due to differences in naturally acquired immunity.

Naturally acquired immunity to malaria protects against the development of high density infections and clinical symptoms rather that infection per se (reviewed in Marsh and Kinyanjui, Reference Marsh and Kinyanjui2006). Naturally acquired immunity develops after repeated exposure, with faster rates of acquisition in high compared with low transmission areas. Antibodies are an important component of naturally acquired immunity to malaria as evidenced by experimental animal models and, most importantly, passive transfer studies in which antibodies from malaria-immune adults were successfully used to treat patients with symptomatic malaria (Cohen et al. Reference Cohen, Mc and Carrington1961; Sabchareon et al. Reference Sabchareon, Burnouf, Ouattara, Attanath, Bouharoun-Tayoun, Chantavanich, Foucault, Chongsuphajaisiddhi and Druilhe1991). Serum antibodies, which are made by plasma cells, mediate protection by acting predominantly against parasites of the asexual blood-stages that cause the clinical symptoms of malaria. Blood-stage targets include those expressed by the merozoite stage of P. falciparum and P. vivax, which invades the erythrocyte (specific targets reviewed in Richards and Beeson, Reference Richards and Beeson2009; Mueller et al. Reference Mueller, Galinski, Tsuboi, Arevalo-Herrera, Collins and King2013), as well as variant surface antigens (predominantly P. falciparum erythrocyte membrane protein 1; PfEMP1) expressed on the surface of the P. falciparum-infected erythrocyte (IE) (specific targets reviewed in Chan et al. Reference Chan, Fowkes and Beeson2014) and potentially vir on the surface of P. vivax-IE (reviewed in Mueller et al. Reference Mueller, Galinski, Tsuboi, Arevalo-Herrera, Collins and King2013). The combination of responses as well as breadth and magnitude of blood-stage responses are important in terms of the protective responses, with individuals who possess a greater repertoire are more protected against clinical malaria, although specific targets or patterns of responses are important (Gray et al. Reference Gray, Corran, Mangia, Gaunt, Li, Tetteh, Polley, Conway, Holder, Bacarese-Hamilton, Riley and Crisanti2007; Osier et al. Reference Osier, Fegan, Polley, Murungi, Verra, Tetteh, Lowe, Mwangi, Bull, Thomas, Cavanagh, McBride, Lanar, Mackinnon, Conway and Marsh2008; Crompton et al. Reference Crompton, Kayala, Traore, Kayentao, Ongoiba, Weiss, Molina, Burk, Waisberg, Jasinskas, Tan, Doumbo, Doumtabe, Kone, Narum, Liang, Doumbo, Miller, Doolan, Baldi, Felgner and Pierce2010; Richards et al. Reference Richards, Arumugam, Reiling, Healer, Hodder, Fowkes, Cross, Langer, Takeo, Uboldi, Thompson, Gilson, Coppel, Siba, King, Torii, Chitnis, Narum, Mueller, Crabb, Cowman, Tsuboi and Beeson2013; Rono et al. Reference Rono, Osier, Olsson, Montgomery, Mhoja, Rooth, Marsh and Farnert2013). Antibodies to blood-stages can act by directly blocking merozoite invasion and inhibiting growth (Cohen et al. Reference Cohen, Butcher and Crandall1969; Brown et al. Reference Brown, Anders, Mitchell and Heywood1982; Guevara Patino et al. Reference Guevara Patino, Holder, McBride and Blackman1997; O'Donnell et al. Reference O'Donnell, de Koning-Ward, Burt, Bockarie, Reeder, Cowman and Crabb2001; Singh et al. Reference Singh, Miura, Zhou, Muratova, Keegan, Miles, Martin, Saul, Miller and Long2006; Dutta et al. Reference Dutta, Sullivan, Grady, Haynes, Komisar, Batchelor, Soisson, Diggs, Heppner, Lanar, Collins and Barnwell2009; Duncan et al. Reference Duncan, Hill and Ellis2012), acting together with complement to inhibit invasion and lyse merozoites (Boyle et al. Reference Boyle, Reiling, Feng, Langer, Osier, Aspeling-Jones, Cheng, Stubbs, Tetteh, Conway, McCarthy, Muller, Marsh, Anders and Beeson2015), and by clearing merozoites and P. falciparum-IE by antibody dependent cellular mechanisms (Khusmith and Druilhe, Reference Khusmith and Druilhe1983; Druilhe and Perignon, Reference Druilhe and Perignon1994) and opsonic phagocytosis (Celada et al. Reference Celada, Cruchaud and Perrin1982; Hill et al. Reference Hill, Eriksson, Li Wai Suen, Chiu, Ryg-Cornejo, Robinson, Siba, Mueller, Hansen and Schofield2013; Chan et al. Reference Chan, Fowkes and Beeson2014; Osier et al. Reference Osier, Feng, Boyle, Langer, Zhou, Richards, McCallum, Reiling, Jaworowski, Anders, Marsh and Beeson2014a), thereby reducing parasite density and clinical symptoms. Additional Plasmodium spp. targets include the pre-erythrocytic sporozoite stage (specific targets reviewed in Dups et al. Reference Dups, Pepper and Cockburn2014) and sexual gametocyte stage (Riley et al. Reference Riley, Williamson, Greenwood and Kaslow1995; Milek et al. Reference Milek, Roeffen, Kocken, Jansen, Kaan, Eling, Sauerwein and Konings1998; Chan et al. Reference Chan, Fowkes and Beeson2014). Antibodies to sporozoites would be expected to prevent infection; however, available evidence suggests that the acquisition of protective pre-erythrocytic immunity may be limited (Hoffman et al. Reference Hoffman, Oster, Plowe, Woollett, Beier, Chulay, Wirtz, Hollingdale and Mugambi1987; Webster et al. Reference Webster, Brown, Chuenchitra, Permpanich and Pipithkul1988; Wongsrichanalai et al. Reference Wongsrichanalai, Webster, Permpanich, Chuanak and Ketrangsri1991; Michon et al. Reference Michon, Cole-Tobian, Dabod, Schoepflin, Igu, Susapu, Tarongka, Zimmerman, Reeder, Beeson, Schofield, King and Mueller2007; Tran et al. Reference Tran, Li, Doumbo, Doumtabe, Huang, Dia, Bathily, Sangala, Kone, Traore, Niangaly, Dara, Kayentao, Ongoiba, Doumbo, Traore and Crompton2013). Antibodies to gametocytes are thought to prevent the infectious spread between human and mosquito vectors (Healer et al. Reference Healer, McGuinness, Hopcroft, Haley, Carter and Riley1997; Bousema et al. Reference Bousema, Drakeley and Sauerwein2006, Reference Bousema, Roeffen, Meijerink, Mwerinde, Mwakalinga, van Gemert, van de Vegte-Bolmer, Mosha, Targett, Riley, Sauerwein and Drakeley2010b).

In this era of increased malaria control and goals of malaria elimination, many endemic areas are transitioning from high-to-low-to-no malaria transmission (e.g. O'Meara et al. Reference O'Meara, Bejon, Mwangi, Okiro, Peshu, Snow, Newton and Marsh2008a, Reference O'Meara, Mwangi, Williams, McKenzie, Snow and Marshb; O'Meara et al. Reference O'Meara, Mangeni, Steketee and Greenwood2010; Kalayjian et al. Reference Kalayjian, Malhotra, Mungai, Holding and King2013; Snow et al. Reference Snow, Kibuchi, Karuri, Sang, Gitonga, Mwandawiro, Bejon and Noor2015) and the impact on immunity is evident epidemiologically; rebounds of malaria in previously eliminated areas and shifts in case distribution to older age groups in areas where malaria incidence is decreasing (Ceesay et al. Reference Ceesay, Casals-Pascual, Erskine, Anya, Duah, Fulford, Sesay, Abubakar, Dunyo, Sey, Palmer, Fofana, Corrah, Bojang, Whittle, Greenwood and Conway2008; Brasseur et al. Reference Brasseur, Badiane, Cisse, Agnamey, Vaillant and Olliaro2011). To fully understand the impact of changing transmission on immunity, a greater knowledge of the development and maintenance of naturally acquired immunity to malaria immunity is required. Here we review the key processes that underpin this knowledge; the amount of Plasmodium spp. exposure required to generate effective immune responses, the longevity of antibody responses and the ability to mount an effective response upon re-exposure through memory responses. This review will focus largely on humoral immunity, since knowledge on longevity of T-cell responses and the potential impact of declining transmission on these responses has been greatly under-studied.

HOW MUCH PLASMODIUM SPP. EXPOSURE IS REQUIRED TO BECOME IMMUNE?

The acquisition of immunity to malaria in populations is evident by the declining incidence of uncomplicated and severe malaria with increasing age, and a reduction in average malaria parasite density with increasing age (reviewed in Marsh and Kinyanjui, Reference Marsh and Kinyanjui2006). It is believed that immunity that reduces the risk of severe and life-threatening malaria is acquired more quickly than robust immunity that protects against all forms of malaria, including uncomplicated illness (reviewed in Marsh and Kinyanjui, Reference Marsh and Kinyanjui2006). This paradigm of the acquisition of immunity is supported by epidemiologic evidence and modelling studies (e.g. Snow et al. Reference Carneiro, Roca-Feltrer, Griffin, Smith, Tanner, Schellenberg, Greenwood and Schellenberg1997; Gupta et al. Reference Gupta, Snow, Donnelly, Marsh and Newbold1999; Reyburn et al. Reference Reyburn, Mbatia, Drakeley, Bruce, Carneiro, Olomi, Cox, Nkya, Lemnge, Greenwood and Riley2005; Michon et al. Reference Michon, Cole-Tobian, Dabod, Schoepflin, Igu, Susapu, Tarongka, Zimmerman, Reeder, Beeson, Schofield, King and Mueller2007; Carneiro et al. Reference Carneiro, Roca-Feltrer, Griffin, Smith, Tanner, Schellenberg, Greenwood and Schellenberg2010; Griffin et al. Reference Griffin, Hollingsworth, Reyburn, Drakeley, Riley and Ghani2015). However, the rate at which immunity to severe malaria is acquired relative to broader immunity, and the extent to which this is influenced by the intensity and nature of malaria transmission and host factors is not entirely clear. More rapid acquisition of immunity to severe malaria may be because some level of immunity is sufficient to prevent severe disease, or there may be specific immune mechanisms mediating protection against severe disease; studies also suggest that host-age is an important factor in susceptibility to severe malaria (e.g. Reyburn et al. Reference Griffin, Hollingsworth, Reyburn, Drakeley, Riley and Ghani2005; Griffin et al. Reference Griffin, Hollingsworth, Reyburn, Drakeley, Riley and Ghani2015). In populations exposed to stable malaria transmission of medium-high intensity, malaria is typically uncommon in older children and adults, and severe malaria is rare (reviewed in Marsh and Kinyanjui, Reference Marsh and Kinyanjui2006; Carneiro et al. Reference Carneiro, Roca-Feltrer, Griffin, Smith, Tanner, Schellenberg, Greenwood and Schellenberg2010). In settings of high transmission, severe malaria is largely restricted to young children (under 5 years), and severe malaria continues to occur later in childhood in settings where malaria transmission is lower (Snow et al. Reference Snow, Omumbo, Lowe, Molyneux, Obiero, Palmer, Weber, Pinder, Nahlen, Obonyo, Newbold, Gupta and Marsh1997; Carneiro et al. Reference Carneiro, Roca-Feltrer, Griffin, Smith, Tanner, Schellenberg, Greenwood and Schellenberg2010). Reductions in malaria transmission would be expected to shift the peak incidence of severe malaria to later in childhood or adulthood (Griffin et al. Reference Griffin, Ferguson and Ghani2014), and in populations where malaria transmission is low and unstable people may remain at risk of severe malaria throughout their life.

Interestingly, epidemiological studies of immunity where P. falciparum and P. vivax are co-endemic have suggested that the rate of acquisition of immunity to P. vivax is faster than for P. falciparum (Maitland et al. Reference Maitland, Williams, Bennett, Newbold, Peto, Viji, Timothy, Clegg, Weatherall and Bowden1996; Bruce et al. Reference Bruce, Donnelly, Packer, Lagog, Gibson, Narara, Walliker, Alpers and Day2000b; Michon et al. Reference Michon, Cole-Tobian, Dabod, Schoepflin, Igu, Susapu, Tarongka, Zimmerman, Reeder, Beeson, Schofield, King and Mueller2007; Lin et al. Reference Lin, Kiniboro, Gray, Dobbie, Robinson, Laumaea, Schopflin, Stanisic, Betuela, Blood-Zikursh, Siba, Felger, Schofield, Zimmerman and Mueller2010). Immunity to P. vivax malaria is evident at a younger age compared with P. falciparum in populations exposed to similar levels of transmission of the two species. Recent studies have suggested that this may be explained by a higher force-of-infection for P. vivax, compared with P. falciparum, for a given entomologic inoculation rate, due to the ability of P. vivax to cause relapses from dormant hypnozoites (Koepfli et al. Reference Koepfli, Colborn, Kiniboro, Lin, Speed, Siba, Felger and Mueller2013). It is possible that there are also underlying differences in the acquisition and nature of protective responses to the two species, but these are yet to be defined and immunity to P. vivax malaria has been greatly under-studied.

Currently there is not a clear understanding of how much exposure is required for the development of immunity. In broad terms, the age at which effective immunity is evident is related to the level of transmission, such that substantial immunity may be acquired by age 10 in areas of medium-high transmission, whereas immunity may not be acquired until teenage years or early adulthood in areas with lower transmission (e.g. Mwangi et al. Reference Mwangi, Ross, Snow and Marsh2005; Marsh and Kinyanjui, Reference Marsh and Kinyanjui2006; Carneiro et al. Reference Carneiro, Roca-Feltrer, Griffin, Smith, Tanner, Schellenberg, Greenwood and Schellenberg2010). While this pattern of acquisition has been reported from many settings, there are some exceptions to this general picture and the development of protective immunity after relatively few infections. For example, studies of transmigrants moving from malaria-free areas to malaria-endemic regions of Indonesia suggested that effective immunity against symptomatic P. falciparum malaria could be acquired within 2 years and was dependent on a threshold number of episodes over that time, and was acquired more rapidly among adults than children (Baird et al. Reference Baird, Jones, Danudirgo, Annis, Bangs, Basri, Purnomo and Masbar1991, Reference Baird, Purnomo, Basri, Bangs, Andersen, Jones, Masbar, Harjosuwarno, Subianto and Arbani1993). Repeated exposure does result in more clinical immunity as studies have shown that within the same region, individuals repeatedly exposed have lower parasite densities and less frequent clinical episodes than less exposed individuals (Thomas and Lindsay, Reference Thomas and Lindsay2000; Bejon et al. Reference Bejon, Warimwe, Mackintosh, Mackinnon, Kinyanjui, Musyoki, Bull and Marsh2009, Reference Bejon, Williams, Nyundo, Hay, Benz, Gething, Otiende, Peshu, Bashraheil, Greenhouse, Bousema, Bauni, Marsh, Smith and Borrmann2014; Mosha et al. Reference Mosha, Sturrock, Greenhouse, Greenwood, Sutherland, Gadalla, Atwal, Drakeley, Kibiki, Bousema, Chandramohan and Gosling2013; Ndungu et al. Reference Ndungu, Marsh, Fegan, Wambua, Nyangweso, Ogada, Mwangi, Nyundo, Macharia, Uyoga, Williams and Bejon2015). In many regions of low transmission, the extent and frequency of exposure may be low to result in effective immunity, such that adults may remain at risk of symptomatic and severe malaria (e.g. Creasey et al. Reference Creasey, Giha, Hamad, El Hassan, Theander and Arnot2004; Reyburn et al. Reference Reyburn, Mbatia, Drakeley, Bruce, Carneiro, Olomi, Cox, Nkya, Lemnge, Greenwood and Riley2005), which is particularly relevant in the context of intensified malaria control efforts and declining malaria globally. The implication from these observations is that extensive exposure to blood-stage infection (e.g. dozens of episodes, or persistent low-grade blood-stage infection) is required for the development of protective immunity. It has also been proposed that chronic parasitemia, or repeated and frequent parasitemia is required to maintain robust immunity (a phenomenon referred to as premunition). Infections can persist for many months in the absence of symptoms (Bruce et al. Reference Bruce, Donnelly, Alpers, Galinski, Barnwell, Walliker and Day2000a; Franks et al. Reference Franks, Koram, Wagner, Tetteh, McGuinness, Wheeler, Nkrumah, Ranford-Cartwright and Riley2001; Njama-Meya et al. Reference Njama-Meya, Kamya and Dorsey2004; Nsobya et al. Reference Nsobya, Parikh, Kironde, Lubega, Kamya, Rosenthal and Dorsey2004) and asymptomatic infections have been reported to provide protection against symptomatic disease (Farnert et al. Reference Farnert, Rooth, Svensson, Snounou and Bjorkman1999). However, several studies have shown that asymptomatic infections predict symptomatic disease and teasing out the effects of exposure and protective immunity have been challenging (Njama-Meya et al. Reference Njama-Meya, Kamya and Dorsey2004; Bejon et al. Reference Bejon, Williams, Liljander, Noor, Wambua, Ogada, Olotu, Osier, Hay, Farnert and Marsh2010; Greenhouse et al. Reference Greenhouse, Ho, Hubbard, Njama-Meya, Narum, Lanar, Dutta, Rosenthal, Dorsey and John2011; Liljander et al. Reference Liljander, Bejon, Mwacharo, Kai, Ogada, Peshu, Marsh and Farnert2011; Loucoubar et al. Reference Loucoubar, Grange, Paul, Huret, Tall, Telle, Roussilhon, Faye, Diene-Sarr, Trape, Mercereau-Puijalon, Sakuntabhai and Bureau2013). Some longitudinal data supports the notion that asymptomatic infections may be important in maintaining antibody responses (Shekalaghe et al. Reference Shekalaghe, Alifrangis, Mwanziva, Enevold, Mwakalinga, Mkali, Kavishe, Manjurano, Sauerwein, Drakeley and Bousema2009; Fowkes et al. Reference Fowkes, McGready, Cross, Hommel, Simpson, Elliott, Richards, Lackovic, Viladpai-Nguen, Narum, Tsuboi, Anders, Nosten and Beeson2012; Ibison et al. Reference Ibison, Olotu, Muema, Mwacharo, Ohuma, Kimani, Marsh, Bejon and Ndungu2012; Proietti et al. Reference Proietti, Verra, Bretscher, Stone, Kanoi, Balikagala, Egwang, Corran, Ronca, Arca, Riley, Crisanti, Drakeley and Bousema2013; Daou et al. Reference Daou, Kouriba, Ouedraogo, Diarra, Arama, Keita, Sissoko, Ouologuem, Arama, Bousema, Doumbo, Sauerwein and Scholzen2015; Rono et al. Reference Rono, Farnert, Murungi, Ojal, Kamuyu, Guleid, Nyangweso, Wambua, Kitsao, Olotu, Marsh and Osier2015), but this has not been extensively studied or clearly established. Removal of these antigenic stimuli may have significant effects on the maintenance of immunity in malaria endemic populations (detailed below).

Acquisition of immunity is influenced by multiple factors, which are reflected in different rates of immune acquisition reported across various populations. Clearly the extent and frequency of exposure is important (reviewed in Doolan et al. Reference Doolan, Dobano and Baird2009), but immunity is not simply determined by the total number of episodes of infections, or the cumulative exposure to blood-stage infection. Host and parasite factors may influence acquisition of immunity. Specific genetic traits influence susceptibility to malaria and potentially immune responses (Edozien et al. Reference Edozien, Boyo and Morley1960; Marsh et al. Reference Marsh, Otoo, Hayes, Carson and Greenwood1989; Cabrera et al. Reference Cabrera, Cot, Migot-Nabias, Kremsner, Deloron and Luty2005; Verra et al. Reference Verra, Simpore, Warimwe, Tetteh, Howard, Osier, Bancone, Avellino, Blot, Fegan, Bull, Williams, Conway, Marsh and Modiano2007), and the prevalence of these traits vary substantially between populations (Howes et al. Reference Howes, Patil, Piel, Nyangiri, Kabaria, Gething, Zimmerman, Barnadas, Beall, Gebremedhin, Menard, Williams, Weatherall and Hay2011, Reference Howes, Piel, Patil, Nyangiri, Gething, Dewi, Hogg, Battle, Padilla, Baird and Hay2012; Reference Howes, Dewi, Piel, Monteiro, Battle, Messina, Sakuntabhai, Satyagraha, Williams, Baird and Hay2013; Piel et al. Reference Piel, Howes, Patil, Nyangiri, Gething, Bhatt, Williams, Weatherall and Hay2013a, Reference Piel, Patil, Howes, Nyangiri, Gething, Dewi, Temperley, Williams, Weatherall and Hayb, Reference Piel, Patil, Howes, Nyangiri, Gething, Williams, Weatherall and Hay2010). Parasite diversity is also important and immunity may be acquired more quickly where genetic diversity is limited, since many of the key targets of protective immunity are polymorphic. Reduced parasite population genetic diversity may be a consequence of intensified malaria control activities with implications for development of immunity (Gray et al. Reference Gray, Dowd, Bain, Bobogare, Wini, Shanks and Cheng2013; Kaneko et al. Reference Kaneko, Chaves, Taleo, Kalkoa, Isozumi, Wickremasinghe, Perlmann, Takeo, Tsuboi, Tachibana, Kimura, Bjorkman, Troye-Blomberg, Tanabe and Drakeley2014). Significantly, most studies of immune acquisition have studied people living in malaria-endemic regions since birth, but the impact of a shift upwards in the age of first exposure with declining transmission on the acquisition of immunity is unclear. Firstly, infection of mothers during pregnancy can influence immune responses in infants (Desowitz, Reference Desowitz1988; Desowitz et al. Reference Desowitz, Elm and Alpers1992; King et al. Reference King, Malhotra, Wamachi, Kioko, Mungai, Wahab, Koech, Zimmerman, Ouma and Kazura2002; Malhotra et al. Reference Malhotra, Dent, Mungai, Wamachi, Ouma, Narum, Muchiri, Tisch and King2009) and the development of immunity in the absence of in utero exposure is unknown. Secondly, the acquisition of immunity may vary according to age of first exposure. Early studies in transmigrants demonstrated that antibody responses are acquired more quickly in adults than children (Baird et al. Reference Baird, Jones, Danudirgo, Annis, Bangs, Basri, Purnomo and Masbar1991, Reference Baird, Purnomo, Basri, Bangs, Andersen, Jones, Masbar, Harjosuwarno, Subianto and Arbani1993), and analyses have suggested that age is an important factor influencing susceptibility to severe malaria (Carneiro et al. Reference Carneiro, Roca-Feltrer, Griffin, Smith, Tanner, Schellenberg, Greenwood and Schellenberg2010; Griffin et al. Reference Griffin, Hollingsworth, Reyburn, Drakeley, Riley and Ghani2015). However, whether this relates to the nature of the immune response is not presently known. Conversely, studies in malaria endemic areas of Africa have suggested that infants acquire immunity faster than older children (Aponte et al. Reference Aponte, Menendez, Schellenberg, Kahigwa, Mshinda, Vountasou, Tanner and Alonso2007), however other studies in the region have demonstrated no association between the rate of antibody acquisition and age of first exposure (Guinovart et al. Reference Guinovart, Dobano, Bassat, Nhabomba, Quinto, Manaca, Aguilar, Rodriguez, Barbosa, Aponte, Mayor, Renom, Moraleda, Roberts, Schwarzer, Le Souef, Schofield, Chitnis, Doolan and Alonso2012; Moncunill et al. Reference Moncunill, Mayor, Jimenez, Nhabomba, Puyol, Manaca, Barrios, Cistero, Guinovart, Aguilar, Bardaji, Pinazo, Angov, Dutta, Chitnis, Munoz, Gascon and Dobano2013b). The effect of the age shift with declining transmission on the development of immunity across the age spectra warrants further investigation.

The challenges in predicting and measuring immunity in populations highlights the need for immune correlates or biomarkers of immunity that would enable the immune status of populations to be monitored to evaluate the impact of interventions and identify populations or sub-groups at risk. With intensified control activities and reducing malaria in many regions, research to address this needs to be a high priority. The utility of serosurveillance has been assessed in several studies using enzyme-linked immunosorbent assay (ELISA) and microarray assays (Drakeley et al. Reference Drakeley, Corran, Coleman, Tongren, McDonald, Carneiro, Malima, Lusingu, Manjurano, Nkya, Lemnge, Cox, Reyburn and Riley2005; Satoguina et al. Reference Satoguina, Walther, Drakeley, Nwakanma, Oriero, Correa, Corran, Conway and Walther2009; Stewart et al. Reference Stewart, Gosling, Griffin, Gesase, Campo, Hashim, Masika, Mosha, Bousema, Shekalaghe, Cook, Corran, Ghani, Riley and Drakeley2009; Bousema et al. Reference Bousema, Drakeley, Gesase, Hashim, Magesa, Mosha, Otieno, Carneiro, Cox, Msuya, Kleinschmidt, Maxwell, Greenwood, Riley, Sauerwein, Chandramohan and Gosling2010a; Cook et al. Reference Cook, Kleinschmidt, Schwabe, Nseng, Bousema, Corran, Riley and Drakeley2011, Reference Cook, Reid, Iavro, Kuwahata, Taleo, Clements, McCarthy, Vallely and Drakeley2010; Crompton et al. Reference Crompton, Kayala, Traore, Kayentao, Ongoiba, Weiss, Molina, Burk, Waisberg, Jasinskas, Tan, Doumbo, Doumtabe, Kone, Narum, Liang, Doumbo, Miller, Doolan, Baldi, Felgner and Pierce2010; Badu et al. Reference Badu, Afrane, Larbi, Stewart, Waitumbi, Angov, Ong'echa, Perkins, Zhou, Githeko and Yan2012; Elliott et al. Reference Elliott, Fowkes, Richards, Reiling, Drew and Beeson2014). Simple immunoassays may be broadly informative, but the use of simplified and standardized functional assays may be more indicative of immunity. A growing body of data supports the utility of antibodies to merozoite antigens as biomarkers of immunity (Osier et al. Reference Osier, Fegan, Polley, Murungi, Verra, Tetteh, Lowe, Mwangi, Bull, Thomas, Cavanagh, McBride, Lanar, Mackinnon, Conway and Marsh2008, Reference Osier, Mackinnon, Crosnier, Fegan, Kamuyu, Wanaguru, Ogada, McDade, Rayner, Wright and Marsh2014b; Fowkes et al. Reference Fowkes, Richards, Simpson and Beeson2010; Richards et al. Reference Richards, Arumugam, Reiling, Healer, Hodder, Fowkes, Cross, Langer, Takeo, Uboldi, Thompson, Gilson, Coppel, Siba, King, Torii, Chitnis, Narum, Mueller, Crabb, Cowman, Tsuboi and Beeson2013; Cutts et al. Reference Cutts, Powell, Agius, Beeson, Simpson and Fowkes2014), and recent studies have reported that opsonic phagocytosis and complement fixation with anti-merozoite responses may be valuable functional assays (Hill et al. Reference Hill, Eriksson, Li Wai Suen, Chiu, Ryg-Cornejo, Robinson, Siba, Mueller, Hansen and Schofield2013; Osier et al. Reference Osier, Feng, Boyle, Langer, Zhou, Richards, McCallum, Reiling, Jaworowski, Anders, Marsh and Beeson2014a; Boyle et al. Reference Boyle, Reiling, Feng, Langer, Osier, Aspeling-Jones, Cheng, Stubbs, Tetteh, Conway, McCarthy, Muller, Marsh, Anders and Beeson2015). Recent longitudinal studies in Kenya and Papua New Guinea, comparing children with different levels of malaria exposure, support the hypothesis that there is a threshold level of immunity required to mediate protection from malaria (Murungi et al. Reference Murungi, Kamuyu, Lowe, Bejon, Theisen, Kinyanjui, Marsh and Osier2013; Stanisic et al. Reference Stanisic, Fowkes, Koinari, Javati, Lin, Kiniboro, Richards, Robinson, Schofield, Kazura, King, Zimmerman, Felger, Siba, Mueller and Beeson2015). In these studies older children, or children from higher transmission settings, had substantially higher antibodies and significant clinical immunity compared with young children who had low levels of antibodies which were not associated with protection against from malaria (Murungi et al. Reference Murungi, Kamuyu, Lowe, Bejon, Theisen, Kinyanjui, Marsh and Osier2013; Stanisic et al. Reference Stanisic, Fowkes, Koinari, Javati, Lin, Kiniboro, Richards, Robinson, Schofield, Kazura, King, Zimmerman, Felger, Siba, Mueller and Beeson2015). These approaches may be valuable in defining threshold antibody levels that act as markers of protective immunity. Determining what constitutes a protective immune response, together with factors required to generate it (number of exposures, antigenic diversity), is critical in understanding how changes in malaria transmission will impact on the acquisition of naturally acquired immunity to malaria.

HOW LONG DO PLASMODIUM SPP. ANTIBODY RESPONSES LAST?

Once immunity is acquired the next question is how long does it last? Investigations into the longevity of antibody responses have spanned almost 50 years and have included studies in malaria endemic areas and studies in previously immune immigrants and neurosyphilis patients, which offer a classic experimental approach to analyse antibody longevity in the absence of intermittent exposure to infection. The question of the longevity of antibody responses is hotly debated, with evidence for and against short- and long-lived antibody responses. It became dogma that antibody responses were relatively short-lived from a number of studies in malaria-endemic areas; P. falciparum and P. vivax antibody responses (sporozoites, merozoites, P. falciparum-IE and gametocytes) have been shown to rapidly decline after a few months following drug treatment and parasite clearance in subjects with symptomatic infections in both children and adults (Cavanagh et al. Reference Cavanagh, Elhassan, Roper, Robinson, Giha, Holder, Hviid, Theander, Arnot and McBride1998; Fonjungo et al. Reference Fonjungo, Elhassan, Cavanagh, Theander, Hviid, Roper, Arnot and McBride1999; Giha et al. Reference Giha, Staalsoe, Dodoo, Elhassan, Roper, Satti, Arnot, Theander and Hviid1999; Soares et al. Reference Soares, da Cunha, Silva, Souza, Del Portillo and Rodrigues1999; Kinyanjui et al. Reference Kinyanjui, Conway, Lanar and Marsh2007; Weiss et al. Reference Weiss, Traore, Kayentao, Ongoiba, Doumbo, Doumtabe, Kone, Dia, Guindo, Traore, Huang, Miura, Mircetic, Li, Baughman, Narum, Miller, Doumbo, Pierce and Crompton2010; Bousema et al. Reference Bousema, Roeffen, Meijerink, Mwerinde, Mwakalinga, van Gemert, van de Vegte-Bolmer, Mosha, Targett, Riley, Sauerwein and Drakeley2010b) and rapidly declined over the dry season in areas of seasonal P. falciparum transmission (Fruh et al. Reference Fruh, Doumbo, Muller, Koita, McBride, Crisanti, Toure and Bujard1991; Ramasamy et al. Reference Ramasamy, Nagendran and Ramasamy1994; Cavanagh et al. Reference Cavanagh, Elhassan, Roper, Robinson, Giha, Holder, Hviid, Theander, Arnot and McBride1998; Perraut et al. Reference Perraut, Mercereau-Puijalon, Diouf, Tall, Guillotte, Le Scanf, Trape, Spiegel and Garraud2000; Akpogheneta et al. Reference Akpogheneta, Duah, Tetteh, Dunyo, Lanar, Pinder and Conway2008; Weiss et al. Reference Weiss, Traore, Kayentao, Ongoiba, Doumbo, Doumtabe, Kone, Dia, Guindo, Traore, Huang, Miura, Mircetic, Li, Baughman, Narum, Miller, Doumbo, Pierce and Crompton2010). These studies suggest that regular exposure is needed for maintenance of antibodies (premunition) and, in the absence of exposure, antibody responses are relatively short-lived. Indeed in the context of declining transmission, P. falciparum and P. vivax antibodies have also been shown to decline with the implementation of malaria control interventions such as chemoprophylaxis, indoor residual spraying and bednets (Warren et al. Reference Warren, Collins, Jeffery and Skinner1983; Staalsoe et al. Reference Staalsoe, Shulman, Dorman, Kawuondo, Marsh and Hviid2004; Aitken et al. Reference Aitken, Mbewe, Luntamo, Maleta, Kulmala, Friso, Fowkes, Beeson, Ashorn and Rogerson2010, Reference Aitken, Mbewe, Luntamo, Kulmala, Beeson, Ashorn and Rogerson2012; Cook et al. Reference Cook, Kleinschmidt, Schwabe, Nseng, Bousema, Corran, Riley and Drakeley2011; Diop et al. Reference Diop, Richard, Diouf, Sokhna, Diagne, Trape, Faye, Tall, Diop and Balde2014) and antibody levels have reflected declines in malaria transmission in studies spanning several years (Migot et al. Reference Migot, Chougnet, Raharimalala, Astagneau, Lepers and Deloron1993; Ceesay et al. Reference Ceesay, Casals-Pascual, Nwakanma, Walther, Gomez-Escobar, Fulford, Takem, Nogaro, Bojang, Corrah, Jaye, Taal, Sonko and Conway2010; Diop et al. Reference Diop, Richard, Diouf, Sokhna, Diagne, Trape, Faye, Tall, Diop and Balde2014).

However, it is important to note that while the afore-mentioned studies show antibody decline post Plasmodium spp. exposure, antibodies still largely persist, and other studies provide evidence for long-lived antibody responses. Stable antibody responses have been shown over a 4 year period in an area of seasonal P. falciparum transmission (Taylor et al. Reference Taylor, Egan, McGuinness, Jepson, Adair, Drakely and Riley1996), and in areas of low P. falciparum and P. vivax transmission with little/no detectable infections over several months and years (up to 6 years) (Warren et al. Reference Warren, Collins, Skinner and Larin1975; Chougnet et al. Reference Chougnet, Deloron, Lepers, Tallet, Rason, Astagneau, Savel and Coulanges1990; Migot et al. Reference Migot, Chougnet, Raharimalala, Astagneau, Lepers and Deloron1993; Torres et al. Reference Torres, Clark, Hernandez, Soto-Cornejo, Gamboa and Branch2008; Bousema et al. Reference Bousema, Youssef, Cook, Cox, Alegana, Amran, Noor, Snow and Drakeley2010c; Wipasa et al. Reference Wipasa, Suphavilai, Okell, Cook, Corran, Thaikla, Liewsaree, Riley and Hafalla2010; Clark et al. Reference Clark, Silva, Weiss, Li, Padilla, Crompton, Hernandez and Branch2012; Fowkes et al. Reference Fowkes, McGready, Cross, Hommel, Simpson, Elliott, Richards, Lackovic, Viladpai-Nguen, Narum, Tsuboi, Anders, Nosten and Beeson2012; Ayieko et al. Reference Ayieko, Maue, Jura, Noland, Ayodo, Rochford and John2013). Persistent life-long P. falciparum merozoite responses have been reported in population seroconversion studies (antibody response half-life 50 years, 95% CI 36, 73) (Drakeley et al. Reference Drakeley, Corran, Coleman, Tongren, McDonald, Carneiro, Malima, Lusingu, Manjurano, Nkya, Lemnge, Cox, Reyburn and Riley2005). In another instance, long-lived P. vivax merozoite antibodies have been reported more than 30 years after malaria elimination (Lim et al. Reference Lim, Park, Yeom, Lee, Yoo, Oh, Sohn, Bahk and Kim2004), although with P. vivax it is hard to separate out the effect of relapses. Studies of VAR2CSA have suggested antibodies may persist for decades (Fowkes et al. Reference Fowkes, McGready, Cross, Hommel, Simpson, Elliott, Richards, Lackovic, Viladpai-Nguen, Narum, Tsuboi, Anders, Nosten and Beeson2012), and VAR2CSA antibodies have been detected 20 years after the last pregnancy (Ampomah et al. Reference Ampomah, Stevenson, Ofori, Barfod and Hviid2014a). Long-lived responses may explain epidemiological observations of long-lived protection against clinical disease in Madagascan adults during malaria epidemics, despite having been exposed to malaria 30 years previously (Deloron and Chougnet, Reference Deloron and Chougnet1992; Kleinschmidt and Sharp, Reference Kleinschmidt and Sharp2001).

Long-lived responses are also supported in studies of previously immune migrants and neurosyphylis patients (who are infected with malaria as a form of treatment), which offer a classic experimental approach to analyse antibody longevity in the absence of intermittent exposure to infection. Early serological investigations showed detectable P. falciparum and P. vivax antibodies in African and Central Asian immigrants and neurosyphilis patients 3–15 years post exposure (Collins et al. Reference Collins, Skinner and Jeffery1968; Bruce-Chwatt et al. Reference Bruce-Chwatt, Dodge, Draper, Topley and Voller1972; Druilhe et al. Reference Druilhe, Pradier, Marc, Miltgen, Mazier and Parent1986). Responses in immigrants appear not to wane completely, with several studies showing that antibody levels are independent of length of residence, which has ranged from 1 to 4 years to almost 4 decades (Bouchaud et al. Reference Bouchaud, Cot, Kony, Durand, Schiemann, Ralaimazava, Coulaud, Le Bras and Deloron2005; Moncunill et al. Reference Moncunill, Mayor, Jimenez, Nhabomba, Casas-Vila, Puyol, Campo, Manaca, Aguilar, Pinazo, Almirall, Soler, Munoz, Bardaji, Angov, Dutta, Chitnis, Alonso, Gascon and Dobano2013a). These studies show that antibody responses can be long-lived in the absence of exposure, and may still provide a degree of protective immunity; immigrants have milder P. falciparum and P. vivax disease and lower parasite densities compared with naïve non-immigrant travellers (Matteelli et al. Reference Matteelli, Colombini, Gulletta, Castelli and Carosi1999; Jelinek et al. Reference Jelinek, Schulte, Behrens, Grobusch, Coulaud, Bisoffi, Matteelli, Clerinx, Corachan, Puente, Gjorup, Harms, Kollaritsch, Kotlowski, Bjorkmann, Delmont, Knobloch, Nielsen, Cuadros, Hatz, Beran, Schmid, Schulze, Lopez-Velez, Fleischer, Kapaun, McWhinney, Kern, Atougia, Fry, da Cunha and Boecken2002; Bouchaud et al. Reference Bouchaud, Cot, Kony, Durand, Schiemann, Ralaimazava, Coulaud, Le Bras and Deloron2005; Mascarello et al. Reference Mascarello, Allegranzi, Angheben, Anselmi, Concia, Lagana, Manzoli, Marocco, Monteiro and Bisoffi2008; Salvado et al. Reference Salvado, Pinazo, Munoz, Alonso, Naniche, Mayor, Quinto and Gascon2008; Gonzalez et al. Reference Gonzalez, Nicolas, Munoz, Castro, Mas, Valls, Coma, Aibar and Gascon2009; Monge-Maillo et al. Reference Monge-Maillo, Norman, Perez-Molina, Diaz-Menendez, Rubio and Lopez-Velez2012; Farnert et al. Reference Farnert, Wyss, Dashti and Naucler2014; Pistone et al. Reference Pistone, Diallo, Mechain, Receveur and Malvy2014). However, despite having some immunity, a degree of immunity may have been lost, with data from recent studies showing that immigrants contracting malaria from visits to malaria endemic areas have lower levels of antibodies than semi-immune adults (Moncunill et al. Reference Moncunill, Mayor, Jimenez, Nhabomba, Casas-Vila, Puyol, Campo, Manaca, Aguilar, Pinazo, Almirall, Soler, Munoz, Bardaji, Angov, Dutta, Chitnis, Alonso, Gascon and Dobano2013a). Furthermore the fact that they have developed malaria in the first place provides evidence for the shorter-lived responses however, this is hard to separate this from possible temporal changes in circulating Plasmodium spp. strains in the source population.

A major factor contributing to diverging results and conclusions on antibody longevity is the difference in sampling on the antibody response curve between studies. After antigenic stimulus antibody titres rapidly increase peak, and then enter a biphasic decay consisting of a rapid decay from the initial peak followed by a slower decay lasting for several years, or providing life-long immunity depending on antigen (reviewed in Amanna and Slifka, Reference Amanna and Slifka2010). Recently, there have been several longitudinal studies estimating Plasmodium spp. antibody response half-life, which is the product of both antibody production and antibody decay (IgG molecules have a half-life of ~21 days Morell et al. Reference Morell, Terry and Waldmann1970). Estimates on antibody response half-life in African children following the peak antibody response after clinical malaria (the first phase of decay) have estimated short antibody response half-lives for P. falciparum merozoite and gametocyte antigens ranging from <2 weeks (Kinyanjui et al. Reference Kinyanjui, Conway, Lanar and Marsh2007) to ~12 weeks (Bousema et al. Reference Bousema, Roeffen, Meijerink, Mwerinde, Mwakalinga, van Gemert, van de Vegte-Bolmer, Mosha, Targett, Riley, Sauerwein and Drakeley2010b). Similar P. falciparum merozoite antibody response half-lives ranges (2–8 weeks) were found in asymptomatic Gambian children over the dry season (Akpogheneta et al. Reference Akpogheneta, Duah, Tetteh, Dunyo, Lanar, Pinder and Conway2008). Studies sampling in the second phase of decay have reported much longer half-lives. Studies in low transmission areas of Thailand, in the absence of Plasmodium spp. exposure, have shown antibody response half-lives for P. falciparum and P. vivax merozoite antigens to be 1–10 years (Wipasa et al. Reference Wipasa, Suphavilai, Okell, Cook, Corran, Thaikla, Liewsaree, Riley and Hafalla2010; Fowkes et al. Reference Fowkes, McGready, Cross, Hommel, Simpson, Elliott, Richards, Lackovic, Viladpai-Nguen, Narum, Tsuboi, Anders, Nosten and Beeson2012). Given that there are few exposures in low transmission areas the development of long-lived responses may be acquired after relatively few infections. Studies estimating half-life from seroconversion models (which will be at the tail end of the antibody response) have estimated of the antibody response of 49·8 year (95% CI 36·4, 72·7 years for P. falciparum-merozoite surface protein 1 (MSP1₁₉)) (Drakeley et al. Reference Drakeley, Corran, Coleman, Tongren, McDonald, Carneiro, Malima, Lusingu, Manjurano, Nkya, Lemnge, Cox, Reyburn and Riley2005), that is life-long immunity. Recent mathematical modelling of antibody responses in African children have shown that the biphasic decay may be the result of short-lived antibody secreting cells (half-life 2–10 days) which boost antibody levels post infection, and long-lived antibody secreting cells (half-life 3–9 years) which maintain persistent antibody responses (White et al. Reference White, Griffin, Akpogheneta, Conway, Koram, Riley and Ghani2014b).

Other factors will also contribute to varying estimates of antibody longevity such as methodology (study design, analysis, antigen/antibody investigated), malaria transmission and study populations. There may also be differences according to the type of antigen. Within individual studies, similar antibody longevity is observed for merozoite antigens, despite antigenic and species diversity (Kinyanjui et al. Reference Kinyanjui, Conway, Lanar and Marsh2007; Wipasa et al. Reference Wipasa, Suphavilai, Okell, Cook, Corran, Thaikla, Liewsaree, Riley and Hafalla2010; Fowkes et al. Reference Fowkes, McGready, Cross, Hommel, Simpson, Elliott, Richards, Lackovic, Viladpai-Nguen, Narum, Tsuboi, Anders, Nosten and Beeson2012), and merozoites and gametocytes (Bousema et al. Reference Bousema, Roeffen, Meijerink, Mwerinde, Mwakalinga, van Gemert, van de Vegte-Bolmer, Mosha, Targett, Riley, Sauerwein and Drakeley2010b). However there is some evidence that responses to P. falciparum-IE may be longer than P. falciparum merozoite antibody responses (Fowkes et al. Reference Fowkes, McGready, Cross, Hommel, Simpson, Elliott, Richards, Lackovic, Viladpai-Nguen, Narum, Tsuboi, Anders, Nosten and Beeson2012) although this is not consistent in all studies (Perraut et al. Reference Perraut, Mercereau-Puijalon, Diouf, Tall, Guillotte, Le Scanf, Trape, Spiegel and Garraud2000). Further studies are required to validate these findings and to dissect out whether specific antigens elicit longer-lived responses. Different antigens also have different IgG1 to IgG3 ratios and the extent of this can vary with age and exposure (Tongren et al. Reference Tongren, Drakeley, McDonald, Reyburn, Manjurano, Nkya, Lemnge, Gowda, Todd, Corran and Riley2006; Stanisic et al. Reference Stanisic, Richards, McCallum, Michon, King, Schoepflin, Gilson, Murphy, Anders, Mueller and Beeson2009). Differences in this ratio and the kinetics may contribute to some of the variation seen. The longevity of antibody responses are dependent on exposure to Plasmodium spp. Studies have shown that antibody longevity is positively correlated with transmission (Drakeley et al. Reference Drakeley, Corran, Coleman, Tongren, McDonald, Carneiro, Malima, Lusingu, Manjurano, Nkya, Lemnge, Cox, Reyburn and Riley2005) and persist for longer in those with recent documented infection compared with those unexposed to Plasmodium spp (Akpogheneta et al. Reference Akpogheneta, Duah, Tetteh, Dunyo, Lanar, Pinder and Conway2008; Fowkes et al. Reference Fowkes, McGready, Cross, Hommel, Simpson, Elliott, Richards, Lackovic, Viladpai-Nguen, Narum, Tsuboi, Anders, Nosten and Beeson2012). This may reflect a cumulative exposure effect as antibody response longevity increases with age in both high and low transmission areas (Perraut et al. Reference Perraut, Mercereau-Puijalon, Diouf, Tall, Guillotte, Le Scanf, Trape, Spiegel and Garraud2000; Akpogheneta et al. Reference Akpogheneta, Duah, Tetteh, Dunyo, Lanar, Pinder and Conway2008; Torres et al. Reference Torres, Clark, Hernandez, Soto-Cornejo, Gamboa and Branch2008; Bousema et al. Reference Bousema, Roeffen, Meijerink, Mwerinde, Mwakalinga, van Gemert, van de Vegte-Bolmer, Mosha, Targett, Riley, Sauerwein and Drakeley2010b; Clark et al. Reference Clark, Silva, Weiss, Li, Padilla, Crompton, Hernandez and Branch2012; Diop et al. Reference Diop, Richard, Diouf, Sokhna, Diagne, Trape, Faye, Tall, Diop and Balde2014).

Determining the longevity of antibody responses is challenging due to the dynamic kinetics and biphasic nature of antibody responses, particularly in high transmission areas with multiple reinfections. Further longitudinal studies, with accurate data on duration of infections, encompassing both parts of biphasic antibody decay are warranted. These studies should be performed in areas of varying transmission intensities to ensure generalizability and enable cross-transmission comparisons. Data from these studies will provide a better understanding for how changing transmission will impact on circulating antibodies and their ability to protect against symptomatic disease. Importantly, a greater understanding of the types of cells underpinning antibody dynamics remain to be elucidated as well as the contribution of immunological memory to antibody responses which remains hotly debated (Struik and Riley, Reference Struik and Riley2004; Hviid et al. Reference Hviid, Barfod and Fowkes2015; Portugal et al. Reference Portugal, Tipton, Sohn, Kone, Wang, Li, Skinner, Virtaneva, Sturdevant, Porcella, Doumbo, Doumbo, Kayentao, Ongoiba, Traore, Sanz, Pierce and Crompton2015). Presently, it is unclear whether immunity depends on the maintenance of antibodies above a threshold level, or by B-cell memory and the ability to recall responses when re-infected. To date, longitudinal dynamics of immunity have been performed almost entirely with standard immunoassays (e.g. ELISA). Studies that measure functional activity of antibodies may reveal different patterns if there is a threshold antibody level for functional activity. A recent study of pregnant women suggested that functional antibodies may be more resilient to changes in malaria transmission (Teo et al. Reference Teo, Hasang, Randall, Feng, Bell, Unger, Langer, Beeson, Siba, Mueller, Molyneux, Brown and Rogerson2014).

IMMUNOLOGICAL MEMORY TO MALARIA

The antibodies first produced in response to a new antigen are of relatively low affinity and are secreted by short-lived plasma cells generated following interaction between naïve B cells and antigen-specific helper T cells. Once the infection is controlled, the population of antigen-specific effector T cells and memory B cells (MBC) contracts leaving behind a small number of long-lived memory T and B cells, but the pool of these cells is thought to increase with each subsequent infection to increase capacity for subsequent responses. Some of the plasma cells differentiate into long-lived plasma cells that migrate to the bone marrow and continue to produce antibodies even in the absence of antigen. Evidence for malaria memory responses comes from the observation of rapid boosting of P. falciparum-specific antibody responses upon re-exposure to malaria following the dry season or prolonged periods of low transmission (Vande Waa et al. Reference Vande Waa, Jensen, Akood and Bayoumi1984; Migot et al. Reference Migot, Chougnet, Raharimalala, Astagneau, Lepers and Deloron1993) and data from controlled human malaria infection (CHMI) studies showing that previously exposed (>5 years prior) aparasitemic individuals showed a stronger increase in antibody titres than naïve volunteers, (Obiero et al. Reference Obiero, Shekalaghe, Hermsen, Mpina, Bijker, Roestenberg, Teelen, Billingsley, Sim, James, Daubenberger, Hoffman, Abdulla, Sauerwein and Scholzen2015) suggesting that individuals can generate and retain P. falciparum-specific MBC.

There is a paucity of studies on MBC, and the ability of Plasmodium spp. to produce a long-lived memory response is debated (Struik and Riley, Reference Struik and Riley2004; Hviid et al. Reference Hviid, Barfod and Fowkes2015; Portugal et al. Reference Portugal, Tipton, Sohn, Kone, Wang, Li, Skinner, Virtaneva, Sturdevant, Porcella, Doumbo, Doumbo, Kayentao, Ongoiba, Traore, Sanz, Pierce and Crompton2015). Much of the debate has originated from the afore-mentioned reports of both short- and long-lived antibody responses in the absence of exposure. Epidemiological studies have shown that MBC, like malarial antibodies, are acquired slowly after repeated infections in malaria endemic areas suggesting that they play an important role in the acquired immune response (Dorfman et al. Reference Dorfman, Bejon, Ndungu, Langhorne, Kortok, Lowe, Mwangi, Williams and Marsh2005; Wipasa et al. Reference Wipasa, Suphavilai, Okell, Cook, Corran, Thaikla, Liewsaree, Riley and Hafalla2010; Nogaro et al. Reference Nogaro, Hafalla, Walther, Remarque, Tetteh, Conway, Riley and Walther2011; Weiss et al. Reference Weiss, Clark, Li, Traore, Kayentao, Ongoiba, Hernandez, Doumbo, Pierce, Branch and Crompton2011, Reference Weiss, Crompton, Li, Walsh, Moir, Traore, Kayentao, Ongoiba, Doumbo and Pierce2009, Reference Weiss, Traore, Kayentao, Ongoiba, Doumbo, Doumtabe, Kone, Dia, Guindo, Traore, Huang, Miura, Mircetic, Li, Baughman, Narum, Miller, Doumbo, Pierce and Crompton2010, Reference Weiss, Ndungu, McKittrick, Li, Kimani, Crompton, Marsh and Pierce2012; Ndungu et al. Reference Ndungu, Olotu, Mwacharo, Nyonda, Apfeld, Mramba, Fegan, Bejon and Marsh2012; Illingworth et al. Reference Illingworth, Butler, Roetynck, Mwacharo, Pierce, Bejon, Crompton, Marsh and Ndungu2013; Muellenbeck et al. Reference Muellenbeck, Ueberheide, Amulic, Epp, Fenyo, Busse, Esen, Theisen, Mordmuller and Wardemann2013). However, the afore-mentioned epidemiological studies have shown that antigen-specific antibodies and MBC are not always correlated. Notably, studies of the immune response to VAR2CSA have shown that multigravid pregnant women may lack VAR2CSA-specific antibodies early on in pregnancy despite having detectable MBC spanning decades (Ampomah et al. Reference Ampomah, Stevenson, Ofori, Barfod and Hviid2014a). Similarly, antibodies to pre-erythrocytic and merozoite antigens have been shown to decline after P. falciparum transmission/exposure ceases, whereas MBC were maintained for decades (Wipasa et al. Reference Wipasa, Suphavilai, Okell, Cook, Corran, Thaikla, Liewsaree, Riley and Hafalla2010; Ndungu et al. Reference Ndungu, Olotu, Mwacharo, Nyonda, Apfeld, Mramba, Fegan, Bejon and Marsh2012). Furthermore, studies of apical membrane antigen 1 (AMA1) in Swedish travellers have also shown the presence of MBC up to 16 years post infection in the absence of antibody (Ndungu et al. Reference Ndungu, Lundblom, Rono, Illingworth, Eriksson and Farnert2013). These studies provide evidence for long-lived, potentially life-long, MBC responses and the observation that MBCs are detectable decades later in previously naïve individuals would suggest that MBC responses are generated and maintained despite any re-exposure or persistent antigen stimulation. However, the role of MBCs in contributing to protective immunity has not yet been quantified.

The number of infections required to generate MBCs is unclear but evidence from MBC studies in CHMI and returned travellers would suggest that MBC can develop after a brief single exposure to malaria (Ndungu et al. Reference Ndungu, Lundblom, Rono, Illingworth, Eriksson and Farnert2013; Nahrendorf et al. Reference Nahrendorf, Scholzen, Bijker, Teirlinck, Bastiaens, Schats, Hermsen, Visser, Langhorne and Sauerwein2014). These findings are supported by data on MBC in low transmission areas. A study in Peru showed that even one reported prior infection was sufficient to generate antigen-specific MBC and to maintain a positive antibody response for at least 5 months, in the absence of reinfection (Clark et al. Reference Clark, Silva, Weiss, Li, Padilla, Crompton, Hernandez and Branch2012). Another study conducted in an area of very low malaria transmission in Northern Thailand showed that malaria-naïve individuals or those who had not had clinical episodes of malaria over the past 6 years had MBC response to P. falciparum (and P. vivax) (Wipasa et al. Reference Wipasa, Suphavilai, Okell, Cook, Corran, Thaikla, Liewsaree, Riley and Hafalla2010). This suggests that the number of infections required to develop long-lived antibody responses is likely to be quite small.

However, protective antibody and MBC responses appear to be slow to develop and ineffectively maintained. This could be due to genetic and antigenic variation of the parasite (Scherf et al. Reference Scherf, Lopez-Rubio and Riviere2008; Takala and Plowe, Reference Takala and Plowe2009) so that repeated exposure leads to a gradual expansion of the repertoire of P. falciparum-specific MBC (Weiss et al. Reference Weiss, Traore, Kayentao, Ongoiba, Doumbo, Doumtabe, Kone, Dia, Guindo, Traore, Huang, Miura, Mircetic, Li, Baughman, Narum, Miller, Doumbo, Pierce and Crompton2010). However, recent evidence also shows that MBC responses may be dysregulated in malaria infections. Malarial MBC responses appear to be sub-optimally produced compared with other MBC responses. While the magnitude of MBC responses in naturally exposed individuals are in the same range as childhood vaccine-induced MBCs in the same populations, MBC prevalence is much lower (30–60% compared with 60–100% for vaccine antigens) even after many decades of exposure (Dorfman et al. Reference Dorfman, Bejon, Ndungu, Langhorne, Kortok, Lowe, Mwangi, Williams and Marsh2005; Wipasa et al. Reference Wipasa, Suphavilai, Okell, Cook, Corran, Thaikla, Liewsaree, Riley and Hafalla2010; Nogaro et al. Reference Nogaro, Hafalla, Walther, Remarque, Tetteh, Conway, Riley and Walther2011; Ndungu et al. Reference Ndungu, Olotu, Mwacharo, Nyonda, Apfeld, Mramba, Fegan, Bejon and Marsh2012; Weiss et al. Reference Weiss, Ndungu, McKittrick, Li, Kimani, Crompton, Marsh and Pierce2012, Reference Weiss, Traore, Kayentao, Ongoiba, Doumbo, Doumtabe, Kone, Dia, Guindo, Traore, Huang, Miura, Mircetic, Li, Baughman, Narum, Miller, Doumbo, Pierce and Crompton2010). It has been shown that repeated exposure to infection, can lead to the large expansion of phenotypically similar ‘atypical’ MBC in children, adults and pregnant women in geographically diverse regions (West and East Africa, Peru, Papua New Guinea) of varying transmission (Weiss et al. Reference Weiss, Crompton, Li, Walsh, Moir, Traore, Kayentao, Ongoiba, Doumbo and Pierce2009, Reference Weiss, Clark, Li, Traore, Kayentao, Ongoiba, Hernandez, Doumbo, Pierce, Branch and Crompton2011; Illingworth et al. Reference Illingworth, Butler, Roetynck, Mwacharo, Pierce, Bejon, Crompton, Marsh and Ndungu2013; Muellenbeck et al. Reference Muellenbeck, Ueberheide, Amulic, Epp, Fenyo, Busse, Esen, Theisen, Mordmuller and Wardemann2013; Nogaro et al. Reference Nogaro, Hafalla, Walther, Remarque, Tetteh, Conway, Riley and Walther2011; Ampomah et al. Reference Ampomah, Stevenson, Ofori, Barfod and Hviid2014b; Requena et al. Reference Requena, Campo, Umbers, Ome, Wangnapi, Barrios, Robinson, Samol, Rosanas-Urgell, Ubillos, Mayor, Lopez, de Lazzari, Arevalo-Herrera, Fernandez-Becerra, del Portillo, Chitnis, Siba, Bardaji, Mueller, Rogerson, Menendez and Dobano2014; Subramaniam et al. Reference Subramaniam, Skinner, Ivan, Mutimura, Kim, Feintuch, Portugal, Anastos, Crompton and Daily2015). CHMI studies have shown that out of all the MBC types, atypical MBCs have the strongest proliferative response and peak either immediately after blood-stage infection or convalescence (Scholzen et al. Reference Scholzen, Teirlinck, Bijker, Roestenberg, Hermsen, Hoffman and Sauerwein2014). Evidence suggests that P. falciparum exposure drives the expansion of atypical MBC; atypical MBC are correlated with P. falciparum transmission intensity (Weiss et al. Reference Weiss, Crompton, Li, Walsh, Moir, Traore, Kayentao, Ongoiba, Doumbo and Pierce2009) and a study in Kenya showed that P. falciparum exposure drove the differential expansion of atypical MBCs in age-matched children (Illingworth et al. Reference Illingworth, Butler, Roetynck, Mwacharo, Pierce, Bejon, Crompton, Marsh and Ndungu2013). Atypical MBC are also found more frequently in children who asymptomatically carry P. falciparum as compared with children who are P. falciparum free (Weiss et al. Reference Weiss, Crompton, Li, Walsh, Moir, Traore, Kayentao, Ongoiba, Doumbo and Pierce2009). Furthermore, in the absence of P. falciparum infection (for 1 year) atypical and activated MBC subsets decrease (Ayieko et al. Reference Ayieko, Maue, Jura, Noland, Ayodo, Rochford and John2013) and studies in pregnant women also show a retraction of VAR2CSA-specific atypical MBCs in the postpartum period (Ampomah et al. Reference Ampomah, Stevenson, Ofori, Barfod and Hviid2014b).

The phenotype of atypical MBCs is unclear but recent studies have shown that atypical MBCs contribute to P. falciparum-specific IgG (Muellenbeck et al. Reference Muellenbeck, Ueberheide, Amulic, Epp, Fenyo, Busse, Esen, Theisen, Mordmuller and Wardemann2013; Portugal et al. Reference Portugal, Tipton, Sohn, Kone, Wang, Li, Skinner, Virtaneva, Sturdevant, Porcella, Doumbo, Doumbo, Kayentao, Ongoiba, Traore, Sanz, Pierce and Crompton2015) but they may have a significantly reduced signalling and effector function through a range of potential molecular modulators (Scholzen and Sauerwein, Reference Scholzen and Sauerwein2013; Scholzen et al. Reference Scholzen, Teirlinck, Bijker, Roestenberg, Hermsen, Hoffman and Sauerwein2014; Portugal et al. Reference Portugal, Tipton, Sohn, Kone, Wang, Li, Skinner, Virtaneva, Sturdevant, Porcella, Doumbo, Doumbo, Kayentao, Ongoiba, Traore, Sanz, Pierce and Crompton2015; Zinocker et al. Reference Zinocker, Schindler, Skinner, Rogosch, Waisberg, Schickel, Meffre, Kayentao, Ongoiba, Traore and Pierce2015). The precise function and specificity of atypical MBCs, their potential regulatory role and the factors that drive their expansion are yet to be elucidated and requires further research. Furthermore the dynamics of MBCs and their contribution to antibody dynamics are unknown, and the basal levels and breadth of circulating antibodies needed for protection against disease need to be defined. The majority of research into MBC responses has centred on P. falciparum, and the role of MBC in P. vivax is also warranted.

The role of T cells in immunological memory is even less clear. T cells are thought to be involved in immune responses that clear pre-erythrocytic stages, and in providing T-cell help and regulatory responses in immunity against blood-stages (Beeson et al. Reference Beeson, Osier and Engwerda2008). The majority of T-cell literature pertains to murine models which have been the feature of several recent reviews (Corradin and Levitskaya, Reference Corradin and Levitskaya2014; Doll and Harty, Reference Doll and Harty2014; Krzych et al. Reference Krzych, Zarling and Pichugin2014; Van Braeckel-Budimir and Harty, Reference Van Braeckel-Budimir and Harty2014). There is a paucity of data on T cell responses in humans, and data from CHMI and epidemiological studies are not complementary. In a CHMI study, DNA prime/adenovirus boost immunization of circumsporozoite protein (CSP) and AMA1 could induce sterile immunity which was associated with higher effector to central memory CSP- and AMA1-specific cytotoxic CD8+ T cells ratios, than non-protected volunteers (Sedegah et al. Reference Sedegah, Hollingdale, Farooq, Ganeshan, Belmonte, Kim, Peters, Sette, Huang, McGrath, Abot, Limbach, Shi, Soisson, Diggs, Chuang, Tamminga, Epstein, Villasante and Richie2014). T cell responses to both P. falciparum sporozoites and IE are induced rapidly and remain almost undiminished up to 14 months even after a single malaria episode (Teirlinck et al. Reference Teirlinck, McCall, Roestenberg, Scholzen, Woestenenk, de Mast, van der Ven, Hermsen, Luty and Sauerwein2011). However, in naturally exposed individuals, CD8+ T cell responses to sporozoite epitopes are of a much lower magnitude than in experimentally infected individuals and CD8+ T cell memory appears to wane with time. The poor natural immunity to sporozoites could be due to the interference of erythrocytic stages on existing T cell responses described in rodent models of malaria (Ocana-Morgner et al. Reference Ocana-Morgner, Mota and Rodriguez2003). Alternatively, irradiated sporozoites (used in experimental infections) could also be more immunogenic than live sporozoites and their antigens could persist for longer, helping to maintain effector T cells in the liver (Scheller and Azad, Reference Scheller and Azad1995). CD4+ helper T cells primary function is to help the development of CD8+ T and B cell responses and MBCs. Murine studies have shown that CD4+ T cells retract and form a resting memory T cell population able to respond rapidly on re-infection, especially in the presence of antigen in the form of ongoing, sub-patent infection (reviewed in Stephens and Langhorne, Reference Stephens and Langhorne2006). Phase 2a RTS,S/AS trials in malaria-naïve individuals have shown that protected individuals can induce higher frequencies of effector and central memory T cells compared with unprotected individuals (Lumsden et al. Reference Lumsden, Schwenk, Rein, Moris, Janssens, Ofori-Anyinam, Cohen, Kester, Heppner and Krzych2011) and in malaria-exposed individuals vaccinated with the RTS,S vaccine, protection from re-infection (within 4 months of immunization) was associated with central memory CD4+ T cells reactive against an epitope of CSP (Reece et al. Reference Reece, Pinder, Gothard, Milligan, Bojang, Doherty, Plebanski, Akinwunmi, Everaere, Watkins, Voss, Tornieporth, Alloueche, Greenwood, Kester, McAdam, Cohen and Hill2004).

There is a clear deficit of studies of T cell immunity in individuals living in malaria endemic areas. Given the importance of T cell responses in immunity against the transmissible sporozoite stage, further investigation in individuals living in malaria endemic areas of varying transmission is warranted. The available evidence suggests that continuous antigenic stimulation may help to maintain a protective effector T cell population, and therefore ongoing exposure to infectious bites may help maintain protection. Further studies in areas of declining transmission are warranted so that the full impact of malaria control on sporozoite immunity can be realized. This is particularly pertinent in the context of vaccination with RTS,S, the most advanced malaria vaccine, which is based on a construct containing B-cell and T-cell epitopes of the CSP protein (RTS, 2015). The efficacy of RTS,S is dependent on the level of pre-vaccination anti-CSP titres, with improved efficacy in individuals with higher anti-CSP titres (Bejon et al. Reference Bejon, White, Olotu, Bojang, Lusingu, Salim, Otsyula, Agnandji, Asante, Owusu-Agyei, Abdulla and Ghani2013; White et al. Reference White, Bejon, Olotu, Griffin, Bojang, Lusingu, Salim, Abdulla, Otsyula, Agnandji, Lell, Asante, Owusu-Agyei, Mahama, Agbenyega, Ansong, Sacarlal, Aponte and Ghani2014a), and exposed to higher transmission intensity (Bejon et al. Reference Bejon, White, Olotu, Bojang, Lusingu, Salim, Otsyula, Agnandji, Asante, Owusu-Agyei, Abdulla and Ghani2013; Campo et al. Reference Campo, Aponte, Skinner, Nakajima, Molina, Liang, Sacarlal, Alonso, Crompton, Felgner and Dobano2015).

CONCLUSIONS AND RESEARCH PRIORITIES

It is clear that further research is required to fully understand the acquisition and maintenance of the naturally acquired immune response to malaria to fully comprehend the impact of changing transmission on malarial immunity (research priorities summarized in Box 1). We know that multiple Plasmodium spp. exposures are required but the number and frequency of exposures required is unknown, as is how this process varies across genetically diverse populations experiencing varying transmission and parasite species and genotypes. What constitutes an effective immune response is also not clear together with which assays can accurately assess and define protective clinical immunity. Identifying immune correlates of exposure and protective immunity are keys to further developing serological tools (including functional assays) that can be used to track the immunological consequences of declining malaria transmission. Furthermore these approaches can also be used to determine whether detectable long-lived antibodies still function to provide effective protection against clinical disease. Longitudinal studies encompassing the biphasic decay of antibodies to identify the cell types which produce long-lived antibody responses are also warranted as are investigations into the sub-optimal effectiveness of MBC and T cell responses. How all of the above differs according to species, life-cycle stage and antigenic diversity also remains to be teased out. This is particularly important in the context of developing vaccine candidate antigens and understanding the efficacy, longevity and impact of vaccines in populations with declining immunity.

Importantly, when we think of declining transmission and less ‘exposure’ we simply think of a reduced number of infections, which we typically measure epidemiologically as an episode of clinical malaria or by detection of parasites by light microscopy (which was the case for many studies featured in this review). However, submicroscopic infections (diagnosed by polymerase chain reaction) often exceeds those detectable by light microscopy by several fold (Okell et al. Reference Okell, Ghani, Lyons and Drakeley2009; Satoguina et al. Reference Satoguina, Walther, Drakeley, Nwakanma, Oriero, Correa, Corran, Conway and Walther2009; Mosha et al. Reference Mosha, Sturrock, Greenhouse, Greenwood, Sutherland, Gadalla, Atwal, Drakeley, Kibiki, Bousema, Chandramohan and Gosling2013; Baum et al. Reference Baum, Sattabongkot, Sirichaisinthop, Kiattibutr, Davies, Jain, Lo, Lee, Randall, Molina, Liang, Cui, Felgner and Yan2015; Tadesse et al. Reference Tadesse, Pett, Baidjoe, Lanke, Grignard, Sutherland, Hall, Drakeley, Bousema and Mamo2015; Thanh et al. Reference Thanh, Hong, Van Van, Van Malderen, Obsomer, Rosanas-Urgell, Grietens, Xa, Bancone, Chowwiwat, Duong, D'Alessandro, Speybroeck and Erhart2015). Sub-microscopic infections will provide an antigenic stimulus to maintain immune responses but the role of submicroscopic infections in maintaining immunity is yet to be quantified. It is critical that a greater understanding of submicroscopic infection in the maintenance of immunity is performed in the context of declining transmission given that the highest prevalences of submicroscopic carriage are found in areas of lowest transmission (Okell et al. Reference Okell, Ghani, Lyons and Drakeley2009) including areas which have recently transitioned from high to low malaria transmission intensity (Satoguina et al. Reference Satoguina, Walther, Drakeley, Nwakanma, Oriero, Correa, Corran, Conway and Walther2009; Kalayjian et al. Reference Kalayjian, Malhotra, Mungai, Holding and King2013). However, whether the prevalence of submicroscopic infections increases when transmission declines in a geographical area is yet to be determined and will be pivotal to our understanding of how immunity changes in this era of increased malaria control and declining malaria transmission.