The immunoepidemiology of human hookworm infection

SUMMARY

Advances in hookworm immunoepidemiology are reviewed. Recent studies demonstrate a mixed Th1/Th2 response in human hookworm infection, with immunosuppression of specific and nonspecific IFN-γ responses. There is increasing evidence for protective immunity in human hookworm infection, including anti-larval IL-5- and IgE-dependent mechanisms, and for immunological interactions between hookworm infection and other diseases.

INTRODUCTION

The human hookworms Necator americanus and Ancylostoma duodenale are directly transmitted nematode parasites of the small intestine. Hookworms are abundant parasites, infecting an estimated 740 million people world-wide, with highest prevalences in sub-Saharan Africa and Papua New Guinea (1). Blood-feeding adult worms live for 1–5 years in the small intestine, and are an important cause of morbidity due to iron-deficiency anaemia. Though mortality due to infection is rare, the global burden of hookworm disease is high, with an estimated 22 million disability-adjusted life years (DALYs) lost each year (2). The epidemiology, pathology and control of hookworm infection have been extensively reviewed (3–6). Despite its clinical significance, the immunology of human hookworm infection has received relatively little attention, in common with that of other human soil-transmitted helminths (STHs). However, interest has increased recently, in part associated with initiatives to develop a human hookworm vaccine (7,8), and with increasing interest in the potential importance of helminth infection in modulating the immunology and pathology of other infectious and non-infectious diseases such as HIV/AIDS, tuberculosis, malaria and asthma (9–11).

Immunoepidemiology is the study of the distribution of immune responses and infection in populations, and of the factors influencing this distribution (12), i.e. taking an epidemiological approach to immunology. The aims of immunoepidemiology are to understand the processes affecting immune responses, and the influence of immunity on infection. One important question is the existence and importance of protective immunity, which has been much debated for human hookworm infection (6,13,14). Immunoepidemiological data can also provide evidence for parasite-mediated immunosuppression, and for immunological interactions between hookworm infection and other diseases. Immunoepidemiological studies are particularly important for pathogens such as the human hookworms, for which there is no suitable animal model. A number of advances have been made recently, such as the characterization of lymphocyte and cytokine responses in human hookworm infection, and of antibody responses to defined hookworm antigens. Here, we focus on the results of such field studies; laboratory and molecular studies associated with vaccine development have been reviewed elsewhere (7). We first consider evidence for immunosuppression in hookworm infection, using results from immunoepidemiological and experimental studies. We then assess the epidemiological and immunoepidemiological evidence for protective immunity, and discuss the potential mechanisms of immunity. Finally, interactions between hookworms and other infections are examined. It should be noted that, although there are two human hookworms, nearly all immunological studies have been carried out on populations infected with N. americanus, or are studies where the species involved have not been reported. Thus we somewhat erroneously use ‘hookworm’ to refer to N. americanus, while results specific to other species are highlighted.

IMMUNE RESPONSES TO HUMAN HOOKWORM INFECTION

Immune responses to hookworm infection in both humans and animal models have been the subject of detailed reviews (14,15). Here, we summarize results from experimental and natural human infections, including recent studies of cellular immunity.

Antibody responses in experimental infection

Results from experimental infections can prove valuable in interpreting immunoepidemiological data. However, although a large number of individuals have been experimentally infected with human hookworms, for instance as an early treatment for polycythaemia vera, detailed immunological data are only available from nine volunteers (16). Of these, seven were infected with N. americanus (17–19), and two with the zoonotic Ancylostoma ceylanicum (20) (Table 1). Antibody responses in these studies can be summarized as follows (16). All infected individuals mounted a detectable IgG or uncharacterized antibody response 2–8 weeks after infection, which rose slowly in primary infection and reached maximal levels only after the 2nd or 3rd infection. Specific and total IgE responses were generally low or undetectable in primary infection, but rose progressively after repeated infection. In repeated infection, antibody levels reflected cumulative exposure to parasites rather than current infection levels.

Table 1.  Immunological studies of experimental human hookworm infection

Study	Species	Number of volunteers	Dose of infective larvae (T – chemotherapy)	Total duration of infection
Ball & Bartlett (18)	N. americanus	1	300, 2 × 100, 2 × 25, T, 4 × 25, T	3 years
Ogilvie et al. (19)	N. americanus	1	250, T, 250, T, 250, T, 250, T	26 months
Maxwell et al. (17)	N. americanus	5	50, T	40–60 days
Carroll & Grove (20)	A. ceylanicum	2	1200, T	30 weeks

Antibody responses in natural infection

Naturally infected populations mount a strong antibody response to hookworm antigens. Serum antibody responses of all five isotypes were elevated in a Papua New Guinean community, using either adult excretory–secretory (ES) or larval homogenate antigens (21). Similarly, IgG, IgG1, IgG3, IgG4, IgM, IgA and IgE responses to adult worm homogenate were elevated in a Zimbabwean community, although IgG2 levels were similar to those in uninfected controls (22). In Australian patients infected with human hookworm, or the zoonotic Ancylostoma caninum, all IgG subclasses reacted to A. caninum antigens (23). Anti-hookworm antibodies of most isotypes show evidence of cross-reactivity with other helminths, such as Ascaris lumbricoides and Schistosoma mansoni (22,24,25), but IgG4 or IgE responses are more species-specific (23,26,27). In addition to hookworm-specific responses, there is also a marked up-regulation of polyclonal IgE in hookworm infection (28,29). This is accompanied by increased levels of sCD23 and IgG anti-IgE autoantibodies (30,31). Levels of serum polyclonal IgG, and intestinal IgG, IgM and IgE, are also elevated in hookworm infection (26,32). In contrast, total levels of intestinal IgA are reduced (32).

Eosinophil responses

Eosinophilia is commonly observed in experimental and natural hookworm infection. All of at least 114 individuals experimentally infected with N. americanus, A. duodenale or A. ceylanicum became eosinophilic (16), and eosinophils showed evidence of an activated state during infection (33). Eosinophilia usually developed 25–35 days after exposure, prior to patency, and reached a peak 35–65 days (N. americanus/A. ceylanicum) or 90–120 days (A. duodenale) after exposure (16).

Lymphocyte and cytokine responses

Proliferative lymphocyte responses to hookworm antigens were first demonstrated in experimental infections (16). Results from N. americanus infections were variable (17,34), while both individuals infected with a high dose of A. ceylanicum developed strong proliferative responses to adult antigens 4–6 weeks after infection (20). Responses in natural infection have only been investigated recently, but there are now a number of studies from around the world that have assessed both lymphocyte proliferation and cytokine responses to hookworm antigens in naturally infected individuals (35–38). Moderate proliferative responses were seen in three studies (36–38), though responses were very weak in others (Bethony unpublished data) (35). The cytokine results clearly show that hookworm infection induces a mixed Th1/Th2 response, with significant production of both Th1 (IFN-γ, IL-12) and Th2 cytokines (IL-4, IL-5, IL-13). Infection is also accompanied by high levels of background or antigen-specific IL-10 and TNF-α. A similar mixed cytokine profile has been reported for Trichuris trichiura infection (39), whereas a more polarized Th2 response is seen in A. lumbricoides infection (40).

IMMUNOMODULATION IN HOOKWORM INFECTION

Blood-feeding adult hookworms live for up to 18 years in close contact with the host immune system. Thus, in common with other helminths, they are likely to have evolved defences against host immunity. These defences may include suppression of antigen-specific immune responses, modulation of the host immune response from protective to non-protective responses, and inactivation of immune effector mechanisms. The difficulty in demonstrating protective immunity highlights the possibility that hookworms possess molecular strategies to subvert host protective responses, and a repertoire of potentially subversive molecules has been identified, including an eotaxin metalloproteinase (41), calreticulin (42,43), antioxidants (44,45), neutrophil inhibitory factor (46) and pro-apoptotic processes involving activated T cells (47). It has been difficult to define a precise role for each of these molecules in vivo in humans. However, evidence is accumulating from immunoepidemiological studies that immunomodulation occurs during hookworm infection. Several immunoepidemiological patterns can indicate immunosuppression during infection. These include a decline in immune responsiveness during infection, and an increase in responsiveness after removal of parasites by chemotherapy. Negative correlations between immune responses and worm burdens are consistent with both immunosuppression and protective immunity, and are discussed later. Such field studies have clearly demonstrated the suppression of specific immune responses in infections with tissue-dwelling helminths such as filarial worms and schistosomes (48).

Dynamics of immune responses during infection

Simple mathematical models of anti-helminth immune responses, assuming constant exposure with age, predict that immune responses will generally increase with host age to a plateau in adults, though declines across some age-classes are possible (49). Consistent with this, monotonic increases in antibody levels through time were seen in experimental infections (16). In contrast, eosinophil levels typically reach a peak in early experimental infection but then decline, even after repeated infection (18), and lymphoproliferative responses declined sharply from an early peak in A. ceylanicum infection, though egg counts also declined during this period (20). Results from natural infections are more variable. A number of antibody responses, and eosinophilia, in the Papua New Guinea and Zimbabwe studies did increase with host age, in a manner similar to worm burden (21,22,50,51). These included the overall IgG responses, some IgE responses and eosinophilia. However, levels of anti-larval IgA, and anti-adult IgD and IgE, in Papua New Guinea, and IgG1 and IgM in Zimbabwe, decreased with age. The dynamics of cytokine responses have only been assessed in one study, in which no relationship was seen between cytokine responses and host age (38). The declines in some immune responses during the course of infection are consistent with immunosuppression. There is also evidence that intestinal pathology declines with repeated infection and is less severe in chronic than acute infection (5,6). However, there are potential confounding factors in field studies, and studies which compare communities with different intensities of infection would be useful (49).

Changes in immune response after chemotherapy

Increases in immune responses after chemotherapy-induced loss of parasites provide stronger evidence for immunosuppression. Such studies are experimental rather than observational, so bias and the influence of confounding factors can be removed by randomization and placebo-controls (though this has yet to be done in hookworm studies). Changes in antibody responses after chemotherapy have been assessed in several studies. In general, hookworm-specific antibody levels decline sharply after chemotherapy, even in the presence of continued reinfection (Table 2a). However, a significant increase in anti-adult IgD was seen in one study (21). Total IgE levels were reduced after chemotherapy in four of five studies (26,32,38,52,53), and eosinophilia fell by 45–50% 6 weeks after treatment (32,52,53). In contrast, intestinal nonspecific IgA levels increased significantly after treatment (32). Only one study has investigated changes in cytokine responses after chemotherapy (38). There was a significant increase in IFN-γ responses to hookworm antigen, whereas IL-4 and IL-5 responses were unchanged (Table 2b). Proliferative responses increased slightly but not significantly after treatment, whereas both IFN-γ and proliferative responses fell significantly with increasing worm burden, consistent with immunosuppression. These data suggest that parasite-mediated immunosuppression occurs for some immune responses. An alternative explanation, that rises in responsiveness result from increased antigenic exposure after worm death, is unlikely for hookworm infection, though is important in schistosome infection. However, host-mediated immunoregulatory changes are also possible, where the removal of antigen might reduce cross-regulatory mechanisms.

Table 2.  Changes in anti-hookworm immune responses after anthelminthic treatment. The sample size (n), hookworm species, time after initiation of chemotherapy that post-treatment samples were taken, worm burden at time of post-treatment sampling (expressed as a percentage of pre-treatment epg), and mean levels of each response pre- and post-treatment. Significant increases after chemotherapy are highlighted in bold

(a) Antibody responses (OD units)
Country (reference)	n	Hookworm species	Time post- treatment	Worm burden post-treatment	Antigen	Antibody	Mean level pre-treatment	Mean level post-treatment
India (26)	25	nd	1 month	0%	L3 ES	IgE	1·510	0·476***
Papua New Guinea (21)	172	N. americanus	1 year	47%	Adult ES	IgG	0·842	0·524***
						IgM	0·870	0·442***
						IgA	0·468	0·429
						IgD	0·088	0·202***
						IgE	0·296	0·304
					L3 SOM	IgG	0·522	0·460*
						IgM	0·810	0·561***
						IgA	0·290	0·093***
						IgD	0·115	0·144
						IgE	0·400	0·099***
Brazil (37)	12	N. americanus	6 months	0%	Adult SOM	IgG	1·01	0·65
						IgG1	0·35	0·18**
						IgG4	1·19	0·53**
						IgA	0·93	0·35**
						IgE	0·33	0·18

(b) lymphoproliferative (cpm) and cytokine responses (pg/mL)
Country (reference)	n	Hookworm species	Time post- treatment	Worm burden post-treatment	Antigen	Response	Mean level pre-treatment	Mean level post-treatment
* P < 0·05; ** P < 0·01; *** P < 0·001; ES, excretory–ecretory; SOM, somatic; nd, not determined.
Papua New Guinea (38)	54	N. americanus	1 month	17%	Adult ES	Proliferative	10339	13888
	42					IFN-γ	166	322**
	42					IL-4	20	17
	37					IL-5	163	151

Thus the available evidence suggests that there is worm-associated suppression of IFN-γ responses. Suppression was not antigen-specific, as there was a worm burden-dependent decline in IFN-γ responses to mycobacterial PPD (38), and suppression of nonspecific cellular responses has also been reported (54). Such bystander suppression is also seen in other helminth infections. In contrast, there is little evidence for suppression of antibody responses or eosinophilia, which typically fall after chemotherapy. However, anti-adult IgD levels increased significantly after treatment, and declined with age pre-treatment (21). Similarly, total intestinal IgA levels increased after treatment, and were negatively related to serum IgE levels pre-treatment (32). The mechanisms of immunosuppression are unknown, though the high levels of IL-10 in hookworm infection suggest a role for regulatory T_reg cells (48), and high levels of IL-10 were associated with reduced levels of IFN-γ, IL-5 and IL-13 in Brazilian patients (37). Levels of another down-regulatory cytokine, TGF-β, have not been assessed in hookworm infection. Recent data from Brazil implicate a block in entry of cells to the G1 phase of the cell cycle in immunosuppressed individuals. Stimulation with larval or adult hookworm antigens failed to activate extracellular-signal-regulated-kinase (ERK) in PBMC from hookworm-infected individuals, indicating that the cells remained blocked in G0. In contrast, activated ERK was produced in response to PPD, and in response to specific stimulation in S. mansoni and A. lumbricoides infected patients (Bethony and Bottazzi, unpublished observations).

A number of other immunomodulatory changes are apparent in helminth infections. Most obvious is the Th2 bias induced by hookworms and other helminths. Various hookworm products may contribute to this Th2 bias (55). Whether the Th2 response reflects a host response to control worms, or is partly parasite-driven to aid parasite survival, remains controversial (48). However, the increasing evidence for Th2-mediated protective immunity (see below) suggests the former. As cytokine studies have shown, there is also a significant Th1 response in hookworm infection. Observations from other helminth systems and allergic disease suggest that complex interactions occur between Th1 and Th2 responses. Individuals with schistosomiasis or filariasis can have a modified Th2 response, with high IL-10 and IgG4, a balanced Th1/Th2 response, with high IgE, or a predominantly Th1 response (48). These different responses may reflect the course of infection, variation in exposure, or host differences. The identification of such variation in hookworm infection will require more detailed immunoepidemiological studies, particularly of changes in antibody and cytokine production with age, and between communities with different intensities of infection. Both Th2 and Th1 stimulatory products have been described from hookworms (55,56).

Further understanding of the mechanisms of immune modulation by hookworms, and identification of the molecules responsible, may lead to the development of novel, much needed therapeutics for immunopathological diseases. The existence of hookworm-mediated immunosuppression also has some practical implications for hookworm vaccine development (7,8,57). Vaccine-induced immunity may itself be susceptible to immunosuppression, so that protective responses may not develop unless pre-existing infections are removed by chemotherapy. More importantly, individuals vaccinated with a partially protective vaccine will be subject to reinfection, which might then suppress vaccine efficacy. The use of immunomodulatory molecules as vaccine candidates could potentially avoid such problems, since successful vaccination might prevent future immunosuppression. It is thus of interest that vaccination with immunomodulatory hookworm calreticulin induced successful protection in a murine model (58), reinforcing the belief that hookworm defence molecules may make good vaccines.

EPIDEMIOLOGICAL EVIDENCE FOR PROTECTIVE IMMUNITY

The epidemiological features of human hookworm infection have been well-described (4,59). Here, we concentrate on those aspects of hookworm epidemiology that may provide evidence for acquired immunity. Mathematical models have shown a number of ways in which immunity may affect the epidemiology of helminth infection (49,60,61). However, in the absence of measures of exposure to infective stages, the interpretation of epidemiological data is difficult. Comparison across communities with differing intensities of infection is likely to be most informative (49).

Experimental infection

Repeated experimental infections have not provided evidence for protective immunity to human hookworm (6,18,19,62,63). In particular, the two individuals repeatedly infected with a high dose of N. americanus did not show any resistance to reinfection, despite developing anti-hookworm immune responses (18,19). However, the number of individuals tested has been very small, and all have been from non-endemic populations, which show increased susceptibility to hookworm infection (15,64).

Age–intensity profiles of infection

Age–intensity profiles for hookworm infection are usually monotonic, with peak intensities occurring in adults, in contrast to the convex age–intensity profiles of infection seen for other human STHs (59). These patterns will reflect variation in both exposure and immunity with host age. An important difference between hookworms and other STHs is that exposure to hookworm infection often increases, rather than decreases, with age. The estimated force of infection with hookworm increased markedly with age in Taiwan (60), and this is likely to be true in many areas where exposure is occupation-related. The monotonic age–intensity profiles often seen for hookworm are not consistent with strong protective immunity in adults. However, mathematical models show that they are entirely consistent with a moderate degree of protective immunity (49). If exposure is constant with age, partial immunity (e.g. a 25–50% reduction in infection rate or mortality in adults) may not produce any observable convexity, especially if immunological memory is short (49). An increase in exposure with age may further reduce the likelihood of convexity being generated.

Peak shift

The strongest epidemiological evidence for acquired immunity in human helminth infection has come from observations of a ‘peak shift’, a reduction in the age at peak intensity of infection with increasing exposure (65). A peak shift is predicted by many mathematical models of acquired immunity, but is not predicted by models without acquired immunity, unless the relationship between age and exposure varies with the intensity of exposure in a precise (and unlikely) way (49). A peak shift has been reported for hookworm infection from a world-wide comparison of age–intensity profiles (65), and comparison of a smaller number of studies suggested that the degree of convexity increases with exposure (61). However, the profiles used were highly variable with respect to geographical location, methodology and hookworm species, which may have confounded the results. For instance, the shape of age–intensity profiles may vary geographically (4). Further comparisons of age–intensity profiles would be useful.

Worm weight, fecundity and pathology

Protective immunity against hookworm in animal models produces not only a reduction in worm burden, but also smaller, less fecund worms, and reduced pathology. Indeed, some vaccination studies have shown pronounced effects on worm size/fecundity (66) or pathology (67), in the absence of any effect on worm burden, suggesting that these may be more sensitive indicators of a protective response. Thus protective immunity may produce a decline in worm size or fecundity with age in human infection. Only one recent study has measured expelled hookworms: male worm weight increased with host age pre-treatment, which may partly reflect worm growth, whereas worm dry weight was negatively correlated with host age after reinfection (68). Analysis of published data on per capita faecal egg production (51,69) shows no relationship with host age (70). These studies did not include young children, in which age-related changes may be most apparent.

Variation in worm burden between individuals

Hookworm burdens are aggregated in the human population, and predisposition to high or low hookworm burdens is well known (59). Predisposition is consistent with host variation in protective immunity, but could equally be explained by variation in exposure. However, recent studies have demonstrated a role for host genetics in generating variation in worm burdens between hosts, with the heritability of hookworm burden (faecal egg count) being 26–37% (Bethony unpublished observations) (71). The genes responsible have yet to be identified, but results from studies of other human and sheep STHs suggest that immune response genes are likely to be important (64). Similarly, the demonstration of predisposition to high or low hookworm weight, as well as worm burden, indicates a role for variation in host susceptibility (72).

IMMUNOEPIDEMIOLOGICAL EVIDENCE FOR PROTECTIVE IMMUNITY

The demonstration of protective immunity in observational field data is not straightforward. Typically, a negative correlation between immune responses and worm burden, or between pre-treatment immune responses and the extent of reinfection after chemotherapy, is taken as evidence for protective immunity. As with all such observational studies, alternative explanations need to be considered. Confounding variables are possible, though known potential confounders such as age and sex can be controlled for statistically. ‘Reverse causation’ due to parasite-mediated suppression of immune responses is also possible, but the likelihood of immunosuppression can be assessed from other evidence. Reverse causation is less likely in reinfection studies (but not impossible if predisposition is strong). Moreover, mathematical modelling of immune responses reveals further complexities. In particular, models predict both positive and negative correlations between worm burdens and protective immune responses (49). Positive correlations are expected in young hosts, and with short-lived immune responses. These models show that an age-structured analysis is more likely to detect protective immunity, but that positive correlations across all ages are not inconsistent with protective immunity (49).

Several studies have found either a positive correlation or no relationship between antibody responses and hookworm burden (18,22,51). However, these studies did not examine correlations in different age-classes. The advantages of an age-structured analysis were shown in analysis of data from Papua New Guinea, where overall correlations were again positive or weak (73). Further analysis demonstrated significant negative correlations between anti-adult ES IgG, IgM and IgE and worm burden, and between anti-larval IgG and both worm burden and the degree of reinfection, in older age-classes (73). These results are highly consistent with those expected from mathematical models of protective immunity, and suggest that antibody responses against both adult and larval hookworms are protective. As discussed earlier, there is no evidence for suppression of these antibody responses during infection (Table 2a).

Further evidence for a protective role of antibody responses comes from recent cross-sectional studies in China and Brazil (8). These demonstrate a significant negative relationship between the intensity of infection and levels of IgE against a recombinant larval protein, ASP-2. In contrast, anti-ASP-2 IgG4 antibody levels were positively associated with worm burdens, whereas other IgG subclasses showed no relationship. Protective effects were specific to ASP-2, and were not seen with antibodies to the related protein ASP-1, nor to crude hookworm extracts. Vaccination with ASP-2 has also been shown to be protective against A. caninum in dogs (8).

Protective effects of cellular immune responses have recently been examined (37,38). In a reinfection study in Papua New Guinea, there was a significant negative relationship between pre-treatment IL-5 production to adult ES antigen and the extent of reinfection (Figure 1a) (38). There was no evidence for parasite-mediated suppression of IL-5 production in this study (Table 2b). The negative correlation between IL-5 and reinfection burden, but not pre-treatment burden, strongly points towards an effect on incoming larvae (38,74). In Brazil, Geiger et al. compared proliferative and cytokine responses in seven uninfected endemic normal and 23 hookworm-infected individuals of similar age (37). Potential effects of immunosuppression were minimized by assessing cytokine production after treatment of infected individuals. Hookworm-specific IFN-γ, IL-5 and IL-13 were significantly higher in endemic normals, whereas IL-10 production was lower. These data are consistent with a role for one or more of these responses in protective immunity.

Figure 1. Immunoepidemiological evidence for protective immunity in human N. americanus infection. (a) Hookworm reinfection 33 months after chemotherapy and IL-5 production in response to adult ES antigen; (b) female worm fecundity and total IgE levels, in Papua New Guinea. Redrawn from (38,68).

Only one study has examined the association between immune responses and parasite size and fecundity. Pritchard et al. found a highly significant negative relationship between female worm weight or fecundity and total IgE production pre-treatment (Figure 1b), and between worm weight and total IgE after reinfection (68). There were also negative correlations between specific IgE and female worm weight at both times, and between eosinophilia and fecundity/male worm weight pre-treatment. These correlations provide strong evidence for a protective effect of Th2 responses acting to reduce worm size and fecundity. The protective effects may act directly on adult worms, though immune damage during the larval stage is also possible. The strongest correlations were seen with total IgE, which may be acting as a marker for Th2 activation. The results were not confounded by either host age or hookworm burden, and are unlikely to reflect immunosuppression, which will depend on worm burden rather than worm size. Future field studies of worm size or fecundity would be useful, since this could provide both a more robust and more sensitive indicator of anti-worm effects.

These results show that there is increasingly strong evidence for protective immunity to human hookworm infection. Strong negative correlations between immune responses and the extent of infection or reinfection have been seen in several studies. Alternative explanations such as immunosuppression are unlikely, though the influence of confounding variables cannot of course be entirely ruled out (14). A key requirement now is replication of these various results, in different communities and epidemiological settings. Consistency of results across studies will be important, but studies to date have generally measured different responses.

MECHANISMS OF IMMUNITY

The immunoepidemiological studies discussed above provide evidence for anti-adult ES antibody effects on worm survival, and anti-larval antibody and IL-5 effects on establishment during reinfection. In addition, reduced growth and fecundity of adult worms is associated with high nonspecific IgE, specific IgE and eosinophilia. The evidence to date thus suggests that protection is associated with Th2 responses. Similarly, up-regulated Th2 cytokine responses are associated with reduced reinfection with T. trichiura (75). Inferences about protective mechanisms from field data are difficult, as the observed responses may correlate with other unmeasured protective responses, rather than being protective themselves. However, several potential mechanisms for protective immunity are suggested. The IL-5 data show that eosinophils may be an important effector mechanism against larvae, as observed in some animal models, though pleiotropic effects of IL-5 may also be important (76). There is evidence that both IgG and particularly IgE responses are also involved in larval killing. IgE and eosinophils may be acting together to kill larvae, e.g. by antibody-dependent cell-mediated cytotoxicity. Antibodies may also inhibit skin penetration or larval feeding: anti-ASP-2 canine sera have been shown to reduce larval entry through skin in vitro (8). Similarly, anti-adult ES antibodies may have direct effects on a number of secreted feeding and immunomodulatory molecules, thus reducing adult worm survival.

However, the mechanisms and site(s) of immune attrition in humans remain elusive. Larval killing may occur in the skin or lungs, and reduced adult worm survival or fecundity may reflect either intestinal immunity, or previous damage to the larval stage. Similarly, despite an effective canine vaccine against A. caninum, the site and mechanisms of vaccine-induced protection are unknown (15). In murine models, as in humans, the Th2 phenotype is associated with immunity to larval stages, and is particularly operative between skin invasion and lung transit (77–79). Immunity can be generated with irradiated larvae or subunit vaccines, although the Th2 phenotype may not be a total prerequisite (58). There is clearly the potential for multiple immune effectors to be involved, at several life-cycle stages. As hookworm larvae enter the skin, they gain essential sustenance (80), and shed a sheath. The shed sheath and a collagen binding protein (81) may divert immune responses away from migrating larvae, particularly if they enter the skin of a primed host. Nevertheless, the Th2-driven IgE response, eosinophils and mast cells may play a role in cuticular attrition at this stage (82,83). The partially damaged hookworm L3 may suffer further eosinophil-driven attrition in the lungs (77); hence the evolution of an eotaxin metalloproteinase (41). Those parasites which reach the gut to undertake a haematophagous existence will be vulnerable to antibody-mediated effector mechanisms operative against parasite feeding (84). The role of calreticulin in inhibiting C1q may be an important defence mechanism in this context (42).

In summary, a combination of larval attrition in the skin, by Type 1 hypersensitivity and resulting secondary inflammation; pulmonary trapping, and eotaxin, IL-5 and the resultant eosinophil mediated attrition; and blood-borne immunity affecting feeding, growth and fecundity in the gut will lead to a reduced and physiologically stunted worm burden. These diverse immune effector mechanisms appear to be counteracted by the parasite's immune suppressive repertoire, resulting in only partial protection. Indeed, this possibility was first demonstrated in mice, where immune stunted Heligmosomoides polygyrus were returned to normal size by the in vivo administration of immune suppressive parasite extracts (85,86).

HOOKWORMS AND CO-INFECTION

The profound immunological changes induced by hookworm infection may have important effects on susceptibility to other infections, and to non-infectious diseases with an immunopathological aetiology. Equally, most hookworm infections occur in hosts exposed to wide range of other immunosuppressive pathogens. There has been much recent interest in immunological interactions in helminth infection. There is increasing evidence for a protective role of helminth infection against asthma and malaria (9,11), whereas helminth infection may predispose to susceptibility for HIV/AIDS or tuberculosis (10), or result in an impaired response to vaccination with bacterial antigens (87).

A large number of epidemiological studies have examined associations between hookworm and other STHs. Individuals with hookworm infection are more likely to be infected with A. lumbricoides and T. trichiura in multiple studies world-wide (88). Positive associations between hookworm and S. mansoni infection have also been seen within communities (89,90), though not across communities (91). While some of these observed associations may be explained by overlapping patterns of exposure (89), variation in susceptibility to multiple species infection may also be important. The latter hypothesis is strengthened by the observations that anti-hookworm (92) and anti-schistosome (Bethony, unpublished observations) lymphoproliferative responses are reduced in individuals co-infected with N. americanus and S. mansoni, suggesting that immunosuppression by both species is important. As discussed earlier, high hookworm burdens also suppress IFN-γ responses to the mycobacterial antigen PPD (38). Conversely, hookworm-specific Th2 responses are down-regulated by concurrent filarial and plasmodial infection (38). Interestingly, two studies have reported a negative association between prior BCG vaccination and hookworm intensity (93) or intestinal nematode (mostly hookworm) infection (94). If confirmed, this would suggest the possibility of protective Th1 responses in hookworm infection, up-regulated by BCG vaccination. However, further studies have not replicated these results (95,96).

The range of potential interactions will require a large number of detailed immunoepidemiological studies. However, the results so far may already have some clinical potential. Hookworm infection has been shown to protect against asthma in Ethiopia (97), and there is a case report of experimental infection alleviating the symptoms of hay fever. These results, and a number of similar studies on other helminths (Cooper this volume), have led to the belief that parasitic infection may be beneficial in the treatment of allergic and autoimmune diseases, because of their anti-inflammatory activity or ability to restore balance to the immune system. Consequently, hookworms are currently on trial in the UK (Nottingham) and Australia (Townsville) for the treatment of seasonal rhinitis and inflammatory bowel disease, respectively (98).

CONCLUSION

This review has highlighted a number of recent developments in human hookworm immunoepidemiology. Despite the difficulties in drawing hard inferences from field data, there is now increasing evidence for the existence of both immunosuppression and protective immunity in human hookworm infection. However, the number of published studies is still very small, and there is a clear need for further work. The immunoepidemiological studies to date provide a patchwork of coverage, with each study examining a limited selection of immunological and parasitological phenotypes. Thus replication of results, and information about relationships between immune responses, has been limited. More detailed studies, including measurements of cellular and antibody responses to a range of parasite antigens, and pre-treatment and reinfection worm burdens/worm weight, in a range of communities of differing intensities of infection, would be ambitious but invaluable.

The immunoepidemiological results, which provide evidence for multiple protective responses to hookworm, appear to contrast with epidemiological patterns, from which evidence for protection has been hard to find. Mathematical modelling has provided important insights here, showing that partial protective immunity (e.g. as a result of parasite-mediated immunosuppression) need not produce declines in worm burden in adults. Extension of these models to include more details of host immunity, such as immunomodulation, is now indicated. The identification of the genes underlying variation in worm burden may provide unambiguous evidence for the importance and mechanisms of protective immunity. Moreover, as noted in the Introduction, we know very little about immune responses to A. duodenale, though the major differences in life-history and epidemiology (99) between the two hookworm species are likely to be reflected in immunological differences. Finally, the potential importance and number of immunological interactions between hookworm infection and other diseases suggests a multitude of profitable investigations. The ease of anthelminthic treatment allows powerful experimental designs comparing infected and uninfected groups. To date, these have largely addressed the effects of killing worms in individuals with already established immune responses. Studies comparing the acquisition of natural immune responses in infected and uninfected groups are logistically harder, but equally important.

ACKNOWLEDGEMENTS

We gratefully acknowledge the support of The Wellcome Trust, the Medical Research Council, and the Human Hookworm Vaccine Initiative for our immunoepidemiological research; the invaluable assistance of the Papua New Guinea Institute of Medical Research; Alan Brown for help in preparing the manuscript, and the cooperation of all the communities studied. JB is a Fogarty International Fellow.