Do hookworms elicit protective immunity in man?

The two main species of human hookworm, Ancylostoma duodenale and Necator americanus (Table 1), are together believed to infect about 900 million people - mainly in tropical countries where adequate sanitary facilities may be lacking. But interactions between the two species, and their relative contributions to observed age-related infection patterns and seasonal cycles of transmission, continue to engender controversy. People tend to remain susceptible to infection through-out life, even with constant exposure to the infective stages. So what role does human immunity or resistance play in the epidemiology and control of infection? In this article, Jerzy Behnke reviews the epidemiology of hookworm infection in the light of current understanding of mechanisms involved in host responses to infection and hookworm evasion of those responses. As he stresses, much further work is required.

It is often stated that A. duodenale is a northern species -the principal agent of human hookworm disease north of 20 ° latitude1, 2. In contrast, N. americanus is considered a southern species. However, this distinction is not so simple because foci ofA. duodenale have been found in many parts of Africa 3-6. Eggs of the two species are not easily distinguishable 1, so estimates of the relative intensity of infection in mixed infections depend either on the recovery of adult worms from the faeces after chemotherapy or on culture of eggs to the infective stage, when the larvae can be individually identiffed. Both are labour intensive exercises and it is not surprising that many workers have not confirmed the identity of the parasites beyond their recognition as hookworms 1. This is a continuing problem in hookworm research, but nevertheless it is quite clear that there are important differences in the biology and host-parasite relationships of N. americanus and A. duodenale. These differences may account known from taxonomic studies on animal parasites but none of these has been studied in the laboratory.
for some of the inconsistencies in epidemiological findings from 'different parts of the World 7, and there is an urgent need to define accurately the extent of infection with both organisms in future studies.

Longevity and Egg Output
Hookworms are considered to be longlived parasites. Few longitudinal studies have been carried out but age-intensity profiles of affected communities reveal that children tend to acquire infection within the first 4 years of life (Fig. 1). This is followed by increasing intensity of infection (monitored through faecal egg counts and expressed as the mean number of eggs per gram of faeces-EPG) which peaks or plateaus in the 20-30 age group. Thereafter EPGs tend to remain stable into late life or even rise further in old age s-10. One study in Nigeria reported a decline in mean EPG in the 45 + age group to a level comparable to that of 1-4 year old children 11. The identity of the hookworm was not given but other studies in Nigeria report that N. americanus accounts for around 88% of the hookworm infections. No other survey to date has yielded quite comparable results, and it is more common for the age-intensity curve to stabilize in adult life-particularly in regions where only Necator occurs, such as Gambia 9 and Puerto Rico 12. A study in Taiwan, where A. duodenale is dominant, found that the mean EPG fell in females aged 45 + but continued to rise in men among whom the heaviest infections were in 55-60 year age group 8. Similar observations have also been made in Indial°, 13. Collectively these studies do not lend support for the existence of immunity to hookworms under field conditions. However, longitudinal studies across 1-2 transmission seasons report fluctuations in EPG (~)1987. Elsevier Publications, Cambridge 0169-4758/87/$0200 which suggest that parasites are regularly lost and subsequently reacquired in the following transmission season. It is exceptional for hookworm larvae to be continuously available throughout the entire year, but even then the density' of infective larvae in contaminated soil oscillates with the seasons. In NigeriaS, 14 the density of hookworm larvae increased over 10-fold from very few in the dry months (January-February) to a maximum in July-October when rainfall was frequent. In Tanzania 3 and West Bengal 13 transmi,;sion seasons are sharply delimited by the hot dry climate between monsoons.
Early studies in India 15 first suggested that faecal egg counts rose during the early part of the rainy season, peaked, and declined in the dry season (Fig. 2). This feature of hookworm epidemiology has been recently coil.firmed 16 and attributed to population changes in A. duodenale rather than N. americanus. In this work 16, the population studied was about equally affected by both species, and the authors concluded that A. duodenale was a relatively short-lived species whilst N. americanus gave rise to more stable, long-lasting infections. This conclusion is supported by several other studies which reveal that N. americanus infections are more stable in the field under endemic conditions. Longitudinal surveys in Brazil 17, Gambia 9 and East  Africa 4 all suggest that where N. americanus is the dominant species, there is no obvious pattern of regular worm loss and reacquisition, despite short term fluctuations in EPG. The exception is amongst children still acquiring worms, where an upward trend is usually discernible in the transmission season 9. Further evidence for the stability of N. arnericanus infections comes from monitoring individuals who have been isolated by hospitalization or imprisonment or have been self-infected. The worms can survive for up to 15 years is and loss is gradual over this period. Nevertheless there is still some disagreement. Nwosu and Anya 5 found seasonal oscillations in EPG coinciding with rainfall, but their data were based on a population in which there was 88% prevalence with N. americanus. At the low point in the dry season, EPGs were still quite high (4000) and the fluctuation in egg counts above this baseline may therefore represent loss and gain of A. duodenale, which although in a minority in this community, is known to be considerably more fecund than N. americanus 16. In areas where A. duodenale is the dominant species, EPGs fluctuate in regular annual cycles in relation to rainfall and transmission season 8,16.
The interpretation of annual cycles in EPG is not straightforward. It is widely assumed that egg counts reflect the number of parasites harboured by the host, but this is not a reliable indicator of parasite bur-den19. Infections of only male worms would not be recognized through faecal egg counts, and larval burdens would also be missed. Furthermore there is evidence for a density-dependent reduction in female fecundity, and the possibility that female worms reduce or even cease egg output in the dry season when transmission is virtually abolished. The relationship between EPG and worm burdens (differential counts for both species) urgently requires examination, and a comparison of worm burdens during dry season and transmission season, following chemotherapy, would go a long way to providing relevant answers.

Arrested development
Another factor which may contribute to fluctuating egg counts is arrested development (hypobiosis) in A. duodenale. Individuals infected with the 'West Bengal strain took 22-38 weeks to develop mature worms considerably longer thorn the normal prepatent period2°, 21. These subjects were not exposed to further infection and it is conceivable that the parasite.'; had lain dormant in the intervening period. Reinterpretation of Maplestone's earlier work 22 reinforces the concept that in West Bengal, A. duodenale may have the capacity for arrested development. Egg counts on prisoners who commenced their sentences just after the monsoon season showed that EPGs fell steadily until February, but from March onwards increasing quantities of eggs were passed, peaking in May despite the absence of reinfection. Field observations provide further support 20. The increase in egg production usually associated with the monsoon season actually began to be apparent in West Bengal in late April when ground conditions were still inappropriate for transmission. A temporary reduction in the fecundity of female worms could explain these observations, but it seems more likely that arrested larvae from the: previous season began to mature in advance of the new transmission season, synchronizing their reproduction to enviromnental conditions most likely to ensure further infections. A. duodenale in the field may therefore have a relatively short survival time, adult worms declining in the dry season leaving newly acquired larvae to survive: the months least suitable for transmission in an arrested state. In contrast, N. americanus continue to produce eggs steadily throughout the year and so, in regions where both species coexist, faecal cultures in the dry season yield a greater proportion of N. americanus larvae 16.

Worm Loss and Immunilty
Does the regular loss of adult hookworms imply immune mediated expulsion of worms and acquired irrmmnity? Evidence for the induction of pI~xective immune responses by hookworm,; in man is still scarce, although recent work supports the idea that some people are predisposed to reinfection following treammnt with anthelmintics 23,24. It may be that A. duodenale burdens decline each year through natural senility, but such a short life span is not compatible with Kendrick's study on experimentally infected prisoners in Madras penitentiary 25. All the prisoners monitored for 12 months or more remained infected, and in some, egg output was maintained for over 5 years. A. duodenale therefore can live for longer than 12 months under conditions free from reinfection. In endemic regions the life span of adult worms is shorter -Nawalinski et al. 16 calculated a 60% annual turnover for A. duodenale in West Bengal.
These observations imply that although a significant proportion of adult A. duodenale are lost in the post monsoon season, some would survive together with arrested larvae (and N. americanus) to continue the cycle into the next season. Perhaps worm loss is selective -acting on individual worms rather than on the entire worm burden. Possibly there are highly localized responses to individual parasites in the intestine, or alternatively age-related changes within the parasites may increase their susceptibility to host effector mechanisms.
It is likely that humans express a range of response levels to hookworm infection, as do sheep to infection with Haemonchus contortus 3o, and inbred mouse strains to intestinal nematodes 26-29. Studies in Tanzania for example, where A. duodenale and N. americanus coexist in almost equal proportions, show that in some children, egg production was rapidly curtailed after the transmission season -while in others, more stable EPG patterns emerged with varying degrees of fluctuation. Individuals who do not pass hookworm eggs have been recognized in all studies despite the generally high prevalence in affected communities. But it is extremely difficult to distinguish between variation in exposure to hookworm larvae, and genetically determined capacity to respond immunologically to infection. Carefully combined epidemiological and immunological observations, continued over several seasons on populations showing contrasting hookworm burdens, single and mixed species infections, would provide much needed information.

Predisposition to Infection
Volunteer studies with N. americanus show no evidence of acquired resistance to reinfection in terms of reduced parasite establishment, delayed maturation or shorter patent period3k Field studies in which populations were treated with anthelmintic suggest that reinfection soon follows after chemotherapy but precontrol levels in the intensity of infection are reached only after several years (Fig. 3). Schad   son 23 found that only 30% of the precontrol worms involving systemically circulating level was restored 630 days after antibody. chemotherapy -in agreement with earlier Among this array of responses there may work32, 33. This slow reacquisition of adult be host protective responses which are hookworms contrasts with the relatively masked by non-protective responses to the more rapid reinfection rates observed after numerous antigens in the parasite preparaelimination of Ascaris lumbricoides, where tions. Localintestinal responses have generpretreatment intensity has been reported ally been ignored but there is some evidence within 12months of chemotherapy ~4.
that IgA may figure prominently in the Two recent studiesZ3, 24 found significant host-parasite relationship. Kumar et al.35 correlation between worm burdens before, reported that IgA in the intestinal aspirates and many months after, treatment. Indi-of hookworm patients was depressed, but viduals heavily infected before treatment, normal levels were restored after chemosubsequently reacquired above average therapy. It may be that the presence of worm burdens -supporting the idea that hookworms led to altered IgA secretion or they are predisposed to heavy infections, reduced IgA half-life in the gut lumen, or But although it is tempting to suggest that the depressed IgA may reflect inactivation predisposition may be linked to genetically of this immunoglobulin by worm products. determined susceptibility and resistance to Hookworms are known to secrete proteases infection, this hypothesis has not yet been which are believed to aid in digesting the specifically addressed in appropriately bolus of host tissues taken in during feeddesigned epidemiological studies, ing31,36. Some of these products may be directed at IgA, as occurs among other Antibody Responses to Hookworms microorganisms living on mucosal sur-Antibody responses to hookworms have faces 37. been documented by a variety of assays Experiments with A. ceylanicum in dogs using L3 and adult worm homogenates 31. lend support for the importance of IgA in Naturally infected patients and volunteers hookworm infection31, 38. A vigorous parainfected with N. americanus show elevated site-specific IgA response was detected in haemagglutinating antibody, complement dogs which resisted a challenge infection fining antibody, raised IgE and anti-parasite administered 28 weeks after primary expoacetylcholinesterase responses. Yet they sure to infective larvae. In contrast, no such remain susceptible to infection 31. There is response was evident in dogs vaccinated no evidence for resistance to human hook-with adult worm antigen, even though they were equally resistant to challenge. Specific IgG levels however, were considerably higher in both groups and this may have reflected responses to the somatic migratory stages: the IgA response detected in reinfected dogs may have been the consequence of enhanced secondary responses in the intestine. Vaccinated animals which did not show an elevated IgA n.~sponse to challenge would not have experienced local mucosal stimulation during vaccination and therefore would not be primed for an anamnestic response to challenge.

Evasion of Immunity
If IgA is involved in protective immunity to hookworms, it appears that the parasites have evolved a counter-strategy to inactivate this immunoglobulin class in the gut. Hookworms may also secrete immunomodulatory components to inactivate local immune responses. Other species causing chronic infections are l~own to do this 39, but there is yet no evidence that comparable strategies have been evolved by hookworms 31. Antigens presented orally can induce life long tolerarLce of cell-mediated responses to dermal challenge and longlasting tolerance of antibody responses, whilst initiating local immunity in the form of specific IgA 4°. Hookworms reside in the gut lumen but their antigenic presentation is not directly akin to the,' oral route. Hookworm mouthparts penetrate deep into the gut mucosa and their excretory products probably enter the circulation -bypassing the normal mucosal-Peyer's patch route of antigen entry from the intestine. The possibility that hookworms interfere with the immunoregulatory circuits in the intestine has not yet been examined experimentally but the hamster models use of hamsteradapted strains of N. americanus and A. ceylanicum31, 44 now offers the possibility of exploring these aspects of the hookworm host-parasite relationship.
Observations on the feeding activity of the dog hookworm, A. caninum, have shown that parasite attachment to the gut mucosa is temporary, lasting for 4-6 hours at most 41. The lesions created during feeding are rapidly repaired by the host but areas of infiltration by inflammatory cells have been observed around parasites embedded in host mucosa 31. The parasites regularly change their feeding sites and may thus avoid potential damage which could be inflicted by inflammatory cells. It may be that the ability of worms to relocate is diminished with age so that, for example, older A. duodenale may be more readily caught in lesions around feeding sites than younger worms or N. americanus.
There is some evidence that hookworms have evolved strategies for evading antibody responses. Adult A. caninum shed immunoglobulin from their cuticle, whilst dead parasites bind antibody which can be detected using immunofluorescent techniques 42. Studies with N. americanus surface-labelled with 125I suggest that they also shed surface antigens 31. This may be an evasive mechanism preventing antibody from adhering to the parasite, although it is difficult to envisage how antibody bound to cuticle could affect hookworm viability.
A specific and continuous intestinal antibody response may explain epidemiological and experimental observations for the slow loss of hookworms. Such a response may cause the gradual deterioration of older worms but not affect young parasites freshly recruited from the environment which still have potent evasive mechanisms. It would also explain why A. duodenale can be lost in the post monsoon season leaving N. americanus unscathed. A separate response against the latter species would be required, and this may be less efficient because the parasite is more robust or employs more effective evasive strategies than A. duodenale.

Strategies for Vaccination
Whatever the nature of immune involvement in hookworm infection, it is clear that the parasites have the upper hand in most people living in endemic areas. A vaccine against hookworms must therefore be a considerable improvement on natural human responses to these parasites. However, it is quite possible that antigens will be identiffed which are not normally made available to the host during natural infection. Two such possibilities are apparent -high affinity antibodies against key components of the various nematode sensory receptors 43 may block sensory input and so inhibit parasite attachments, and, because hookworms pass large quantifies of serum through their intestines, a serum antibody response against the parasite intestine may cause sufficient damage for the worms to be lost. During natural infection, such antigens may never be exposed or may present insufficient antigenic challenge to the host, but their isolation may be feasible using fresh parasites grown in hamstersenabling vaccine production through gene cloning and antigen synthesis by recombinant DNA methods 46.
There is considerable scope for future

Medical Research Council
Laboratories, Fajara, Banjul, The Gambia research on hookworms, using recently developed animal models 44 and by epidemiological projects linking immunological assays with resistance and worm loss in the field. Understanding the mechanisms involved in worm loss will be an important step in potential vaccine production, and I hope that a concerted effort to apply the tools of modern experimental medicine will help to elucidate aspects of hookworm biology which have eluded experimenters in the past.