Complex modes of transmission and sources of infection

One of the most challenging aspects of preventing exposure to Toxocara infection is the complexity of the routes of transmission and sources of infection, a number of which we still know remarkably little about.

Adult worms of Toxocara spp. reside in a wide range of domestic and feral definitive hosts. For example, Toxocara canis infects domestic dogs, foxes, wolves and coyotes (O'Lorcain, Reference O'Lorcain1994a ; Segovia et al. Reference Segovia, Torres, Miquel, Llaneza and Feliu2001; Roddie et al. Reference Roddie, Holland, Stafford and Wolfe2008a ; Wapenaar et al. Reference Wapenaar, Barkema and O'Handley2013) and Toxocara cati (syn. Toxocara mystax) infects cats and other felids (Fisher, Reference Fisher2003). In contrast, Toxascaris leonina infects both dogs and cats (Miyazaki, Reference Miyazaki1991). Environmental contamination with Toxocara ova is extensive as a consequence of faecal deposition by both domestic and feral definitive hosts and under appropriate conditions of temperature and moisture, such eggs can embryonate and become potentially infective (Traversa et al. Reference Traversa, Frangipane di Regalbono, di Cesare, La Torre, Drake and Pietrobelli2014). Eggs containing infective larvae can be transmitted to humans via contaminated soil, food or utensils. Somatic larvae in the tissues of definitive hosts (particularly dogs – see Coati et al. Reference Coati, Schneider and Epe2004) can re-activate and resume migration and eventually develop into adult worms in the intestine (Schnieder et al. Reference Schnieder, Laabs and Welz2011).

Toxocara eggs can infect numerous paratenic hosts including humans. Toxocara do not develop to adulthood in such hosts but remain as third-stage larvae in their tissues (Brunaska et al. Reference Brunaska, Dubinsky and Reiterova1995; Strube et al. Reference Strube, Heuer and Janecek2013). Paratenic hosts can act as food items for both humans and definitive hosts. Meat-borne transmission via the consumption of raw or undercooked liver has been implicated in human infection (Salem and Schantz, Reference Salem and Schantz1992; Yoshikawa et al. Reference Yoshikawa, Nishiofuku, Moriya, Ouji, Ishizaka, Kasahara, Mikasa, Hirai, Mizuno, Ogawa, Nakamura, Maruyama and Akao2008). In a fascinating recent report, an elderly male patient from France was infected with Toxocara after ingestion of live slugs (apparently a long-standing daily intake of slugs as an alternative therapy for gastro-oesophageal reflux) (Fellrath and Magnaval, Reference Fellrath and Magnaval2014). Terrestrial molluscs are not recognized as paratenic hosts for Toxocara, however the authors hypothesize that the slugs captured embryonated eggs in their mucus thereby playing a phoretic role. More recently, another potential source of infection has been identified with the discovery of the presence of Toxocara eggs on the hair of domestic definitive hosts such as dogs and cats.

There are, therefore, four key epidemiological reservoirs of Toxocara: intestinal infections in definitive hosts (dogs, cats and foxes), eggs in the environment, larvae in paratenic hosts and somatic larvae in the definitive host (Overgaauw and Van Knapen, Reference Overgaauw and Van Knapen2013). As outlined below, there is also the presence of Toxocara eggs on the hair of definitive hosts, but this physical extension of the environmental reservoir of eggs would appear to be less important for transmission. What emerges from this complexity of infection sources is that we still lack knowledge of the relative importance of some sources; but from the point of view of control, it is quite clear that tackling intestinal infections in domestic dogs and cats is by far the easiest way of reducing environmental contamination and therefore exposure of both definitive and paratenic hosts to infection. However, as revealed by high levels of seroprevalence in humans, especially in tropical and subtropical regions and among the disadvantaged, it is clear that such an approach is not being implemented in a sufficiently rigorous manner to reduce exposure successfully.

The role of the veterinarian with respect to key sources of infection

Reducing infection in domestic animals such as dogs and cats underlines the key front-line role that veterinarians play in the provision of anthelmintic treatment and the education of pet owners and the general public. However, this is a challenge because, for the most part, Toxocara infection is not pathogenic in adult dogs and cats and other infections are of greater immediate concern to the clients of veterinarians. Despite this lack of concern, Toxocara is a parasite of significant zoonotic potential (Holland and Smith, Reference Holland and Smith2006) and therefore, veterinarians need to extend their role, beyond that of the immediate benefit to their patients, to the education of pet owners and the broader society. This requires communication between medical clinicians and veterinarians in supporting each other with the common goal of reducing overall incidence of this disease. However, toxocariasis is also low on the priorities of most medical clinicians, even ophthalmologists and this is not just a problem restricted to Toxocara but also for other neglected diseases (Parise et al. Reference Parise, Hotez and Slutsker2014).

One recent informative study undertaken by veterinarians in the Netherlands highlights a number of challenges for pet owner education and awareness. A comprehensive survey of the owners of over 900 household dogs over 6 months of age was conducted with epidemiological data collected on prevalence of gastrointestinal parasites and a detailed questionnaire on risk factors. The prevalence of Toxocara was 4·6% overall, peaking in 6–12 month-old dogs (7·8%). Risk factors included age, ranging freely, coprophagy and recent kenneling. Only 16% of the dogs were dewormed four times per year (see European Scientific Counsel Companion Animal Parasites (ESCCAP) guidelines, 2010) and only 14% of these were dewormed for public health reasons (other reasons included the dog's health, dogmatic (‘because we must’), or a combination of dog's health and public health) (Nijsse et al. Reference Nijsse, Ploeger, Wagenaar and Gras-Mughini2015a ). Among non-coprophagic, kenneled and leashed dogs that were treated four times per year, no Toxocara infection was detected. The authors concluded that owner knowledge was insufficient to expect sound decisions on routine deworming.

In a very useful study that should be emulated in other countries, Palmer et al. (Reference Palmer, Thompson, Traub, Rees and Robertson2008) undertook a national survey of the gastrointestinal parasites of dogs and cats in Australia. This ambitious study was undertaken in the context of a very high pet ownership with 53% of all households owning a dog or cat. A total of 2463 faecal samples (1400 canine and 1063 feline) were collected from both urban and rural locations within three climatic zones – tropical, arid and temperate. The prevalence of T. canis and T. cati in dogs and cats was low compared with previous data collated from the 1970–80s with values of 1·2% (CI 95% 0·6–1·8) and 3·2% (CI 95% 2·1–4·3), respectively. The authors concluded that the frequent administration of anthelmintic treatment had a significant impact on gastrointestinal parasitism in Australia.

Although it is quite clear that puppies and kittens represent the most important source of Toxocara infection, there has been some debate about the significance of older dogs as sources of infection and a lack of knowledge of the dynamics of infection in such animals. The role of older animals as a reservoir of infection should not be underestimated, hence the ESCCAP guidelines of average frequency of treatment four times per year for adult dogs and cats (Overgaauw and Van Knapen, Reference Overgaauw and Van Knapen2013). Experimental data has established that low doses of 100 embryonated eggs can induce patency in adult dogs, whereas previous experiments that utilized high doses failed to do so (Fahrion et al. Reference Fahrion, Staebler and Deplazes2008). This introduces what Overgaauw and Van Knapen (Reference Overgaauw and Van Knapen2013) describe as the therapeutic paradox whereby with declining exposure, there is an increasing probability that older animals will get infected and harbour patent infections.

The epidemiological significance of Toxocara eggs on the hair of definitive hosts

Contact with soil contaminated with infective (embryonated) eggs is considered to be the primary route of transmission of Toxocara spp. to humans. This is in part due to the fact that Toxocara eggs are not infective on shedding but require a period of time, under appropriate environmental conditions, to develop to infectivity. However, a recent focus of interest has been the epidemiological significance of the presence of Toxocara eggs on the hair of domestic pets, particularly dogs, raising the possibility of direct transmission of infection to humans through close contact with their pets.

Some much earlier work by Hasslinger et al. (Reference Hasslinger, Jonas and Berger1973) examined the hair of 17 owned and stray cats and found a single animal (6% prevalence) to have eggs of Toxocara mystax (= T. cati) on its hair. No attempt was made to count the number of eggs or to determine whether the eggs were embryonated or even viable. No further observations were reported until 30 years later, when Wolfe and Wright (Reference Wolfe and Wright2003) detected a higher prevalence of Toxocara eggs on the hair of 60 dogs and designated the viability and embryonation status of the eggs. Since then a number of studies from a range of geographical locations have provided data on dog (and in few cases cat) hair as a potential source of infection. Despite variation in sample size, dog status (stray vs owned, breed, coat type, age and sex), the location on the body from where the samples were taken and the methods of detection, a number of clear trends have emerged. The prevalence of Toxocara eggs on hair is higher in stray vs owned dogs, with a peak prevalence of 67% (rising to 100% in the sampled puppies) among a sample of stray dogs reported by Roddie et al. (Reference Roddie, Stafford, Holland and Wolfe2008b ) (see Table 1). As might be expected, the total numbers of eggs detected on hair vary considerably but the pattern for higher numbers on stray animals remains (Table 1).

Table 1. Summary of studies on the presence of Toxocara eggs on the hair of dogs and cats

Two studies provided important refinements to the approach. One examined the relationship between eggs on the hair and the worm burden of Toxocara at postmortem (Roddie et al. Reference Roddie, Stafford, Holland and Wolfe2008b ). The relationship varied significantly with age. Puppies demonstrated a highly significant relationship between eggs on the hair and worm burden. In contrast, no such relationship existed for adult dogs. This provides indirect evidence for the likelihood that puppies are deriving hair-borne eggs from their own worm infections, whereas adult dogs may be picking up eggs from the environment. The second study provided an important standardization of the egg detection method (Overgaauw et al. Reference Overgaauw, Van Zutphen, Hoek, Yaya, Roelfsema, Pinelli, Van Knapen and Kortbeek2009).

The most significant observation that emerges from the work undertaken so far is the very low numbers of embryonated eggs detected on host hair. Most studies categorized eggs detected as being non-viable, viable, embryonating or embryonated (Roddie et al. Reference Roddie, Stafford, Holland and Wolfe2008b ). Proportions of embryonated eggs ranged from the highest value of 8·1% (Aydenizoz-Ozkayhan et al. Reference Aydenizoz-Ozkayhan, Yagci and Erat2008) to 0% (Overgaauw et al. Reference Overgaauw, Van Zutphen, Hoek, Yaya, Roelfsema, Pinelli, Van Knapen and Kortbeek2009; Amaral et al. Reference Amaral, da, Rassier, Pepe, Gallina, Villela, Nobre, de, Scaini and Berne2010; Keegan and Holland, Reference Keegan and Holland2010; Oge et al. Reference Oge, Oge, Ozbakis and Gurcan2014). It is of interest to note that even in the study by Roddie et al. (Reference Roddie, Stafford, Holland and Wolfe2008b ) – where a very large number of eggs (39 120 in total) were detected – only 0·3% were embryonated (0·312% in puppies, 0·120% in adult dogs). In this context, the proportion of embryonated eggs is a key epidemiological measure and the data suggest that embryonation on hair is a rare phenomenon, particularly among well-cared-for dogs such as those described by Keegan and Holland (Reference Keegan and Holland2010). Therefore the risk of exposure of humans to such a source of infection is likely to be negligible.

However, an experimental approach undertaken by Keegan and Holland (Reference Keegan and Holland2013) tested the embryonation rates of T. canis eggs in soil vs hair under laboratory conditions. T. canis eggs were exposed to two temperature (10 and 20 °C) and two moisture regimes (with and without the addition of water) in the contrasting media over an 8-week period. Embryonation is possible in the medium of hair but the rate of development is significantly lower than that observed in the medium of soil. Temperature is an essential factor determining the rate of embryonation. For example, no embryonation whatsoever occurred at 10 °C with no water added. However, at the higher temperature of 20 °C in the presence of water, embryonation rates were higher in both soil and hair but significantly higher still in soil compared with hair (Keegan and Holland, Reference Keegan and Holland2013).

Although the presence of Toxocara eggs on the hair of foxes is unlikely to be of any epidemiological significance to humans, comparative data from stray dogs and foxes do provide a possible insight into the epidemiology of Toxocara eggs on the hair of dogs. The prevalence and mean eggs per gram of hair on adult dogs are higher than that of adult foxes (prevalence 45 vs 25%, mean eggs per gram of faeces (epg) 279 vs 1·3). In contrast, higher worm burdens were detected in foxes compared with dogs (prevalence 22·5 vs 61%, mean worm burden 3·55 vs 0·5) (Roddie et al. Reference Roddie, Holland, Stafford and Wolfe2008a , Reference Roddie, Stafford, Holland and Wolfe b ). The relationship between eggs on the hair and worm burden in foxes was not significant, analogous to that found for adult dogs. This provides indirect evidence that in comparison with foxes, adult dogs are acquiring higher numbers of eggs on their hair. This may be explained by the fact that such dogs inhabit more contaminated environments (due to the presence of other dogs). Lee et al. (Reference Lee, Schantz, Kazacos, Montgomery and Bowman2010) speculated about the role of scent-rolling among dogs in the acquisition of eggs on the hair. To conclude, it would appear, based upon the available evidence, that there is a low risk of transmission associated with the very low numbers of embryonated eggs found on hair, but the suitability of hair as a medium for oval development should not be ignored.

Variation in exposure to Toxocara infection and the implications for disease

Humans act as paratenic hosts for Toxocara infection. On ingestion, embryonated eggs hatch in the small intestine and release larvae that migrate through the tissues and some organs, but do not successfully develop to adulthood in the small intestine as they do in definitive hosts. Definitive diagnosis of Toxocara infection is by histopathological examination and morphological and morphometric or molecular identification of larvae in tissue samples (Smith and Noordin, Reference Smith, Noordin, Holland and Smith2006). However, such diagnosis requires biopsy material, the collection of which is invasive and may not even contain larvae of Toxocara. It is not a practical approach for routine examination. Therefore, the mainstay of diagnosis of human toxocariasis is serology (Smith et al. Reference Smith, Holland, Taylor, Magnaval, Schantz and Maizels2009).

There is significant variation in seroprevalence values worldwide and a number of potentially confounding factors that may influence the observed values. These include the specificity of the serological test employed, the choice of Toxocara antigen, what cut-off titre was employed to designate seropositivity and the composition of the population surveyed (Holland et al. Reference Holland, O'Lorcain, Taylor and Kelly1995). However, despite these caveats, some trends can be observed which include a generally higher seroprevalence in the tropics and among the disadvantaged. For example, a recent seroprevalence survey of primary school children from an urban slum in Lagos state, Nigeria, revealed an 86·1% seropositivity (Western blotting) with risk factors including child age, contact with young dogs, the feeding location of the dogs, consumption of raw vegetables and drinking unboiled water (Gyang et al. Reference Gyang, Akinwale, Lee, Chuang, Orok, Ajibaye, Liao, Chen, Chou, Huang, Barghouth and Fan2015). However, in the same country, but from Jos, a seroprevalence value of 29·6% (ELISA, enzyme-linked immunosorbent assay) was reported (Ajayi et al. Reference Ajayi, Duhlinska, Agwale and Njoku2000).

Even within Europe, where seroprevalence values are much lower than those reported from Nigeria, significant variation is observed. A study of over 3000 serum samples from Denmark reported a seroprevalence of 2·4% (ELISA followed by western blotting) (Stensvold et al. Reference Stensvold, Skov, Moller, Jensen, Kapel, Peteresen and Nielsen2009). This contrasts markedly with two values reported from Ireland – a large-scale study of schoolchildren (from randomly selected schools) with a value of 31% (ELISA cut-off titre 1:50) (Holland et al. Reference Holland, O'Lorcain, Taylor and Kelly1995), and a more focused (non-random) investigation of asthmatics and their families with a seroprevalence value of 52·1% (ELISA cut-off titre 1:50) (Taylor et al. Reference Taylor, Keane, O'Connor, Mulvihill and Holland1988).

In a clinic-based case control study conducted in the USA to explore the relationship between exposure to Toxocara and asthma, investigators found a striking difference in seroprevalence among children, aged 2–15 years, from two towns in Connecticut – 6·1% in New Haven and 28% in Bridgeport (ELISA cut-off titre 1:32) (Shargi et al. Reference Shargi, Schantz, Caramico, Ballas, Teague and Hotez2001). Furthermore, children who were seropositive for Toxocara were 12 times more likely to be Puerto Rican, eight times more likely to be of other Hispanic origin and seven times more likely to be of Negroid than Caucasian origin.

In a significant departure in terms of scale, a number of investigators have taken advantage of the US National Health and Nutrition Examination Survey (NHANES), which is a nationally representative cross-sectional survey of over 33 000 people. The third such survey took place between 1988 and 1994, and serum samples were analysed for the presence of Toxocara antibodies. The age-adjusted seroprevalence for toxocariasis was 13·9% and was higher in non-Hispanic blacks (21·2%) than both non-Hispanic whites (12%) and Mexican Americans (10·7%). Increased Toxocara seropositivity was associated with a number of variables: the levels of education of the head of household, poverty, elevated blood lead concentrations and dog ownership (Won et al. Reference Won, Kruszon-Moran, Schantz and Jones2008). This large scale national study was possible because the evaluation of exposure to Toxocara was linked to the NHANES survey which was designed to collect health statistics from a large sample of people representative of the civilian, non-institutionalized general US population. The authors highlighted that the striking differences in seroprevalence could be used to target health education messages. Such data from an equivalent European population would be very valuable.

In further work utilizing the NHANES data set, Congdon and Lloyd (Reference Congdon and Lloyd2011) evaluated the relative risk of Toxocara infection for 20 396 survey subjects using a binary regression model that incorporated demography, family poverty and geographic location. It is of interest that even after allowing for the elevated risk associated with poverty, ethnicity still played an important role in explaining increased risk of exposure. Prevalence estimates were particularly elevated among non-Hispanic blacks, most notably in the South and Northeast. Hotez (Reference Hotez2008) speculated that this elevation might reflect differential contextual exposures linked inter alia to ethnic residential clustering and segregation, as distinct from the impact of family poverty per se. These observations are particularly potent when linked to the findings of Walsh and Haseeb (Reference Walsh and Haseeb2012) as discussed in the section on cerebral toxocariasis.

A major gap in our understanding of the epidemiology of human toxocariasis is our persistent inability to distinguish exposure to T. canis vs T. cati using serological methods. In many respects our knowledge of T. cati is depauperate (Fisher, Reference Fisher2003), but its relative contribution to human exposure remains one of the most pressing issues. In a recent report by Poulsen et al. (Reference Poulsen, Skov, Yoshida, Skallerup, Maruyuma, Thamsbourg and Nejsum2015) western blotting was used to attempt to distinguish between sera obtained from pigs infected with T. canis and T. cati. No proteins were observed that could be used to discriminate between the two ascarid species. The authors emphasized the pressing need to develop species-specific serological methods in order to evaluate the relative significance of T. canis vs T. cati as aetiological agents of human exposure and consequent disease.

Since the discovery by Wilder (Reference Wilder1950) of granulomata in the eyes of children that had been mistakenly diagnosed with retinoblastoma, and the subsequent description of what are now known to be third-stage larvae of T. canis in histological sections from such granulomata by Nichols in 1956 that led to the description of OT, the number of syndromes associated with human toxocariasis has expanded. At present there are four distinct clinical entities – visceral larva migrans (VLM), OT, covert toxocariasis (CT) and cerebral toxocariasis or neurotoxocariasis (NT). However, the relationships between these clinical entities and specific symptoms or clinical features are not well understood particularly because of the non-specific nature of most of symptoms (Smith et al. Reference Smith, Holland, Taylor, Magnaval, Schantz and Maizels2009).

One of the key difficulties in evaluating the public health significance of toxocariasis is what exposure actually means. In other words, we can say that an individual is exposed to the Toxocara parasite as a consequence of detecting a positive titre at a particular cut-off, but what is the significance of such a diagnosis in terms of disease or symptomatology? In a study of 221 individuals (comprising both patients attending an outpatients clinic that had been identified as having high titres and asthmatic patients and their families), the relationship between titre and clinical and laboratory features was explored (Taylor et al. Reference Taylor, Keane, O'Connor, Mulvihill and Holland1988). Titres were divided into three categories (I = low or negative titre 0–0·29 n = 41; II = moderate, titre 0·30–0·69 n = 51; and high titre ⩾ 0·7 n = 129) and the highest proportion of patients with 9–16 clinical features was found amongst those with the highest titres (51%). The clinical features most commonly associated with Toxocara titres of 0·30 and above were abdominal pain, hepatomegaly, anorexia, nausea, vomiting, lethargy, sleep and behaviour disturbance, pneumonia, cough, wheeze, pharyngitis, cervical adenitis, headache, limb pains and fever. All of these non-specific clinical features are commonly reported in childhood and this study provides support for the view that a proportion of cases of ‘idiopathic abdominal pain of childhood’ is due to toxocariasis (Taylor et al. Reference Taylor, Keane, O'Connor, Girdwood and Smith1987). In the Taylor study, 61% of the subjects that reported abdominal pain had raised Toxocara titres.

In an important study providing the kind of data that are sorely lacking, Wisniewska-Ligier et al. (Reference Wisniewska-Ligier, Wozniakowska-Gesicka, Sobolewska-Dryjanska, Markiewicz-Jozwiak and Wieczoek2012) evaluated the clinical course of toxocariasis in 103 Polish children who were treated at a hospital's zoonotic diseases outpatient facility. Among clinically diagnosed children ELISA absorbance values between 32 and 100% were deemed to confirm infection. Children were aged between 1·4 and 14·7 years and lived in towns or villages. The clinical symptoms associated with toxocariasis are shown in Table 2. The vast majority of children were diagnosed with the covert form of the disease (95·1%) with 4·9% with OT. The most common symptom was abdominal pain (Table 2). Children were treated with either albendazole or mebendazole or a combination of both anthelmintics while those children with eye disease were treated in consultation with ophthalmologists and received thiabendazole, albendazole and/or mebendazole in combination with steroids. Some children required repeated treatment with 45% receiving three courses of anthelmintic treatment. After treatment, the mean titre and the number of children with abdominal pain and lymphatic node enlargement declined, but no decline in headaches was observed after one and two courses of treatment. Even after the third course, the decline did not attain statistical significance (Table 2). In some children, despite the observation of negative titres, symptoms persisted and this was particularly marked for headaches. The authors suggest that this diminishes the value of using headache as a symptom of covert toxocariasis. The authors conclude that due to the risk of eye disease, anti-parasitic treatment should be implemented, but their observations provide important evidence of the long-term persistence of symptoms despite several rounds of treatment.

Table 2. Characteristics and symptomatology of children with toxocariasis (Wisniewska-Ligier et al. Reference Wisniewska-Ligier, Wozniakowska-Gesicka, Sobolewska-Dryjanska, Markiewicz-Jozwiak and Wieczoek2012)

* Values differ statistically significantly before and after treatment, P ≤ 0·05.

To conclude, our knowledge to date indicates that seroprevalence values can be high and vary significantly even within the same country, that higher titres are associated with greater symptomatology and that symptoms can persist despite rounds of anthelmintic treatment. However, we still do not know the relative contribution of T. canis vs T. cati to human exposure and there is a paucity of studies linking exposure to disease. The exemplary large-scale study by Won et al. (Reference Won, Kruszon-Moran, Schantz and Jones2008) indicates that certain high-risk populations can be identified by means of seroprevalence surveys and that public health education could be targeted to such groups.

The extent of ocular toxocariasis

Toxocara larval involvement in the eye, with consequent visual impairment, remains potentially the most devastating of all human sequelae (Good et al. Reference Good, Holland, Taylor, Larragy, Moriarty and O'Regan2004). Ocular toxocariasis is generally described as a relatively rare disease primarily observed in children (Taylor, Reference Taylor, Holland and Smith2006). In this context, one major gap in our knowledge relates to population-based assessments of the prevalence of infection. One of the reasons for this is that toxocariasis is not a reportable disease. A population-based study of 121 156 school children in Ireland reported a prevalence of consultant-diagnosed toxocaral eye disease of 6·6 cases per 100 000 persons that increased to 9·7 cases per 100 000 persons once both definite and strongly suspected cases were included. Geophagia and a history of convulsion were associated with toxocaral eye disease for both school and county-based case control studies (Good et al. Reference Good, Holland, Taylor, Larragy, Moriarty and O'Regan2004). This study provided important data on the prevalence of eye disease among a defined population. It indicates that OT is a rare disease among Irish children (aged 3–19 years) – a country where the previously-recorded seroprevalence rate among a similar schoolchild population (31%) was relatively high compared with other European countries (Holland et al. Reference Holland, O'Lorcain, Taylor and Kelly1995; Smith and Noordin, Reference Smith, Noordin, Holland and Smith2006).

To our knowledge, only one other documented estimate of the prevalence of OT has been published (as an abstract). It reports data collected from eye clinics in Alabama over a 6-month period and yields an estimate of 1 case per 1000 persons (increasing to 11 cases per 1000 persons once ophthalmoscopy had been carried out) (Maetz et al. Reference Maetz, Kleinstein, Federico and Wayne1987). This value indicates that the prevalence of eye disease is higher in Alabama than in Ireland; indeed it is of interest to note that a more recent web-based survey of American ophthalmologists reported that among 68 patients diagnosed with OLM, 57% emanated from the South (Centres for Disease Control and Prevention, 2011). The most common symptom was vision loss (83%) (n = 37) with permanent vision loss among 68% (n = 25). The preponderance of neglected parasitic infections, including toxocariasis, in the Southern USA was highlighted by Hotez (Reference Hotez2014b ).

Most reports of OT tend to focus on children since the disease is more common in those age groups (see Taylor, Reference Taylor, Holland and Smith2006). However, a recent paper described the clinical features and course of ocular infection in adults in South Korea (Ahn et al. Reference Ahn, Woo, Jin, Chang, Kim, Ahn, Heo, Yu, Chung, Park and Hong2014). This retrospective cohort study included 101 adult patients diagnosed clinically and serologically with OT. The vast majority of patients (92·1%) were diagnosed by means of the presence of a retinal granuloma and 17 (16·8%) had severe vision loss. Ingestion of raw cow liver and meat was significantly more common among OT cases compared with controls. Combined treatment with albendazole and corticosteroids reduced intraocular inflammation and recurrence.

To conclude, we particularly lack information on the prevalence and pathological implications of OT in populations from the tropics where seroprevalence values can be as high as 93% (Smith and Noordin, Reference Smith, Noordin, Holland and Smith2006).

Cerebral toxocariasis: the most cryptic of all disease syndromes?

The presence of Toxocara larvae in the human brain, now described as cerebral toxocariasis or NT, has always been a source of fascination for parasitologists. But of all the disease entities so far described, it is the most cryptic. However, due to a combination of factors – such as improved diagnosis, greater awareness and the use of animal models – our understanding of the condition is becoming better understood and its public health significance, is improving. Cerebral toxocariasis was first described in an autopsy study of a child, in whom a larva was found in the left thalamus. Initially it was described as a larval Ascaris, but after the work of Nichols (Reference Nichols1956) it was correctly identified as a larva of T. canis (Beautyman and Woolf, Reference Beautyman and Woolf1951, Beautyman et al. Reference Beautyman, Beaver, Buckley and Woolf1966). In a recent review of the literature, Fan et al. (Reference Fan, Holland, Loxton and Barghouth2015) found reports of 86 patients with various neurological manifestations. Nevertheless, given the high levels of exposure to Toxocara, the extent of neurological involvement and its manifestations are likely to be significantly underestimated.

One key gap in our knowledge relates to the public health significance of the presence of what are likely to be small numbers of larvae in a human brain. Clearly in some cases the impact can be profound. For example, the symptoms of eosinophilic meningitis were described by Vidal et al. (Reference Vidal, Sztajnbok and Seguro2003) in a 2 year-old boy and included mental confusion, fever, headache, tachycardia, hyperreflexia, dyspnea, lethargy, irritability, motor weakness and nuchal rigidity. Toxocara specific immunoglobulin G (IgG) antibodies were detected in both cerebrospinal fluid (CSF) and serum and most symptoms declined in response to treatment with albendazole and corticosteroids. However, many other infections may be asymptomatic or manifest as non-specific behavioural alterations.

There is a paucity of studies exploring the relationship between exposure to Toxocara and neuropsychological parameters in young children (Holland and Hamilton, Reference Holland and Hamilton2013). However in an important contribution, Walsh and Haseeb (Reference Walsh and Haseeb2012) measured different components of cognition in a large number of seropositive and seronegative children from the USA, again utilizing the NHANES survey database. Seropositive children (n = 688) had significantly lower measures of cognitive function compared with seronegative children (n = 3261), after controlling for important potentially confounding variables such as socio-economic status, gender, ethnicity, residence, cytomegalovirus and blood lead levels. Despite being an associational study, these findings are important, particularly as evidence from mouse models has demonstrated clearly reduced learning and memory in infected animals (Hamilton et al. Reference Hamilton, Stafford, Pinelli and Holland2006).

The logistical and ethical challenges associated with conducting studies on the relationship between cognition and parasitism are manifest (Bundy et al. Reference Bundy, Kremer, Bleakley, Jukes and Miguel2009). Consequently, we need good quality laboratory studies in animal models to unravel the relationship between behavioural deficits, neuropathology and larval burden (Fan et al. Reference Fan, Holland, Loxton and Barghouth2015). Mice are useful in this respect because of the ease of manipulation and the availability of inbred murine strains (Holland and Hamilton, Reference Holland and Hamilton2013).

Early work established the fact that T. canis larvae accumulate in the murine brain (Dunsmore et al. Reference Dunsmore, Thompson and Bates1983) and can remain there for long periods of time (Bardon et al. Reference Bardon, Cuellar and Guillen1994). Cox and Holland (Reference Cox and Holland2001a ) explored the optimum dose of Toxocara eggs required in experimental infections and concluded that low dose infections mimic those found in wild mice, whereas higher doses may be selected by the experimenter in order to achieve marked accumulation over time. A wide range of behavioural alterations was observed in both outbred and inbred mice experimentally infected with Toxocara (Cox and Holland, Reference Cox and Holland1998; Cox and Holland Reference Cox and Holland2001a , Reference Cox and Holland b ). However, of particular relevance to human infections was the observation of reduced learning and memory in inbred susceptible BALB/c (inbred strain of mouse) mice infected with Toxocara (Hamilton et al. Reference Hamilton, Stafford, Pinelli and Holland2006).

Among mouse model studies of the link between cerebral toxocariasis and memory impairment, Holland and Hamilton (Reference Holland and Hamilton2013) identified a variety of important host and parasite variables, some of which may be confounders. These include inbred mouse strain, infective dose, duration of infection, method of larval recovery, behavioural tests employed and choice of immunological/neuropathological measures.

To conclude, we now require both data from longitudinal studies in humans to examine the etiological connection between toxocariasis and impaired cognitive function (Walsh and Hasseb Reference Walsh and Haseeb2012), and from animal model work dissecting simultaneously the relationship between learning and memory deficits, neuropathology and immunological responses (Holland and Hamilton, Reference Holland and Hamilton2013; Fan et al. Reference Fan, Holland, Loxton and Barghouth2015).

Improved integration of our understanding of environmental contamination with eggs of Toxocara spp.

The presence of potentially infective or infective eggs of Toxocara spp. in the environment is one of the key routes of transmission to humans. Infected definitive hosts such as dogs, cats and to a lesser extent foxes release their feces, and the helminth eggs within them, into the environment. However, our understanding of the relative importance of the different host sources is incomplete and more hypothesis-driven studies with better integration are urgently required. One of the difficulties is that there are many studies conducted every year on the extent of environmental contamination with Toxocara eggs, but because they tend to vary methodologically, valid comparisons between regions, countries and continents are problematic. For example, in a recent review of environmental contamination with helminth eggs, Traversa et al. (Reference Traversa, Frangipane di Regalbono, di Cesare, La Torre, Drake and Pietrobelli2014) tabulated 28 different surveys from both temperate and tropical countries worldwide, with contamination rates for roundworms ranging from a low value of 0·5% to a high of 79·4%. However, when looking at these data it is very difficult to identify any clear trends and considerable variation is exhibited, even within the same geographical location. For example, in Poland soil contamination varied from 3·2 to 26·1% and in Italy from 0·7 to 33·6%. Most studies collected soil samples, but how such samples were collected (site, number and volume, season, representation, depth, soil quality, etc.) and analysed can vary considerably (Mizgajska-Wiktor and Uga, Reference Mizgajska-Wiktor, Uga, Holland and Smith2006).

Furthermore, the identification of Toxocara eggs to species level (i.e. T. canis vs T. cati vs T. leonina) is not always performed and discrimination is not easy (Mizgajska-Wiktor and Uga, Reference Mizgajska-Wiktor, Uga, Holland and Smith2006). In a survey of public playgrounds in Dublin, Ireland, O'Lorcain (Reference O'Lorcain1994b ) recorded almost entirely T. canis eggs with no T. cati and a single T. leonina egg. In contrast, from Krakow and two nearby villages, Mizgajska (Reference Mizgajska2000) recorded, 90% of the eggs recovered as T. cati with only 10% being T. canis. At the light microscope level, T. canis and T. cati can be distinguished by size (O'Lorcain, Reference O'Lorcain1994b ). Recently, a duplex quantitative real-time polymerase chain reaction (PCR) assay has been described that can detect and discriminate between the eggs of T. canis and T. cati in soil samples (Durant et al. Reference Durant, Irenge, Fogt-Wyrwas, Dumont, Doucet, Mignon, Losson and Gala2012).

As is the case for the studies of eggs on the hair of definitive hosts, the proportion of eggs that are embryonated is an important additional piece of information required from soil contamination studies, but such data are not always reported. In a summary of 13 studies of soil contamination that did provide such data, embryonation rates varied considerably from 0·7 to 94·8%. Indeed these two widely diverging figures came from the climatically similar countries of the UK and Ireland (Roddie et al. Reference Roddie, Stafford, Holland and Wolfe2008b ). Furthermore, the relationship between soil contamination and seropositivity in the human population is not conclusive (Traversa et al. Reference Traversa, Frangipane di Regalbono, di Cesare, La Torre, Drake and Pietrobelli2014 but see Manini et al. Reference Manini, Marchioro, Colli and Nishi2012). This is not particularly surprising, given the wide range of possible sources of infective eggs that a human being might be exposed to (gardens, allotments, public parks, sports fields etc.) Therefore, the question should be asked – how useful are such surveys and what are they telling us?

In a significant advance, an improved quantitative framework for the epidemiology of egg contamination was developed in the city of Bristol, UK (Morgan et al. Reference Morgan, Azam and Pegler2013). The authors collected both empirical data (parasitology – prevalence, egg density and host data – faecal output, population density, age and status (owned vs stray)), and utilized a modelling approach. Their key conclusion was that in the absence of a large population of stray dogs and cats, pet dogs (especially those less than 12 weeks of age), dominate total egg output. This output can be modified by the degree of dog access to public areas and the removal of feces from these areas. However, under certain circumstances foxes can also play a role as contributors to egg contamination. Unsurprisingly, patterns of egg contamination are likely to vary significantly by location and be influenced by the relative proportion of different definitive hosts and their status (stray vs owned), the age structure of such hosts, rates of anthelmintic treatment and habitat use. Interestingly, the authors use their data to emphasize the need to eliminate infection in younger dogs in tandem with improved social responsibility in removing dog feces from public places.

Pertinent to this is the publication of a systematic review of interventions to prevent dog fouling (Atenstaed and Jones, Reference Atenstaed and Jones2011). Of 68 interrogated articles, none fulfilled the authors' inclusion criteria and the conclusion was that no good-quality studies had been undertaken to assess interventions to prevent dog fouling. Clearly such interventions, in tandem with the approach adopted by Morgan et al. (Reference Morgan, Azam and Pegler2013) would be extremely beneficial.

Building upon the work of Morgan et al. (Reference Morgan, Azam and Pegler2013), Nijsse et al. (Reference Nijsse, Mughini-Gras, Wagenaar, Frannssen and Ploeger2015b ) developed a stochastic model to quantify the relative contribution of household dogs, household cats, stray cats and foxes (all older than 6 months of age) to environmental contamination with Toxocara spp. eggs in the Netherlands (stray dogs were not included as the Netherlands is free of such animals). The choice of older animals was linked to the debate about whether these animals shed many eggs and require treatment four times per year (Nijsse et al. Reference Nijsse, Ploeger, Wagenaar and Gras-Mughini2015a ). Both parasite variables (prevalence and intensity of infection) and host factors (density, coprophagic behaviour, feces disposal by owners, cats' outdoor access) were included in the model. Scenario analyses were performed to evaluate the impact of different deworming strategies and feces clean-up compliances on the expected reduction in dog's egg output. Dogs were found to be the main contributors to environmental contamination with Toxocara ova. However, stray cats, owned cats and foxes also contribute eggs to the environment, and in urban areas egg output is dominated by stray cats. Furthermore, intervention scenarios revealed that only a very high compliance with the four-times-per-year deworming recommendations would yield a reduction in the contribution of dogs to egg output.

The authors made an important observation that due to the role of stray cats and foxes (and stray dogs in other contexts) control focused upon household pets alone is not sufficient to reduce environmental contamination to very low levels. Models of this kind are a very useful tool in quantifying the sources of Toxocara eggs in a given locality so as to prioritize control interventions and to assess the impact of such interventions. Studies such as those of Morgan et al. (Reference Morgan, Azam and Pegler2013) and Nijsse et al. (Reference Nijsse, Mughini-Gras, Wagenaar, Frannssen and Ploeger2015b ) also provide a framework for a more hypothesis-driven approach to the study of environmental contamination.

The role of paratenic hosts in the transmission of Toxocara: the biggest gap of all?

Our knowledge of the significance of paratenic hosts, and in particular feral paratenic hosts, as sources of infection for definitive hosts, both domestic and feral, is probably one of the largest gaps in our understanding of the epidemiology of Toxocara. Paratenic hosts are defined as hosts in which development does not occur, but may serve to bridge an ecological, or trophic, gap in a parasite's life cycle (Bush et al. Reference Bush, Fernandez, Esch and Seed2001). Furthermore, paratenic hosts are likely to disseminate infective stages of the parasite or aid these stages in avoiding unfavourable conditions such as the temporary absence of a definitive host. In the case of Toxocara, large numbers of eggs are released into the environment by fecund adult worms inhabiting the intestinal tract of domestic and feral definitive hosts and these eggs can be consumed by paratenic hosts. Our knowledge of the relative infective capacity of a range of vertebrate and invertebrate paratenic hosts is virtually non-existent (Holland and Hamilton, Reference Holland, Hamilton, Holland and Smith2006).

There are only a handful of published studies on the seroprevalence and larval burden of Toxocara in feral paratenic hosts and these are all confined to small mammal hosts. One of the most comprehensive studies is that of Dubinsky and colleagues (1995) from Slovakia, but it is of concern to note that this was published 20 years ago! Eleven small mammal species were investigated for the presence of parasite-specific antibodies in sera and the presence of Toxocara larvae in the brain and the hind leg femoral muscles. Considerable variation between hosts was observed with no detectable seropositivity in Rattus norvegicus and the highest level recorded (32%) in the house mouse, Mus musculus. The intensity of larval burden was low relative to the proportion of seropositive animals, in the order of one–three larvae per brain, with a higher intensity observed in mammals from suburban locations. These numbers are similar to those recovered from outbred laboratory mice exposed to a dose of 100 eggs under experimental conditions (Cox and Holland, Reference Cox and Holland2001a ). Dubinsky et al. (Reference Dubinsky, Havasiova-Reiterova, Petko, Hovorka and Tomasovicova1995) concluded that small mammals could act as important foci for the circulation and maintenance of Toxocara in the environment. To my knowledge this is the only published study that reported intensity of larval burden in a feral host. More recently, Antolova et al. (Reference Antolova, Reiterova, Miterpakova, Stanko and Dubinsky2004) reported seropositivity from 10 non-commensal rodents from the Slovak Republic, confirming the higher seropositivity from suburban locations but identifying the highest seropositivity in Apodemus agrarius (21%).

The role of small mammals as sentinels for the degree of environmental contamination, with Toxocara eggs and other important parasitic infections, such as Echinococcus multilocularis and Toxoplasma gondii, was highlighted by Reperant et al. (Reference Reperant, Hegglin, Tanner, Fischer and Deplazes2009). The highest Toxocara seroprevalence (13·2%) was reported in an urban area of Switzerland among four species of non-commensal rodents. In parts of the world where raccoons are present, small mammals could also act as important sentinels for the presence of the highly pathogenic and emerging zoonotic infection, Baylisascaris procyonis.

Unfortunately, no data whatsoever exist on the species identity of the Toxocara larvae found within the tissues of feral paratenic hosts such as house mice. It would be very interesting and epidemiologically useful to know if the larvae are T. canis, T. cati or T. leonina as this would shed light upon the relative importance of different ascarid species within the tissues of paratenic hosts and therefore, what species are most likely to be disseminated to definitive hosts via this route of transmission. The first and second internal transcribed spacers (ITS-1 and ITS-2) of nuclear ribosomal DNA (rDNA) were first used by Jacobs et al. (Reference Jacobs, Zhu, Gasser and Chilton1997) to identify and differentiate between three species of adult worms – T. canis, T. cati and T. leonina. Such diagnostic PCR could also be used on homogenized tissue to detect and identify Toxocara larvae in paratenic hosts (Gasser, Reference Gasser2013). This is particularly interesting given the so-far unproven supposition that infected rodent prey may be a more important source of infection for cats than dogs.

Laboratory models of paratenesis. In contrast to the paucity of data from feral paratenic hosts, our knowledge of experimental infection with (mainly) T. canis, and the consequent larval migration, under laboratory conditions is more comprehensive. Paratenic hosts infected under these conditions include mice, rats, guinea pigs, hamsters, gerbils, chickens, quail, pigeons, rabbits, pigs and monkeys (Fenoy et al. Reference Fenoy, Ollero, Guillen and del Aguila2001; Holland and Hamilton, Reference Holland, Hamilton, Holland and Smith2006; Strube et al. Reference Strube, Heuer and Janecek2013). Some species are more useful than others in terms of reflecting the risk of infection to humans (as food items) or particular aspects of human infection. For example, gerbils are particularly susceptible to eye involvement and are, therefore, good model organisms to explore the pathogenesis of OT (Takayangi et al. Reference Takayangi, Akao, Suzuki, Tomoda, Tsukidate and Fujita1999; Akao et al. Reference Akao, Takayanagi, Suzuki, Tsukidate and Fujita2000; Alba-Hurtado et al. Reference Alba-Hurtado, Tortora, Tsutsumi and Ortega-Pierres2000). Holland and Hamilton (Reference Holland, Hamilton, Holland and Smith2006) argued that gerbils are less satisfactory model organisms for cerebral toxocariasis, compared with mice, due to the development in gerbils of irreversible brain damage after chronic infection with T. canis (Akao et al. Reference Akao, Tomoda, Hayashi, Suzuki, Shimizu-Suganuma, Shichinohe and Fujita2003). In some parts of the world, there is evidence to suggest that humans can become exposed to Toxocara larvae as a result of the consumption of raw or undercooked chicken (Ito et al. Reference Ito, Sakai, Okajima, Quchi, Funakoshi, Nishimura, Ibayashi and Tsuji1986; Nagakura et al. Reference Nagakura, Tachibana, Kaneda and Kato1989). Experimental infections of chickens with Toxocara have demonstrated long-term survival of larvae even at low temperatures thereby underlining the public health risk to humans (Sprent, Reference Sprent1953; Taira et al. Reference Taira, Saitoh and Kapel2011, Reference Taira, Saitoh, Okada, Sugiyama and Kapel2012).

Following the original work of Done et al. (Reference Done, Richardson and Gibson1960), several authors have investigated the pig as a model for human infection (Helwigh et al. Reference Helwigh, Lind and Nansen1999; Taira et al. Reference Taira, Saeed, Lind, Murrell and Kapel2003). Toxocara larvae were eliminated early in infection and little eye or brain involvement was observed, thereby diminishing the usefulness of the pig as a model system. Furthermore, the large size of the porcine organs makes investigation less tractable. However, consumption of pork may pose a potential zoonotic risk for humans.

The model system that has received more attention than any other is that of the mouse. Significant advantages include ease of manipulation, small organ size for easy detection of larvae, availability of inbred and knockout strains and the fact that mice form part of the natural life-cycle of Toxocara in the wild (Holland and Hamilton, Reference Holland and Hamilton2013). Furthermore, larvae are known to accumulate in the murine brain over time (Dunsmore et al. Reference Dunsmore, Thompson and Bates1983) and show significant variation in larval burdens between individual outbred mice (Skerrett and Holland, Reference Skerrett and Holland1997), thereby suggesting a role for host immunity and genetic resistance/susceptibility to infection (Dold and Holland, Reference Dold and Holland2011). The contribution of factors such as strain (outbred vs inbred), dose, days post-infection and larval burden in the brain have been explored both with respect to larval migration and impact on host behaviour (Epe et al. Reference Epe, Sabel, Schneider and Stoye1994; Cox and Holland, Reference Cox and Holland2001a , Reference Cox and Holland b ; Holland and Cox, Reference Holland and Cox2001; Hamilton et al. Reference Hamilton, Stafford, Pinelli and Holland2006). Interesting differences between the larval migration of T. canis vs T. cati have been observed in mice; T. canis larvae accumulate to a greater extent in the murine brain whereas T. cati larvae show greater concentration in the muscles (Havasiova-Reiterova et al. Reference Havasiova-Reiterova, Tomasovicova and Dubinsky1995). Some authors have argued that because T. cati larvae are smaller they may be able to exit the arteries more easily whereas T. canis, being larger, are more likely to be trapped in the brain tissue (Strube et al. Reference Strube, Heuer and Janecek2013). A comparison of the evidence for brain involvement and subsequent accumulation of Toxocara larvae was undertaken by Holland and Hamilton (Reference Holland, Hamilton, Holland and Smith2006) in a range of experimentally-infected animals and it is mice that account for most of the evidence. Such studies underline the important biological differences that exist between the migratory pathway of Toxocara spp. and larval involvement in key organs such as the eye and the brain in different paratenic hosts.

Toxocara cati: even more enigmatic?

In 2003, Maggie Fisher described T. cati as ‘an underestimated zoonotic agent’. Our understanding of the role of T. cati in exposure to humans, environmental contamination and within paratenic hosts has been hindered by our inability to distinguish between the two species in these various media. Serological differentiation remains a problem but ova and larval identification using molecular methods is now possible (Durant et al. Reference Durant, Irenge, Fogt-Wyrwas, Dumont, Doucet, Mignon, Losson and Gala2012; Gasser, Reference Gasser2013). However, as outlined previously, there are no data on the species of third-stage larvae detected in paratenic hosts from the wild. Recently, T. cati eggs, identified by PCR, have been detected in dog feces and the explanation for this phenomenon is likely to be coprophagy (Fahrion et al. Reference Fahrion, Schnyder, Wichert and Deplazes2011).

Despite receiving less attention than the dog ascarid T. canis, a number of interesting biological differences between the two Toxocara spp. have emerged. There is evidence to suggest that T. cati may be more resistant to cold temperatures. Experimental evidence showed that T. cati eggs are more resistant to freezing compared with those of T. canis (O'Lorcain, Reference O'Lorcain1995) and T. cati infections were used to establish the ability of larvae to withstand cold temperatures in chicken tissue (Taira et al. Reference Taira, Saitoh, Okada, Sugiyama and Kapel2012). Furthermore, Akao et al. (Reference Akao, Takayanagi, Suzuki, Tsukidate and Fujita2000) demonstrated that T. cati larvae can migrate to the eye and cause significant pathology in a relatively novel animal model – the Mongolian gerbil. In eight cats and their offspring, the mode of transmission following both natural and experimental infection was explored (Coati et al. Reference Coati, Schneider and Epe2004). Lactogenic transmission of larvae occurred after an acute infection of the queen during late pregnancy but did not occur during chronic natural infection. No evidence of arrested somatic larvae was observed in adult cats.

Differential pathogenesis at the level of the transcriptome was explored for T. canis and T. cati, respectively (Janecek et al. Reference Janecek, Wilk, Schughart, Geffers and Strube2015). Major differences between the two species were observed, with T. canis-infected brains demonstrating significant disruption of lipid biosynthetic processes. The authors suggest that such disruption may lead to dysfunction in signal transduction and neurogenerative disease. On day 42 post-infection, T. canis mice exhibited partial paralysis of hind limbs as well as ataxia, in contrast to T. cati-infected mice that exhibited no neurological lesions at post-mortem.

Toxascaris leonina. Experimental infections (from eggs derived from both canine and feline hosts) were originally established in laboratory mice by Sprent (Reference Sprent1959) demonstrating that this parasite can infect paratenic hosts. However, the role of T. leonina in human disease remains unknown, particularly given our current inability to distinguish Toxocara spp. serologically (Gasser et al. Reference Gasser, Zhu, Min, Jacobs, Chilton, Holland and Smith2006). It has been suggested that the zoonotic potential of T. leonina is limited because somatic migration in definitive hosts does not occur as part of the normal life-cycle and larvae are not vertically transmitted (Overgaauw and Van Knapen, Reference Overgaauw, Van Knapen, Macpherson, ., Meslin and Wandeler2000).

Concluding remarks