Toxoplasma seroprevalence in a rural population in France: detection of a household effect

Toxoplasma gondii (T. gondii), the agent of toxoplasmosis, is one of the most widespread parasite species worldwide: in 2000, Tenter et al. estimated that one third of the human population might be parasitized [1]. In healthy humans,T. gondii infection is generally subclinical or mild. However, during pregnancy, the risk of materno-foetal transmission may have serious consequences for the foetus: abortion or congenital infection [2]. Most children with congenital toxoplasmosis have no symptoms at birth but are at risk of developing retinal diseases or neurological abnormalities later in life [3]. T. gondii may also cause severe encephalitis among immunosuppressed patients via acute infection or reactivation of a latent infection: encephalitis is the most frequent neurological infection among patients with acquired immunodeficiency syndrome [4].

Prevention of human toxoplasmosis relies on the knowledge of the disease epidemiology. However, the life cycle of T. gondii is complex: post-natal human infection may result from ingestion of tissue cysts contained in raw or undercooked meat from intermediate hosts or from ingestion or inhalation of oocysts shed by definitive hosts (felids) and disseminated in soil and water. Because virtually all mammals and birds may serve as intermediate hosts and because of the high resistance of T. gondii sporulated oocysts in the environment [5], possible sources of infection are multiple.

Given the multiplicity of those sources, the epidemiological situation of toxoplasmosis is spatially contrasted: incidence and prevalence values vary among countries and even among regions within countries [6, 7]. The risk factors already identified differ also among studies. In a multicentre study in Europe, ingestion of tissue cysts has been estimated to represent 30 to 63% of the risk of infection, according to the locality, whereas contact with soil would represent 6 to 17%; finally, 14% to 49% of that risk remained unexplained [8]. Other studies found no relationship with ingestion of raw meat [9] or identified risk factors that were not clearly related to a known source of risk (e.g., the marital status) [7]. This variability reflects the fact that exposure to toxoplasmosis is probably variable at a local scale, depending on the climate, the agricultural methods, and on cultural and individual differences concerning eating habits or exposure to cats. Thus, searching for risk factors at a local spatial scale may be very informative.

The present survey on toxoplasmosis in the human population of two adjacent rural villages is a part of an analysis of toxoplasmosis dynamics. The aims were to provide a population-based estimate of Toxoplasma seroprevalence in a geographically-defined population, to describe the relationship between seroprevalence and factors like age and sex, and to search for other factors associated with prevalence at the scale of that population.

In our specific rural setting, the overall seroprevalence and relationship with age were comparable to the values recently observed in neighboring geographical areas. In a recent population-based survey in the Netherlands, 40.5% of persons carried anti-T. gondii antibodies, the higher risks being those of people living in non-urbanized areas [7]. In France, the last perinatal survey indicated that 44% of women at the time of delivery were immune [13]. Although a standardization on age should be performed for direct comparison with other surveys, this seroprevalence estimate is higher than the values observed in other European countries: recent studies found a seroprevalence of 23.4% among blood donors living in rural areas in Slovakia [14] and of 23% among military personnel in the Czech Republic [15]. The relationship between seroprevalence and age is frequently observed and reflects the fact that the risk of toxoplasmosis infection remains present during whole life [14, 15]. In the population-based survey in the Netherlands, the steepest rise in seroprevalence occurred in persons aged 25 to 44 years [7]. Here, we also show that the slope of the relationship between seroprevalence and age is steeper in young persons up to 40 years of age. This may reflect a higher risk in this age class, but this hypothesis cannot be disentangled from a birth-cohort effect.

There was a trend, although only close to statistical significance, towards a higher probability of seropositivity in men than in women. This result has not been found in other large population-based surveys [7, 14, 16] or, when found, the relationship seen in a univariate approach was no longer significant when other factors were considered [17]. However, a higher risk for men was also found in a previous study on a rural population in Illinois, USA [18] and in the middle and upper socio-economic classes in Brazil [19] suggesting that a difference exists but only in specific populations or settings. That higher risk for men was attributed either to contact with soil or to improper hygiene [17, 18]. Thus, a difference between men and women might appear in populations with high exposure to soil. In the present study, men and women differed significantly as to the time spent gardening (4.5 h/week for men versus 2.8 for women); thus, as to exposure to soil.

Toxoplasma seroprevalence was not significantly associated with any of the factors related to meat consumption. In the univariate analysis, consumption of rabbit meat was the meat consumption factor with the highest estimated Odds Ratio, but this effect was far from statistical significance (Table 2), possibly because, in our population, few subjects never ate rabbit meat (only 27 subjects). This factor needs further investigation because it was found to be the major risk factor in a recent study carried out in the Czech Republic [15]. Besides the low sample size, the absence of any other factor concerning meat consumption may reflect the fact that this source of infection has become less important than others. Indeed, the latest evolutions of the agricultural methods have brought changes in the risk of infection of domesticated animals. For example, the seroprevalence in slaughter pigs is now under 8% in most countries; thus, the importance of pork meat as source of infection is probably declining [1]. The frequent consumption of frozen meat may also lower the risk of infection.

Consumption of raw vegetables was the only factor related to the eating habits that had a strong effect, close to statistical significance. This factor was also found associated with toxoplasmosis in Slovakia [14], in Norway (raw vegetables and fruits) [20], and in France (raw vegetables eaten outside home) [21]. Raw vegetables may be contaminated by soil oocysts or, in case of inappropriate handling, during meal preparation [21]. Checking and further investigating the link between contamination of raw vegetables and soil contamination will involve isolation of T. gondii oocysts from the soil and from locally available raw vegetables. A study conducted in an urban area with high densities of stray cat populations showed that high concentrations of oocysts can be detected in soil samples using real-time PCR [22].

Contrarily to the hypothesis that owning cats constitutes a significant risk factor, contact with cats, measured by the number of cats bred during life, was not found clearly associated with Toxoplasma seroprevalence. Actually, previous studies have found contradictory results about the link between contact with cats and the risk of toxoplasmosis. One study found that an increase in the number of seropositive cats trapped on the farm increased the risk of human infection [18]. Some have shown that owning a cat increased the risk of T. gondii seropositivity [15, 21] while others found no association between daily contact with cats or the presence of cats in the environment and the risk of toxoplasmosis [8, 20]. Kortbeek et al. [7] found only a weak association and concluded that owning a cat is not a risk factor whereas contamination of the environment with oocysts is. The present results suggest that the risk of contamination by oocysts may be more related to cats roaming outside shedding oocysts than to contact with household cats.

An important factor not reported before is the household effect: the probabilities of seropositivity for subjects belonging to the same household were closer than for subjects belonging to different households. This effect persisted after taking into account the mean age of household members; thus, it was not due to close ages within households. The heterogeneity, already found at country and region scales [6, 7], was also found among households. An explanation of this similarity among members of the same household may be a common susceptibility due to genetic relatedness. However, that heterogeneity was also found in two-member households, most probably couples with no specific genetic relatedness (though this information was not checked). For example, among 57 two-member households in which the serological status of both members was known, 41 had concordant results and 16 discordant ones. The household effect resulted most probably from common risk factors within households. Common sources of exposure may be the household diet, the level of hygiene [21], or the contact with oocysts present in the soil around the house, which could be an important risk factor in this rural context [18]. In our study, none of the factors we aggregated at household level was able to explain heterogeneity among households. Because explaining the risk at the household level was not our primary aim, the household factors we built were probably not accurate measures of common exposures of households to the same risk factors. Except the mean age, those factors were built by aggregation of binary risk factors at household level; thus, the information they provided about the exposure to common risk factors was poor and their value could depend on the household size. We advocate that defining exposure at a household level should help highlighting relevant risk factors useful to predict individual risks and direct prevention.

The major finding of our study was the household effect, with a strong heterogeneity of seroprevalence among households. This effect could be explained by common exposure of subjects sharing the same household to local risk factors. Our future work will be directed towards building quantitative factors measuring the presence of oocysts in the soil and towards quantifying the link between these factors and the seroprevalence of exposed households using a spatial analysis.