Epidemiology of Concomitant Infection Due to Loa loa and Mansonella perstans in Gabon

Background The filarial parasites Loa loa and Mansonnella perstans are endemic in the central and western African forest block. Loa loa is pathogenic and represents a major obstacle to the control of co-endemic filariae because its treatment can cause fatal complications such as encephalitis. Methodology/Principal Findings 4392 individuals aged over 15 years were studied both by direct examination and a concentration technique. The overall prevalence rates were 22.4% for Loa loa microfilaremia, 10.2% for M. perstans microfilaremia, and 3.2% for mixed infection. The prevalence of both filariae was higher in the forest ecosystem than in savannah and lakeland (p<0.0001). The intensity of microfilariae (mf) was also higher in the forest ecosystem for both parasites. The prevalence and intensity of microfilaria were both influenced by age and gender. Correlations were found between the prevalence and intensity of Loa loa microfilariae (r = 0.215 p = 0.036), and between the prevalence of Loa loa and the prevalence of individuals with microfilaria >8000 mf/ml (r = 0.624; p<0.0001) and microfilariae >30 000 mf/ml (r = 0.319, p = 0.002). In contrast, the prevalence of pruritis and Calabar swellings correlated negatively with the prevalence of Loa loa microfilaria (r = −0.219, p = 0.032; r = −0.220; p = 0.031, respectively). Pruritis, Calabar swellings and eye worm were not associated with L. loa mf intensity (r = −0.144, p = 0.162; r–0.061, p = 0.558; and r = 0.051, p = 0.624, respectively), or with the prevalence or intensity of M. perstans microfilariae. Conclusions/Significance This map of the distribution of filariae in Gabon should prove helpful for control programs. Our findings confirm the spatial uniformity of the relationship between parasitological indices. Clinical manifestations point to a relationship between filariae and allergy.

Interest in loiasis has grown during the last 30 years, for several reasons. First, in endemic areas loiasis is the second reason for medical visits, after malaria [1,9]. Second, this infection mainly affects active young individuals, who contribute to agricultural productivity [10], and their health is often aggravated by coinfection by other parasites. Two-thirds of infected individuals are amicrofilaremic, despite subconjunctival migration of adults worms, suggesting immunological elimination of microfilariae [1,11]. Severe adverse events can occur during treatment with diethylcarbamazine (DEC) and ivermectin in individuals with high-level microfilaremia, requiring close treatment monitoring and hindering mass administration of antifilarial drugs aimed at controlling other filariae in areas where Loa loa is co-endemic. This is not the case with M. perstans [12].
Many epidemiological studies of loiasis and Mansonellosis have been carried out throughout the western and central African forest block. These studies mainly focused on the distribution of loiasis and on the possible relationship between the prevalence and intensity of microfilaremia, in order to estimate the risk of adverse events during mass chemotherapy.
The prevalence of L. loa microfilaremia varies from country to country [13], as well as within a given country and even a given geographic area [14]. The highest prevalence is observed in forest areas and the lowest in savannah areas of both Gabon [15,16] and Cameroon [17,18], for example. Differences within a given geographic zone are directly linked to the bioecological specificity of a microzone [19]. These observations were recently used to create a predictive geographical model of loiasis endemicity based on satellite, vector habitat, prevalence, vegetation, temperature, relief, pluviometry and topography data [20]. However, when compared to field data, this model showed certain limitations [21].
A linear relationship between the prevalence and intensity of loiasis has been established. A high prevalence is indicative of intense L. loa infection and therefore a high risk of adverse events [22,23]. The 20% threshold prevalence of microfilaremia at the community level corresponds to about 5% of high microfilaremia loads (.8000 mf/ml) and 2% of very high microfilaremia loads (.30000 mf/ml), the latter being the cut-off point above which there is a risk of severe adverse events during ivermectin treatment [24]. Owing to the difficulties of drawing regional maps based on microscopic analysis, a rapid method for evaluating the prevalence and intensity of Loa loa infection at the community level has been developed (RAPLOA: Rapid Assessment of Prevalence of Loa loa) [25]. RAPLOA is based on interviews assisted by photographs of adult worms in the eye, to detect subconjunctival migration of adult worms (which lasts 1 to 7 days), as reported by interviewees. A 40% prevalence of a history of eye worm corresponds to a 20% threshold prevalence of microfilaremia at the community level, 5% of high microfilaremia loads (.8000 mf/ml) and 2% of very high microfilaremia loads (.30000 mf/ml) [25]. Another clinical manifestation, Calabar swellings, was used to evaluate the risk of adverse events. This sign has shown to correlate with the prevalence of highly microfilaremic individuals [25].
The use of eye worm and Calabar edema to assess the risk of fatal side effects in patients with loiasis suggests a relationship between clinical symptoms and parasitological indices.
The aim of the present study was to obtain a fuller picture of the distribution of blood-borne filariae in Gabon, using both the wet blood film and concentration techniques, and to detect a linear relationship between the prevalence and intensity of loiasis and between clinical symptoms and parasitological indices. We therefore conducted a large survey, including all the country's ecological niches and recording the main clinical manifestations of Loa loa infection.

Area of study
We surveyed rural Gabonese populations. The country is 800 km long and 20 to 300 km wide, consists of 80% rain forest, and is bordered to the west by the Atlantic Ocean. The forest zone extends from west to east, from the coastal basin with the grassland forest to the interior and north-eastern forest plates band, through a wide mountainous forest band from 60 to 100 km parallel to the coast. The south and southeast contain isolated areas of savannah and steppe. A coastal and continental marine ecosystem named lakeland is located around the mouth of River Ogooué ( Figure 1) [29]. The population is about 1.5 million and there are 2048 villages located in 9 provinces. Rural populations are located along roads and rivers, and few villages have more than 300 inhabitants.

Study population
This survey was conducted during nine-month field missions between June 2005 and September 2008. For this survey, a stratified random sampling method was used, based on the 9 provinces. Twenty to 30 villages per province were randomly selected. The required sample size was calculated on the basis of an estimated prevalence of 5 to 10% (using n = e 2 [p (12p)]/e 2 ; with e = 1.96 (alpha risk = 5%), e (precision) = 2% and p = expected prevalence; with n varying from 188 to 864). Within each village, individuals over 16 years of age having lived for at least one year in their village and who accepted blood sampling where included in the study. A free medical examination was offered and basic medicines were provided to all participants and non participants, if appropriate. All the villages were georeferenced ( Figure 1).

Questionnaire
The rationale of the study was explained and a one-page questionnaire was administered to all participants. We collected demographic data (age, sex and occupation), geographic data (name of the village, length of residence, department and province) and the medical history (eye worm, Calabar swellings, chronic arthralgia, pruritus, etc.) ( Figure S1).

Ethical considerations
The study protocol was approved by the Ministry of Health. The Health Director, the governor of each province and the chiefs of each village received written information. Individual written consent was required before blood sampling. The results of the study were transmitted to the Ministry of Health.

Blood collection
Field laboratory facilities were set up in regional hospitals. Blood samples were collected, usually in the villages' healthcare centers, on a daily basis, into two 7-ml Vacutainer tubes containing EDTA (VWR International, France). The tubes were stored in the dark at +4uC before transportation to the field laboratory.

Author Summary
Loa loa and Mansonella perstans are blood filarial parasites, endemic in the central and western African forest block, and transmitted by chrysops and culicoides flies, respectively. Loa loa is pathogenic and represents a major obstacle to the control of co-endemic filariae. Treatment of individuals with .8000 Loa loa microfilariae/ml can result in severe adverse reactions. M. perstans is prevalent in the tropics, with undefined clinical symptoms. We screened 4392 individuals for these infections in 212 Gabonese villages. The overall prevalence rates were 22.4% for Loa loa microfilariae, 10.2% for M. perstans, and 3.2% for mixed infection. These rates varied across the different ecosystems: forest, savannah, Lakeland, river (Ogouée), and equator. A correlation was found between the prevalence and intensity of microfilariae, while a negative relationship was found between clinical symptoms (pruritis, Calabar swelling) and the prevalence of Loa loa microfilaremia. This study confirms the spatial uniformity of the relationship between parasitological indices, and provides a map and baseline data for implementation of mass chemotherapy for these infections.

Parasitological analyses
Due to the variability of microfilarial load, the analysis started systematically by direct examination of a wet blood film, followed by a concentration technique. Two experienced technicians read the slides separately, and the results were controlled by a parasitologist. Briefly, microfilariae were counted directly in a 10-ml wet blood film between microscope slide and coverslip, using an optical microscope equipped with a 106 objective. Parasitemia was expressed in microfilariae per milliliter (mf/ml) of blood. A modified Knott's concentration technique [30] was applied routinely to each sample, as follows: 1 ml blood was diluted with 9 ml PBS in a conical tube and 200 ml of saponin (2% w/v) was added to lyze red cells. The tubes were centrifuged (10 min, 500 g) and the supernatants discarded. The entire pellet was then examined under the microscope (106 objective) and microfilariae were counted. Parasite species were identified by their size and motility, and by the absence or presence of a sheath.

Data analysis
Loa loa prevalence rates were estimated nationwide. As mentioned above, the 20% threshold prevalence of microfilaremia is the cut-off above which serious adverse events are likely to occur, and corresponds to 5% of high microfilaremia loads (.8000 mf/ml) and 2% of very high microfilaremia loads (.30000 mf/ml). Thus, prevalence rates were calculated in each province, village and ecosystem as prevalence rates for microfilaremia loads .8000/ml and .30000/ml. The intensity of infection was estimated as described elsewhere [25]. The difference between the results of the two laboratory tests was  calculated. The Chi2 test and Fisher's exact test were used as appropriate. Minitab 16 software was used to calculate Spearman's correlation coefficient for the association between parasitological and clinical parameters, and the Mann Whitney U test was used to compare mf intensity among groups. Univariate crude conditional maximum likelihood estimates of odds ratios (OR) and exact 95% confidence intervals (CI) were determined for each potential risk factor, using STATA software version 9.0 (Stata Corporation, College Station, USA). Multivariate logistic regression models stratified by the ecosystem were constructed from risk factors with a significance of #0.10 in univariate analysis, using a backwards stepwise elimination procedure. P values below 0.05 were considered statistically significant.

Characteristics of the study population
In total, 4392 individuals from 15 to 85 years old were enrolled in 212 villages, representing 10.7% of all villages in the country. The distance between villages ranged from 5 to 30 km. The sex ratio (M/F) was 0.88 (47.4% men and 52.6% women). About 58% of individuals were more than 45 years old and 63.9% had spent more than 10 years in their village. Farmers represented 69.8% of the population and hunters 10.2%. Around 80% of individuals were surveyed in the forest area, 10% in the savannah and the lakeland. The reported proportions of eye worm, Calabar swellings and pruritis were 29.3%, 11.2% and 22.4% ( Table 1).

Comparison of microfilariae counts with and without concentration
The wet blood smear identified 790 Loa loa and 116 M. Perstans microfilaremic subjects while the concentration technique detected 984 L. loa and 447 M. perstans microfilaremic subjects (difference of 19.7% for L. loa and 74% for M. perstans) ( Table 2).
Most of these individuals who were positive only after concentration had microfilaremia below 100/ml, for both species (Table 3).

Analysis of risk factors
In univariate analysis, males had a significantly higher risk of Loa loa infection than females (OR: 2.38, 95%CI: 2.05-2.75, p,00001), and the prevalence of Loa loa parasitemia increased linearly with age (p,0.00001) ( Table 6). The prevalence of Loa loa microfilaremia was higher in hunters than in farmers and other occupational groups (p,0.04), and higher in individuals with eye worm (p,0.001) and those without Calabar swellings (p,0.014) ( Table 6). Only gender was a risk factor for M. perstans microfilaremia, males having a significantly higher prevalence than females (OR: 1.89, 95%CI: 1.54-2.31, p,0.0001) ( Table 7). In multivariate analysis, only age and sex remained significantly associated with Loa loa parasitemia, throughout the country and within the forest ecosystem (Table 8 and 9).  For clinical symptoms, only eye worm and Calabar swellings remained significantly associated with Loa loa parasitemia, both throughout the country and within the forest ecosystem (Table 8 and 9).
The intensity of Loa loa microfilaremia did not vary with age countrywide (r = 0.249, p = 0.634), while it correlated with age in males (r = 0.915 p = 0.011) but not in females (r = 0.684 p = 0.134) (Figure 4). At the district level, the intensity of Loa loa microfilaremia did not vary significantly with age and sex.

Relationship between clinical symptoms and parasitological indices
The prevalence of pruritis correlated negatively with the prevalence of Loa loa microfilaremia (r = 20.219; p = 0.032)

Discussion
We conducted a large-scale survey of two blood-borne filarial parasites, using direct examination and a concentration technique, in rural populations of 212 villages in Gabon, in order to map their distribution throughout the country, to characterize the modalities of population exposure and to explore the relationship between prevalence and intensity, and between clinical symptoms and parasitological indices.  Loa loa and Mansonella perstans in Gabon www.plosntds.org The overall prevalence rates were 22.4% for Loa loa microfilariae, 10.2% for M. perstans, and 3.2% for mixed infection. These rates varied across the different ecosystems, the Ogooue River, and the equator. A correlation was found between the prevalence and the intensity of microfilariae, and between clinical symptoms (eye worm, Calabar swelling) and the prevalence of Loa loa microfilaremia.
As direct microscopic detection of microfilaria in wet blood films is not very sensitive, we combined two techniques for this survey, namely direct examination of 10 ml of blood (wet film) and prior concentration of 1 ml of blood. If we had used direct examination only, 19.7% Loa loa mf carriers and 74% of M. perstans carriers would have been missed. Most of these subjects had fewer than 100 mf/ml. Such underestimation may have implications for estimates of the risk of transmission and even for control programmes. Better sensitivity after sample concentration has been reported [14,30], although this method is more tedious for large-scale surveys. Previous surveys used direct examination with larger volumes (30-50 ml [22], 50 ml [10] or 75 ml [19]).
The prevalence of Loa loa microfilaremia was 22.4% overall (up to 57% in some villages) while that of M. perstans was 10% (up to 67% in some villages). Gabon is thus a highly endemic country and a zone at high risk of fatal treatment complications. These prevalence rates are similar to those reported in southern Cameroon (up to 38% in the district of Elig-Mfomo) [13] and Equatorial Guinea (27%). This contrasts with Central African Republic (CAR) and Chad, where prevalence is lower (11% and 8.4% respectively). In DRC-Congo, Republic of Congo and Nigeria the prevalence rates range from 1.2% to 97% [13]. It should be noted that these prevalence rates are for specific regions of these countries, whereas our survey covered the whole of Gabon. The prevalence of Loa loa remains high in Gabon [15,28].
Loa loa was highly prevalent in the north Equator (.20%), compared to the south (10-20%). Most areas crossed by the Ogooue River from the south-east (its source) to the north-west (towards the Atlantic Ocean) had an M. perstans prevalence of more than 10%, while other areas had a prevalence below 10%.
Among the three major ecosystems, forest had a higher prevalence of both parasites than savannah and lakeland. Differences were also seen among the different types of forest, as previously observed in Cameroon [19]. Geographic factors have been implicated in the prevalence of diseases like arteriosclerosis [31]. Sunlight might have a protective effect on some diseases [32], as ultraviolet B radiation stimulates the synthesis of vitamin D, which plays a role in immunity [33]. Geographic factors may influence filarial distribution by affecting the host immune system or the vector. The environment created by Ogooue River may affect the distribution and transmission of M. perstans. Although no soil studies around Ogooue River are available, studies in other   Loa loa and Mansonella perstans in Gabon www.plosntds.org areas have shown that low-pH soil, low organic soil content, salty soil, and wet soil promote Culicoides fly breeding [34,35] while temperature may affect vector competence [36]. The prevalence of Loa loa microfilaremia was influenced by age in both sexes. In some parts of the country the prevalence continued to increase up to 70 years of age, while in others the prevalence appeared to plateau after 60 years. Males tended to be more microfilaremic than females, possibly because men are more exposed to chrysops bites due to their outdoor occupations (hunting, etc.), which become more intense with age, hence the correlation between age and microfilaremia. Genetic factors may also have a role [37]. Furthermore, the negative correlation of the  intensity of microfilaremia with age in males may due to concomitant immunity against new incoming infection [38] or natural death of existing microfilariae [39].
In some areas of Cameroon where the general prevalence of microfilaremia exceeds 20%, approximately 5% of individuals have 8000 mf/ml and 2% have more than 30 000 mf/ml [24]. Similarly, in an area with a prevalence of 30%, 9% of carriers had .30 000 mf/ml, while in an area with a prevalence of 40%, approximately 16% of carriers had .8000 mf/ml and 5-6% had .30 000 mf/ml. Therefore, areas with a prevalence of more than 20% are considered to be at a high risk of treatment complications. Such studies have only been conducted in Cameroon [24,25]. In this study, we observed a positive relationship between the prevalence and intensity of microfilaria. This suggests that the relationship between these two parasitological indices is spatially stable.
Clinical symptoms have also been used to predict the risk of side effects during mass chemotherapy. As previously described, eye worm and Calabar swelling have been found to correlate strongly with prevalence [25]. Photographs of ocular passage of the eye worm were used in previous studies [25]. Whether the lack of photographs in the present study influenced the accuracy of the patients' answers is not known. Yet, in our opinion, the use of photographs would yield a higher prevalence of amicrofilaremic subjects. Another striking observation is the negative correlation of pruritis and Calabar swelling with the prevalence of Loa loa but not M. perstans. Pruritis is a clinical sign of an allergic reaction. The negative relationship suggests that Loa loa filaria may induce desensitization. In Gabon, skin test reactivity against common allergens is low [40], while treatment of helminth infections increases skin test reactivity to mite antigens [41]. Similar observations have been made with M. perstans in Ugandan women [42]. A previous study in Gabon showed a high level of polyclonal IgE and Loa loa-specific IgG4 in permanent residents [27].
Further investigations are needed to elucidate the relation between filaremia and allergy in Gabon. In conclusion, we provide a map of Loa loa and M. perstans microfilaremia in Gabon, and describe important relationships between parasitological indices and clinical manifestations. A clear and spatially uniform relationship was found between the prevalence and intensity of parasitemia. These data should be of use for planning mass chemotherapy.