Long-term periodic anthelmintic treatments are associated with increased allergen skin reactivity

Background The low prevalence of allergic disease in the rural tropics has been attributed to the protective effects of chronic helminth infections. There is concern that treatment-based control programmes for these parasites may lead to an increase in the prevalence of allergic diseases. Objective We measured the impact of 15–17 years of anthelmintic treatment with ivermectin on the prevalence of allergen skin test reactivity and allergic symptoms in school-age children. Methods The prevalence of allergen skin test reactivity, exercise-induced bronchospasm and allergic symptoms was compared between school-age children living in communities that had received community-based treatments with ivermectin (for onchocerciasis control) for a period of 15–17 years with those living in geographically adjacent communities that had received no ivermectin. Results The prevalence of allergen skin test reactivity was double in children living in treated communities compared with those in untreated communities (16.7% vs. 8.7%, adjusted OR 2.10, 95% CI 1.50–2.94, P<0.0001), and the effect was mediated partly by a reduced prevalence of Trichuris trichiura among treated children. Ivermectin treatments were associated with an increased prevalence of recent eczema symptoms (adjusted OR 2.24, 95% CI 1.05–4.78, P=0.04) but not symptoms of asthma or rhino-conjunctivitis. The effect on eczema symptoms was not associated with reductions in geohelminth infections. Conclusion Long-term periodic treatments with ivermectin were associated with an increased prevalence of allergen skin test reactivity. There was some evidence that treatment was associated with an increased prevalence of recent eczema symptoms but not those of asthma or rhino-conjunctivitis. Cite this as: P. Endara, M. Vaca, M. E. Chico, S. Erazo, G. Oviedo, I. Quinzo, A. Rodriguez R. Lovato, A.-L. Moncayo, M. L. Barreto, L. C. Rodrigues and P. J. Cooper, Clinical & Experimental Allergy, 2010 (40) 1669–1677.


Introduction
The prevalence of allergic diseases appears to be low in rural areas of developing countries [1,2]. In such areas, poverty and the inadequate disposal of faeces are commonplace and there is a high prevalence of geohelminth infections [3]. Geohelminth parasites are estimated to infect 3.8 billion humans world-wide [4], and WHO has endorsed the control of these infections through the provision of periodic treatments with anthelmintic drugs to high-risk groups particularly schoolchildren [4].
Several epidemiologic studies have provided evidence for an inverse association between geohelminth infections and allergen skin test reactivity in areas where these infections are highly endemic [5][6][7][8] and it has been suggested that geohelminths may suppress allergy in these populations [6,9]. This has raised the concern that the mass treatment of helminth infections through anthelmintic treatment programmes may increase the

Epidemiology of Allergic Disease
Re-use of this article is permitted in accordance with Terms and Conditions set out at http://wileyonlinelibrary.com/onlineopen prevalence of allergic disease in populations where these infections are endemic.
Intervention studies of periodic anthelmintic treatment have provided conflicting evidence with three studies showing an increased risk of allergen skin test reactivity after treatment [10][11][12] and one study showing no effect on allergen skin test reactivity or allergic disease [13]. The negative study provided treatment for 12 months while the other three studies with positive findings provided treatment for between 12 and 30 months.
An opportunity to study the long-term effects of anthelmintic treatment on allergy was provided by a control programme for onchocerciasis that has been treating endemic communities with the broad-spectrum anthelmintic drug ivermectin at annual or semi-annual intervals for the past 15-17 years [14]. Because ivermectin also has potent activity against geohelminths [15], we investigated the impact of long-term ivermectin treatments on allergen skin test reactivity and allergic disease by comparing the prevalence of these parameters in children living in communities that had received mass treatments with ivermectin with children living in geographically adjacent communities that had never received ivermectin treatments.

Study population and design
The study was conducted between March 2005 and April 2007 in the Districts of Eloy Alfaro and San Lorenzo in the northern coastal Province of Esmeraldas, Ecuador [16]. Most communities were located along rivers in a tropical rain forest area at altitudes below 100 m above sea level. Greater than 90% of the population in each study community was Afro-Ecuadorian. The main economic activities are agriculture, logging, fishing and the extraction of African Palm oil. Poorly developed infrastructure, untreated sewage and rudimentary solid waste disposal are common characteristics of most communities. All children aged 6-16 years in the communities were eligible to participate. Updated censuses from all communities were used as the basis for recruitment. Informed written consent was obtained from the child's parents or guardian. The study protocol was approved by the Ethics Committee of the Hospital Pedro Vicente Maldonado, Ecuador.

Selection of communities
Treated communities were selected according to the treatment schedule of the Ecuadorian Onchocerciasis Elimination Programme (EOEP). Non-treated communities were selected from maps of the study area and consisted of geographically adjacent communities selected to be as similar as possible with respect to size, ethnicity, and socio-economic characteristics. Onchocerca volvulus infection was never present in non-treated communities and none have received mass ivermectin treatment. The prevalence of O. volvulus infection in treated communities before the introduction of ivermectin was 440% in adults but lower in children [17]. No other filarial helminth infections were present in the study area [18].

Ivermectin treatment
Twice-annual community treatments with ivermectin were initiated between 1991 and 1992. The second annual dose was suspended in 1995-1996 and then reintroduced from 1998. Eligibility criteria for treatment are: weight 415 kg and free of serious illness (e.g. active tuberculosis, terminal cancer, etc.), and for women, not pregnant and not nursing infants up to 3 months of age. Distribution was organized by community health workers and single-dose treatments of 150 mg/kg ivermectin were observed directly. The annual treatment coverage of the programme was 85.2% (range 54.9-97.9%) over the 15year period [14].

Subject evaluations
A questionnaire modified from the ISAAC Phase II questionnaire [16] was administered to the child's mother or guardian to collect information on allergic symptoms and other relevant data. A stool sample was collected and analysed for the presence of eggs and larvae by the modified Kato-Katz and formol-ethyl acetate concentration methods [19]. Skin reactivity was tested to house dust mite (HDM) (Dermatophagoides pteronyssinus; Greer Laboratories, Lenoir, NC, USA), grass pollen mix (Greer Laboratories), American cockroach (Periplaneta americana; Greer Laboratories), fungi mix (Greer Laboratories), Alternaria tenuis (Greer Laboratories), cat (Greer Laboratories) and dog (Greer Laboratories) extracts. Allergens and positive histamine and negative saline controls were pricked onto the volar surface of the forearm, and reactions were recorded after 15 min. A reaction was considered positive if the mean diameter was X3 mm greater than the negative control. All tests were conducted by the same observer (M. V.). Exercise-induced bronchospasm (EIB) was performed in a subgroup of 2040 children, 437 (21.4%) from non-treated and 1603 children (78.6%) from treated communities. Peak expiratory flow rate (PEFR) was measured before and after 6 min of vigorous exercise as described previously [13].

Statistical analysis
Allergen skin test reactivity was defined as a positive skin test reaction to any allergen. Recent wheeze was classified as wheeze within the previous 12 months; recent rhino- conjunctivitis as the presence of rhinitis symptoms accompanied by itchy eyes within the previous 12 months; recent eczema symptoms as the presence of an itchy skin condition affecting the flexures within the previous 12 months; and EIB as a 15% or greater fall in PEFR after exercise. Analyses were performed using multiple logistic regression models allowing for clustering using robust standard errors. Age, sex, monthly household income, maternal educational level and water source were included as a priori confounders in all models. Other potential confounders were included in the final model if inclusion altered OR by 410%. Interactions were assessed using the Wald test. Because all children in the treated communities had received at least one dose of ivermectin, we could not do a separate analysis of the effects of treatment in the treated communities. Analyses were performed using STATA 10 (StataCorp, College Station, TX, USA).

Study population
A total of 3901 children were assessed, 2070 children from 27 non-treated and 1831 children from 31 treated communities. The mean cluster size was 76.7 (range 17-224) and 59.1 (range 14-223) in non-treated and treated communities, respectively. Results of allergen skin prick tests were available for 1983 (95.8%) and 1782 (97.4%) from non-treated and treated children and stool samples were collected from 1953 (94.3%) and 1794 (97.8%) of non-treated and treated children, respectively.
Demographic, socio-economic and environmental characteristics of the study children are shown in Table 1 and show differences between treated and non-treated children with respect to several characteristics. EOEP has achieved extremely high rates of coverage with ivermectin in the study communities over the past 15-17 years [15]. All children living in treated communities had received at least one dose of ivermectin and 79.3% had received 475% of designated treatments over the previous 5 years. Reported treatments with other anthelmintic drugs by parental questionnaire were similar among children from both treated (77.5%) and non-treated (78.7%) communities over the previous 6 months. Most treatments were bought directly by parents, were distributed through schools, or through doctor consultations. During the period of this study, there were no systematic programmes of periodic treatments with other anthelmintic drugs such as albendazole in any of the study communities.
The prevalence of any geohelminth infection was greater among non-treated children (86.2% vs. 62.7%), largely attributable to a higher prevalence of Trichuris trichiura infection (81.5% vs. 31.2%) ( Table 1). There was no difference in the prevalence of Ascaris lumbricoides between treated and non-treated children (48.7% vs. 57.3%), and surprisingly, the prevalence of hookworm was greater in treated than in non-treated children (14.6% vs. 3.9%). Infection intensities with A. lumbricoides did not differ significantly between treatment groups [geometric mean (GM) infection intensities, untreated 34 eggs per gram (epg) vs. treated 30 epg, P = 0.61] but the intensity of T. trichiura was significantly greater in untreated children (GM infection intensities, untreated 132 epg vs. treated 44 epg, Po0.0001).

Treatment area and allergic parameters
The effect of treatment on allergic parameters is showed in Table 2. Children living in treated communities had a greater prevalence of allergen skin test reactivity (SPT1) (16.7%) compared with those living in untreated communities (8.7%) (adjusted OR 2.10, 95% CI 1.50-2.94, Po0.0001). The prevalence of recent wheeze symptoms, rhino-conjunctivitis and EIB did not differ significantly between treated and non-treated children, but treated children had a higher prevalence of recent eczema symptoms than non-treated children (adjusted OR 2.24, 95% CI 1.05-4.78, P = 0.04). The prevalence of SPT1 was greater in 11 of 31 treated communities than in any of the nontreated community (Fig. 1), but no patterns were observed for the other allergic parameters (data not shown).

Effect of geohelminth prevalence on the association between treatment area and SPT1
The effect of treatment on SPT1 was explored in hierarchical analyses by examining whether the association between treatment and SPT1 might be explained by reductions in geohelminth prevalence. To investigate this, the effect on the association between treatment area and SPT1 of inclusion of each of A. lumbricoides, T. trichiura and hookworm was assessed either separately or simultaneously in the model. A reduction in OR will occur if the geohelminth is a mediating factor in the causal pathway between treatment and skin test reactivity. The only helminth that reduced the OR between treatment area and SPT1 was T. trichiura (change of OR from 2.10 to 1.69) ( Table 3). A. lumbricoides and hookworm had negligible effects. Sequential addition of geohelminths to the model showed a strong effect only when T. trichiura was included in the model (Table 3). A similar analysis for the association between treatment area and eczema symptoms did not show effects for any of the geohelminth parasites (Table 3). 0.55-0.91, P = 0.007). This effect appeared to be explained by a strong inverse association between skin test reactivity and T. trichiura infection (OR 0.72, 95% CI 0.58-0.90, P = 0.01) -neither A. lumbricoides nor hookworm infections were significantly inversely associated with SPT1. There was evidence for an inverse association between  (Table 4).
Wheeze, rhino-conjunctivitis, eczema symptoms and EIB. The prevalence and intensities of A. lumbricoides, T. trichiura and hookworm infections were not associated with symptoms of recent wheeze, rhino-conjunctivitis and eczema or with EIB.

Discussion
The present study examined the effect of long-term periodic treatments with a broad-spectrum anthelmintic drug, ivermectin, on the prevalence of allergy in schoolage children living in communities in a rural tropical area of Ecuador. To perform this, we compared the prevalence of allergen skin test reactivity and other parameters of clinical allergy between children who lived in communities that had received ivermectin annually or semiannually for a period of 15-17 years with children living in communities that had never received ivermectin treatment. The data provide evidence that long-term anthelmintic treatment may be accompanied by an increase in the prevalence of allergen skin test reactivity and perhaps recent eczema symptoms but no evidence for effects on asthma or rhinitis symptoms. Strengths of the study are the objective documentation of allergen skin test reactivity, exercise-induced bronchospasm, and geohelminth infections using standardized protocols. Because the study was conducted over a 2-year period, seasonal differences (i.e. dry vs. rainy season) in rates of skin sensitization could have biased the findings; however, an analysis of skin test reactivity by month showed no evidence for seasonal differences in allergen skin test reactivity (data not shown). A high proportion of children living in untreated communities had received other anthelmintic treatments during the previous 6 months -these treatments were generally purchased directly by parents from pharmacies and such treatments were not associated with any of the study outcomes (data not shown). We have shown previously that such sporadic and short-term treatments are not associated with SPT [6]. Treatment contamination of the untreated communities, if such treatment were to mediate an effect, would be expected to increase the prevalence of SPT in untreated communities and reduce the estimate of effect. Definitions of clinical allergy including eczema were questionnaire based and used standardized definitions [16]. Question-naire data could have been associated with possible biases -observation bias because treatment allocation was not concealed and recall bias because the educational levels of the primary respondents to the questionnaires, the mothers, differed by treatment group. Several environmental and socio-economic factors differed between the treated and untreated communities. Although we controlled for these differences by treatment area in the analysis, residual confounding or systematic bias cannot be excluded. However, ORs were stable with respect to controlling for confounding suggesting that these estimates may not be subject to significant confounding. An important potential limitation of the present study was the absence of data from two decades previously on the prevalence of SPT and allergic outcomes in the communities before the start of ivermectin distribution. We believe that the prevalence of these outcomes was unlikely to have been systematically different between these Afro-Ecuadorian communities because 20 years ago, these communities were more homogeneous than they are now given the geographic isolation of the study area, a similar tropical rain forest environment, and a shared lifestyle and ethnicity reinforced by contact between communities by river. Only more recently have differences started to emerge (see Table 1) caused by the building of roads and social and environmental changes associated with the process of modernization [20] that so far have started to intrude upon untreated more than treated communities. The differences between treated and untreated communities observed in Table 1 (e.g. maternal educational level, number of household electrical appliances and sources of drinking water) indicate a greater degree of 'rurality' in treated communities, a factor that is considered to be strongly protective against atopy [21]. Such a bias would be expected to reduce rather than increase the prevalence of SPT among children living in treated compared with untreated communities. The findings that SPT and eczema had a higher prevalence in the treated communities but asthma prevalence was similar between treated and non-treated communities are consistent with previous intervention studies: an elevated prevalence of SPT [10,12] and eczema [22] in treated arms but similar prevalence of asthma in children living in treated and non-treated arms [12,13].
Epidemiological studies have shown strong inverse associations between SPT1 and geohelminth infections in high-prevalence populations [5][6][7][8], and the inverse associations observed in the present studies are similar to those reported in previous studies conducted in rural Ecuador [6,13,23]. The inverse associations have been interpreted to indicate an active suppression of allergen skin test reactivity by active geohelminth infections [23].
Previous anthelmintic treatment studies have documented the effects of between 12 and 30 months of periodic treatments and have provided conflicting findings: (1) a non-randomized intervention study in Venezuela showed that monthly anthelmintic treatment with oxantel-pyrantel over 18 months increased the prevalence of allergen skin test reactivity to HDM from 17% to 68% among 94 children with a high prevalence of infection before treatment [10]; (2) an open-label placebocontrolled randomized intervention study in Gabon treated 165 children with a combination of praziquantel and mebendazole every 3 months for 30 months and showed that anthelmintic treatment increased the rate of developing skin sensitivity to HDM compared with the placebo (hazard ratio 2.51, 95% CI 1.85-3.41) among skin-test-negative children and the effect appeared to be mediated partly by reductions in infections with A. lumbricoides and T. trichiura [11]; (3) a double-blind randomized controlled trial in Vietnam treated 1566 schoolchildren in a hookworm-endemic region to receive placebo or mebendazole/albendazole at 3-monthly intervals and showed no effect of the intervention on clinical allergy but evidence of an increased prevalence of allergen skin test reactivity (OR 1.31, 95% CI 1.02-1.67) [12] and (4) a cluster-randomized study in Ecuador that allocated schools to monthly treatments with albendazole or no treatment showed no effect of treatment on allergen skin test reactivity or clinical allergy after 12 months of follow-up of 68 schools (1632 children) [13].
Possible explanations for differences in the effects of anthelmintic treatment on allergen skin test reactivity between studies are: (1) difference in treatment periodthe treatment period was much longer in the present study (15-17 years) compared with the previous negative Ecuadorian study (1 year). Periodic anthelmintic treatment given over 15-17 years may cut transmission levels, reduce the prevalence of active infections, and attenuate the development of immune regulatory mechanisms associated with chronic infection. The mechanisms by which chronic geohelminth infections modulate allergen skin test reactivity are unclear. Enhanced production of IL-10 by lymphocytes stimulated with parasite antigen has been associated with reduced skin test responses in children infected with schistosomiasis [24], but not geohelminth infections [12,25,26]. (2) Differences in anthelmintic drugs -there is no evidence that any of the drugs used in the intervention studies (albendazole, mebendazole, praziquantel, oxantel-pyrantel and ivermectin) have direct effects on allergic reactivity but they do have differing spectrums of activity against geohelminth parasites [27]. Albendazole given at single doses of 400 mg is extremely effective against A. lumbricoides but has limited effects on T. trichiura infection [27]. In the present study, long-term ivermectin had no effect on A. lumbricoides prevalence, but had a significant impact on T. trichiura prevalence (treated 31.2% vs. untreated 81.5%). These findings are consistent with those of previous studies of the anthelmintic effects of one to four doses of ivermectin [28][29][30]. (3) Differences in endemic geohelminths -the type of geohelminth parasites endemic in a particular geographic region may be an important determinant of findings from different studies [10][11][12][13]. The prevalence of A. lumbricoides (55.9%) was similar in the present study to the previous negative Ecuadorian study [13], although infection intensities were greater in the previous study (GM infection intensity 73 epg) [13]. The prevalence and infection intensities of T. trichiura were lower in the previous study (prevalence 55.8%, GM infection intensity 25 epg) [13]  indicated that the effect of anthelmintic treatment on SPT1 prevalence was mediated partly by this infection. A cohort study in urban Salvador in Brazil showed that infections, particularly high-intensity infections with T. trichiura during the first 3 years of life was associated with suppression of allergen skin test reactivity later in childhood and this effect was independent of later T. trichiura infections [8]. Thus, an explanation for the effect observed of T. trichiura infection in the present study may be the effect of treatment in reducing the prevalence of T. trichiura in early childhood, through a reduction in community transmission of infection. Such early effects could attenuate the development of immune regulatory mechanisms associated with chronic infection [31].
The effect of long-term ivermectin treatments on recent eczema symptoms was surprising, and was not associated with geohelminth infections in schoolchildren. There are four possible explanations for this observation: (1) data on eczema symptoms obtained by questionnaire may overestimate prevalence where pruritic infections such as scabies are common. However, because ivermectin is an extremely effective treatment for scabies [15], the prevalence of scabies would be expected to be much higher among non-treated than treated children and is, therefore, an unlikely explanation for the greater prevalence of recent eczema symptoms observed among treated children. In fact, childhood infections with ectoparasites such as scabies and lice (against which ivermectin is also highly effective [32]), which were ubiquitous before the start of ivermectin and are now less common (EOEP, unpublished data) in treated communities; (2) an increased prevalence of recent eczema symptoms among children living in treated communities could be caused by impaired regulation of skin inflammation arising from the absence of regulatory effects in the skin that could be induced by scabies and lice; (3) effects of anthelmintic treatment on reducing maternal or early infant geohelminth infections -a previous intervention study has provided some evidence that anthelmintic treatment of mothers during pregnancy may be associated with an increased prevalence of eczema in the offspring [22] and (4) pruritus and rash are common reactions to the microfilaricidal effects of ivermectin on dermal microfilariae of O. volvulus [33]. These so-called Mazzotti reactions are self-limiting and resolve within a few days of treatment, but only occur in the presence of active O. volvulus infection. It is unlikely that recent eczema symptoms might have been confused with such reactions in the present study because transmission of O. volvulus infection has been interrupted in all treated study communities for at least 10 years [14], none of the study children had evidence of dermal microfilariae in previous surveys, and no Mazzotti reactions were observed in those study communities that were included in detailed dermatological surveys conducted in 2004 and 2008 (EOEP, unpublished data). Further, serological assays to detect exposure to O. volvulus infection that have been conducted in samples of children from some of the study communities have been all negative (EOEP, unpublished data).
In conclusion, the present analysis comparing allergic parameters between children living in communities that have received periodic mass treatments with ivermectin with children from communities that have not received treatment provides some evidence that long-term anthelmintic treatments -in this case 15-17 years, to our knowledge by far the longest period of continuous treatment so far documented through a real world public health intervention with high rates of treatment coverage sustained over many years [14] -may be associated with an increase in the prevalence of allergen skin test reactivity and possibly also recent eczema symptoms. The increase in allergen skin test reactivity, but not eczema symptoms, was associated with a reduced prevalence of T. trichiura infection.