https://orcid.org/0000-0002-9014-3616
Abstract
Shiga toxin-producing Escherichia coli (STEC) are enteric pathogens that cause hemolytic-uremic syndrome (HUS). Ruminants, especially cattle, are their main reservoir. This study describes the seroepidemiology of STEC in rural and urban populations in Argentina, a country with a high HUS incidence.
A cross-sectional study was performed in patients without gastrointestinal symptoms. IgG antibodies against Stx2 were detected by western blotting.
Anti-Stx2 antibodies were detected in 14.56% of serum samples, more frequently in rural (19.38%) than urban residents (12%). Seropositivity was associated with lower socioeconomic status (SES). Among the other variables considered, thawing homemade hamburgers before cooking them, and the lack of knowledge about HUS were also associated with seropositivity. A multivariate logistic regression analysis performed with the variables that were statistically significant showed that only the SES index remained significant. As SES was measured based on several variables, we further analyzed each one of them and found that the lack of a high education level was statistically associated with seropositivity.
The present findings have implications for STEC prevention efforts, highlighting the importance of considering SES and risks factors linked to different SES levels when targeting consumer-level public health interventions.
Introduction
Shiga toxin-producing Escherichia coli (STEC) can cause typical hemolytic-uremic syndrome (HUS). HUS is more common in children under 16 years old with the highest incidence in children up to 5 years old and is the leading cause of acute renal failure.1 The majority of outbreaks and sporadic cases of illness have been associated with serotype O157:H7; however, a wide range of non-O157 STEC serotypes are increasingly reported.2 Worldwide, there is substantial geographic variation in the prevalence of STEC serotypes as well as in the incidence of HUS.3–6
The most important virulence property of STEC is the production of Shiga toxins (Stxs). Stxs are AB5 toxins composed of one A subunit (32 kDa), which inhibits protein synthesis, and a pentamer of B subunits (7.7 kDa each), which binds to glycolipid receptors. These toxins comprise two major groups that are serologically distinct, called Stx1 and Stx2.1
STEC is commonly found in the intestinal tract of ruminants, particularly cattle. In humans, the main route of infection is fecal-oral, usually because of the consumption of contaminated foods of bovine origin and also to the ingestion of other foodstuffs and contact with recreational water, animals and animal waste, or to person-to-person contact. Although several strategies to lower STEC prevalence in cattle are under development, good hygienic practices at all stages of the food chain and access to safe drinking water are fundamental to prevent STEC infections.
The presence of antibodies against Stx has been detected in patients who developed HUS1,7,8 and a protective role of anti-Stx antibodies has been shown experimentally.9–11 In addition, Ludwig et al.7 suggested that the prevalence of anti-Stx antibodies in healthy individuals could indicate population immunity against HUS.
Karmali et al.1 demonstrated that in Canada antibodies against Stx were more frequent and developed at an earlier age in residents of dairy farms than in those of urban areas, and related this fact to the more frequent exposure to STEC from cattle. Furthermore, contact with farm animals or their environment has been identified in different countries as a risk for STEC infection.12–14
STEC infections are a serious problem worldwide. Particularly in Argentina, HUS is endemic and presents a high incidence, with near 400 new cases reported each year.15 Similar to other countries, most HUS cases are associated with strains carrying Stx2, and notably, this toxin type is the most frequent among STEC strains isolated from bovines and their environment.16–18 In a previous study, we found that the incidence rate of HUS was significantly higher in rural than urban population,19 but information about population immunity was lacking. Therefore, the first objective of the present study was to compare the prevalence of anti-Stx2 antibodies between these two populations. The second objective was to assess the association of different factors with seropositivity to gain epidemiological knowledge useful for developing preventive measures.
Methods
Human serum samples
A cross-sectional study was performed on 460 patients who underwent blood collection for other diagnostic purposes at hospitals and public health centers in rural and urban areas of Buenos Aires Province (Sanitary regions I and VIII, Supplementary Fig. S1). Immunosuppressed patients and those previously treated with antibiotics were excluded. Written informed consent was obtained from each participant or care provider in case of minors. The calculated minimum sample size was 383, assuming a seroprevalence of 50% with an absolute error of 5% and a confidence level of 95%.
Information about clinical and epidemiological data was obtained with a questionnaire designed ad hoc. Socioeconomic status (SES) was measured by scoring the variables proposed by Odriozola et al.,20 i.e. education level, health insurance, number of employed family members, employment of the family’s main breadwinner and household possessions. All serum samples were stored at −20°C until use.
Detection of IgG antibodies against Stx2
Recombinant Stx2 B-subunit was produced and purified by affinity chromatography as described elsewhere.21 It was kindly provided by Dr Yanil Parma (Buenos Aires, Argentina).
Serum samples were tested in a blinded fashion by a western blot (WB) assay adapted from the method described by Karmali et al.1 Briefly, the B subunit of Stx2 was subjected to SDS-PAGE and transferred to a nitrocellulose membrane, which was stained with Ponceau red to confirm protein transfer and then cut into longitudinal strips containing ≅1 μg of the protein. They were blocked with 3% skim milk in Tris-buffered saline and then incubated overnight with each serum sample diluted 1:100 in the blocking solution. Strips were washed, blocked and incubated with a 1:1000 dilution of peroxidase-conjugated anti-human IgG (Vector Laboratories, USA) in blocking solution. The reaction was revealed with H2O2, 4-chloro-1-naphthol and imidazole. Results were qualitatively analyzed.
Data entry and statistical analysis
Patient information and WB results were entered into Epi-Info database version 7 (Centers for Disease Control and Prevention, Atlanta, GA, USA). Quantitative variables were analyzed with t-test or Wilcoxon’s rank sum test.
A univariate analysis was performed to assess associations between the dependent variable (seropositivity to B subunit of Stx2) and the independent variables (potential risk or protective factors) obtained in the survey. All variables were compared by using a χ2-test with an alpha level of 5%. Second, a stepwise-forward logistic regression was performed with seropositivity to B subunit of Stx2 as the dependent variable, and patient characteristics were offered as independent variables. When seeking for the simplest model that could explain the presence of antibodies to B subunit of Stx2, only variables significantly associated (P < 0.05) with the dependent variable after χ2-tests were offered to the model. The estimation method was maximum likelihood with a convergence criterion of 1E−8. To identify possible confusion factors, the association between variables was assessed by χ2-test. In addition, interaction among variables was also evaluated in the logistic regression model. Analyses were performed using Epi-Info database version 7 (Centers for Disease Control and Prevention, Atlanta, GA, USA) and SAS version 9.3 (SAS Institute Inc., Cary, NC, USA).
Rural population included individuals living on farms or in small rural villages (<2000 inhabitants).22
Ethical considerations
The study complied with the revised Declaration of Helsinki for biomedical research involving human subjects and was approved by the Research Ethics Committee of the Penna Hospital (Bahía Blanca city, Buenos Aires Province).
Results
A total of 460 serum samples were analyzed, being 300 from urban and 160 from rural residents. Individuals were aged 1 month to 92 years old (median, 32 years old; mean, 33.6 years old). The age distribution for each population is shown in Fig. 1A.
(A) Age distribution of individuals sampled from each population; (B) age distribution of seropositive individuals.
A positive immune response to the B subunit of Stx2 was observed in 67 of 460 serum samples (14.56%). No differences were found in seroprevalence among age ranges (P = 0.5895), but all tested individuals in the range 81-90 years of age were seronegative (Table 1 and Fig. 1B). The frequency of seropositive individuals was significantly higher in rural (19.38%) than in urban residents (12%) (P = 0.0327). This difference was observed in almost all age ranges except for 0-10 years of age (Fig. 1B).
Multivariate logistic regression analysis of factors associated with seropositivity to B subunit of Stx2 in the bivariate analysis
| Factora | OR | 95% CI | P |
|---|---|---|---|
| Type of population (rural versus urban) | 1.28 | 0.64-2.53 | 0.4846 |
| SES (low versus high) | 3.56 | 1.47-8.6 | 0.0188 |
| Previous information about HUS (no versus yes) | 1.74 | 0.88-3.45 | 0.1134 |
| Factora | OR | 95% CI | P |
|---|---|---|---|
| Type of population (rural versus urban) | 1.28 | 0.64-2.53 | 0.4846 |
| SES (low versus high) | 3.56 | 1.47-8.6 | 0.0188 |
| Previous information about HUS (no versus yes) | 1.74 | 0.88-3.45 | 0.1134 |
aThe ‘way of hamburger thawing before cooking’ was not considered because of the scarce number of samples with complete information available for this analysis.
Multivariate logistic regression analysis of factors associated with seropositivity to B subunit of Stx2 in the bivariate analysis
| Factora | OR | 95% CI | P |
|---|---|---|---|
| Type of population (rural versus urban) | 1.28 | 0.64-2.53 | 0.4846 |
| SES (low versus high) | 3.56 | 1.47-8.6 | 0.0188 |
| Previous information about HUS (no versus yes) | 1.74 | 0.88-3.45 | 0.1134 |
| Factora | OR | 95% CI | P |
|---|---|---|---|
| Type of population (rural versus urban) | 1.28 | 0.64-2.53 | 0.4846 |
| SES (low versus high) | 3.56 | 1.47-8.6 | 0.0188 |
| Previous information about HUS (no versus yes) | 1.74 | 0.88-3.45 | 0.1134 |
aThe ‘way of hamburger thawing before cooking’ was not considered because of the scarce number of samples with complete information available for this analysis.
No differences were found in the mean age of seropositive and seronegative individuals, considering either all the individuals sampled (P = 0.5167) or each type of population independently (P = 0.6112 for rural; P = 0.7756 for urban) (Wilcoxon’s rank sum test) (Fig. 2A and B).
(A) Comparison of age distribution between seropositive and seronegative individuals; (B) comparison of age distribution of seronegative and seropositive individuals between urban and rural populations.
SES for each individual was measured as a combination of education and occupation of family members, household possessions and health insurance. A higher frequency of seropositivity was detected among individuals with low SES (P = 0.0086). This association was also evident in either urban or rural populations when analyzed separately.
Among the other variables considered in the questionnaire, thawing of homemade hamburgers at room temperature or in the refrigerator was a risk factor for Stx2 seropositivity with respect to cooking the hamburgers without thawing them first. The lack of knowledge about HUS was also associated with seropositivity (P = 0.0272). Remarkably, there was no association between seropositivity and any of the other 70 variables included in the questionnaire (Supplementary Table S1).
All the statistically significant variables were included in the logistic regression model, except ‘the way of hamburger thawing before being cooked at home’, because of the scarce number of individuals who provided information on this variable. After the multivariate logistic regression analysis, only the SES index remained significant (P = 0.0188), with an odds ratio (OR) value of 3.56 (Table 1). As this index was measured based on several variables, they were further analyzed individually and only the lack of a high education level showed an association with seropositivity (Table 2).
Analysis of SES variables in relation to seronegative/seropositive results
| SES variable | Level | Negative | Positive | P |
|---|---|---|---|---|
| Educational level | Low | 316 (84.7) | 57 (15.3) | 0.0210 |
| High | 62 (95.4) | 3 (4.6) | ||
| Car possession | Yes | 198 (86.1) | 32 (13.9) | 0.8780 |
| No | 166 (85.6) | 28 (14.4) | ||
| Health insurancea | Yes | 195 (84.1) | 37 (15.9) | 0.2230 |
| No | 179 (88.6) | 23 (11.4) | ||
| Number of family members with employment | 1 or none | 19 (85.5) | 1 (14.7) | 0.2000b |
| >1 | 319 (95.0) | 56 (5.0) | ||
| Employment status of the family’s main breadwinner | Retired/Unemployed | 59 (81.9) | 13 (18.6) | 0.3500 |
| Employee/Employer | 300 (86.2) | 48 (13.8) | ||
| Household possessionsc | 1-3 items | 91 (80.5) | 22 (19.5) | 0.1384 |
| 4-6 items | 148 (87.1) | 27 (12.9) | ||
| 7-10 items | 138 (88.5) | 18 (11.5) |
| SES variable | Level | Negative | Positive | P |
|---|---|---|---|---|
| Educational level | Low | 316 (84.7) | 57 (15.3) | 0.0210 |
| High | 62 (95.4) | 3 (4.6) | ||
| Car possession | Yes | 198 (86.1) | 32 (13.9) | 0.8780 |
| No | 166 (85.6) | 28 (14.4) | ||
| Health insurancea | Yes | 195 (84.1) | 37 (15.9) | 0.2230 |
| No | 179 (88.6) | 23 (11.4) | ||
| Number of family members with employment | 1 or none | 19 (85.5) | 1 (14.7) | 0.2000b |
| >1 | 319 (95.0) | 56 (5.0) | ||
| Employment status of the family’s main breadwinner | Retired/Unemployed | 59 (81.9) | 13 (18.6) | 0.3500 |
| Employee/Employer | 300 (86.2) | 48 (13.8) | ||
| Household possessionsc | 1-3 items | 91 (80.5) | 22 (19.5) | 0.1384 |
| 4-6 items | 148 (87.1) | 27 (12.9) | ||
| 7-10 items | 138 (88.5) | 18 (11.5) |
Data are presented as frequency (%).
aHealth insurance coverage for >50% of the family members.
bFischer exact test. Not all patients provided information on all aspects surveyed.
cThe items considered are telephone, automatic washing machine, microwave oven, video player, cable television, personal computer, internet connection, refrigerator with freezer and cell phone.
Analysis of SES variables in relation to seronegative/seropositive results
| SES variable | Level | Negative | Positive | P |
|---|---|---|---|---|
| Educational level | Low | 316 (84.7) | 57 (15.3) | 0.0210 |
| High | 62 (95.4) | 3 (4.6) | ||
| Car possession | Yes | 198 (86.1) | 32 (13.9) | 0.8780 |
| No | 166 (85.6) | 28 (14.4) | ||
| Health insurancea | Yes | 195 (84.1) | 37 (15.9) | 0.2230 |
| No | 179 (88.6) | 23 (11.4) | ||
| Number of family members with employment | 1 or none | 19 (85.5) | 1 (14.7) | 0.2000b |
| >1 | 319 (95.0) | 56 (5.0) | ||
| Employment status of the family’s main breadwinner | Retired/Unemployed | 59 (81.9) | 13 (18.6) | 0.3500 |
| Employee/Employer | 300 (86.2) | 48 (13.8) | ||
| Household possessionsc | 1-3 items | 91 (80.5) | 22 (19.5) | 0.1384 |
| 4-6 items | 148 (87.1) | 27 (12.9) | ||
| 7-10 items | 138 (88.5) | 18 (11.5) |
| SES variable | Level | Negative | Positive | P |
|---|---|---|---|---|
| Educational level | Low | 316 (84.7) | 57 (15.3) | 0.0210 |
| High | 62 (95.4) | 3 (4.6) | ||
| Car possession | Yes | 198 (86.1) | 32 (13.9) | 0.8780 |
| No | 166 (85.6) | 28 (14.4) | ||
| Health insurancea | Yes | 195 (84.1) | 37 (15.9) | 0.2230 |
| No | 179 (88.6) | 23 (11.4) | ||
| Number of family members with employment | 1 or none | 19 (85.5) | 1 (14.7) | 0.2000b |
| >1 | 319 (95.0) | 56 (5.0) | ||
| Employment status of the family’s main breadwinner | Retired/Unemployed | 59 (81.9) | 13 (18.6) | 0.3500 |
| Employee/Employer | 300 (86.2) | 48 (13.8) | ||
| Household possessionsc | 1-3 items | 91 (80.5) | 22 (19.5) | 0.1384 |
| 4-6 items | 148 (87.1) | 27 (12.9) | ||
| 7-10 items | 138 (88.5) | 18 (11.5) |
Data are presented as frequency (%).
aHealth insurance coverage for >50% of the family members.
bFischer exact test. Not all patients provided information on all aspects surveyed.
cThe items considered are telephone, automatic washing machine, microwave oven, video player, cable television, personal computer, internet connection, refrigerator with freezer and cell phone.
Discussion
Main findings of the study
This paper focuses on the study of the seroprevalence against Stx2, the main virulence factor of STEC, among individuals not suffering from gastrointestinal disorders. The results showed differences in seropositivity between rural and urban residents, indicating a higher exposure to STEC among the former group. To gain epidemiological knowledge useful for the development of prevention measures, the association of different factors with seropositivity was also studied. Interestingly, seropositivity was associated with a low SES, and, particularly, to lower education levels.
What is already known on this topic
In a previous study,19 we found a higher HUS incidence in rural than in urban population (12.7 versus 7.1 cases per 100.000 inhabitants), with significantly lower median age for rural cases. However, there was no information about the serological status of urban and rural residents that could indicate differences in exposure or immunity to STEC. Moreover, studies that address the serological status against this bacterium are scarce worldwide. With that in mind and to identify factors associated with seropositivity, we chose inhabitants from the same region of our previous study (Buenos Aires Province, Argentina), which presents a range of HUS cases near the national average.15
What this study adds
In the present study, anti-Stx2 antibodies were detected in 14.6% of individuals who had no signs of infectious disease or symptoms of gastrointestinal disorders. This seroprevalence value is notably lower than that of a study8 that was also performed in Argentina but only with children (53% of seropositivity to B subunit of Stx2) and that of another1 with people of different ages in Canada (52.2% of seropositivity to the B subunit), but is comparable to the prevalence determined by Ludwig et al.7 in Germany (10% among control sera) and by Guirro et al.3 in Brazil (16.6% of healthy children positive to that subunit).
The percentage of seropositive individuals was higher in rural than in urban residents, in line with the results obtained by Karmali et al.1 but with lower prevalence values. Many studies investigating risk factors for STEC infection and STEC-associated HUS have identified direct zoonotic and environmental transmission,13,23–25 and for Argentine children, Rivas et al.12 found that one of the risk factors for sporadic STEC infection was contact with farm animals and their environment.
Geographical analyses performed in Sweden, the Netherlands, Canada and England showed a positive relationship between cattle density and incidence of STEC O157 infections,14,26–28 and of both O157 and non-O157 infections in Germany.29 According to Elson et al.,14 the association with animal density was most likely due to environmental exposure in areas with high density of farmed animals and poorly managed private water supplies, although other routes of infection, such as the consumption of locally sourced food, could exist. Interestingly, from the analysis of data from several countries, Pianciola and Rivas15 noted that STEC O157:H7 strains from human patients reflected the predominant genotypes present in cattle in each country.
In our previous study19 we found that the distribution of factors hypothetically associated with HUS was different in rural and urban populations. Rural HUS patients had more frequent contact with bovine feces and consumption of raw milk, less access to public water and waste services; they washed their hands less frequently and had a higher proportion of relatives who worked in contact with animals or raw meat. Among urban HUS patients, a higher proportion lived in crowded conditions and consumed minced meat and homemade hamburgers. Taking into account those differences and the existence of multiple transmission routes, risk factors for STEC infection can be different between rural and urban residents, or, furthermore, between sporadic cases and outbreaks, or different age groups. SES has also been shown to be associated with risk for several gastrointestinal infections, possibly having a great impact on risk factors and transmission dynamics within different regions.14
In the present study, seropositivity to Stx2 was 3.56 times more frequently detected among individuals with low SES. As far as we know, there are no studies about the relationship between SES and seropositivity to Stx2, but IgA levels against O157 LPS were reported to be higher among adolescents living in Israeli communities with lower SES.30 A systematic review by Newman et al.31 indicated that studies performed in high-income countries found conflicting results or have failed to find any association between SES and STEC infection. Recent studies detected higher rates of infection among groups with higher SES14,32,33 but another one found no association.34 Particularly, some authors observed a relationship between higher SES and foreign travel which could contribute to explain the association with STEC infection.14,32
Differences in measurement methods for SES could contribute to partially explain dissimilar results among countries or studies. As discussed by Newman et al.,31 different methods are used to measure SES, some rely upon individual-level metrics and others use community-level metrics (census tract-level). In addition, SES can also be determined either based on only one metric, e.g. educational level, home ownership or as an index by combining multiple data. As we used this last strategy, we further analyzed each variable that was considered in SES calculation, and, notably, we found that the lack of a high education level was the only SES variable associated with seropositivity. Considering that the lack of knowledge about HUS was another factor associated with seropositivity, a possible explanation could be that people with lower education levels probably have less information about proper food handling procedures, being consequently more exposed to STEC.
SES can result in both differential exposure and vulnerability to foodborne illness, factors that can interact rendering a differential disease incidence.31 An early study from Southern Africa,35 suggested that the higher HUS incidence found in high SES urban children was related to lower environmental exposure and a supposed lower immunity against the etiological agent. Rural residents are at a higher risk of exposure to E. coli O157:H7 than urban residents, and this frequent antigenic stimulation can lead to immunity against this pathogen and reduce clinical illness.36–38 Furthermore, the presence of antibodies of IgG class, as it was found in the present study, suggests long-term or repeated exposure to STEC.39
According to Karmali,40 serum antibodies against Stx could protect against the development of HUS while antibodies to bacterial colonization factors could prevent pathogen colonization. Interestingly, Karmali et al.1 found that in urban residents, the frequency of seropositivity to Stx at different age ranges followed a pattern, which inversely reflected the age-related incidence of HUS, supporting the role of Stx antibodies in protective immunity. In dairy-farm residents, these authors did not find that pattern, as the frequency of seropositivity to Stx was already elevated in the first decade of life, consistent with the frequent exposure to bovine STEC from an early age. In the present study, however, we did not find differences in seroprevalence among age ranges.
How can we interpret the present results in light of those of our previous study? We previously characterized the epidemiology of HUS in the same region and found a higher incidence of HUS in the rural population, with a lower median age than that of the urban patients.19 As antibody presence reflects a previous contact with the antigen, rural residents appear to be more exposed to STEC, which explains the higher rural incidence of HUS. But what about the protective role of the antibodies? Apparently, the results from our two studies do not support such a role. However, a higher exposure to STEC can lead not only to a higher HUS incidence but also to a higher level of protection in the general population; although the population may have a protective immunity, young children will be at risk in the period when maternal antibodies decline, and children still have not yet produced their own antibodies. A study performed in Canada41 indicated that rural residents are likely to experience subclinical immunizing STEC infections at a young age, finding a high proportion of rural children under 10 years of age who were seropositive to Stx1. A higher exposure to STEC from a young age in rural children and the fact that not all the diverse STEC strains present in animals are necessarily pathogenic to humans could lead to the development of antibodies against STEC at an earlier age in rural children who consequently could become protected. Therefore, this hypothesis could explain our previous finding that HUS rural cases developed at an earlier age than urban cases. Unfortunately, we were unable to confirm this because of the limited number of samples from children (particularly from the rural area), as blood tests are not so frequently performed for regular health checks on children as on adults.
Limitations of the study
The data collection covered a broad spectrum of the Buenos Aires Province geographical area; however, sampling was not probabilistic because a facility-based cross-sectional study with a convenient sampling method was used. As indicated before, children were underrepresented, especially young children from rural areas. Another limitation was that a low response rate was observed for some of the variables, and therefore, they could not be included in the logistic regression analysis. Being aware of the limitations of the study, results were interpreted with caution and some meaningful conclusions could be drawn.
Conclusions
Our results reinforce the recommendation to consider SES and to identify associated exposure factors when developing public health interventions to prevent diseases caused by STEC. Moreover, it is essential to determine personal behaviors and habits of high-risk populations from the region of interest. Further studies are needed to identify appropriate educational methods targeted at different cultural and socioeconomic groups in the prevention of food-borne disease.
Data availability
The data underlying this article will be shared on reasonable request to the corresponding author.
Acknowledgments
Authors thank the technical assistance of Med. Vet. Guillermo Arroyo and M.R. Ortiz, and are grateful for helpful discussions with Dr Héctor Tarabla. Health care workers who performed serum sample collection at Debilio Blanco Villegas Hospital, Rodríguez Larreta Hospital, Penna Hospital and Pérez Cambet Laboratory are also gratefully acknowledged. M.A.R. was holder of a fellowship from Ministerio de Salud de la Nación (Becas Ramón Carrillo-Arturo Oñativia). A.K., M.L.S.P. and P.M.A.L. are members of the Research Career of CONICET (Consejo Nacional de Ciencia y Tecnología).
Funding
This work was supported by Agencia Nacional de Promoción Científica y Tecnológica (grant number PICT bicentenario 1459).
Conflict of interest
The authors declare no conflict of interest.
Mariana Alejandra Rivero, Dr., Assistant Professor
Alejandra Krüger, Dr., Assistant Professor, Adjunct Researcher
Edgardo Mario Rodríguez, Professor
Marcelo Lisandro Signorini Porchietto, Dr., Independent Researcher
Paula María Alejandra Lucchesi, Dr., Associate Professor, Independent Researcher
