Introduction

Hemolytic uremic syndrome (HUS) is a severe acute disease defined by hemolytic anemia, thrombocytopenia, and thrombotic microangiopathy. It is a major cause of acute kidney injury in children. Food-borne infection with Shigatoxin (Stx1/2)-producing Escherichia coli (STEC) is one of the main causes of bacterial diarrhea in typical HUS.1,2,3,4,5 Another form of HUS is due to the neuraminidase of different serotypes of Streptococcus pneumoniae (SP-HUS).6,7,8 The atypical form of HUS (aHUS), represents about 1% of all HUS forms and is frequently associated with a dysregulation of the alternative complement pathway due to genetic mutations affecting the genes encoding the alternative pathway components C3 and Factor B, or, more frequently, the complement regulatory proteins Factor H (FH), Factor I (FI), or CD46 (also called MCP for membrane cofactor protein). Most of these hereditary forms have variable penetrance and occur at all ages, suggesting that a secondary trigger is necessary for the onset of disease. Various pathogens have been reported associated with aHUS.1,2,3,4,5,6,7,8 It has been reported with viral infections like H1N1 or VZV (varicella-zona virus) associated with CD46 mutation9,10 or with Epstein-Barr virus,11 Cytomegalovirus,12 virus of hepatitis A and C13,14 or enterovirus.15

Bacterial pathogens like Streptococcus pneumoniae, Shigella flexneri, Hemophilus influenza, and Bordetella pertussis have also been found in aHUS patients with or without known mutations in the complement genes.16,17,18,19

Lastly, parasites like Plasmodium falciparum or Plasmodium vivax have also been associated with HUS.20,21

One subtype of aHUS, called auto-immune HUS (AI-HUS), is due to functional neutralization of FH by autoantibodies and has the particularity to affect school-age children.22,23,24 The prevalence of AI-HUS seems to be higher in the Asian subcontinent, where it accounts for up to 56% of all pediatric HUS in India, as compared to 10–25% of aHUS in European populations.25,26 All worldwide cohort studies have reported a very strong association of AI-HUS with a genetic condition consisting of a homozygous deletion of two CFH-related genes, CFHR1 and CFHR3.27,28

In India, the frequency of the homozygous CFHR1/R3 deletion in the normal population does not explain the highest frequency of AI-HUS.26 Other factors, infection in particular, may therefore be implicated in disease onset. Several studies have reported the presence of prodromes in AI-HUS patients: febrile illness in 27.3% of cases, gastrointestinal symptoms such as diarrhea or vomiting in 27.7% or respiratory infections in 5.4%.19,25,26,29 Furthermore, some studies identified pathogens isolated from AI-HUS patients such as norovirus (2 cases), varicella or STEC (1 case for each pathogen).22,30

In the present study, we screened for a large number of pathogens in patients presenting with acute-phase HUS and prospectively included after admission to a single Indian pediatric nephrology unit. For this purpose, after validation of the analytical method, we used a multiplex assay allowing the simultaneous detection of 15 gastrointestinal pathogens (9 bacteria, 3 parasites, and 3 viruses) in nucleic acids extracted from stools collected from 35 acute-phase HUS patients. The results were then analyzed according to the presence (n = 16) or not (n = 19) of anti-FH autoantibodies in the patients’ blood.

Patients and Methods

Study population

Stool samples were collected from 35 acute-phase HUS patients between November 2012 and February 2015. Demographic and biological characteristics are presented in Table 1, where patients are assigned to two groups according to the anti-FH antibody status. All patients and their parents were informed about the study purpose and gave their consent.

Table 1 Demographic, clinical and biological findings for the study population
Full size table

Study design

Sera samples (n = 35) were used to evaluate anti-FH status.

Stool samples (n = 35) were submitted to pathogen detection by using xTAG-based multiplex PCR. This assay allows the detection of nine bacteria: Campylobacter (1 probe), Clostridium difficile (2 probes: 1 for toxin A, one for toxin B), E. coli (5 probes: E. coli 0157, enterotoxigenic E. coli LT, enterotoxigenic E. coli ST, Stx 1, Stx 2), Salmonella (2 probes), Shigella (1 probe), Vibrio cholerae (1 probe), Yersinia enterolitica (1 probe); three parasites: Giardia (1 probe), Cryptosporidium (1 probe), Entamoeba histolytica (1 probe); and three viruses Adenovirus 40/41 (1 probe), Rotavirus A (1 probe), Norovirus (2 probes: Norovirus GI and Norovirus GII). Nucleic acid extracts were then evaluated according to the following routine algorithm: real-time PCR for Norovirus GI/GII and Rotavirus (Centre National de Référence des virus Entériques; Laboratoire de Virologie, CHU de Dijon, France); real-time PCR for Campylobacter, Salmonella, Yersinia enterocolitica, ETEC, STEC, E.coli 0157, EIEC/Shigella, (Centre de Référence National des E coli, Hôpital Robert Debré, Paris, France);

Sputum samples, collected from two patients presenting with symptoms of upper respiratory tract infection as prodroms, were sent to a reference laboratory (Centre National de Référence de la Rougeole et des Paramyxoviridae Respiratoires Humains, CHU Caen; Pr A. Vabret) for microbiological investigations by PCR multiplex Respifinder-plus assay (PathoFinder CE, Eurogentec, Madrid, Spain).

Anti-FH antibody and CFHR1 genetic status assessment

These experiments were performed, as previously described.26

CE-IVD xTAG GPP assay

For the xTAG-GPP assay, DNA/RNA was isolated from stool samples of the study using the EasyMag instrument (Biomérieux, Marcy I’Etoile, France). Initially, 100–150 mg of feces and 10 µL of an internal control (xTAG MS2) were added to 1 mL PBS in 2 mL Precellys soil-grinding SK38 tubes (Bertin Technologies, Montigny-le-Bretonneux, France). After vortexing, 10 min of incubation at room temperature and centrifugation, 800 µL of the supernatant was extracted on the EasyMag instrument. PCR amplification and hybridization were performed with the xTAG-GPP (Luminex, Theradiag, Croissy-Beaubourg, France) kit as per the manufacturer’s instructions, and data were acquired on the MagPIX system (Theradiag, Croissy-Beaubourg, France). Further analysis and interpretation of the data were carried out using the xTAG Data Analysis Software GPP (TDAS-GPP, Luminex Corp, Austin, TX). Positive and negative results were determined according to an algorithm taking into account the target median fluorescent intensity (NET-MFI), and the specific/background MFI ratio obtained for each probe.31,32 In addition, for Salmonella spp, a positive result was based on the simultaneous positivity of both specific probes.

Reference results and validation of the Luminex platform

As the Luminex Gastrointestinal Pathogen Panel xTGAG® was not routinely used in standard laboratory testing, we first evaluated the pertinence of this multiplex PCR platform by performing two nucleic acid extraction approaches, including the Promega Maxwell 16 instrument (Promega, Madison, WI) and EasyMag (BioMérieux, Marcy l’etoile, France). In total 22 well-characterized stool samples were used to evaluate the nucleic acid extraction methods including virus-positive samples (n = 9), bacteria-positive samples (n = 8) and negative samples (n = 5). For viruses, sample positivity was determined using real-time PCR with further confirmation by genome sequencing. For bacteria positivity was determined by real-time PCR,

In addition, 11 nucleic acid extracts were used as positive controls (virus, n = 6 and bacteria, n = 5) to evaluate the accuracy of this platform to detect well-known pathogens. Nucleic acids from all these 11 samples were extracted using the EasyMag instrument.

Results

Characteristics of the study population according to the anti-FH Status

Serum samples were available for the entire study population (n = 35). Patients were then classified into two groups according to their anti-FH status. In total 16 patients (45.7%) were positive for the anti-FH antibody. The mean antibody titer was 8653 AU/mL (Range: 250–29,000 AU/mL). Demographic, clinical, and biological findings are presented in Table 1. A homozygous CFHR1 deletion was found in 12 among the 14 tested (86%) anti-FH antibody positive patients.

Validation of the xTAG GPP platform

In the first step, we evaluated the accuracy of the assay for the identification of pathogens using purified nucleic acid extracts. As shown in Table 2, the assay was first validated for sensitivity and accuracy in detecting the pathogens using nucleic acid extracted from positive stools, which were initially characterized with PCR assays. The xTAG multiplex assay was as accurate as the qPCR methods, and allowed detection of multiple pathogens in a single reaction. The assay was also validated for any cross-reactivity: it was very specific for adenovirus 40/41 and did not cross react with other subtypes of adenovirus (Table 2).

Table 2 Pretested positive nucleic acid samples used to validate the sensitivity of xTAG-GPP
Full size table

The second step focused on selecting the best nucleic acid extraction machine. Stool samples are among the most difficult clinical samples to process because of the presence of very potent inhibitors of nucleic acid amplification.33 It is widely accepted that the efficiency of nucleic acid extraction and purification influences the sensitivity, reproducibility and the accuracy of RT-PCR target detection. Our objective was to identify the most suitable nucleic acid extraction machine to prepare stool samples for further detection using the xTAG GPP protocol. Extraction efficiencies of total nucleic acids from 22 stool samples by EasyMAG and Maxwell 16 were 28.1 ± 34.2 and 20.6 ± 27.6 ng/µL, respectively. Even though both extraction methods yielded comparable amount of nucleic acids, Nucleisens EasyMAG allowed the identification of additional pathogens or probes in 6/22 samples (Table 3, Fig. 1).

Table 3 Pathogens detected in the samples used for validation of the assay after extraction by Maxwell 16 or Nuclisens EasyMag and detection using xTAG GPP
Full size table
Fig. 1
figure 1

Results of 22 well-characterized stool samples used to evaluate nucleic acid extraction methods. a Net MFI (Mean of Fluorescence Intensity subtracted from background) after the MagPIX analysis for the extraction performed by Maxwell automated nucleic acid extraction with representation of individual samples and pathogens. b Representation of Net MFI for each sample and pathogen for extractions by Nuclisens EasyMAG. Arrows indicate the additional pathogens detected by Nuclisens easymag when compared to Maxwell extraction for each patient

Full size image

xTAG GPP investigations for pathogens in aHUS samples

The Nuclisens EasyMAG method was used to analyze 35 stool samples. At least one pathogen was detected in 16 samples (45.7%). There was no statistical difference of duration between disease onset and stool collection in the positive (median: 17 days) and the negative (median: 14.5 days) samples (t-test = 1.098).

Samples from anti-FH antibody-positive patients were twice as likely as those from seronegative patients to harbor pathogens. Indeed, at least one pathogen was found in 62.5% (10 out of 16) of the anti-FH antibody-positive patients and in 31.5% (6 out of 19) of negative patients (Fig. 2). Among the two groups of patients, even though the prevalence of pathogens was heterogeneous, Clostridium difficile (6 out of 16) and Giardia spp (4 out of 16) were the most frequently detected pathogens in patients with anti-FH antibody, whereas these pathogens were found in 2, and 3 out of 19 of stools collected from anti-FH antibody-negative patients, respectively. Rotavirus, Norovirus, Salmonella sp, Shigella flexneri, and Entamoeba histolytica were also detected (Tables 4a and 4b). Among the anti-FH-positive patients, no significant difference of antibody titers was found between patients with or without stool pathogens (Fig. 3a), or between patients who were positive or negative for Clostridium difficile (Fig. 3b). However, anti-FH IgG titers were significantly higher in patients who were positive for Giardia than in other patients (Fig. 3c, p = 0.0198).

Fig. 2
figure 2

Results for 35 stools samples collected from acute-phase HUS patients. Representation of Net-MFI obtained in anti-FH-negative patients (a) and in anti-FH antibody-positive patients (b). Results were determined according to an algorithm taking into account the target median fluorescent intensity (MFI) and the specific/background MFI ratio obtained for each probe. For Salmonella spp, a positive result was defined by the simultaneous positivity of both specific probes

Full size image
Table 4a Results of pathogens detected in stool samples in patients with anti-FH-negative HUS
Full size table
Table 4b Results of pathogens detected in stool samples in patients with anti-FH-positive HUS
Full size table
Fig. 3
figure 3

Anti-FH IgG titers according to the presence of pathogens in stools. No significant difference in titers was observed according to the presence of any pathogen (a), for the presence of Clostridium difficile (b). Anti-FH titers were significantly higher in patients with Giardia-positive stools than in patients negative for Giardia (c)

Full size image

In the whole analysis, only one sample from one AI-HUS patient was positive for E. coli O157:H7 but negative for the Stx probes. This sample was also positive for other pathogens (Salmonella, Clostridium difficile Toxin B, Giardia, Entamoeba histolytica). None of the stools collected from the two sets of patients were positive for STEC or for Stx probes (Tables 4a and 4b). This result was confirmed by routinely used real-time PCR. None of the stools were positive for Vibrio cholera, Yersinia enterocolitica, Crypstosporidium, ETEC, and Adenovirus 40/41.

Among the two sputum samples tested, one was positive for Enterovirus/ Rhinovirus and the other for Influenza. No pathogen was found in the stool of these patients

Association between clinical symptoms and biological results

No significant difference was observed in stool results between males and females, in the all cohort (Fischer t-test p = 0.726), or among the anti-FH positive and negative patients (Supplemental Figure S1). In the anti-FH positive group, 9/12 (75%) CFHR1-deleted patients had positive stools and the two non-deleted patients had negative stools (Table 4a).

Fever and diarrhea were more frequently observed in the anti-FH negative patients than in the positive ones (Tables 1, 4a and b). The most frequent symptom was vomiting, observed in 23 out of 34 patients (67.6%), equally distributed between AI-HUS and non-AI-HUS. No association was observed between the presence of any symptom (fever, abdominal pain, vomiting or diarrhea) and the stool test positivity (data not shown).

Discussion

After different case reports of non-STEC infections leading to HUS, this study is the first aiming to identify gastro-intestinal pathogens at disease onset in a cohort of HUS patients who were subsequently classified according to the presence or not of anti-FH autoantibody. It also provides valuable information about prodromal infections, which may possibly act as a secondary hit for the triggering autoimmune response.

On the basis of cohort studies reporting a high frequency of gastrointestinal (GI) symptoms especially in AI-HUS patients (up to 27%), we chose to investigate the presence of GI pathogens in a cohort of pediatric HUS from India, the country with the highest frequency of reported AI-HUS. Furthermore, as STEC-related HUS are not routinely diagnosed in India, the study also wanted to investigate the frequency of this etiology. The xTAG-GPP technique allows multiplex detection of 15 pathogens (9 bacteria, 3 viruses, and 3 parasites) in a single test. The validation experiments allowed us to demonstrate the efficiency of pathogen detection using the multiplex assay chosen and to select the most sensitive reagent for the extraction of nucleic acids. Extraction using Nuclisens EasyMAG demonstrated greater sensitivity than the Maxwell automated extraction method because it allowed the detection of some additional probes in 6 out of 22 samples. This method was therefore used for the subsequent analyses. The results yielded several interesting facts about the differential prevalence of the disease and pathogens at the acute phase of the disease. At least one pathogen was found in 16 of the 35 studied stools (45.7%). Interestingly, the prevalence of the pathogens in the anti-FH antibody-positive patients was twice that in anti-FH-negative patients (62.5% vs. 31.5%). The most frequent pathogens found were Clostridium difficile (37.5% in AI-HUS and 10.5% in non AI-HUS), Giardia intestinalis (25% in AI-HUS and 15.7% in non AI-HUS), Salmonella and Shigella, Rotavirus A, Norovirus GI/GII, and Entamoeba histolytica. Thus, a large variety of pathogens were detected, however no specific distribution between AI-HUS and non-AI-HUS patients. Several pathogens were found in seven patients, with up to five pathogens in one patient belonging to the AI-HUS group. However, it is important to note that all samples that were positive for Clostridium difficile were collected from patients who had received antibiotics. This may explain the high frequency of Clostridium difficile carriers in our study population. There was no difference of antibiotics administration between anti-FH positive and negative patients (Tables 4a and 4b).

Among the pathogens detected in this study, Norovirus,30 Campylobacter,34 Clostridium difficile35,36 have already been reported as being implicated in HUS, suggesting they could be also implicated in HUS here.

Thus, no single pathogen was implicated exclusively as a trigger for one form of HUS and several factors like geographical origin or microbiota probably influence the prevalence of the disease.

Interestingly, in this study, none of the patients was positive for the ETEC and STEC probes contained in this assay. Despite the small number of patients tested, this could suggest that in India, as in several Asian countries like Japan or Taiwan, STEC-induced HUS is not the predominant etiology of the disease.37,38

Other infections may trigger the disease and 2 tested throat samples collected from patients having symptoms of respiratory infection were found to be positive: one sample was positive for Enterovirus/ Rhinovirus and one for Influenza virus. These two samples were collected from patients where no pathogen was found in stools. The association between Influenza A or H1N1 virus and aHUS has been reported earlier in association with a mutation in CD469 suggesting that presence of these viruses should be investigated in HUS patients, mainly during epidemic periods and even in the absence of characteristic symptoms.

Although this cohort is small and explorations of other infections are needed, it provides the first insights into the prevalence of pathogens in both classes of disease. In our study, no association between the digestive clinical symptoms and the stool positivity was observed. However, this is not in contradiction with the hypothesis of a role of these pathogens in the initiation of the disease by inducing complement activation and/or inducing an autoimmune response in the forms with antibody. Indeed, in AI-HUS, the identification of an infectious trigger is of particular interest due to recent advances in our understanding of this disease. Its occurrence mainly in school-age children and its relatively seasonal onset argue for the potential role of minor infections.26 Moreover, its association with the homozygous deletion of the CFHR1 and CFHR3 genes points out the similarity between the C terminal domains of Factor H, which are targeted by the autoantibodies, and CFHR1 protein, which is deficient in the majority of affected people. A predominant epitope has recently been localized in this region at a site close to a binding site of proteins synthesized by some pathogens as an immune evasion strategy39: a surface protein Tuf from P. aeruginosa, a surface protein Psp C from S. pnenumoniae, extracellular complement binding protein (Ecb) from S. aureus, and the outer surface protein Osp E from Borrelia burgdorferi. This has led to the hypothesis that the binding of such proteins to CFH may induce exposure of a neoepitope on CFH leading to autoimmune responses in patients with CFHR1 deficiency.40

In conclusion, the present study shows the practicality of a multiplex PCR approach for the investigation of GI pathogens in stools. It revealed a higher prevalence of GI pathogens in AI-HUS than in non-AI-HUS patients admitted to one Indian pediatric nephrology unit and also showed the absence of STEC-HUS in the latter group. Identification of the triggering infection in AI-HUS should not only be proof of the concept of the mechanism of immunization against Factor H but may also be a way to prevent the onset of the disease or its relapses.