Risk factors for seasonal influenza virus detection in stools of patients consulting in general practice for acute respiratory infections in France, 2014‐2016

Background Previous studies reported detection of influenza RNA in stools of patients with seasonal influenza infection. While this detection may have a clinical significance, other factors may influence the stool positivity for influenza viruses. Objectives The objective of this study was to investigate demographical, clinical, and microbiological factors which could favor the presence of influenza viral RNA in the stools of patients with laboratory‐confirmed influenza infection. Methods Acute respiratory infection (ARI) patients were enrolled by general practitioners (GP) during two winter seasons (2014‐2016). Nasopharyngeal swabs, stool specimens, and clinical data were collected. Samples were tested for 12 respiratory pathogen groups (nasopharyngeal and stool specimens) and for 12 enteric pathogens (stool specimens). Results Among the 331 patients with ARI enrolled by GP, 114 (34.4%) presented influenza infection. Influenza RNA was detected in stool samples of 21% (24/114) of the 114 stool specimens analyzed. Hospitalization (adjusted odds ratio (aOR) = 7.8 (95% confidence interval (CI)) [1.7‐33.7], P = .02), age between 45 and 64 years (aOR = 4.8 [1.7‐14.5], P = .01), consumption of raw shellfish and/or mollusks (aOR = 16.7 [3.6‐90.9], P = .00), and use of antibiotics (aOR = 6.4 [2.1‐19.8], P = .006) or antiviral treatment (aOR = 7.4 [1.9‐29], P = .01) were significantly associated with an increased odds of the detection of influenza RNA in stools. Among the 24 stool samples subjected to viral isolation, no one showed virus growth. Conclusions These findings will be useful to studies investigating the dissemination route of influenza viruses to gastrointestinal tract.


Seasonal and pandemic influenza virus enters and replicates in cells
of the upper respiratory tract where the virus recognizes sialic acid molecules linked to the Gal of glycoprotein on the surface of host epithelial cells. 1 However, viral RNA of these viruses has been detected in stools of patients with confirmed influenza infection [2][3][4][5][6][7] with an overall reported prevalence of 20.6%. 8 Although the main route of human influenza virus infection is respiratory, gastrointestinal (GI) symptoms such as diarrhea, nausea, vomiting, and abdominal pain are not uncommon manifestations. [9][10][11][12][13][14][15] Currently, only avian influenza virus A(H5N1) is known to replicate in human intestinal tissues 16 and to cause severe GI symptoms. 17 While seasonal and pandemic influenza virus is likely to spread to the GI tract of patients after a primary respiratory infection, the route of dissemination remains unknown. Current knowledge explains the detection of human influenza viruses in feces because of (i) swallowing of influenza viruses from the upper respiratory tract; (ii) remnants of infected submucosal intestinal antigen-presenting immune cells; and (iii) virus replication in intestinal cells. 8 The significance of the detection of influenza viral RNA in stools, their dissemination on GI tract, their correlation with GI symptoms, and their viability in stool remain unclear and are still debated. 7,18 A recent study reported that while swallowed influenza seasonal viruses are vulnerable to digestive juices, artificial highly viscous mucus (imitating sputum, nasal discharge, and other mucous membrane secretions) seems to protect viral RNA and virions, allowing their detection in feces. 19 It seems important to identify whether other factors would favor influenza detection in stools with the aim of limit interpretation biases in the future.
Thus, in this prospective study, we investigated for the first time microbiological, demographical, and clinical factors which could influence the presence of influenza viral RNA in stools of patients with laboratory-confirmed influenza infection. Results from this study will help and inform future studies that investigate the role of human influenza virus in stools.

| Ethic
This study was approved by the ethics committee (CPP V, ref number 14.078). The protocol was conducted in accordance with the Helsinki Declaration. All samples were coded and tested anonymously. None of the authors collected samples. Samples were collected and sent to the test laboratory by general practitioners (GPs) involved in the research project. Patient information was stored according to national regulations, and access to such data was restricted (permission CNIL 471393). The patient's identities were not disclosed at any stage.
Written informed consent was obtained from patients by the GP's.
For children under the age of 18, parents or legal guardians gave permission for their participation in this project. Consent from the child was also obtained, depending on her/his age and maturity.

| Design and study population
This is a prospective, observational, cross-sectional study, conducted in patients consulting a GP of the French Sentinelles Network 20 for an acute respiratory infection (ARI) during two winter seasons, from November 2014 to April 2015 and from November 2015 to April 2016. The case definition of ARI was "any person with a sudden onset of symptoms and at least one of the following four systemic symptoms: fever or feverishness, malaise, headache, myalgia, AND at least one of the following three respiratory symptoms: cough, sore throat, or shortness of breath." All patients were recruited within 48 hours of the onset of symptoms.

| Data collected
Two types of samples were obtained for each enrolled patient: a nasopharyngeal swab and a stool sample. The nasopharyngeal specimen was collected by the GP and was sent with the case report form (CRF) to the test laboratory by post in a triple packaging as required by the United Nations class 6.2 specifications. The CRF included variables such as age, sex, professional status, seasonal influenza vaccination, ARI signs and symptoms, GI symptoms, presence of chronic disease, risk factors for severe influenza, travel 15 days before consultation, drug consumption 7 days before consultation (antiviral and/or antibiotic and/or anti-inflammatory and/or antipyretics and/or other), consumption of particular foods (tap water, oysters, mussels and/or shellfish, cooked or raw), depression, drug prescription, and required hospitalizations. Included patients were asked to collect stool specimens and send them to the laboratory within 48 hours after GP consultation by post in triple packaging as required by the United Nations class 6.2 specifications. phages) was added to each extraction tube to assess the quality of the extraction at the end of the amplification. 21

| Quantitative reverse transcriptase polymerase chain reaction (RT-qPCR) tests
A RT-qPCR test was considered positive if it passed internal positive controls and had an exponential curve. Samples were excluded if the internal control value had a cycle threshold (Ct) value >35.

| Influenza virus isolation
Nasopharyngeal and stool specimens that were positive for influenza virus were inoculated onto Madin-Darby canine kidney (MDCK) tissue cells with an aliquot of clinical specimens and incubated for 7 days using a method reported previously. 5

| Microbiological analyses of nasopharyngeal and stool samples of laboratory-confirmed influenza patients
The detailed results from the microbiological investigation are presented in

| Factors associated with laboratory-confirmed detection of seasonal influenza viruses in stools
Bivariate association between the RNA detection of influenza virus in stools (yes/no) in laboratory-confirmed influenza patients and demographical, clinical, and management factors are presented in Table 1.
Patients who had consumed raw shellfish and/or mollusks or had Multivariable analyses reported that hospitalization (aOR = 7.8 [1.7-33.7], P = .02) was significantly associated with the fecal viral RNA detection (Table 1). Three patients (12.5% of 24 fecal viral RNA positive) have been hospitalized (ID patient 4, ID patient 6, and ID patient 9, Table 3 (Table 1). We noted that the confidence intervals were quite large due to small sample sizes. Results of the present study suggest that age, hospitalization, and food and drug consumption seem to modify odds of detection of influenza RNA in stools of laboratory-confirmed influenza patients.

| DISCUSSION
However, microbiological and clinical characteristics such as influenza virus type, Ct mean values, or GI symptoms seem to not be associated with the influenza RNA positivity of stool specimens. However, we did not report successful recovery of viable influenza virus from stools, boosting the hypothesis that intestinal cells could not support replication of influenza seasonal viruses. 5,23,24 Overall, the proportion of influenza viral RNA in stool specimens reported in the present study (21%) was in line with that reported in a recent meta-analysis (20.6% (95% CI, 8.9-35.5)). 8 In our study, fecal viral RNA-positive detection ranged from 8.3% in children aged less than 5 years to 37.5% in adults belonging to the 45-to 64-year age group. We observed that belonging to the age group of 45-64 years and to be hospitalized increased the odds of detection of influenza virus in stools. This is in line with a previous study reporting positive fecal viral RNA association in stools of hospitalized adults patients, 5 with a detection rate of 32% in patients aged between 50 and 65 years.
The source of influenza viruses in feces and how the viruses pass through the gastrointestinal tract are poorly understood.
Swallowing of influenza viruses from the upper respiratory tract and hematogenous dissemination to organs through infected lymphocytes in severe influenza cases are possible explanations for fecal viral RNA detection. 5,25,26 The increased odds of hospitalization in patients with influenza RNA-positive stools seems to be more in agreement with the hematogenous dissemination of influenza viruses in severe cases with high viral load. 4,27 A positive association with the consumption of raw shellfish, and/or mollusks, and/or antibiotic and/or antiviral administration was observed. These three factors are known to increase the risk of having diarrhea, vomiting, or mild stomach symptoms resulting from the disruption of the normal microflora from drug consumption 28,29 or from the presence of enteric pathogens or toxin in food, or even intolerance to certain seafoods. 30 Similar to other studies, no association was found between GI symptoms and influenza detection. 31,32 Interestingly, among the four patients who had influenza RNA exclusively in the stool and had consumed raw shellfish, and/or mollusks, three reported having GI symptoms, but none of the enteric pathogens screened here were detected in their stools. These symptoms could be caused by other enteric pathogens or allergy, 33 intolerance, 34 or toxin 35 responsible for GI symptoms and present on shellfish and mollusks. Among the patients with antibiotic or antiviral drug prescriptions, we did not collect data on GI symptom development after GP consultation, but stool samples were collected 2 days after GP consultation; therefore, patients had started antibiotic and/or antiviral treatment at time of stool collection. We could make the assumption that disruption of the gut microflora leads to an increase in swallowing and gastric flu, which helps influenza viruses reach the lower gastrointestinal tract. A recent study reported that swallowed influenza viruses become resistant to simulated gastric acid and bile/pancreatic juice and could be detected in stools if certain protective mechanisms such as viscous mucus protect it from degradation in the GI tract. 19 Limitations of this study include the observational study design for which residual bias cannot be excluded, and the small sample size resulting in wide confidence intervals. We did not carry out patient follow-up to collect clinical information after GP consultation. The detection of influenza virus in stool specimens may have been reduced due to inhibitory material present in the gastrointestinal tract and stool. Ct values are a semiquantitative measure of viral load; thus, it is impossible to assign specific viral loads by pathogens.
Notwithstanding its limits, this is the first prospective multicenter study, with a standardized patient screening and a central-