Detection of Influenza C Viruses Among Outpatients and Patients Hospitalized for Severe Acute Respiratory Infection, Minnesota, 2013–2016

We detected influenza C viruses mostly in children in both outpatients and inpatient surveillance populations. Our findings suggest that influenza C may be an underrecognized cause of outpatient and severe hospitalized illness in the United States.

Acute respiratory infections (ARIs) are a major cause of morbidity and mortality worldwide [1], and viral pathogens, including influenza viruses, cause many of these infections [2]. Two genera of human influenza-influenza A and B-are well studied and thought to cause most influenza-associated human disease. In contrast, less is known about a third genus, influenza C, which was first described in 1947 as being antigenically distinct from influenza A and B [3]. Like other influenza types, influenza C is a negative-sense, segmented RNA virus that circulates worldwide [4][5][6] and can cause disease in both the upper [7,8] and lower [9][10][11][12][13] respiratory tracts in humans as well as pigs [14] and dogs [15]. In experimentally infected individuals, the virus caused a febrile illness with mild upper respiratory symptoms [7], similar to most described cases of naturally acquired infections [8], and is thought to be less severe than other influenza types. However, some studies have reported episodic occurrences of more serious disease and hospitalizations [9,11,16]. The underlying reasons for these differences in disease severity remain unclear.
Although less well studied than other influenza viruses, influenza C infection appears to be common. Cross-sectional, population-based serological studies show peak prevalence of influenza C-specific antibody responses reaching 78%-100% [6,[17][18][19]. The primary target of influenza C-specific antibody responses is the surface glycoprotein hemagglutinin-esterase-fusion (HEF) protein, which is analogous to the 2 separate hemagglutinin and neuraminidase proteins of influenza A and B [20]. Unlike influenza A viruses, which undergo regular antigenic drift, influenza C viruses appears to be more antigenically stable, with the same antigenic types circulating over a period of multiple years [21]. Interestingly, the presence of influenza C-specific antibodies does not confer complete protection, as adults with serological evidence of past exposure can develop symptoms and shed viruses [8]. A potential explanation for this may be the circulation of at least 6 distinct lineages of influenza C viruses [21]. However, the ways in which these lineages evolve and cocirculate in a population are not well understood.
In addition to the perceived low pathogenicity, limitations of diagnostic testing have hindered surveillance for influenza C virus. While there are commercially available tests for influenza A and B viruses, no such tests exist for influenza C virus. Furthermore, the virus is difficult to culture, even in confirmed outbreaks [22]. The expansion of molecular diagnostic techniques into research and surveillance settings could significantly increase knowledge of the spectrum of influenza C disease and patterns of circulation. Thus, beginning in May 2013, the Minnesota Department of Health (MDH) with the support of the Centers for Disease Control and Prevention (CDC) incorporated molecular testing for influenza C into existing sentinel surveillance systems for outpatient and inpatient ARI, allowing us to study the epidemiology of influenza C virus infection.

Inpatient Surveillance
The Minnesota Severe Acute Respiratory Illness (SARI) sentinel surveillance program was established at 3 hospitals in Minneapolis and St. Paul in May 2013, including 2 general hospitals and a large pediatric hospital system. Patients qualified as a possible SARI case if they were admitted to an inpatient unit for ARI or asthma exacerbation with 1 or more ARI symptoms. Testing of possible cases was encouraged but not required. After completion of clinician-ordered testing at the submitting facility, any residual upper or lower respiratory specimens were placed in viral transport media and submitted to the MDH public health laboratory for testing. Medical records were reviewed for all patients with submitted specimens using a standardized case report form (CRF).

Outpatient Surveillance
The Influenza Incidence Surveillance Project (IISP), which was established in 2009, conducts surveillance for influenza-like illness (ILI) and ARI through 4 primary care clinics that serve patients of all ages, with sites currently in Hennepin, Kandiyohi, Kittson, and Rock counties. ILI is defined as fever with cough or sore throat and ARI as any 2 of the following symptoms: fever (temperature ≥38⁰C or, after 2015, patient/family-reported fevers), cough, sore throat, rhinorrhea, or congestion. Clinical staff collected an upper respiratory specimen for testing in the MDH public health laboratory and a limited CRF from the first 10 patients who presented each week with ILI and with ARI, as described previously [23].

Laboratory Testing
Specimens submitted from SARI between May 2013 and December 2016 and from IISP between September 2014 and December 2016 were tested at the MDH public health laboratory. Total nucleic acids (DNA and RNA) were extracted from specimens. Nucleic acid testing for viral and bacterial pathogens was performed (Supplementary Figure S1). using TaqMan real-time polymerase chain reaction (rPCR) and realtime RT-PCR (rRT-PCR) assays with oligonucleotide primers/ probes obtained from the CDC [24,25] or a Luminex respiratory pathogen panel. Pathogens detected in referring hospital laboratories were also included in the analysis when available.

Sequencing of HEF Genes
HEF genetic sequence analysis was performed by the CDC's Diagnostic Development Team, Influenza Division, on influenza C-positive specimens with ample residual volume and rRT-PCR cycle threshold (CT) values ≤32. The Invitrogen SuperScript III One-Step RT-PCR System with Platinum Taq High-Fidelity kits were used for PCR amplifications. Primers are available upon request. PCR products were purified by ExoSAP-IT for the PCR Product Clean-Up kit (USB Corporation). Sequencing reactions were performed using an Applied Biosystems BigDye Terminator v3.1 Cycle Sequencing Kit and an Applied Biosystems Sequencer 3730 DNA Analyzer. Sequences analyzed were obtained from this study or from the National Center for Biotechnology Information's Influenza Resource (http://www. ncbi.nlm.nih.gov/genomes/FLU/; Supplementary Figure S2) and were aligned using the CLUSTALW program. Phylogenetic analyses were performed using Molecular Evolutionary Genetics Analysis software (version 5.1) [26]. The evolutionary history was inferred using the neighbor-joining method [27].

Data Analysis
Patients were included as influenza C cases if they tested positive for influenza C virus. Data were analyzed using Epi Info 7.2.0.1. The data presented are public health surveillance data and not subject to institutional review board approval for human research protections.

Study Population
During the study period, we completed influenza C testing on 12 484 specimens, including from 10 202 hospitalized SARI patients and from 2282 outpatients with ARI or ILI. The SARI population was younger than the IISP population, with a median age of 2.72 years (interquartile range [IQR], 0.5-31.7 years; range 0 days-101.2 years) compared to 20.29 years (IQR, 10.1-33.9 years; range 27 days-95 years) for IISP. However, older patients are represented as well, with 10.9% of SARI patients and 5.9% of IISP patients aged >65 years.
Seventy individuals tested positive for influenza C virus; 59 were identified among SARI patients and 11 from IISP, giving detection rates of 0.58% and 0.48%, respectively (Table 1). Four hospitalized cases had multiple specimens that tested positive for influenza C: 2 with positive detections within 1 day of each other, 1 with 2 positive detections 21 days apart, and 1 case with 5 positive specimens detected over a 4-month period. In cases with multiple detections, only the data that corresponded to the date of the initial positive specimen were included in the analysis. 7-20 days 6 (9) 6 (10) 0 (0) >20 days 1 (1) 1 (2) 1 (4)

Known Treatment/Outcome
Received influenza anti-viral 3 (5) 1 (5) The median age of all influenza C cases was 20 months (range, 3 weeks-84 years; IQR, 8 months-9 years) and 59% were aged <2 years. Among hospitalized patients, influenza C was detected among 0.80% of children aged <5 years and among 0.41% of children aged ≥5 years. Among outpatients, influenza C was detected among 0.39% of children aged <5 years and among 0.49% of cases aged ≥5 years. Influenza C virus was detected in 1 of 1114 individuals aged >65 years, which is equal to a detection rate of 0.09% (data not shown). Influenza C virus detections were distributed similarly by sex, and 39% were non-white. Of cases with a known county of residence, 57% resided within either Hennepin (Minneapolis) County or Ramsey (St. Paul) County, and the remainder were distributed among 10 surrounding counties (data not shown).

Clinical Characteristics
Among all influenza C cases, cough (60%), fever (47%), and congestion (41%) were most commonly reported. Among the 59 hospitalized cases for whom a more extensive symptom inventory was obtained, 54% reported respiratory distress or shortness of breath. At least 1 comorbid condition was reported for 58% of hospitalized cases (55% of cases aged <18 years; 83% of cases aged ≥18 years). The most common comorbidities included asthma or chronic obstructive pulmonary disease and prematurity. For hospitalized cases, the median length of stay was 2 days (IQR, 1-4). Seven cases were admitted to the intensive care unit (ICU), 4 received mechanical ventilation, and none died. Among the cases admitted to the ICU, all were aged <3 years and had at least 1 underlying condition, the most common of which were prematurity and congenital heart disease. Influenza C virus was the sole pathogen detected for 2 of the ICU-admitted cases, both of whom were premature; 1 had ventilator-dependent chronic respiratory failure and the other had cyanotic congenital heart disease. Of the 2 cases with widely spaced influenza C virus detections suggestive of prolonged shedding, both had underlying medical comorbidities. The case with 5 detections had acute lymphoblastic leukemia and was undergoing maintenance chemotherapy, and the other had multiple chronic medical comorbidities including a history of prematurity and neurological and upper airway abnormalities.

Copathogen Detections
More than 1 pathogen was detected in 46 (66%) influenza C cases (Table 2). Although codetections of multiple pathogens were more frequent in younger children, occurring in 34 (71%) cases aged <5 years, we also found more than 1 pathogen in 12 (55%) cases aged ≥5 years. Rhinovirus/enterovirus and respiratory syncytial virus (RSV) were the most frequently codetected, similar to their overall prevalence among all SARI cases (data not shown). Among influenza C cases aged ≥5 years, influenza A virus, human metapneumovirus, coronavirus NL63, and rhinovirus were most commonly codetected with influenza C virus. Of the 5 cases admitted to the ICU, 2 had only influenza C virus detected and 3 had codetections, including adenovirus, parainfluenza 2, influenza B virus, RSV, or Moraxella catarrhalis.

Seasonality
Overall, influenza C exhibited a peak of infection in 2014-2015 that overlapped with the seasonal pattern of other   Received mechanical ventilation 4 (7) 1 (5) Received extracorporeal membrane oxygenation 0 (0) 0 (0) Died 0 (0) 0 (0) a Patients hospitalized compared to outpatients, with or without co-detection of pathogens. The last column represents patients with influenza C virus infections without any other pathogen detected in respiratory sample. b Signs/symptom data collected only for hospitalized patients Odds ratio of infection statistically elevated relative to reference category of adult, white and male patients with *P < .05, ** P < .01; *** P < .001.

Lineages
To determine which lineages were circulating, we sequenced the HEF genes of the influenza C isolates from 39 cases who tested positive from December 2014 through February 2016. We detected only the C/Kanagawa and C/Sao Paulo lineages in our isolates (Supplementary Figure S2). The C/Kanagawa lineages originated from 8 counties, and the C/Sao Paulo lineages originated from 5 counties without apparent geographic clustering or differences in age (data not shown). Interestingly, the C/Kanagawa lineage was overrepresented among hospitalized cases, whereas the C/Sao Paulo lineage was more evenly split between inpatients and outpatients.

DISCUSSION
Although the first published case of influenza C was described in New York in 1949 [3], few subsequent studies have characterized the burden of disease in the United States. Because other respiratory viruses show seasonal and geographic variability in circulation [28], we reasoned that influenza C might show similar variability and thus merit direct study of the epidemiology. Furthermore, because most routine laboratory tests do not detect influenza C virus, we hypothesized that influenza C may be an underappreciated cause of respiratory illness. To address this deficit, we incorporated molecular testing into 2 existing large-scale surveillance programs drawn from a largely urban and suburban population.
We detected influenza C virus infections in 2 of the 3 years of the surveillance program. The average frequency of influenza C detection over the 44-month study period was similar among outpatients and hospitalized SARI patients (0.48% and 0.58%, respectively) and was consistent with previous reports, ranging from 0.2% [29] to 2.6% [30]. However, we noted substantial year-to-year variability in the number of cases, with 73% of cases occurring during a single season, similar to previous observations [19] and other seasons with few or no detections. The annual variability seen in our study illustrates the risk of missing outbreaks of this disease with intermittent testing, highlighting the value of including influenza C in ongoing surveillance programs. Furthermore, while some previous reports suggested no seasonal pattern [31], studies from Japan [19] and Canada [30] have found peaks in the winter and spring seasons, which are consistent with our findings. We observed a peak from December 2014 to May 2015, overlapping with the seasonal peak for influenza A and B. Additional surveillance is needed to further clarify whether true seasonality exists.
In addition to characterizing the burden and temporal patterns of influenza C virus circulation, our data provide insights into the characteristics of individuals with possible influenza C virus-associated disease. We showed a statistically higher percentage of influenza C virus detection in those aged 6 months to 2 years (1.22%), with lower rates of infection through the remainder of childhood and adulthood. There was also a lower rate of detection in those aged <6 months, corresponding to a time when children have residual maternal antibodies that could provide some immunological protection [32]. The lowest rates of detection occurred among teenagers, with slightly higher rates in adults, though this was not statistically significant. While we cannot rule out the possibility that rates of influenza C detection in adults were artificially low due to a bias against testing for respiratory pathogens in older age groups, another recent study found almost identical rates (0.3%) of influenza C infection in adults systematically tested for viral infection [33].
Although we do not have corresponding serological data from our population, multiple studies from varied locations have consistently demonstrated high levels of seropositivity against influenza C virus in older children and adults [6,[17][18][19]. Our observation that rates of influenza C virus detection decline with age fits with previous studies that demonstrated increased prevalence of influenza C antibodies with increasing age. Interestingly, 20% of our influenza C virus detections were among symptomatic adults. These data may suggest that while many adults have had prior exposure to influenza C virus, the immunological protection from past exposure may wane over time. Alternatively, it may be possible to have serial infections with antigenically distinct circulating lineages. The presence of at least 6 influenza C lineages raises the question of whether exposure to new strains might allow infection of previously exposed individuals. Consistent with reports from Japan [19], we identified 2 lineages-C/Kanagawa and C/Sao Paulo-that were cocirculating in our study population. The sample numbers were small, and we were unable to find any differences between strains with respect to age or geographic distribution (data not shown). Nevertheless, we did find the C/Kanagawa lineage more consistently among hospitalized cases. Ongoing testing through established surveillance systems will make it possible to identify the emergence of new antigenic lineages.
Our data may also challenge the perception that influenza C virus is rarely associated with severe disease [31]. We detected influenza C virus at similar rates in hospitalized cases and outpatients with ARI, raising the possibility that it may be associated with a spectrum of disease. The reasons for the variability of disease severity are likely multifactorial and may include both host and viral factors. From a host perspective, all cases admitted to the ICU had at least 1 underlying medical condition, suggesting that underlying illness is a risk factor for severe disease and consistent with literature on other respiratory infections such as RSV [34]. Specifically, prematurity was noted in 17% of all cases and 80% of cases admitted to the ICU. Interestingly, we observed prolonged shedding of virus in 2 cases with underlying medical illness. From a viral standpoint, many of our cases had coinfections, and we detected 2 circulating influenza C lineages, which may differ in pathogenicity. At this time, no specific antiviral therapy is available [35], so treatment is primarily supportive. Nevertheless, awareness of respiratory viral infections such as influenza C that are not currently detected on routine respiratory pathogen panels may inform hospital infection control policy and would support presumptive respiratory isolation based on symptoms, even in the setting of negative test result.
Although our study provides valuable information about naturally occurring influenza C virus detections, it has several limitations. First, clinical testing was done at the discretion of the treating clinician. Thus, some patients may not have had testing ordered or had a residual specimen for testing at MDH, leading to an underestimation of cases. A second caveat is that children, particularly those aged <1 year, are more likely to be hospitalized for respiratory infections and therefore have respiratory specimens obtained for testing [36]. This is reflected in the fact that 57% of patients from the SARI surveillance program were aged <5 years. Third, no information was obtained about underlying medical conditions or follow-up data for outpatient cases, and thus, a comparative analysis of comorbidities and outcomes with SARI cases was not possible. Finally, influenza C virus was commonly codetected with other viral pathogens. When compared to culture-based methods, molecular detection methods such as ours frequently detect multiple infections, which poses a challenge to interpretation [37]. Viral nucleic acid may be detectable for prolonged periods of time, including in asymptomatic individuals, and it can be difficult to distinguish among symptomatic infection, asymptomatic shedding, and noninfectious viral debris [37]. Therefore, it will also be important to measure influenza C virus prevalence in asymptomatic controls to address the potential for asymptomatic shedding. Nevertheless, by conducting surveillance in a large population and by incorporating multiplexed testing for other known pathogens, we identified 24 cases with influenza C virus but no copathogens detected. This included 2 cases who became ill enough to warrant ICU admission. While we cannot rule out the possibility that another agent is the primary cause of symptoms, these data raise the possibility that influenza C virus has the potential to cause more disease than previously appreciated.
In conclusion, we used a large-scale surveillance system to identify cases of influenza C in the Upper Midwest. Our data suggest that influenza C is detected in a minority of patients with symptomatic respiratory illness but may be a cause of severe disease and periodic outbreaks.

Supplementary Data
Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes
Financial support. This work was supported by the Council of State and Territorial Epidemiologists through a cooperative agreement from the CDC (5U38HM000414).
Potential conflicts of interest. All authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.