Influenza infection in immunosuppressed patients: clinical characteristics, risk factors and effect of antiviral therapy

Background: Influenza infection was a vital threat to immunosuppressed patients with longer viral shedding; however, data on these populations in China are still lacking. We analyzed clinical characteristics, risk factors for admission to intensive care unit (ICU) and death, and effect of antiviral therapy in these populations. Methods: We analyzed 73 immunosuppressed inpatients tested positive for influenza virus using reverse-transcription polymerase chain reaction during the 2018-2019 influenza season. Medical data were analyzed using descriptive statistics. Univariate analysis and multivariate logistics analysis were used to identify risk factors. Results: The most common immunosuppression type was malignancies with chemotherapy 73.9% (54/73), then hematopoietic stem cell transplantation 19.2% (14/73). The most common presenting symptom was fever in 91.8% (67/73) patients, then cough 59.6% (34/57) and muscular soreness 35.1% (20/57). Complications and co-infections were found in 38.4% (28/73) and 17.8% (13/73) patients respectively, which significantly prolonged the hospital stay. Antiviral treatment after 48 hours was significantly associated with admission to ICU, mechanical ventilation and death. Combination and double dose of neuraminidase inhibitors did not significantly reduce the admission to ICU and death. 15.1% (11/73) patients were admitted to ICU and 8.2% (6/73) patients died. Risk factors for admission to ICU were long symptom onset (OR 5.60, P =0.018) and co-infection with other infections (OR 68.66, P =0.019), and presence of dyspnea was independently associated with death (OR 48.00, P =0.003) through multivariate logistics analysis. Seasonal influenza vaccination in preceding 12 months only took up 2.7% (2/73). Conclusion: Fever and other classical symptoms may be absent in immunosuppressed recipients, and conducting influenza virus detection at the first time is a good choice for early diagnosis. Antiviral treatment within 48 hours is of significance; however, patients

may not benefit from combination and double dose of neuraminidase inhibitors.
Immunosuppressed patients with dyspnea, long symptom onset and co-infection with other infections are of note needed, because these people have high-risk to severe cases.
Inactivated influenza vaccination should be taken into account in immunosuppressed patients.

Background
Influenza infection gathered continuous attentions because of its significant morbidity and mortality worldwide. Recently a modelling study estimated about 290 000-650 000 seasonal influenza-associated respiratory deaths each year globally [1]. While a study from China estimated that 88 100 influenza-associated excess respiratory deaths occurred in China annually [2]. Immunosuppressed patients such as patients with malignancies and chemotherapy, haemopoietic stem cell transplants (HSCT), solid-organ transplants (SOT), patients on chronic hemodialysis, and patients receiving systemic corticosteroids increased year by year due to medical advances [3,4]. Compared with the immunocompetent patients, influenza infection in immunosuppressed patients had more morbidity and mortality, longer viral shedding, more frequent complications, and more antiviral resistance, increasing the potential for nosocomial transmission [3,[5][6][7][8][9].
Previous studies abroad most focused on influenza infection in patients with HSCT and SOT, paying less attentions to other types of immunosuppression [10][11][12], while investigations of outcome from influenza virus infection in immunosuppressed patients in China are still lacking. Neuraminidase inhibitors such as oseltamivir and peramivir were mainstay of antiviral therapy, previous study suggested high doses and long duration of antiviral treatment for patients who were immunocompromised [13]; however, specific data on effect of antiviral therapy in immunosuppressed patients in China are scarce.
Furthermore, influenza vaccine responses in immunosuppressed populations were uncertain [3], in this study we tried to obtain the vaccination data and review the literatures on influenza vaccination strategies in immunosuppressed patients.
We did a study of influenza infection in immunosuppressed inpatients, analyzed the clinical characteristics, established risk factors for admission to intensive care unit (ICU) and death, estimated the efficacy of available antiviral drugs, and tried to know the influenza vaccination situation in these populations.

Patients and study design
During the 2018-2019 influenza season (November to the following April), a total of 23343 nasal swab specimens were obtained from influenza-like illness patients of outpatient and inpatient in Peking University People's Hospital (PKUPH), a national influenza surveillance sentinel unit, at which at least 100,000 inpatients from all over China are seen annually.
Influenza A or/and B viruses were screened positive by using colloidal gold method in 4783 samples, including 140 inpatients. There were 73 immunosuppressed populations in these 140 inpatients, who were enrolled in our study, and these 73 inpatients' samples were immediately placed in virus transport media tubes and were stored at -80 °C within 24 hours until analyzed. Immunocompromised populations were defined as patients with HIV infection, recipients of solid-organ transplants, recipients of haemopoietic stem cell transplants, patients with malignancies and chemotherapy, patients on chronic hemodialysis, and patients receiving systemic corticosteroids [3]. Influenza infections were defined as samples detected positive by using reverse-transcription polymerase chain reaction (RT-PCR) method [14].

Data Collections
Data on demographic factors (age, sex, type of immunosuppression, comorbidity, smoking, rejection in preceding 3 months, interleukin-2 receptor antagonists in preceding 6 months, monoclonal antibody in preceding 6 months, application of corticosteroids in preceding 3 months, seasonal influenza vaccination in preceding 12 months, application of neuraminidase inhibitors before admission), clinical presentation and complications (

Statistical analysis
All statistical analyses were performed using SPSS statistical software version 22.0 (SPSS Inc., Chicago, IL, USA). Categorical variables were expressed as counts (percentage) and compared using the χ 2 test, while continuous variables were as means ± SD or median (interquartile range) and compared using the independent samples t-test. Univariate analysis and multivariate logistics analysis were used to identify risk factors for admission to the intensive care unit and death. The removal probability for multivariate stepwise logistic regression analysis was 0.10. P < 0.05 was considered to be statistically significant.

Epidemics characteristics of study patients
A total of 73 immunosuppressed inpatients tested positive for influenza virus using colloidal gold method were enrolled in our study. These samples were verified and classified using RT-PCR (50 influenza A viruses, and 23 influenza B viruses). Weekly distribution of immunosuppressed and immunocompetent inpatients with influenza infections in PKUPH during November 2018 to April 2019 was shown in Fig. 1. Compared with positive rate of influenza from Chinese Center for Disease Control and Prevention in northern China, which data mainly from the outpatients, we could see that influenza symptom onset of outpatients was earlier than inpatients; however, inpatients peaked earlier, especially in immunosuppressed patients.

Clinical Characteristics Of Study Patients
From the Table 1, we could see that the most common immunosuppression type was malignancies with chemotherapy 73.9% (54/73), then hematopoietic stem cell    Table 3, blood routine was characterized by lymphopenia in immunosuppressed populations with influenza infections. Erythrocyte sedimentation rate and C-reactive protein could be a little higher than normal. Albumin was usually lower than normal.   The median day of hospital stay was 15 days (range 1-90 days), and it significantly associated with organ transplantation, application of corticosteroids in preceding 3 months, complications, and co-infection with other infections (Fig. 2).
95.6% (65/68) patients received standard dosing antiviral treatment that adult patients received oseltamivir at the equivalent of 75 mg twice or peramivir 300 mg once per day, and children received standard dosing according to weight, while 4.5% (3/68) patients received double dose. Double dose of neuraminidase inhibitors was also not significantly associated with admission to the ICU and death through the χ 2 test (P 0.05).
Median duration of antiviral treatment was 6 days (range 1-32). Antiviral treatment began within 48 hours of symptom onset in 83.8% (57/68), and greater than 48 hours in 16.2% (11/68). Antiviral treatment after 48 hours was significantly associated with admission to the ICU, mechanical ventilation and death through the χ 2 test (P < 0.05). Median day of virus detection turned negative was 3 days (range 1-11); however, our data was available only for 34 patients, because some mild illness did not recheck and some severe illness lasted positive persistently even to death. Analyzed these 34 patients' data, days of virus detection turned negative were associated with complications and antiviral treatment after 48 hours (Fig. 3).

Risk Factors For Admission To Icu
In our study 15.1% (11/73) patients were admitted to the ICU, and we performed univariate analysis of all variables in Table 1

Discussion
In this study, half of inpatients with influenza infection in PKUPH were immunosuppressed cases. Compared with the outpatients and immunocompetent inpatients, immunosuppressed patients peaked earliest, which indicated their susceptibility to influenza infection (Fig. 1). Previous studies most focused on influenza infection in patients with HSCT and SOT; however, more attentions have been paid to other types of immunosuppression in recent years [4,15,16], the most common immunosuppression type in this study was malignancies with chemotherapy accounted for 73.9%, even with two different types of tumors. HSCT was the most common transplant type, the median time of symptom onset after transplant was 11 months, nearly half of whom had rejection in preceding 3 months.

The common presenting symptoms of influenza infection in immunosuppressed patients
were variable in different studies; however, most patients exhibited fever and cough [5,17]. It was worth noting that in immunosuppressed recipients, fever and other classical symptoms such as cough, muscular soreness, sore throat, headache, rhinorrhea may be absent, instead of gastrointestinal symptoms, dyspnea and altered mental status [5,18].
As to complications, pneumonia was the most common (40.3%), consistent with previous literature from 32-56% [11,15,17]; however, other relatively rare complications such as acute respiratory distress syndrome, shock, acute kidney injury, and viral encephalitis were also should be taken seriously. Co-infections with bacterial, fungal, and other viral pathogens were found in 17.8% patients, which have been reported in previous studies [3,15,17]. Co-infection with other infections significantly prolonged the hospital stay and increased risks of admission to ICU, especially with Aspergillus spp. infection. The hospital stays of two patients with Aspergillus spp. were 90 and 51 days respectively (the latter died). Dunbar et al. discussed influenza and invasive aspergillosis in immunocompromised patients, and summarized probable mechanisms for influenza and aspergillus mixed infections were disruption of bronchial mucosal barrier and phagocytosis, T-cell dysfunction and apoptosis, and neuraminidase inhibitors use and so on [19]. Laboratory test results of influenza infection in immunosuppressed patients were characterized as lymphopenia, a little higher erythrocyte sedimentation rate and C-reactive protein, and lower albumin than normal, and partial results were consistent with previous studies, furthermore, several studies point that lymphopenia was a risk factor for severe cases [20][21][22]. Based on the clinical characteristics above, greater consideration should be given to early identification of influenza infection in immunosuppressed patients, conducting influenza virus detection at the first time.
We described the use and effect of neuraminidase inhibitors in these people, and found antiviral treatment after 48 hours was significantly associated with admission to the ICU, mechanical ventilation and death, according with guideline suggestion clinicians should start antiviral treatment as soon as possible for immunocompromised patients [16,23,24]. Immunocompromised patients were susceptible to suffering from antiviral drug resistance due to prolonged viral shedding [17,25,26], which also contributed to more influenza-related death and nosocomial transmission [12,27,28]. Previous studies recommended higher doses and/or prolonged courses of antiviral therapy [7]. In our study combination and double dose of neuraminidase inhibitors did not significantly reduce the risk for admission to the ICU and death. Median duration of treatment was 6 days (range 1-32) in our study, and longer duration of antiviral treatment maybe more benefit for immunocompromised patients, as influenza viral replication was often protracted [14].
Data on virus detection turned negative was not available for more than half, and some severe illness lasted positive persistently even to death. We should pay attention to complications and early antiviral treatment by analyzing these 34 patients' data.
In our study 15.1% patients were admitted to the ICU and mortality rate was 8.2%, similar or even higher than some other large sample studies [11,15,16]. Presence of organ transplantation, dyspnea, altered mental status, application of corticosteroids in preceding 3 months, co-infection with other infections, and antiviral treatment after 48 hours were both associated with admission to the ICU and death in univariate analysis.
Furthermore diabetes, long symptom onset, and complications were risk factors for admission to the ICU. In multivariate logistics analysis risk factors for admission to ICU were long symptom onset and co-infection with other infections, and presence of dyspnea was independently associated with increased risks of death. Part of above risk factors have been reported in previous literatures [11,15,[29][30][31], and some were the first reported; however, all analyses reminded us of identification of patients at risk of a severe course. Corticosteroids therapy may decrease inflammation, but at the cost of longer duration of viral shedding [30,[32][33][34].
Recently Chinese scholars pointed out that absence of an active immunization program could explain influenza-associated mortality was higher than other countries with higher vaccination coverage, in our study seasonal influenza vaccination rate was lower than 3%.
The suggestions of influenza vaccines depended on types of immunosuppression [3].
There were evidences that HSCT population could benefit from vaccination [35][36][37][38], and the latest recommendation was to start inactivated vaccines from 3 months after transplant [39]. However, a systematic review showed that seasonal inactivated influenza vaccine remained suboptimal in SOT recipients [40]. No consensus guidelines on influenza vaccination for patients with malignancies, but timing of vaccination was recommended more than two weeks before receiving chemotherapy or between chemotherapy cycles in order to enhance humoral responses [3].
The data obtained from our study was essential to understand outcomes of influenza infection in immunosuppressed patients, but this study had limitation that data collection was limited to one influenza season of one hospital. Therefore, further studies with a larger sample size will be needed to confirm and extend our findings.

Availability of data and materials
The datasets analyzed during the current study are available from the corresponding author on reasonable request.

Ethics approval and consent to participate
This study has been carried out in accordance with the Declaration of Helsinki and was approved by the ethics committees of the Peking University People's Hospital (PKUPH, IRB No. 2016PHB100-01). Written informed consents were obtained prior to their inclusion in this study. Nasal samples and medical data were collected and analyzed anonymously.

Consent for publication
Not applicable.