Routine blood parameters can detect early inuenza virus infection in children

Purpose We aimed to explore the value of Routine blood parameters, such as the lymphocyte (LYM) count, platelet (PLT) count, lymphocyte-to-monocyte ratio (LMR), neutrophil-to-lymphocyte ratio (NLR), LYM*PLT and mean platelet volume-to-platelet ratio (MPV/PLT), are widely used to predict the prognosis of infectious diseases, for predicting inuenza virus infection in children. We conducted a single-center, retrospective, observational study on fever with inuenza-like symptom in pediatric outpatients in different age groups and evaluated the predictive value of various routine blood parameters within 48 hours of the onset of fever after inuenza virus infection.


Introduction
In uenza is an acute respiratory infectious disease caused by in uenza viruses. There are 1 billion patients with seasonal in uenza each year worldwide, among whom 3 to 5 million have severe cases and as many as 500,000 die [1]. Although most children recover spontaneously from infection, morbidity and mortality are higher in children with underlying diseases who are younger than 5 years, especially those younger than 2 years [2].
However, previously healthy children are also at risk. In the USA, the admission rate of non-high-risk children due to in uenza was estimated to be 9 per 10000 children younger than 5 years old [3]. According to the WHO, in the past 11 in uenza epidemic seasons, the annual infection rate of children was as high as approximately 50% [4].
Complications of in uenza, including pneumonia, myocarditis, septic shock and multiple organ dysfunction, are the main causes of death in children [5]. Early (within 48 hours after infection) use of anti-in uenza drugs can signi cantly relieve symptoms, shorten the course of disease, and reduce complications. Therefore, the early and rapid diagnosis of in uenza and the early use of anti-in uenza drugs are essential to improve the prognosis of in uenza in children.
The diagnosis of in uenza depends on the detection of in uenza virus nucleic acid in the respiratory tract, the isolation of in uenza virus or the detection of a level of serum-speci c antibodies that is at least 4 times the normal level. Common detection methods include virus isolation and culture, RT-PCR and serological detection, all of which have advantages and disadvantages. Virus isolation and culture, RT-PCR and serological detection are time-consuming and di cult and are not suitable for outpatient screening. Viral antigen detection, such as rapid in uenza detection, is rapid and simple, with good speci city; however, the sensitivity is low, and it is prone to false negative results [6].
Routine blood tests are the rst choice in pediatric fever clinics. They are easy to perform and inexpensive. In recent years, studies have found that the lymphocyte-to-monocyte ratio (LMR), neutrophil-to-lymphocyte ratio (NLR), mean platelet volume-to-platelet ratio (MPV/PLT) and lymphocytes multiplied by platelets (LYM*PLT) can be used as new in ammatory markers to predict the prognosis of infectious diseases [7][8][9], tumors [10,11] and cardiovascular diseases [12,13]; these markers have been widely studied in clinical practice.
There have been some studies on routine blood index values and in uenza A infection in adults, but there have been few studies in children. In this study, we collected early routine blood test results from children suspected of having in uenza and further explored the value of the LMR, NLR, LYM*PLT, MPV/PLT, lymphocyte (LYM) count and platelet (PLT) count for the prediction of in uenza in children.

Patients
Children with fever and in uenza-like symptoms who were 0.2 to 14 years old and presented at the First A liated Hospital of Wenzhou Medical University from January 2018 to February 2020 were included in this study.
In uenza-like symptoms were de ned as follows: fever (temperature ≥ 38℃), cough or sore throat [14]. The exclusion criteria were as follows: (1) systemic chronic diseases, such as diseases of the blood, heart, lung, liver and kidney; (2) immunode ciency, due to a tumor, HIV infection, or the long-term use of oral hormones or immunosuppressive agents; (3) severe or critical illness; (4) bacterial infections, such as sepsis and suppurative tonsillitis; and (5)

Detection of Routine Blood Parameters
Finger prick blood samples taken from children with suspected in uenza were subjected to routine blood tests within 48 h of the onset of fever. A routine analyzer (XN-350, SYSMEX, Japan) was used for detection. The LYM count, monocyte (MON) count, PLT count and mean platelet volume (MPV) were recorded. Additionally, other hematological parameters were calculated: the LMR is the ratio of lymphocytes to monocytes, the NLR is the ratio of neutrophils to lymphocytes, the MPV/PLT is the MPV divided by PLT count, and the LYM*PLT is the lymphocytes multiplied by the platelets.

Detection of In uenza Virus Nucleic Acid by RT-PCR
A fully automatic nucleic acid extractor and the associated reagents (Shanghai ZJ Bio-Tech Co., Ltd) were used to extract all nucleic acid from pharyngeal swabs. Throat swab specimens obtained for the purpose of in uenza virus nucleic acid determination were subjected to RT-PCR, and in uenza A and B virus nucleic acid detection kits were used (Z-RR-0097-02, Shanghai ZJ Bio-Tech Co., Ltd). The ampli cation system used a nal volume of 25 µl, consisting of 19 µl of a mixture of in uenza A and B virus nucleic acid uorescent probes, 1 µl of enzymes, and 5 µl of the sample. The ampli cation conditions were as follows: 45℃ reverse transcription for 10 min, predenaturation at 95℃ for 15 min, denaturation at 95℃ for 15 s, annealing elongation and uorescence detection at 60℃ for 60 s for 45 cycles. All ampli cation reactions were performed with an ABI7500 quantitative PCR instrument (Applied Biosystems, Inc. USA).
Statistical Analysis SPSS 23.0 (SPSS Inc., Chicago, IL, USA) was used for data analyses. The measurement data are expressed as the means ± standard deviations (x±s). The measurement data were tested by one-way variance tests. The pairwise comparisons of the mean were performed by the LSD method. The data were tested with the chi-square test.
Receiver operator characteristic curve (ROC) analysis was used to evaluate the diagnostic value of the LYM count, PLT count, LMR, NLR, LYM*PLT and MPV/PLT for in uenza A and B virus infection. Statistical signi cance was indicated by P < 0.05.

Patient Characteristics
A diagnosis of in uenza infection was made based on the occurrence of in uenza-like symptoms with a positive RT-PCR for in uenza A or B [15]. In this study, a total of 388 children with in uenza A virus infection (A+ group), There were no statistically signi cant differences in the age and sex distributions between groups within the three age groups (P>0.05) ( Table 1).

Differences in Routine Blood Parameters in the Three Age Groups
The red blood cell count (RBC) and hemoglobin (Hb) level in the A+ group, B+ group, A-B-group and H group were not signi cantly different in the groups of patients <1 year old, 1-6 years old and >6 years old. In all three age groups, compared with the H group, the A+ group, B+ group, and A-B-group had lower LYM counts, PLT counts, LMRs and LYM*PLT values, and higher NLRs and MPV/PLT values ( Figure 1). In the <1-year-old group, there were no signi cant differences in the LYM count, PLT count, LMR, NLR, LYM*PLT and MPV/PLT between the A+ group and A-B-group. There were only 3 patients in the B+ group, so no statistical analysis was performed. In the 1-to 6-year-old group, the LYM count, PLT count, LMR, LYM*PLT and MPV/PLT were signi cantly different between the A+ group and the A-B-group and the LYM count, LMR and LYM*PLT were signi cantly different between the B+ group and the A-B-group ( Figure 1). In the >6-year-old group, the LYM count, LMR, NLR and LYM*PLT were signi cantly different between the A+ group and the A-B-group; the PLT count, LMR and LYM*PLT were signi cantly different between the B+ group and the A-B-group ( Figure 1 The variable that best predicted positivity for in uenza virus A based on the area under the curve (AUC) was the LMR. When the A-B-group was taken as a reference, the cutoff value was 3.75, the AUC was 0.886, and the sensitivity and speci city were 81.87% and 84.31%, respectively. When the H group was used as the reference, the maximum AUC was obtained with the LYM*PLT (followed by the LMR); the cutoff value of the LYM*PLT was 680.48, the AUC was 0.958, and the sensitivity and speci city were 90.67% and 89.66%, respectively ( Figure 2).

1-to 6-year-old B+ group
The variable that best predicted positivity for in uenza virus B based on the AUC was the LMR. When the A-Bgroup was taken as a reference, the cutoff value was 3.71, the AUC was 0.843, and the sensitivity and speci city were 73.58% and 84.31%, respectively. When the H group was used as the reference, the cutoff value for the LMR was 4.47, the AUC was 0.918, and the sensitivity and speci city were 86.79% and 89.66%, respectively ( Figure 3).

>6-year-old A+ group
The maximum AUC was obtained with the LMR. When the A-B-group was taken as a reference, the cutoff value was 3.05, the AUC was 0.949, and the sensitivity and speci city were 89.27% and 89.61%, respectively. When the H group was a reference, the cutoff value was 3.09, the AUC was 0.975, and the sensitivity and speci city were 90.40% and 95.24%, respectively ( Figure 4).

>6-year-old B+ group
The maximum AUC was obtained with the LMR. When the A-B-group was taken as a reference, the cutoff value was 2.88, the AUC was 0.924, and the sensitivity and speci city were 83.19% and 92.21%, respectively. When the H group was a reference, the cutoff value of the LMR was 3.48, the AUC was 0.954, and the sensitivity and speci city were 90.27% and 92.38%, respectively ( Figure 5).

Discussion
According to the nucleocapsid protein and the matrix protein antigen, in uenza viruses can be divided into three types: A, B and C. There is no cross-immunity among these types. The type with the greatest antigen variability is in uenza A, which often causes regional outbreaks and epidemics and could even cause a worldwide pandemic.
Over the past 100 years, in uenza A outbreaks occur seasonally, and there have been several global pandemics. The most serious was the Spanish H1N1 in uenza pandemic in 1918, which killed 50 million people [16,17].
In uenza B has weak antigen variability, and it often causes moderate epidemics or local outbreaks. It is rare for in uenza C to infect humans [18]. The clinical symptoms of in uenza virus infections in children are similar to those of infections with other respiratory pathogens and are nonspeci c; these include a high fever, chills, muscle aches, sore throat, cough, and runny nose. Gastrointestinal symptoms such as vomiting, abdominal pain, and diarrhea are relatively common in children infected with in uenza B [15]. Patients with mild symptoms can recover within a short time, and patients with severe symptoms rapidly develop dyspnea accompanied by refractory hypoxemia and can eventually develop acute respiratory distress syndrome, septic shock, heart failure, acute necrotizing encephalopathy, and multiple organ dysfunction, which are life-threatening and even fatal conditions [19]. Therefore, it is important to seek a rapid and simple index for the early diagnosis of in uenza infection in children.
The diagnosis of in uenza mainly depends on the detection of viral nucleic acids and antibodies. Virus isolation and culture used to be the " gold standard " for the diagnosis of in uenza, but it is time-consuming and expensive, has high technical requirements and hard to perform. RT-PCR is the most effective nucleic acid detection technology for in uenza, with a sensitivity and speci city as high as 98.5% and 100% [20], respectively. RT-PCR has now replaced as the " gold standard " for the diagnosis of in uenza. Although RT-PCR takes less time than virus isolation and culture, it is still expensive and takes several hours, making it di cult to use as a routine means of screening for in uenza in pediatric fever clinics. Serological testing requires two serum samples from both the acute phase and the convalescent phase. Convalescent blood samples should be collected 2-4 weeks after the onset of the disease. If the antibody level is more than 4 times higher in the convalescent phase than in the acute phase, the patient can be diagnosed with in uenza. Obviously, this is not suitable for in uenza screening in outpatients. In uenza virus antigen detection, such as rapid in uenza diagnostic tests, can be performed within 30 minutes, but the sensitivity is only 40-70% [6]. Routine blood tests are the most common tests in pediatric fever clinics and can be used as the primary means of identifying bacterial and viral infections. In recent years, researchers have performed an in-depth analysis of various routine blood parameters, these parameters have been found to be useful for the early diagnosis and prognostic assessment of other diseases [7][8][9][10][11][12][13].
In uenza strains infect respiratory epithelial cells and attachment of the virus to cells via sialic acid receptors enables uptake of the virus into the cells, followed by recognition of the virus via pattern recognition receptors (PRRs). PRRs trigger cytokine responses and the induction of protective immunity, but they might also contribute to immune pathology [21]. Lymphocytes are the main immune cells involved in the elimination of viruses. In conventional viral infections, the proportion of lymphocytes in the circulation is usually increased. Previous studies have suggested that there is a signi cant decrease in LYM counts in patients infected with in uenza A [9,22,23], but there have been few studies on LYM counts in patients with in uenza B. Nichols et al. [24] found that LYM can induce self-apoptosis by regulating the expression of FasL on the cell surface and the release of soluble FasL after in uenza infection, leading to a decrease in the LYM count. In this study, it was found that the LYM count in children with in uenza A or B infection was signi cantly lower than that in children from 1-6 years old with in uenza-like symptoms who tested negative for in uenza A and B viruses, and there was no signi cant difference in the LYM count between children with in uenza A and in uenza B infections. In children >6 years old, the LYM count in those infected with in uenza A was signi cantly lower than that in those not infected with in uenza viruses. The LYM count was not signi cantly different between children infected with in uenza virus B and those who were not infected with either in uenza A or B viruses. Lewis et al [25] suggested that lymphopenia is mainly due to a reduction in T cells and, to a lesser extent, B cells and is of short duration.
Leukocytes such as neutrophils and monocytes provide anti-in uenza host protection by releasing preformed cytokines, and the granule contents help hosts eliminate the threat posed by replicating viruses. Coskun O et al. [26] suggest that monocytosis may be considered a surrogate marker for infection with in uenza A virus. In this study, there was no statistically signi cant difference in MON count in children older than 1 year who were infected with either in uenza A or B viruses and those who were not infected with in uenza viruses. The LMR is the ratio of the LYM count to the MON count, and a LMR<2 has been used as a surrogate marker for in uenza A infection [27]. This study suggested that the LMR is the best index for the prediction of in uenza virus infection.
At the same time, the study showed that the AUCs for the prediction of in uenza A and in uenza B infections were higher in children who were >6 years old than in children who were 1-6 years old, suggesting that the diagnostic value of the LMR for in uenza is greater in children over 6 years old than in children under 6 years old. In addition, the results showed that the AUC of the LMR for the prediction of in uenza A infection was higher than that for in uenza B in the >1-year-old group, indicating that the predictive value of the LMR was greater for in uenza A than for in uenza B.
Prior studies demonstrated that PLT can regulate host immunities and complement responses in the initial intrinsic defense against in uenza virus infection [28] and explored the different mechanisms by which virus infection can interfere with PLT production and might trigger PLT destruction [29], thus decreasing the PLT count.
In this study, the PLT count in the in uenza A group was signi cantly lower than that in the group of children who were negative for both in uenza A and in uenza B among those 1-to 6 years old, which was consistent with the results in the study by Fei et al. [9]. However, there was no signi cant difference between the group infected with in uenza B and the group of children who were negative for both in uenza A and in uenza B. Among the children older than 6 years, the PLT count was not statistically signi cant between the group infected with in uenza A, the group infected with in uenza B and the group of children who were negative for both in uenza A and in uenza B.
Our study suggests that the PLT count has predictive value for only children who are 1-to 6 years old and are infected with in uenza A; furthermore, even in that group, its predictive value is low, with an AUC of 0.615 and sensitivity and speci city of 58.03% and 63.73%, respectively. In recent years, researchers have attempted to combine the PLT count with other indicators, such as LYM*PLT and MPV/PLT, for disease prediction [30]. Fei et al.
reported [9] that the LYM*PLT and MPV/PLT had better predictive value for children under 6 years old who were infected with in uenza A, and the predictive value was higher than that of the LMR. However, our study showed that the LMR had the highest predictive value for in uenza infection, followed by the LYM*PLT, PLT count, and MPV/PLT among children over the age of 1 year.
Recently, studies have shown that the NLR is positively associated with systemic in ammation [7], acute pancreatitis [31], liver disease [32] and rheumatic diseases [33]. The NLR was found to have a high sensitivity for the detection of in uenza virus infection [34]. In this study, there was no signi cant difference in the NLR among the group infected with in uenza A, the group infected with in uenza B and group of children who were negative for both in uenza A and in uenza B among those who were 1-6 years old. In children who were >6 years old, the NLR was signi cantly higher in the group infected with in uenza A than those who were negative for both in uenza A and in uenza B. However, there was no signi cant difference between the group infected with in uenza B and the group of children who were negative for both in uenza A and in uenza B among those >6 years old. Our study suggested that the NLR has predictive value only for children over 6 years old who are infected with in uenza A, with an AUC of 0.657 and a sensitivity and speci city of 58.19% and 70.13%, respectively. In addition, it was found in this study that there was no signi cant difference in the LYM count, PLT count, LMR, NLR, LYM*PLT and MPV/PLT between the group infected with in uenza A and the group of children who were negative for in uenza A and in uenza B among those <1 year old. Due to the small sample size, no statistical analysis was performed to compare the in uenza B group with the other groups among children <1 year old. This suggested that the routine blood parameters had scant predictive value for in uenza in the <1-year-old age group. We believe that this may be related to immune function and the development of blood cells in children, and this nding needs further investigation.

Conclusion
This study showed that the LMR was signi cantly lower in children older than 1 year who had in uenza, especially children older than 6 years infected with in uenza A, compared to children without in uenza. The LMR can be used as an early predictor of in uenza A infection in children older than 6 years, with an AUC of 0.949, a sensitivity of 89.27% and a speci city of 89.61%.

Declarations
Ethics approval and consent to participate

Authors' contributions
Ronghe Zhu: study design, managed the experiments, analyzed the results, was involved in manuscript preparation.
Qiu Wang: data collection, analyzed the results.
Cuie Chen: analyzed the results.
Xixi Zhang: analyzed the results, and was involved in manuscript preparation.
Chaosheng Lu: was involved in data collection and analysis.