Diagnostic Value of Different Biomarkers to Identify Bacterial Coinfection in Vietnamese Children with Severe Rhinovirus Pneumonia

Abstract This study aims to assess the diagnostic value of high-sensitivity C-reactive protein (hs-CRP), procalcitonin (PCT), and interleukin (IL)-6 in the diagnosis of pneumonia caused by Rhinovirus alone or with bacterial coinfection in Vietnamese children under 5 years of age. A cross-sectional study was conducted on 26 children under 5 years of age with severe pneumonia due to Rhinovirus at the National Pediatric Hospital. IL-6, hs-CRP, and PCT tests were performed. The diagnostic values of PCT, IL-6, and hs-CRP in classifying those with viral alone and those with bacterial coinfection were determined. Of 26 children, 10 children were diagnosed to have bacterial coinfections (38.5%). The optimal cutoff point for PCT was > 2.30 ng/mL (sensitivity 50%, specificity 94%, positive predictive value 83%, and negative predictive value 75%). The optimal cutoff point for hs-CRP was > 1.53 mg/dl (sensitivity 90%, specificity 56%, positive predictive value 56%, and negative predictive value 90%). Finally, the optimal cut-off point for IL-6 was > 441.5 pg/mL (sensitivity 20%, specificity 100%, positive predictive value 100%, and negative predictive value 60%). The accuracy rate of PCT was the highest with 69.2%, followed by hs-CRP with 65.4%. Inflammatory biomarkers such as PCT and hs-CRP were able to distinguish children with severe pneumonia caused by Rhinovirus alone and those with bacterial coinfection.


Introduction
Rhinovirus is a leading cause of respiratory infection in children, accounting for approximately 20 to 50%. [1][2][3][4] Rhinovirusrelated pneumonia may be asymptomatic, but this virus may worsen respiratory infection or cause outbreaks. [5][6][7][8] Rhinovirus infection causes severe consequences for hospitalized children, similar to infection caused by respiratory syncytial virus or influenza viruses. 9 During a severe outbreak of acute respiratory infections in Vietnam caused by Rhinovirus, the mortality rate of hospitalized children was 58.3%. 8 Rhinovirus also causes a significant burden on resources, costs, and health in children with acute respiratory failure. [10][11][12] The rate of bacterial coinfection in pediatric pneumonia caused by Rhinovirus is over 40%. 13 Compared with patients with viral pneumonia alone, coinfected pneumonia patients had a higher rate of mechanical ventilation and death. 13,14 Therefore, determining whether infants have coinfection is very important. 15 Difficulties in confirming the diagnosis of viral pneumonia with or without bacterial coinfection also contribute to the increase in antibiotic resistance. 16 Unfortunately, clinical features, routine tests, and X-ray results are not specific to distinguishing between viral or bacterial pneumonia. [17][18][19] Isolating the pathogen directly for a comprehensive diagnosis in a patient is not feasible. 16 However, recent studies have shown the advantages of using inflammatory biomarkers such as IL-6, hs-CRP, and PCT, in the diagnosis of concurrent bacterial pneumonia. 18,[20][21][22][23] Although the figures show high rates of pneumonia in children and its burden 24 in Vietnam, there is a lack of research related to identifying the methods to diagnose pneumonia with and without bacterial coinfection. This study aimed to assess the diagnostic value of hs-CRP, PCT, and IL-6 in the diagnosis of pneumonia caused by Rhinovirus alone or bacterial coinfection in Vietnamese children under five years of age.

Study Designs and Setting
A cross-sectional study was conducted on children at the National Hospital of Pediatrics from January 2015 to March 2017. Twenty-six children under five years of age diagnosed with Rhinovirus severe pneumonia were included in this study. A child was determined to have severe pneumonia according to WHO 2013 standards when he or she had a cough or trouble breathing and had, at least, one of the following symptoms: 1) cyanosis or SpO2 <9 0%; 2) severe respiratory distress; and 3) could not drink or vomit everything, coma or not awake, or convulsions. 25 PCR tests were used to confirm Rhinovirus infection. We excluded children with congenital chronic diseases and those whose parents or guardians who did not agree to participate in the study. The study was approved by the institutional review board of Vietnam Military Medical University (Code 92/QD-HVQY).

Variables and Measurement
Physicians clinically examined all patients admitted to the hospital and asked their parents or guardians to report demographic, medical, and clinical characteristics. We collected 2 mL of venous blood from each child for biochemical tests. Quantification of biomarkers, including Hs-CRP, IL-6, and PCT, were also performed. Olympus AU 2700 was used to quantify Hs-CRP by turbidity measurement method. PCT concentration was determined by the luminescent immunization method, using Siemens ADVIA Centaur. Samples were placed in tubes with Li-Heparin and K3-EDTA or without anticoagulants before centrifugation. Biprad's Bio-Plex Protein Array System was used to quantify IL-6. The Vitek 2 machine was used to determine bacteria by identifying the chemical and biological properties of bacteria through the changing color of environmental wells. The minimum inhibitory concentration was measured to monitor the growth of bacteria. The system used visible light at the wavelengths of 660 nm, 568 nm, 428 nm showed off the environment wells, and then measured the blocked light intensity to identify the decrease in light intensity.

Statistical Analysis
We used STATA 15.0 software to analyze the data. Clinical and subclinical characteristics were compared between children with severe pneumonia caused by Rhinovirus with and without bacterial coinfection by the Mann-Whitney test and Chi-squared test. The receiver operating characteristic (ROC) was calculated to determine the diagnostic value of the Hs-CRP, IL-6, and PCT. Then, sensitivity, specificity, positive predictive value, negative predictive value, positive likelihood ratio (LR þ ), negative likelihood ratio (LR À ), and accuracy rate of each diagnostic value were calculated. We used the Youden index to determine the optimal cutoff point. Statistical significance was determined if the p value was less than 0.05.

Results
►Fig. 1 shows that 10/26 children with severe pneumonia due to Rhinovirus have coinfection with bacteria, accounting for 38.5%. The most common organisms for coinfection bacteria were Pseudomonas aeruginosa (four cases), Hemophilus influenza (three cases), Streptococcus pneumoniae (two cases), and Klebsiella pneumoniae (one case).

Discussion
Our study demonstrates that bacterial coinfections often present in children with severe pneumonia caused by Rhinovirus. This study also demonstrates the potential role of different inflammatory biomarkers, especially PCT and hs-CRP, in the detection and classification of severe Rhinovirus pneumonia with and without concurrent bacterial infection.
Children with viral pneumonia may have either bacterial coinfection or secondary bacterial infection. The incidence of bacterial coinfection with Rhinovirus in the study was 38.5%, which was similar to the rate in an earlier report by Hwa Sik Jung et al. (42.5%). 13 The bacteria such as Pseudomonas aeruginosa, Hemophilus influenza, Streptococcus pneumoniae, and Klebsiella pneumoniae were found in severe pneumonia patients in this study. Other studies reported other types of bacteria, such as Staphylococcus aureus, Chlamydia pneumoniae, Mycoplasma, etc., which might also be present in pediatric patients with severe Rhinovirus pneumonia. 13,26 This difference might be likely due to the limited sample size in this study. The determination of bacterial coinfection in patients with severe Rhinovirus pneumonia is important in predicting the outcome and assisting in the development of the appropriate treatment plan. 15,16 Fever, rapid pulse rate, runny nose, wheezing, and diarrhea are common clinical symptoms in both Rhinovirus patients with severe pneumonia with and without bacterial coinfection. These symptoms are also common in people with pneumonia due to other causes. 27 Our study showed that the rapid pulse  rate in children with bacterial coinfection was significantly higher than that in patients with Rhinovirus alone. Previous studies have also shown that tachycardia has a high-diagnostic value for identifying serious infections in children, 28,29 while coinfection with bacteria increases the severity of the viral disease pneumonia. 13,14 However, there has been no evidence that a rapid pulse rate can be used as a criterion to classify the cause of pneumonia. There was no clinical difference between children with Rhinovirus alone and bacterial coinfection. This result is consistent with previous literature that the clinical features are nonspecific to distinguish pneumonia agents. [17][18][19] Recent evidence has shown the diagnostic effectiveness of biomarkers in distinguishing bacterial pneumonia from bacterial superinfection. 18,[20][21][22][23] Although prior study indicated a significant increase of IL-6 in experimental mice with cobacterial pneumonia, 30 our research results show that IL-6 has poor diagnostic values with AUC of 0.400 (95% CI ¼ À 0.01-0.807, p ¼ 0.63). According to a study by Jaana Karhu et al, IL-6 is not different in groups of patients with bacterial, viral, or both pneumoniae. 31 In this study, the CRP values were significantly higher among the coinfection group than others, which was similar to the results of previous reports. 20,21,[32][33][34] The optimal cutoff point for hs-CRP was 1.53 mg/dl (sensitivity 90%, specificity 56%), and the AUC for hs-CRP was 0.772, indicating this diagnostic value was valid. Although some authors were arguing that CRP is not useful in distinguishing coin-fected bacterial infections in patients with Rhinovirus community pneumonia, 26,35 other studies were confirming the ability to distinguish between viral and bacterial agents that cause pneumonia with CRP limit point > 0.8mg/dl. [36][37][38] Although controversies exist surrounding the use of the hs-CRP index to identify severe pneumonia caused by coinfection with the virus, this biomarker can still be used to support decisions about antibiotic use. The diagnostic value may be increased if hs-CRP is combined with other criteria. 21,33 Recent studies suggested that increased PCTwas the result of bacterial infection, since PCT is a calcitonin precursor produced by transcription of the gene CALC-1, which was activated to increase PCT production in the endocrine gland, and also epithelial when being bacterially infected. 39 On the other hand, in a patient with viral infection alone, PCT only increases to a certain limit because the viral infection can cause a decrease in the induction of PCT. 20 The results also suggest that PCT has the potential to be a valuable diagnostic indicator to detect children with severe pneumonia caused by coinfection with the virus. The optimal limit is defined as PCT > 2.30 ng/mL for 50% sensitivity and 94% specificity. According to Hoshina et al, the limit was defined as PCT > 0.2 ng/mL with 86% sensitivity and 80% specificity when diagnosing pathogens with bacterial agents. 40 Thomas reported about children with community-acquired pneumonia who also identified PCT > 0.07 points with 48.7% sensitivity and 81.1% specificity. 18 Although there are differences in cutoff points, it is undeniable that the diagnostic value of PCT presented in previous studies is relatively high. The heterogeneity among these studies can explain differences in critical values. Children with pneumonia caused by different viruses and bacteria may need to have diagnostic values of different biomarkers. Moreover, the severity of illness (severe or mild) can also affect the PCT index. Our research results also show that PCT has the highest accuracy rate of among biological markers, followed by hs-CRP. This finding is consistent with previously published data. 18,20,[41][42][43][44]   Our research has some limitations. First, the small sample size and the study subjects selected by convenient sampling methods reduced the representativeness and generalization of the results. Second, the limitation of a cross-sectional study design is that it is not possible to identify the causative agent accurately. As the bacterial culture has low-sensitivity, the failure to find a bacterium after the culture is not sufficient to conclude the absence of that bacterium. In contrast, the detection of a well-developed colony is not adequate to conclude that it is the cause of the disease. Besides, the determination of a viral infection based on a positive serological test does not rule out the possibility of bacterial coinfection.
In some cases, one may see an unspecific increase in IgG/IgM in the inflammatory response. Therefore, the use of biomarkers such as PCT or CRP with the above cutoff recommendations is necessary to support the decision of whether to use antibiotics. Recently, it has been suggested that decision-making regarding antibiotic use based on PCT levels should not be used solely at hospital admission. However, there has been some opinion that making decisions about the use of antibiotics based on PCT concentration should not rely solely on PCT levels when hospitalized. 45,46 Conclusion This study highlights that inflammatory biomarkers such as PCT and hs-CRP can distinguish children with severe pneumonia caused by Rhinovirus alone and coinfected Rhinovirus. This result can assist clinicians in making decisions about appropriate antibiotic use.

Conflicts of Interest
None declared.