PRTN3/FCER1A Transcriptomic Ratio Predicts Hospitalization in Primary Care Attenders With Respiratory Infection

Early detection of patients with respiratory infection at risk of deteriorating could help to improve their outcome by facilitating immediate transfer to the hospital to receive the adequate level of care. In this regard, gene expression proling is emerging as a promising tool to identify patients with infection at risk of suffering a complicated outcome. In a cohort of patients with respiratory infection attending to an Emergency Room at a community health centre, we quantied expression levels in blood of ve genes involved in the granulocyte biology that have been previously described to be linked to infection severity: MMP8 (matrix metallopeptidase 8), LCN2 (lipocalin-2), LTF (lactotransferrin) and PRTN3 (proteinase 3) and FCER1A (receptor for Fc fragment of IgE, high anity I). Expression levels of these genes were evaluated to predict hospitalization. Multivariate analysis adjusted by the National Early Warning Score (NEWS), neurovascular disease, hypertension and age revealed that all these genes independently predicted hospitalization. Nonetheless, the ratio between PRTN3/FCER1A outperformed individual genes to predict necessity of hospitalization (OR [CI95%], p: 8.36 [2.02-34.52],0.003). In conclusion, quantication of PRTN3/FCER1A gene expression ratio could represent a useful test to early identify those patients with respiratory infection at risk of deterioration in extra-hospital settings.


Introduction
Respiratory infection is one of the major causes of morbimortality worldwide (1) (2). Early detection of patients at risk of deteriorating could help to improve their outcome by facilitating immediate transfer to the hospital to receive the adequate level of care. In this regard, gene expression pro ling is emerging as a promising tool to guide clinical decisions in respiratory (3) (4) and other kind of severe infections such as sepsis (5). Expression levels of genes involved in the granulocyte biology has been described to be altered in severe respiratory infections of viral (6) (7) or bacterial origin (8), responding to a phenomenon called "emergency granulopoiesis" (9). In consequence, granulocyte-related genes are potentially useful to detect patients with respiratory infections at risk of developing a complicated outcome. Nonetheless, current studies pro ling gene expression in respiratory infections have been developed late in the course of the disease, once the patient is already hospitalised.
In the present study, we quanti ed the expression levels in blood of ve of these granulocyte genes early after the onset of a respiratory infection, in attenders to an Emergency Room sited at a community health centre. Expression levels of MMP8 (matrix metallopeptidase 8 /neutrophil collagenase), LCN2 (lipocalin-2, also known as neutrophil gelatinase-associated lipocalin, NGAL), LTF (lactotransferrin), PRTN3 (proteinase 3) and FCER1A (receptor for Fc fragment of IgE, high a nity I) was pro led by using droplet digital PCR (ddPCR), a next-generation PCR method, which offers absolute quanti cation with no need of standard curve and greater precision and reproducibility than currently available qRT-PCR methods (10). Expression levels of these genes were tested for its ability to predict hospitalization using a multivariate analysis.

Materials And Methods
Design this is a new work from our PREVISE study, which prospectively recruited patients presenting with respiratory infection at a Community Health Centre in Spain (Servicio de Urgencias de Atención Primaria, SUAP, Salamanca) from December 2018 to January 2019. While in a previous article from this study we evaluated transcriptomic signatures for detecting viral infection in these patients (11), in the present manuscript we test other totally different gene expression signatures to predict hospitalization. Patients and setting: recruited patients met the following criteria: age >45 years, clinical signs of upper or lower respiratory infection according to the physician in charge, and at least one of the following items of the National Early Warning Score (NEWS score) (12): Respiration rate < 11 rpm or ≥ 21 rpm, oxygen saturation < 95%, any supplemental oxygen, temperature < 36.0 or > 38.1 o C, systolic blood pressure < 110 mmHg or ≥ 220 mmHg, heart rate < 50 or > 91 or altered level of consciousness. A total of 129 patients were enrolled, of which 29 were nally excluded from the study (14 of them had incomplete follow up data and 15 showed infections of other source than respiratory one).

Ethics declarations
informed consent was obtained from all individuals. The research complied with all the relevant national regulations and institutional policies and was developed in accordance to the tenets of the Helsinki Declaration, and was approved by the "Comité Ético de Investigación con Medicamentos" of the Instituto de Investigación Biomédica de Salamanca (IBSAL) (code PI 2018 11 138).
Gene expression pro ling: 2.5 mL of blood were collected by using PaxGene (BD) venous blood vacuum collection tubes. Total RNA was extracted from blood samples using the PAXgene Blood RNA System (PreAnalytix, Hombrechtikon, Switzerland). The evaluation of concentration and quality was performed by spectrometry (Nano-Drop ND1000, NanoDrop Technologies,Wilmington, DE). Gene expression was quanti ed by ddPCR (BioRad). Granulocyte genes assessed were MMP8 (reference Hs01029057_m1); LCN2 (reference Hs01008571_m1); LTF (Hs00914334_m1); PRTN3 (Hs01597752_m1) and FCER1A The obtained volume of cDNA (20mL) was further diluted (1/25), and 2.5mL (5 ng of total mRNA) were employed for quanti cation of target gene expression according to the manufacturer instructions. Brie y, ddPCR was performed using the BioRad QX200 ddPCR system, ddPCR Supermix for Probes (no dUTP), and BioRad standard reagents for droplet generation and reading. End-point PCR with 40 cycles was performed by usingC1000Touch Thermal Cycler (BioRad) after splitting each sample into approximately 20,000 droplets. Next, the droplet reader used at least 10,000 droplets to determine the percentage of positive droplets and calculation of copy number of cDNA per nanogram of initial mRNA.

Statistical analysis
For the descriptive analysis of the patients 'characteristics, the differences between groups were assessed using the Chi-square test or Fisher's Exact Test for categorical variables. For continuous variables, differences between groups were assessed with the Mann-Whitney U test. A multivariate logistic regression analysis was employed to evaluate the association between gene expression levels and the risk of hospitalization in the next 48 hours following rst contact with the family doctor. The potential confounding factors were identi ed by using a univariate analysis followed by multiple testing correction (false discovery ratio FDR-Benjamini-Hochberg) of the p-values (a complete list of the variables considered for this analysis is showed in the supplementary material 1). Those variables yielding FDR/q-values < 0⋅200 were further introduced in the multivariate analysis as adjusting variables for gene expression levels. In the case of the categorical variables, only those with a frequency > 5% were considered for the analysis. The predicted values from the multivariate model were used to calculate the area under the receiver operating curve (AUC), which summarizes its predictive power. Statistical analysis was performed using IBM SPSS Statistics 26.0 (SPSS INC, Armonk, NY, U.S.A). The level of signi cance was set at 0·05 (2-tailed).

Results
Patients' characteristics demographic and clinical characteristics of the study subjects strati ed by the need of hospitalization are provided in Table 1. Patients who nally were admitted to the hospital were older than those not needing hospitalization. The former showed more frequently an antecedent of a neurovascular disease and/or hypertension. The prevalence of viral infections was similar between groups (64% in the patients needing of hospitalization and 65% in the group not needing it, p > 0.05), with no differences between both groups in the pro le of viruses causing the respiratory infection (supplementary material 1). Mean blood pressure and O2 saturation was lower in the hospitalised patients, while they presented with higher heart and respiratory rates. NEWS scores were higher in this group. Treatment received for the current episode of respiratory infection was similar between groups, except that non hospitalised patients received more frequently amoxicillin-clavulanic acid. 76% of patients that were nally hospitalised needed of O2 at rst contact with the Primary Health Center, for 13% in the non-hospitalised group. The four patients of this cohort who nally died had been hospitalised. Gene expression levels in patients needing of hospitalization vs those who did not need to be hospitalised levels of MMP-8, LCN-2, PRTN-3 and LTF were higher in those patients nally admitted to the hospital, while, on the contrary, levels of FCER-1A were lower ( gure 1).
Multivariate logistic regression analysis to predict hospitalization: univariate analysis revealed that all the granulocyte related genes predicted need of hospitalization (MMP8, LCN2, PRTN3, LTF, FCER1A) (supplementary material 2). As shown in Table 2, multivariate analysis demonstrated that, after adjusting by potential confounding factors, all the genes still predicted risk of hospitalization, with ORs even higher than that showed by the NEWS score for ruling in  when the multivariate models were tested for their ability to differentiate between hospitalized and non-hospitalized patients using AUCs, the ratio between PRTN3 and FCER1A yielded the highest AUC from all the genes or potential gene ratios when compared to the AUC obtained from the model not including neither individual gene expression levels or gene expression ratios ( gure 2).

Discussion
Our results evidenced that the hyperexpression of MMP-8, LCN2, PRTN3, LTF and the hypo-expression of FCER1A in blood is an early predictor of complicated outcome in patients with a respiratory infection acquired at the community. Expression levels of these genes is known to be altered in the context of other kinds of severe respiratory infections. Hyper-expression of MMP-8 and LTF is a signature of ventilator-associated pneumonia ( (19). FCERIA is constitutively expressed in mast cells and basophils and mediates transmission of stimulatory signals upon engagement of IgE-bound allergens (20). Hypo-expression of FCER1A has been repeatedly described in severe infection (6) (21), but its biological implications are unknown.
While the studies evidencing the link between expression levels of these genes and the severity of an infection have been developed at the hospital, our study is the rst conducted in a pre-hospital setting, being pioneer in evidencing that this transcriptomic signature involving MMP-8, LCN2, PRTN3, LTF and FCER1A is already altered in the early moments of the disease, at the onset of the symptoms. Early and accurate detection of patients with respiratory infection at risk of deterioration could help to improve outcome of these patients, by accelerating transfer of the patient to the hospital, where further complementary tests and speci c vital/organ support can be administered. The multivariate analysis evidenced that the PRTN3/FCER1A transcriptomic ratio was the gene combination yielding the best results to predict hospitalization, outperforming NEWS. While calculation of NEWS implies scoring several items (some of them showing important inter-observer variations, such as the level of consciousness) (12), gene expression quanti cation could inform on the prognosis of the patient in a more objective and reproducible manner. Nonetheless, the quanti cation of gene expression has classically represented a major challenge for translating transcriptomic biomarkers into the clinical practice. New technologies which allows rapid quanti cation of mRNA transcripts, such as loop-mediated isothermal ampli cation (LAMP) (22), along with the implementation of laboratories speci cally designed to respond to the necessities of family doctors, will facilitate application of these kind of biomarkers to guide clinical decisions in primary care (23).
Finally, a limitation of our study is the limited sample size of our cohort, and the over-representation of respiratory infections of viral origin. Another limitation is that the study was developed before the current COVID-19 pandemics. Further studies recruiting larger cohorts with a more balanced composition of bacterial and viral infections (including SARS-CoV-2) are needed to con rm the predictive performance of the PRTN3/FCER1A ratio.
In conclusion, the gene expression ratio between PRTN3 and FCER1A could represent a useful test to early identify those patients with respiratory infection at risk of deterioration in extra-hospital settings.
Declarations Figure 1 Granulocyte`s gene expression levels in patients needing of hospitalization vs those who did not need to be hospitalised. Results are provided as copies of cDNA/ng of initial RNA.

Figure 2
AUROC analysis to predict hospitalization based on the logistic regression models including or not the PRTN3/FCER1A gene expression ratio.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. 7Supplementarymaterial.pdf