Prognosis of lower respiratory tract infected patients with virus detected in bronchial alveolar lavage uid: a retrospective observational study

Background and Objectives The inuence of virus detected in BALF is still debating. This study aimed to compare the prognosis of lower respiratory tract infected patients with virus detected in bronchial alveolar lavage uid (BALF) and patients with virus undetected by using metagenomic next-generation sequencing technology. Methods This was a retrospective cohort study. 53 patients with lower respiratory tract infection were enrolled. BALF samples were collected from each patient and sent to perform mNGS pathogenic test in the study. According to the results of mNGS test, patients were divided into virus-detected group and virus-undetected group. In the meanwhile, patients’ clinical information, medical history, disease severity scores, parameters of organ function at the day of ICU admission, prognosis, hospital length of stay, ICU length of stay and needs for medical support were also collected. Results 39.6 percent (21/53) of the BALF samples were virus nucleic acid positive. Mortality rate, tracheotomy rate, mechanical ventilation supporting time, blood transfusion rate were signicantly higher in virus-detected group than that in virus-undetected group. Virus-detected was closely related to hospital and ICU survival time. Conclusions Patients with virus detected in BALF were prone to a poorer prognosis. The detection of virus was a high-risk factor of death for LRTI patients. Virus-detected patients required more medical resources. It was to that higher was determined by put together into a regression model and found that only detected in BALF” had an effect on patient ICU time and hospital survival time. Virus-detected was supposed to be an independent risk factor for


Introduction
Community acquired lower respiratory tract infection (LRTI ) is the most common reason of requiring mechanical ventilation support among various kinds of cases admitted to an Intensive Care Unit ( ICU ).
In the past, bacterial infection was considered to be the leading cause of community acquired pneumonia ( CAP ), and Streptococcus Pneumoniae accounted nearly half of the criminal pathogen. However, since the outbreak of In uenza A ( H1N1) in 2009, there was an increasing number of viral pneumonia cases admitted to ICU [1]. Studies on American adult or pediatric CAP patients who required hospitalization indicated that respiratory viral infection accounted for the largest proportion [2,3]. A prospective study on the etiology of adult LRTI in an European primary hospital founded that bacterial infection accounted for one fth of the total while viral infection accounted for nearly half [4].
Isolation and identi cation of the virus is the "gold standard" for virus detection, however, complicated procedure and long time for cultivation limit the clinical application. Molecular diagnostic technologies such as Polymerase Chain Reaction (PCR) can rapidly and directly detect the pathogens from a clinical sample, which improving the diagnostic e ciency greatly. But the shortcoming is its relatively narrow pathogenic detection range and requiring to prede ne the possible pathogen according to the clinical manifestation in priority [5]. Only one pathogen or a limited array of pathogens can be check once a time.
Metagenomic next-generation sequencing (mNGS) is an e cient molecular diagnostic tool used in various clinical setting, from lab to clinics. Based on next-generation sequencing technology, it can detect genetic composition of all the microorganism in a sample, sequencing billions of DNA or RNA fragments simultaneously. Although it is not as fast as PCR for pathogen detection, it can comprehensively detect all the pathogenic nucleic acid in a sample or even explore the microbial community and human host response through transcriptome analysis at the same time [6].
Bronchial alveolar lavage uid (BALF) is a commonly used sample type of mNGS. In the clinical practice, we have noticed that virus nucleic acid fragments are often detected positive in the BALF when sequenced. Whether virus detected in BALF mean something is still controversial among researchers. Vergera et al [7] founded that cytomegalovirus (CMV ) detected positive in BALF from critically ill patients was associated with longer hospital length of stay ( HLOS ) and higher mortality rate. Meanwhile, another study on pediatric patients whose CMV DNA was detected from BALF concluded that the CMV was commonly found in BALF, which didn't mean acute dominant CMV infection and wouldn't be regarded as the rst primary diagnose [8].

Inclusion criteria
We screened all consecutive patients with LRTI admitted to or transferred to the ICU from Nov 2017 to Nov 2019. LRTI mainly included acute exacerbation of chronic obstructive pulmonary diseases, community-acquired pneumonia and acute bronchitis. These diagnoses were made based on clinical characteristics, test reports and chest radiography. The inclusion criteria were (1) age 18 years or above, (2) patient was clinically diagnosed as lower respiratory tract infection (3) BALF sample had been obtained for mNGS test.

Exclusion criteria
Patient was excluded when (1) younger than 18, (2) hadn't been discharged yet within the research time, (3) the cases was impossible to do accurate statistics due to data losing. Due to traditional habits, patients in our country would not be followed up regularly. The study lacked clinical data related to patients' follow-up information.
Mortality means patients died in ICU, or was dying when the client required being discharged.
Metagenomic next-generation sequencing BALF samples were obtained by an operator who had at least two years full-time working experience,following standardized operational ow. 96.2% participants were intubated. The specimen was then sent to a quali ed pathogenic microorganism detection center for DNA sequencing. After the procedure of nucleic acid extraction, library preparation, DNA sequencing and bioinformatic analysis, a compliant report was issued.
Statistical analysis SPSS 25 was used for data processing and analysis. The continuous variables were expressed as median and quartile, while categorial variables were expressed in quantity and percentage. Due to small amount of data collected, non-parametric test was used for the comparison of continuous variables while Chi-square test was used for the comparison of categorical variables between two groups. Multivariate Cox proportional hazard regression model [Method = Forward stepwise (Conditional LR)] was used to identify an independent prognostic factor. Kaplan-Meier method was used for survival analysis between two groups. P < 0.05 was considered statistically signi cant.

Patient demographic characteristics
A total of 53 patients were enrolled according to the criteria ( Fig. 1), who had various complications, such as hypertension or diabetes mellitus. The majority were male and the median age was 68 years old. The most frequent detected virus was herpes simplex virus 1, followed by Epstein-Barr virus and cytomegalovirus ( Table 1 ). In total, 21 patients with virus nucleic acid detected in BALF made up the virus-detected group, 32 patients with virus nucleic acid undetected in BALF made up the virus-undetected group. Abbr. BALF, bronchial alveolar lavage uid.

Comparison between virus-detected group and virusundetected group
There were no differences between the virus-detected group and the virus-undetected group in age, gender and medical history. The two groups were comparable ( Table 2 ).  ( Table 3 ). There were both 6 participants with missing data for CD 4 %,CD 8 % and CD 4 /CD 8 ratio in each group. All other variables were completed.    ICU survival functions showed that virus-undetected group had signi cantly longer ICU survival time than virus-detected group, which shared similarity in hospital survival functions ( Fig. 2 -Fig. 3 ).

Discussion
The detection rate for virus by traditional methods is extremely low. Since the advent of molecular diagnostic technologies such as PCR, the ability of diagnosing respiratory virus have been improved greatly. Next generation sequencing has better diagnostic performance on viral infection than traditional methods which mainly consist of isolation and identi cation, serology tests, and PCR which targets only a limited panel of speci c viruses. mNGS technology can unbiasedly sequence all the DNA fragments and detect all pathogens in a sample in one time [9]. Many researches had con rmed that diagnostic e ciency of mNGS is superior to that of traditional methods. The pathogenic information obtained by mNGS could also help guiding the usage of clinical antibiotic and antiviral drugs, including initiation, deescalation, withdrawal of the drugs [10,11]. So, in this study, we chose the versatile mNGS to detect whether there were virus nucleic acid fragments in BALF instead of large amounts of conventional tests, avoiding missed diagnosis.
HSV can commonly be detected in BALF. The most frequent detected virus in our study was exact the HSV1. It's still unclear of the clinical signi cance of HSV detection in lower tract since the relationship between HSV detection and mortality in clinic were inconsistent in different studies [12]. At present, it was uncertain that HSV detected in lower respiratory tract of critically ill patients were related to poor prognosis, or it was just an epiphenomenon associated with disease severity and immune paralysis [13].
German researchers discovered that HSV-positive patients had longer HLOS and ICULOS compared to HSV-negative patients, there were no signi cant difference in ICU mortality rate or hospital mortality rate between two groups [14]. In our study, we found virus-positive patients had higher mortality rate and tracheotomy rate. Time required for ventilator support was longer in virus-detected group. The proportion of patients in virus-detected group who needed blood transfusion was also higher. It seemed that virus in BALF was not merely an innocent bystander, it did in uence the patient to some extent.
Theoretically, viral infection could stimulate tremendous cytokine reactions, causing disorders in immune system. This kind of cytokine storm could also damage endothelial cells, trigger coagulation disorders, change microvascular permeability, induce tissue edema or even shock [1]. But patients in virus-detected group had higher APACHE II scores and TB value. It was hard to say that higher mortality rate was just determined by viral infection. We put "APACHE II scores ", "TB value" and "virus-detected" together into a Cox regression model and found that only "virus detected in BALF" had an effect on patient ICU survival time and hospital survival time. Virus-detected was supposed to be an independent risk factor for mortality It is still uncertain about the in uence of virus nucleic acid detected in critically ill patients with LRTI on the whole clinic procedure. To our knowledge, this was the rst research discussing the prognosis of lower respiratory tract infected patients with virus detected in BALF by using mNGS. Samples used in previous research on correlation of virus and prognosis were mostly throat swabs and sputum, BALF was seldom used. So our results were relatively more convincing.
There were some limitations in our study. It was a just single-center, retrospective, observational study. The sample size was relatively small, it might lead to bias when generalized to a broader condition.
Because of economic and technological reasons, the mNGS performed in this study were mainly aimed at DNA virus, instead of RNA virus. So common RNA virus such as in uenza virus or respiratory syncytial virus was not shown in the reports, which could caused the deviation. Virus nucleic acid detected positive didn't mean virus infection. As long as there were virus DNA fragments in BALF, it could be detected positive no matter whether the virus was alive or not. Inapparent infection couldn't be distinguished from dominant infection. False positive results could easily appear because contamination might exit during sample collection, transportation, or in test procedure. Last but not least, as the limited laboratory conditions in our hospital, the results couldn't be veri ed by the PCR method for each patient. More largescale, multi-center, randomized, prospective researches on this topic are needed.

Conclusion
In a conclusion, patients with virus detected in BALF might relate to a poor prognosis and require more clinical resources. Availability of data and materials The datasets used and analyzed during the current study are available in supplementary material.

Competing interests
All the authors declared that there had no competing interest to this work.

Consent for publication
All the authors agreed to publish the study.

Authors' contributions
All the authors contributed equally in this work and approved the nal manuscript. Abbr. BALF, bronchial alveolar lavage fluid.  The two groups had statistically difference in APACHE II score and TB value (P < 0.05) Virus-detected was found to be an independent prognostic factor for hospital survival by Cox regression model, the relative hazard HR was 2.787