Comparison Of Diagnostic Criteria For Invasive Pulmonary Aspergillosis In Critically Ill Populations And Clinical Diagnostic Value Of Bronchoalveolar Lavage Fluid Metagenomic Next-generation Sequencing And Host Transcriptome For Invasive Pulmonary Aspergi

Background:Invasive pulmonary aspergillosis(IPA)is a common infection in intensive care units(ICUs).Rapid and accurate diagnosis of IPA is essential for outcomes.There are no consensus criteria for defining IPA in ICU,and several criteria are currently used.Metagenomic Next-Generation Sequencing(mNGS)of bronchoalveolar lavage fluid(BALF)and host transcriptome have applications in identifying the pathogenic,however,the role of these new technologies in diagnosing IPA are not well established.Objective:To compare the diagnosis and prognosis performances of three criteria(2019 EORTC/MSG criteria(EORTC 2019),2021 EORTC/MSG ICU criteria(EORTC-ICU),modified AspICU criteria(M-AspICU))for IPA in the ICU.To evaluate the application of six different criteria of mNGS(SDSMRN>1,SDSMRN≥3,SMRN≥10;SMRN>50;RPM ratio≥10;Relative abundance of genus>30%)for the diagnosis of IPA.To analyze lung microbiome via mNGS data and the host transcriptome via PBMC in IPA patients,and to develop the classifier for the diagnosis of IPA.Methods:In the first part of our study,we retrospectively applied three criteria(the EORTC 2019,the EORTC-ICU,and the M-AspICU criteria)for IPA in patients with suspected pneumonia and undergoing at least one mycological test between 2016 and 2021.We compared the diagnosis and prognosis performances of the three criteria in the ICU.In the second part of our study,we used the clinical composite criteria as gold criteria for diagnosing IPA,and applied six BALF mNGS-derived criteria(SDSMRN>1,SDSMRN≥3,SMRN≥10;SMRN>50;RPM ratio≥10;Relative abundance of genus>30%)in patients with suspected pneumonia undergoing BALF mNGS and Aspergillus mycological tests between January 2022,and February 2023.We compared the diagnostic agreement and performances of these six criteria.The lung microbiome detected by mNGS was characterized and developed a pathogen classifier for diagnosing of IPA.We assessed whether the combined mNGS and microbiome had a perfect diagnostic performance.In the third part of our study,human transcriptomes were compared between patients with IPA and those without IPA.We developed a gene classifier to identify IPA.Results:Overall,In the first part,2,403 patients were included.The rates of IPAaccording to the EORTC 2019,EORTC-ICU,and M-AspICU were 3.37%,6.53%,and23.10%,respectively.Diagnostic agreement among these criteria was poor(Cohen’s kappa 0.208-0.666).IPA diagnosed by either the EORTC 2019(odds ratio=2.709,P<0.001)or the EORTC-ICU(odds ratio=2.086,P=0.001)criteria was independently associated with 28-day mortality.IPA diagnosed by M-AspICU became an independent risk factor of 28-day mortality(odds ratio=1.431,P=0.031)only after excluding the population missed by the EORTC-ICU criteria but diagnosed as IPA using the MAspICU because of both broader host criteria and radiological factors.In the second part,Taking the clinical composite criteria as gold criteria,a total of 115 patients were analyzed,of which 28 were identified as IPA.Diagnostic agreement among the six mNGS-derived criteria was moderate,with a Cohen’s kappa of 0.577(P<0.001).The AUC for the four parameters of the six mNGS-derived criteria was around 0.6.All mNGS-derived criteria had low sensitivity ranging from 21.4%to 57.1%and high specificity ranging from 65.5%to 92%.The optimal cut-off value of SDSMRN,SMRN,RPM ratio,and Relative abundance of genus for diagnosing IPA were 5,0.25,8,and 20%,respectively.The pulmonary microbiome of patients with IPA significantly differed from those without IPA(p<0.001).Next,we developed a 19-pathogen classifier that showed a perfect AUC of 0.902 for diagnosing IPA.When tested in the validation cohort,the pathogen classifier performed with an AUC of 0.87.We also found when combing the Shannon index with the relative abundance of gene level increased AUC from 0.746(95%CI:0.656-0.823)up to 0.834(95%CI:0.753-0.897)for the diagnosis of IPA.In the third part,a total of 34 patients were analyzed.There are a total of 115 differential genes in patients with IPA and those without IPA,and these differential genes were enriched to ribosomes,cadherin binding,as well as cell adhesion molecule binding.We built a 7-genes-classifier to identify IPA and performed with an AUC of 0.851.Conclusions:There is poor agreement among the three criteria for diagnosing IPA,Although M-AspICU criteria has the highest‘sensitivity’,IPA diagnosed by M-AspICU was not an independent risk factor of 28-day mortality.Caution is required when using the M-AspICU criteria in ICU,especially in patients with non-specific infiltration and non-classical host factors.The diagnostic agreement among different BALF mNGS-derived criteria for IPA was moderate.All mNGS-derived parameters and criteria had low sensitivity but high specificity for the diagnosis of IPA.Patients with IPA showed a significant decrease in the diversity of the pulmonary microbiome.Combining lung microbiome diversity can improve the diagnostic performance of mNGS.Pathogen classifier is also a good tool for diagnosing IPA.Although there are several differential host transcriptomes between patients with IPA and those without,the different genes cannot be enriched to specific inflammatory pathways.The gene classifier showed perfect prediction value to diagnosis IPA but was not robust,which needs to be further explored in the future.