Identification of prognostic burns-related indicators and microRNA biosignatures in burns patients with inhalation injury

Abstract

Burn injuries remain a major global public health concern, disproportionately affecting low- and middle income countries (LMICs). The high incidence and adverse outcomes stem from both socioeconomic disparities and the complexity of thermal injury, which is further influenced by co-factors such as inhalation injury. This condition, caused by toxic smoke inhalation during thermal events, is strongly associated with increased mortality and morbidity due to its variable clinical presentation and delayed onset. Despite being one of the most important mortality co-factors, there are no standard diagnostic criteria, and those that do exist have numerous shortcomings that particularly impact resource-poor clinical settings. Early identification of inhalation injury could improve burn management by leveraging readily available clinical markers from medical records. Additionally, most burn patients require intravenous fluid administration as part of standard protocol, presenting an opportunity for non-invasive blood sampling for biomarker analysis. Ideal biomarkers require that they be stable, disease-specific, and quantifiable, a criterion met by small, single-stranded RNA molecules with the ability to regulate gene expression called miRNAs. The primary and overall aim of this study was therefore to identify burn-related clinical markers extracted from a selected cohort of burn patient’s medical records and miRNA biomarkers from the patients’ whole blood samples using sequencing and bioinformatics tools. Following ethical approval and patient consent, the medical record data were extracted, and corresponding whole blood samples were collected from burn patients (n=59) admitted to the Western Cape Provincial Adult Tertiary Burns Centre (WCPATBC) at Tygerberg Hospital, South Africa, between 23 April 2016 and 15 August 2017. Recorded burns-related variables included sociodemographic factors (i.e. gender, age, referral level, etc.), burn severity (i.e. % total body surface area [TBSA] burns, inhalation injury, complications, etc.), and clinical factors (i.e. burns intensive care unit length of stay [BICU LOS], partial gas pressures, lactate, etc.). Inhalation injury was stratified based on total ventilation days, with mild inhalation injury defined as ≥4 days and severe inhalation injury as ≥5 days. The first aim was to evaluate whether the recorded medical data aligned with reported findings in South Africa and if it was comparable to data reported from LMIC clinical settings. This analysis facilitated the identification of trends associated with increased mortality risk and inhalation injury in these vulnerable groups. Descriptive statistics were reported as mean (95% CI) and/or frequency (%) and analyzed using IBM SPSS Statistics version 28. Results showed that the majority of patients admitted were male, of working-age adults (21-39 years), referred from local areas within the district, who had sustained flame burns, primarily on weekdays and during colder months. The majority had serious to maximum abbreviated burn severity index (ABSI) scores, <40% TBSA burn and severe inhalation injury. While most arterial blood gas (ABG) measurements were found to be within normal ranges, elevated PO2 and lactate levels were observed. The overall mortality rate was 25.4%, with higher rates found among females, younger and older age groups, and in patients referred from distant districts and clinics or community health centres. Elevated mortality rates were also observed in patients with flame burns, larger TBSA, severe inhalation injury, complications, longer ICU LOS, pre-admission ventilation, and ABG values with elevated lactate; and reduced pH, PaO2, PCO2, and base excess levels. These findings strongly aligned with previously reported South African and LMIC-based data, with worse outcomes demonstrated compared to high income countries (HICs). The findings reported in this section contribute to the limited national literature and data on burn patients and highlights the continued burden of burns (and disease) in LMICs, potentially due to economic constraints and limited specialized burn care. Identifying these trends may help identify prevention strategies adapted from HICs to reduce burn incidence. Although some of the burn injury findings varied compared to the reported literature, their prognostic value remains crucial in assessing survivability. The second aim was to (i) assess burn-related factors in relation to mortality to determine specifically whether inhalation injury reported in this patient cohort was a potential significant mortality co-factor, and (ii) whether the findings corroborated previous reported studies. Relationships between burn-related variables, mortality, and inhalation injury were analyzed using Fisher’s Exact test (association), Pearson’s point biserial (rpb)/Spearman’s correlation (rho) coefficient (correlation strength/direction), and partial least squares regression (predictive contribution via variable importance in prediction [VIP] values). Correlation strength was categorized from none to very strong, with VIP scores >0.5 indicating significant contribution and >1 demonstrating greater predictive power. Statistical analyses were conducted in IBM SPSS Statistics version 28, with significance set at p<0.05 (two-tailed). An average of 2.7 patients (CI: 2.3–3.4) sustained mild inhalation injury, while 11.2 patients (CI: 9.5–12.9) had sustained severe inhalation injury, with the latter associated with 93.3% mortality, a strong positive correlation (rho = 0.441), and a significant predictive contribution (VIP = 0.819). Additional mortality correlated factors included referral setting, TBSA, complications, and BICU LOS. The multifactorial nature of burn outcomes would predictably see mortality correlate with multiple parameters, of which inhalation injury had a particularly strong impact. In alignment with previous reported studies, these findings confirm inhalation injury as a critical mortality co-factor, supporting the need for further research. The third aim was to evaluate the relationship between burn-related variables and inhalation injury using the same statistical methods employed with the second aim. Identifying significant relationships would indicate which variables could potentially serve as the most effective prognostic markers. Analysis displayed notable positive correlations observed for lactate (rho=0.331), %TBSA (rho=0.357), complications (rho=0.690), and BICU LOS (rho=0.908), with complications and BICU LOS showing the strongest correlations and the highest predictive contributions (VIP=1.229 and 1.372, respectively). Initial in-hospital prognosis could benefit or be improved by considering markers that would immediately present on admission or shortly thereafter. Complications may therefore be more suitable for early prognosis, while prolonged BICU LOS could provide insight into the degree of progression. Lactate levels, measured shortly after injury, may offer valuable prognostic information, based on the positive correlation observed with inhalation injury. However, further studies are required to determine the specific lactate level changes that correlate with particular pathological sequelae. The fourth aim was to identify potential miRNA biomarkers from whole blood samples of the burns patients, which had been collected in parallel with the patients’ medical records. Total RNA was extracted, and its quantity and quality were assessed using a NanoDrop and a Bioanalyzer, respectively. Thirty exemplars representing mild and severe inhalation injury were selected for sequencing, with 22 passing quality control (mild: n=8, severe: n=14) and sequenced using the Illumina NextSeq 550 platform. Reads were aligned to the human genome (GRCh38) with Bowtie in sRNAbench, and miRNAs were quantified as counts per million using sRNAde. Differential abundance analysis was conducted with EdgeR in R v4.1.2 and validated by DESeq2. Fisher’s Exact test compared differentially expressed (DE) miRNAs between groups, and significance was set at Padj.<0.05 with fold changes |log2 (FC)| >1.5. Results displayed ten overlapping DE miRNAs that met the significance threshold, comprising nine up-regulated in severe injury and one down-regulated in mild injury. MiR-30a-5p, miR 15a-5p, and miR-21-5p had the highest degree rankings, targeting 734, 717, and 612 genes, respectively. These findings present a potential panel of miRNAs that may present a cocktail of inhalation injury prognostic markers. The identified miRNAs have also been demonstrated as having key roles in mechanisms of inflammation and apoptosis, as well as in conditions closely related to inhalation injury, despite varying methodology. The fifth aim was identify the mRNA target genes of the DE miRNAs and determine the associated pathways that were regulated. This could demonstrate biological significance of the identified miRNAs relative to the pathways involved. Protein-protein interaction networks, analyzed using STRING and Cytoscape, identified key hub genes of the up-regulated (i.e. TP53, AKT1, MYC, CTNNB1, EGFR, PTEN, JUN, STAT3, EP300 and TNF) and down-regulated miRNA (i.e. TP53, MDM2, BCL2L11, CDK6, BBC3, GADD45A, BAX, FAS and RHOA). Gene ontology, KEGG, Reactome, and PANTHER analysis revealed that the hub genes were mainly enriched in inflammatory and apoptotic pathways. The findings highlight the role of miRNAs and their target genes in pathways impacting potential hallmarks of inhalation injury. While further validation with larger cohorts and additional techniques is required, these preliminary results provide a strong foundation for identifying miRNAs with prognostic potential in inhalation injury. Overall, the findings presented in this thesis demonstrated several key firsts in the field of inhalation injury prognostication, particularly within the South African and broader LMIC context. For the first time, prognostic burn-related indicators (TBSA, complications, BICU LOS, and lactate) were retrospectively identified as specific predictors of inhalation injury, rather than general burn mortality. This research also represents the earliest application of human whole blood from burn patients to investigate miRNA expression profiles, proposing a novel panel of 10 differentially expressed miRNAs relating not only to the presence of inhalation injury but also to its severity (mild versus severe). While further validation is warranted, the strength of these preliminary findings was supported by consistent overlaps across statistical outcomes for burns-related medical-file-based indicators, while reinforced miRNA expression patterns were confirmed through dual bioinformatics pipelines, threshold filtering, and four independent enrichment platforms to ensure robust biological interpretation. Moreover, the use of whole blood, a readily accessible and clinically relevant biospecimen, may better represent systemic responses to injury, thereby improving the translational potential of the results. Collectively, this research presents the first integrated evaluation of retrospective clinical data alongside prospective molecular profiling within a single burn patient cohort for the purpose of inhalation injury prognostication. By combining these two dimensions of analysis, the study lays critical groundwork for incorporating both clinical and circulating biological markers—many of which have not been previously explored in this context—into existing diagnostic or prognostic frameworks. This approach not only broadens the spectrum of potential indicators but also sets the stage for the future development of comprehensive, multi-modal tools capable of enhancing the timely and precise assessment of inhalation injury.