Multidimensional analysis of the host response reveals prognostic and pathogen-driven immune subtypes among adults with sepsis in Uganda

Molecular phenotypes of critical illness confer prognostic and biological enrichment in sub-Saharan Africa: a prospective cohort study from Uganda

The generalisability of critical illness molecular phenotypes to low- and middle-income countries (LMICs) is unknown. We show that molecular phenotypes derived in high-income countries (hyperinflammatory and hypoinflammatory, reactive and uninflamed) stratify sepsis patients in Uganda by physiological severity, mortality risk and dysregulation of key pathobiological domains. A classifier model including data available at the LMIC bedside modestly discriminated phenotype assignment (area under the receiver operating characteristic curve (AUROC) 0.80, 95% CI 0.71 to 0.90 for hyperinflammatory vs hypoinflammatory; AUROC 0.74, 95% CI 0.65 to 0.83 for reactive vs uninflamed). Our findings highlight the potential for a globally relevant, clinicomolecular classification of critical illness and may support the inclusion of diverse populations in phenotype-targeted critical care trials. Improved laboratory capacity and access to rapid biomarker assays are likely necessary to optimise phenotype stratification in LMIC settings.

A host immune-related LncRNA and mRNA signature as a discriminant classifier for bacterial from non-bacterial sepsis in children

Background

The variations in non-coding RNA alterations and the host immune response between patients with bacterial and non-bacterial sepsis, along with their clinical characteristics, are largely unknown.

Methods

The landscape of long non-coding RNA (lncRNA) and mRNA in whole blood cells from pediatric patients with bacterial sepsis or non-bacterial sepsis were characterized using an Arraystar human LncRNA microarray. Weighted correlation network analysis (WGCNA) were conducted to identify immune-related LncRNA-mRNA signatures. Least absolute shrinkage and selection operator (Lasso) regression and Ridge regression analysis were employed to develop a specific LncRNA-mRNA signature, serving as a discriminant classifier for bacterial and non-bacterial sepsis in children.

Results

A total of 33 differentially expressed lncRNAs and 52 mRNAs were identified in pediatric patients with either bacterial sepsis or non-bacterial sepsis. Among these, 69 lncRNAs and mRNAs were pinpointed using WGCNA and found to be significantly correlated with clinical parameters. Further intersection analysis identified 12 lncRNAs and 16 mRNAs as immune-related signature for discerning bacterial infections in children with sepsis. Additionaly, the lncRNA-mRNA co-expression network highlighted the key lncRNAs (AC090159.1 and AC080129.2) and mRNAs (S100A8 and TCF7L2) as an infection score model. Lasso regression analysis revealed that this infection score model achieved an area under the received operating curve (AUROC) of 0.96 in the training set and 0.86 in the validation set. Ultimately, the expression levels of these 4 key lncRNAs and mRNAs showed significant correlation with CRP or PCT levels.

Conclusion

The machine learning model, developed utilizing key lncRNAs (AC090159.1 and AC080129.2) and mRNAs (S100A8 and TCF7L2), demonstrates robust discrimination and calibration capabilities for distinguishing between bacterial and non-bacterial sepsis in children.

The Universal Vital Assessment (UVA) score at 6 hours post-resuscitation predicts mortality in hospitalized adults with severe sepsis in Mbarara, Uganda

Sepsis is the leading cause of global death with the highest burden found in sub-Saharan Africa (sSA). The Universal Vital Assessment (UVA) score is a validated resource-appropriate clinical tool to identify hospitalized patients in sSA who are at risk of in-hospital mortality. Whether a decrease in the UVA score over 6 hours of resuscitation from sepsis is associated with improved outcomes is unknown. We aimed to determine (1) the association between 6-hour UVA score and in-hospital mortality, and (2) if a decrease in UVA score from admission to 6 hours was associated with improved in-hospital mortality. We analyzed data from participants with severe sepsis aged ≥14 years enrolled at the Mbarara Regional Referral Hospital in Uganda from October 2014 through May 2015. Among 197 participants, the median (interquartile range) age was 34 (27-47) years, 99 (50%) were female and 116 (59%) were living with HIV. At 6 hours, of the 65 participants in the high-risk group, 28 (43%) died compared to 28 (30%) of 94 in the medium-risk group (odds ratio [OR] 0.56, 95% confidence interval [CI] 0.29,1.08, p = 0.086) and 3 (9%) of 33 in the low-risk group (OR 0.13, 95% CI 0.03, 0.42, p = 0.002). In a univariate analysis of the 85 participants who improved their UVA risk group at 6 hours, 20 (23%) died compared to 39 (36%) of 107 participants who did not improve (OR 0.54, 95% CI 0.27-1.06, p = 0.055). In the multivariable analysis, the UVA score at 6 hours (adjusted OR [aOR] 1.26, 95%CI 1.10-1.45, p<0.001) was associated with in-hospital mortality. When adjusted for age and sex, improvement in the UVA risk group over 6 hours was associated with a non-statistically significant 43% decrease in odds of mortality (aOR 0.57, 95%CI 0.29-1.07, p = 0.08). Targeting a decrease in UVA score over 6 hours from admission may be a useful clinical endpoint for sepsis resuscitation in sSA, but this would need to be proven in a clinical trial.

A Transcriptomic Classifier Model Identifies High-Risk Endotypes in a Prospective Study of Sepsis in Uganda

OBJECTIVES

In high-income countries (HICs), sepsis endotypes defined by distinct pathobiological mechanisms, mortality risks, and responses to corticosteroid treatment have been identified using blood transcriptomics. The generalizability of these endotypes to low-income and middle-income countries (LMICs), where the global sepsis burden is concentrated, is unknown. We sought to determine the prevalence, prognostic relevance, and immunopathological features of HIC-derived transcriptomic sepsis endotypes in sub-Saharan Africa.

DESIGN

Prospective cohort study.

SETTING

Public referral hospital in Uganda.

PATIENTS

Adults ( n = 128) hospitalized with suspected sepsis.

INTERVENTIONS

None.

MEASUREMENTS AND MAIN RESULTS

Using whole-blood RNA sequencing data, we applied 19-gene and 7-gene classifiers derived and validated in HICs (SepstratifieR) to assign patients to one of three sepsis response signatures (SRS). The 19-gene classifier assigned 30 (23.4%), 92 (71.9%), and 6 (4.7%) patients to SRS-1, SRS-2, and SRS-3, respectively, the latter of which is designed to capture individuals transcriptionally closest to health. SRS-1 was defined biologically by proinflammatory innate immune activation and suppressed natural killer-cell, T-cell, and B-cell immunity, whereas SRS-2 was characterized by dampened innate immune activation, preserved lymphocyte immunity, and suppressed transcriptional responses to corticosteroids. Patients assigned to SRS-1 were predominantly (80.0% [24/30]) persons living with HIV with advanced immunosuppression and frequent tuberculosis. Mortality at 30-days differed significantly by endotype and was highest (48.1%) in SRS-1. Agreement between 19-gene and 7-gene SRS assignments was poor (Cohen's kappa 0.11). Patient stratification was suboptimal using the 7-gene classifier with 15.1% (8/53) of individuals assigned to SRS-3 deceased at 30-days.

CONCLUSIONS

Sepsis endotypes derived in HICs share biological and clinical features with those identified in sub-Saharan Africa, with major differences in host-pathogen profiles. Our findings highlight the importance of context-specific sepsis endotyping, the generalizability of conserved biological signatures of critical illness across disparate settings, and opportunities to develop more pathobiologically informed sepsis treatment strategies in LMICs.

Distinct host-response signatures in circulatory shock: a narrative review

Circulatory shock is defined syndromically as hypotension associated with tissue hypoperfusion and often subcategorized according to hemodynamic profile (e.g., distributive, cardiogenic, hypovolemic) and etiology (e.g., infection, myocardial infarction, trauma, among others). These shock subgroups are generally considered homogeneous entities in research and clinical practice. This current definition fails to consider the complex pathophysiology of shock and the influence of patient heterogeneity. Recent translational evidence highlights previously under-appreciated heterogeneity regarding the underlying pathways with distinct host-response patterns in circulatory shock syndromes. This heterogeneity may confound the interpretation of trial results as a given treatment may preferentially impact distinct subgroups. Re-analyzing results of major 'neutral' treatment trials from the perspective of biological mechanisms (i.e., host-response signatures) may reveal treatment effects in subgroups of patients that share treatable traits (i.e., specific biological signatures that portend a predictable response to a given treatment). In this review, we discuss the emerging literature suggesting the existence of distinct biomarker-based host-response patterns of circulatory shock syndrome independent of etiology or hemodynamic profile. We further review responses to newly prescribed treatments in the intensive care unit designed to personalize treatments (biomarker-driven or endotype-driven patient selection in support of future clinical trials).

The CD56-CD16+ NK cell subset in chronic infections

Long-term human diseases can shape the immune system, and natural killer (NK) cells have been documented to differentiate into distinct subsets specifically associated with chronic virus infections. One of these subsets found in large frequencies in HIV-1 are the CD56-CD16+ NK cells, and this population's association with chronic virus infections is the subject of this review. Human NK cells are classically defined by CD56 expression, yet increasing evidence supports the NK cell status of the CD56-CD16+ subset which we discuss herein. We then discuss the evidence linking CD56-CD16+ NK cells to chronic virus infections, and the potential immunological pathways that are altered by long-term infection that could be inducing the population's differentiation. An important aspect of NK cell regulation is their interaction with human leukocyte antigen (HLA) class-I molecules, and we highlight work that indicates both virus and genetic-mediated variations in HLA expression that have been linked to CD56-CD16+ NK cell frequencies. Finally, we offer a perspective on CD56-CD16+ NK cell function, taking into account recent work that implies the subset is comparable to CD56+CD16+ NK cell functionality in antibody-dependent cell cytotoxicity response, and the definition of CD56-CD16+ NK cell subpopulations with varying degranulation capacity against target cells.

Brief Report: Detection of Urine Lipoarabinomannan Is Associated With Proinflammatory Innate Immune Activation, Impaired Host Defense, and Organ Dysfunction in Adults With Severe HIV-Associated Tuberculosis in Uganda

BACKGROUND

The immunopathology of disseminated HIV-associated tuberculosis (HIV/TB), a leading cause of critical illness and death among persons living with HIV in sub-Saharan Africa, is incompletely understood. Reflective of hematogenously disseminated TB, detection of lipoarabinomannan (LAM) in urine is associated with greater bacillary burden and poor outcomes in adults with HIV/TB.

METHODS

We determined the relationship between detection of urine TB-LAM, organ dysfunction, and host immune responses in a prospective cohort of adults hospitalized with severe HIV/TB in Uganda. Generalized additive models were used to analyze the association between urine TB-LAM grade and concentrations of 14 soluble immune mediators. Whole-blood RNA-sequencing data were used to compare transcriptional profiles between patients with high- vs. low-grade TB-LAM results.

RESULTS

Among 157 hospitalized persons living with HIV, 40 (25.5%) had positive urine TB-LAM testing. Higher TB-LAM grade was associated with more severe physiologic derangement, organ dysfunction, and shock. Adjusted generalized additive models showed that higher TB-LAM grade was significantly associated with higher concentrations of mediators reflecting proinflammatory innate and T-cell activation and chemotaxis (IL-8, MIF, MIP-1β/CCL4, and sIL-2Ra/sCD25). Transcriptionally, patients with higher TB-LAM grades demonstrated multifaceted impairment of antibacterial defense including reduced expression of genes encoding cytotoxic and autophagy-related proteins and impaired cross-talk between innate and cell-mediated immune effectors.

CONCLUSIONS

Our findings add to emerging data suggesting pathobiological relationships between LAM, TB dissemination, innate cell activation, and evasion of host immunity in severe HIV/TB. Further translational studies are needed to elucidate the role for immunomodulatory therapies, in addition to optimized anti-TB treatment, in this often critically ill population.

Development of a Novel Clinicomolecular Risk Index to Enhance Mortality Prediction and Immunological Stratification of Adults Hospitalized with Sepsis in Sub-Saharan Africa: A Pilot Study from Uganda

The global burden of sepsis is concentrated in sub-Saharan Africa (SSA), where epidemic HIV and unique pathogen diversity challenge the effective management of severe infections. In this context, patient stratification based on biomarkers of a dysregulated host response may identify subgroups more likely to respond to targeted immunomodulatory therapeutics. In a prospective cohort of adults hospitalized with suspected sepsis in Uganda, we applied machine learning methods to develop a prediction model for 30-day mortality that integrates physiology-based risk scores with soluble biomarkers reflective of key domains of sepsis immunopathology. After model evaluation and internal validation, whole-blood RNA sequencing data were analyzed to compare biological pathway enrichment and inferred immune cell profiles between patients assigned differential model-based risks of mortality. Of 260 eligible adults (median age, 32 years; interquartile range, 26-43 years; 59.2% female, 53.9% living with HIV), 62 (23.8%) died by 30 days after hospital discharge. Among 14 biomarkers, soluble tumor necrosis factor receptor 1 (sTNFR1) and angiopoietin 2 (Ang-2) demonstrated the greatest importance for mortality prediction in machine learning models. A clinicomolecular model integrating sTNFR1 and Ang-2 with the Universal Vital Assessment (UVA) risk score optimized 30-day mortality prediction across multiple performance metrics. Patients assigned to the high-risk, UVA-based clinicomolecular subgroup exhibited a transcriptional profile defined by proinflammatory innate immune and necroptotic pathway activation, T-cell exhaustion, and expansion of key immune cell subsets including regulatory and gamma-delta T cells. Clinicomolecular stratification of adults with suspected sepsis in Uganda enhanced 30-day mortality prediction and identified a high-risk subgroup with a therapeutically targetable immunological profile. Further studies are needed to advance pathobiologically informed sepsis management in SSA.

HIV infection drives pro-inflammatory immunothrombotic pathway activation and organ dysfunction among adults with sepsis in Uganda

BACKGROUND

The global burden of sepsis is concentrated in high HIV-burden settings in sub-Saharan Africa (SSA). Despite this, little is known about the immunopathology of sepsis in persons with HIV (PWH) in the region. We sought to determine the influence of HIV on host immune responses and organ dysfunction among adults hospitalized with suspected sepsis in Uganda.

DESIGN

Prospective cohort study.

METHODS

We compared organ dysfunction and 30-day outcome profiles of PWH and those without HIV. We quantified 14 soluble immune mediators, reflective of key domains of sepsis immunopathology, and performed whole-blood RNA-sequencing on samples from a subset of patients. We used propensity score methods to match PWH and those without HIV by demographics, illness duration, and clinical severity, and compared immune mediator concentrations and gene expression profiles across propensity score-matched groups.

RESULTS

Among 299 patients, 157 (52.5%) were PWH (clinical stage 3 or 4 in 80.3%, 67.7% with known HIV on antiretroviral therapy). PWH presented with more severe physiologic derangement and shock, and had higher 30-day mortality (34.5% vs. 10.2%; P  < 0.001). Across propensity score-matched groups, PWH exhibited greater pro-inflammatory immune activation, including upregulation of interleukin (IL)-6, IL-8, IL-15, IL-17 and HMGB1 signaling, with concomitant T-cell exhaustion, prothrombotic pathway activation, and angiopoeitin-2-related endothelial dysfunction.

CONCLUSIONS

Sepsis-related organ dysfunction and mortality in Uganda disproportionately affect PWH, who demonstrate exaggerated activation of multiple immunothrombotic and metabolic pathways implicated in sepsis pathogenesis. Further investigations are needed to refine understanding of sepsis immunopathology in PWH, particularly mechanisms amenable to therapeutic manipulation.

Performance of the Universal Vital Assessment (UVA) mortality risk score in hospitalized adults with infection in Rwanda: A retrospective external validation study

Background

We previously derived a Universal Vital Assessment (UVA) score to better risk-stratify hospitalized patients in sub-Saharan Africa, including those with infection. Here, we aimed to externally validate the performance of the UVA score using previously collected data from patients hospitalized with acute infection in Rwanda.

Methods

We performed a secondary analysis of data collected from adults ≥18 years with acute infection admitted to Gitwe District Hospital in Rwanda from 2016 until 2017. We calculated the UVA score from the time of admission and at 72 hours after admission. We also calculated quick sepsis-related organ failure assessment (qSOFA) and modified early warning scores (MEWS). We calculated amalgamated qSOFA scores by inserting UVA cut-offs into the qSOFA score, and modified UVA scores by removing the HIV criterion. The performance of each score determined by the area under the receiver operator characteristic curve (AUC) was the primary outcome measure.

Results

We included 573 hospitalized adult patients with acute infection of whom 40 (7%) died in-hospital. The admission AUCs (95% confidence interval [CI]) for the prediction of mortality by the scores were: UVA, 0.77 (0.68–0.85); modified UVA, 0.77 (0.68–0.85); qSOFA, 0.66 (0.56–0.75), amalgamated qSOFA, 0.71 (0.61–0.80); and MEWS, 0.74 (0.64, 0.83). The positive predictive values (95% CI) of the scores at commonly used cut-offs were: UVA >4, 0.35 (0.15–0.59); modified UVA >4, 0.35 (0.15–0.59); qSOFA >1, 0.14 (0.07–0.24); amalgamated qSOFA >1, 0.44 (0.20–0.70); and MEWS >5, 0.14 (0.08–0.22). The 72 hour (N = 236) AUC (95% CI) for the prediction of mortality by UVA was 0.59 (0.43–0.74). The Chi-Square test for linear trend did not identify an association between mortality and delta UVA score at 72 hours (p = 0.82).

Conclusions

The admission UVA score and amalgamated qSOFA score had good predictive ability for mortality in adult patients admitted to hospital with acute infection in Rwanda. The UVA score could be used to assist with triage decisions and clinical interventions, for baseline risk stratification in clinical studies, and in a clinical definition of sepsis in Africa.