Utility of laboratory biomarkers as mortality predictors for melioidosis: A systematic review

INTRODUCTION

Melioidosis is a frequently overlooked and often severe disease caused by the Gram-negative soil and water saprophyte Burkholderia pseudomallei. It is a significantly fatal re-emerging infectious disease with a high burden in Southeast Asia and Northern Australia.^[1] An estimated 165,000 cases and 89,000 deaths due to melioidosis occur annually worldwide.^[2] In lieu of its substantial impact on both the healthcare system and economies of low- and middle-income countries, combined with a mortality rate higher than many recognized neglected tropical diseases (NTDs), recent calls are being made to the World Health Organization to officially recognize melioidosis as an NTD.^[3]

Epidemiological trends suggest a risk of spread to non-endemic regions, with current data from Africa and America reflecting a potential under-reporting of the disease in these regions.^[4] The Centers for Disease Control recently reported the first confirmation of the presence of B. pseudomallei in environmental soil and water samples in the Gulf Coast Region of Mississippi.^[5] Increased global travel, environmental changes, potential zoonotic spread, and rising rates of comorbidities like diabetes may enhance its geographical spread and emergence in new areas.^[6]

Clinically, melioidosis presents as a diverse spectrum of illnesses, ranging from localized skin infections and abscesses to fulminant septicemia, pneumonia, and even neurological manifestations. It is often referred to as the “great mimicker” due to its immense range of non-specific manifestations, such as localized abscesses, septicemia, pneumonia, septic arthritis, osteomyelitis, and encephalomyelitis.^[7] Difficulty in diagnosis, combined with mortality rates as high as 30-50% associated with melioidosis, underscores the urgent need to identify factors that can help predict prognosis and mortality outcomes.^[7,8]

The high mortality rates and diverse clinical presentations have attracted significant interest in understanding the factors that may help predict mortality outcomes in melioidosis patients. Numerous studies have investigated a diverse array of risk factors for mortality in melioidosis. Routine laboratory-based investigations are among the many factors that have shown promise as mortality predictors in melioidosis patients. However, a systematic analysis of these numerous potential biochemical markers is lacking.

This systematic review aimed to identify consistent biochemical predictors of mortality among patients with melioidosis and to qualitatively summarize the findings.

MATERIALS AND METHODS

RESULTS

Search results

Database searches using the pre-determined search string yielded 987 records in total (220 from PubMed, 617 from Scopus, and 150 from Embase). After eliminating 50 duplicates, 937 unique entries were screened. Initial screening of titles and abstracts led to the exclusion of 894 studies. Full-text analysis was performed on 43 articles, of which 34 did not meet eligibility criteria. No further studies were found on the evaluation of the references of the individual studies. The process has been illustrated in the PRISMA flow diagram [Figure 1].

Close

Figure 1:

Preferred Reporting Items for Systematic Reviews and Meta-analysis flowchart for the study.

Export to PPT

All nine studies that met the inclusion criteria underwent quality evaluation using the Joanna Briggs Institute tool. Each was found to be of adequate quality for inclusion in the final synthesis.

DISCUSSION

Melioidosis has been reported to have one of the highest mortality rates among emerging tropical diseases, with a possible threat of expansion to non-endemic regions. This underscores the need to evaluate factors influencing mortality among these patients. This study was conducted with the aim of identifying basic laboratory investigations that can serve to identify melioidosis patients at a higher risk of death. The routine and inexpensive nature of these investigations can help make them easily translatable into clinical practice for effective risk prediction.

In our review, we identified hypoalbuminemia at presentation, commonly defined as serum albumin <3 g/dL, as the most commonly reported mortality risk factor across the literature. Hypoalbuminemia as a risk factor for mortality has been reported in other infectious diseases as well. Several authors have reported increased mortality among severe acute respiratory syndrome coronavirus, two patients with hypoalbuminemia.^[20-24] Serum albumin concentrations also were shown to be an independent predictor of poor outcomes among patients admitted to the emergency department for an infection.^[25] Interestingly, similar to our findings, both bicarbonate and albumin were found to predict in-hospital mortality in Clostridioides difficile infection using a machine learning approach.^[26]

Mechanistically, the production of albumin, a negative phase reactant, is decreased during inflammatory processes as a result of an increase in levels of cytokines such as interleukin (IL)-1, IL-6, and tumor necrosis factor-alpha.^[27] Acute hypoalbuminemia is most commonly triggered by infections, resulting from a combined effect of decreased synthesis, increased tissue utilization, and increased transcapillary leakage.^[28] The non-oncotic properties of albumin, such as its anti-oxidative nature, are hypothesized to be of critical importance in our body’s defense against microbes.^[29-31] Systemic inflammation associated with infections can significantly alter albumin dynamics, impairing immune responses and exacerbating cytokine-driven immune dysregulation, ultimately impacting prognosis and clinical outcomes.^[32]

The second most commonly implicated laboratory parameter was low serum bicarbonate at presentation. Low serum bicarbonate has been shown to independently predict acute kidney injury (AKI) in intensive care unit (ICU) patients.^[33] Among melioidosis patients, AKI has been reported to occur in as many as 36% of patients, likely due to pre-renal or acute tubular necrosis. At the same time, the need for intensive care and mortality rates have both been shown to be higher in melioidosis patients with AKI as compared to those without AKI.^[34] This complex interplay between melioidosis, AKI, and serum bicarbonate levels may be a possible mechanistic explanation for the increased mortality rates in patients with low serum bicarbonates.

These batteries of investigations, combined with epidemiological and clinical risk factors, may be used to predict comprehensive scoring systems to predict prognosis. The varied clinical presentations underscore the importance of developing effective scoring systems for melioidosis to help clinicians make informed decisions. Gassiep et al.^[17] used a scoring system by Cheng et al.^[19] to predict melioidosis mortality, where they scored patients 0-11 based on factors such as pneumonia, age, and serum markers. Cheng et al.^[19] found 8.6% mortality for scores ≤3 and 44.6% mortality for scores ≥4.^[19] Gassiep et al.^[17] reported 32% mortality for scores >3 and 9% for ≤3, with high sensitivity (85%) and negative predictive value (91%) but low specificity (47%) and positive predictive value (32%). Thus, this system effectively identifies low-risk patients but is less accurate for high-risk predictions.

Recently, Chayangsu et al.^[11] in their retrospective study included a quick SOFA score of ≥2, abnormal chest X-ray findings, elevated creatinine (≥1.5 mg/dL), elevated aspartate aminotransferase (≥50 U/L), and low bicarbonate (≤20 mEq/L) for the prediction of mortality. The model demonstrated strong discrimination between survivors and non-survivors, with a receiver operating characteristic curve exceeding 0.80, underscoring its potential for early risk assessment in clinical practice.

A major limitation of our study is the wide heterogeneity observed in the reported outcomes (ORs/RRs, HRs), which prevented a formal meta-analysis and pooling of data in the current study. It is imperative to understand the various sources that may have given rise to heterogeneity in our studies before being able to draw any meaningful conclusions on the utility of laboratory biomarkers for predicting melioidosis mortality.

Varying study designs may have been a major source of heterogeneity. In the context of laboratory biomarkers, the timing of collection of the samples is crucial. While most of the studies included in our review evaluated the laboratory results at baseline/admission, the study by Menon et al.^[15] did not mention the timing of sample collection, thereby preventing a pooled analysis without clarity on the time of sample collection. Another area where study designs varied was the timelines for mortality. While the majority of the studies being retrospective in nature evaluated in-hospital mortality, the study by Nisarg et al.^[12] explicitly mention that they have retrospectively evaluated risk factors for 28-day mortality. Such variation in the mortality definition again hinders a pooled analysis without significant heterogeneity.

A major contribution to the heterogeneity of the findings can also be attributed to the heterogeneous nature of the presentations that melioidosis presents with. Being a “great mimicker,” identifying a single laboratory parameter as a mortality risk for the entire spectrum of presentations in melioidosis might not be the best approach. Each study population differs in its underlying composition with respect to various factors such as underlying population distribution, comorbidities, clinical presentations, and baseline risk factors. Analysis of risk factors while taking into account the predominant clinical presentation, such as pulmonary melioidosis, renal melioidosis, and central nervous system melioidosis, might be a more effective approach. However, the present systematic review did not succeed in achieving the same due to the paucity of data available in the reported literature.

Another significant limitation of our study was the nature of the included studies. All included studies were retrospective in nature. A true identification of risk factors would need prospective studies that evaluate the utility of these laboratory investigations in predicting prognosis, not just in terms of mortality outcomes but overall disease severity, including ICU admissions, septicemia, relapses, and recurrences. There was no analysis of follow-up data.

Another limitation is that our findings may be subject to publication bias. Several investigators may not have reported their results if the findings were not significant for mortality. Our report thus highlights the direction for future researchers to pursue while evaluating predictors of mortality in melioidosis.

While the findings from our study may not be ideal or definitive, they do provide a significant direction for future research to focus on and highlight several lacunae that need to be addressed.

CONCLUSIONS

Hypoalbuminemia at presentation emerged as the most consistently reported biochemical risk factor for in-hospital mortality, followed by low serum bicarbonate at presentation.