Introduction

Infection in critical care medicine

Infectious pathology represents a leading cause of admission to the intensive care units (ICU). Sepsis (defined by the presence of a dysregulated host response to infection inducing organ dysfunction) is present in up to 30% of all ICU patients, as recently reported by Sakr et al. in a large study with 10,000 patients from 730 ICUs [1]. One of the leading causes of sepsis is severe community acquired pneumonia (sCAP) of bacterial or viral origin [2]. Current Coronavirus disease 2019 (COVID-19) pandemics has largely boosted the cases of sCAP all over the world.

In turn, infection is one of the most frequent complications in patients who are critically ill. Compromise of body’s physical barriers by invasive devices, surgical aggression or traumatic injury, disruption of the mucosa, pressure sores, ventilator-induced lung injury, immune suppression, poor nutritional state, the use of broad-spectrum antibiotics which alter the commensal microbiota, combined with the increased exposition to opportunistic (often multi-drug resistant, MDR) pathogens [3], all represent predisposing factors favouring ICU acquired infections [4]. In fact, approximately 19.2% of ICU patients develop infections compared to approximately 5.2% of infections developed by patients staying in all other hospital wards [3, 5]. Ventilator-associated pneumonia (VAP) affects 10–25% of all ventilated patients after at least 48 h on mechanical ventilation [6]. Other frequent nosocomial infections affecting critically ill patients are catheter-associated urinary tract infection, bloodstream infection (BSIs), skin and wound infections, sinusitis, and gastrointestinal infection (often with Clostridium difficile) [4]. Clinical management of these infectious diseases or complications of the critically ill patient faces several challenges, including early diagnosis with microorganism identification, severity stratification, prognosis assessment and treatment guidance. Digital polymerase chain reaction (dPCR) is a next-generation PCR method that represents an opportunity to address these challenges.

dPCR: technical principles and applications

dPCR has emerged as a promising technology that might fill in the current gaps of other standard or emerging diagnostic technologies employed in microbiology (Table S1 – additional file 1). dPCR is based on the division of the PCR mastermix (all components including DNA or RNA targets) into thousands of partitions. PCR amplification of target genes occurs in each individual partition, acting as an individual microreactor [7]. These partitions can be created using a number of different mechanisms, such as emulsified microdroplets suspended in oil (droplet digital PCR, ddPCR), manufactured microwells, or microfluidic valving [8]. The distribution of target sequences in the partitions is detected by fluorescence at endpoint. Quantification of target genes is estimated based on Poisson’s distribution, by calculating the ratio of positive partitions (presence of fluorescence) over the total number of partitions [9]. This technology has several advantages: i) it is less affected by PCR inhibitors than other standard or real-time PCR (qPCR) methods, as target sequences are concentrated in the microreactors; ii) it also offers a high reproducibility of the results; iii) it provides an absolute quantification of the target sequence without the need for standard curves; and iv) it has an improved analytical sensitivity ideal for detecting microbial genes, for species identification or for genes conferring antimicrobial resistance or higher pathogenicity. dPCR also presents some limitations: i) it is unable to distinguish between viable and non-viable microorganisms (an inconvenient which affects all PCR-based methods); ii) it might have different sensitivity for different types of microorganisms; iii) it needs specialized training; and iv) it has a high cost, particularly to acquire the devices. This represents a major drawback for applications in low or middle income countries, for example during the COVID-19 pandemics [7, 9]. In spite its limitations, the previously mentioned dPCR properties make it an ideal tool for clinical applications in the field of microbiology and infectious diseases [7]. In this article we reviewed the existing evidence on the use of dPCR to improve the clinical management of infection in critical care medicine.

Methods

We searched PudMed combining the following MESH terms: “digital PCR”, “digital droplet PCR”, “droplet digital PCR” and “droplet PCR” with “ICU”, “critical AND infection”, “critically ill AND infection”, “severe infection”, “critical care”, “pneumoniae”, “ventilator-associated pneumoniae”, “ventilator”, “ventilation”, “sepsis”, “septic shock”, “bloodstream infections”, “skin and soft tissue infections”, “necrotizing fasciitis”, “peritonitis”, “invasive pulmonary aspergillosis”. We found a total of 487 PubMed articles. Only articles in English were considered. Using PMID we excluded duplicated articles appearing in more than one search, obtaining a total of 198 articles. We screened these articles for relevance and excluded 166 for the reasons mentioned in Fig. 1. Finally, thirty-two articles were included in this review (Table 1). These articles were further divided in two groups, one focused on the pathogens—diagnosis of infection, prognosis and treatment guidance (n = 23)—and the other focused on the host response to infection (n = 9).

Fig. 1
figure 1

PRISMA diagram

Full size image
Table 1 Articles included in the review depicting the applications of dPCR for infection diagnosis and management in critical care medicine
Full size table

Evidence on the use of dPCR for the diagnosis and management of infection in critical care medicine (Fig. 2)

Fig. 2
figure 2

Summary of the existing evidence on the applications of dPCR in the field of infection in Critical Care Medicine

Full size image

Applications of dPCR targeting microbial genes

The gold standard for the detection of bacterial and fungal pathogens still relies on culture based methods that present a long turnaround time, and often yield low positivity rates [10]. For viral pathogens the reference method is frequently qPCR [11,12,13]. As previously stated, dPCR presents several advantages making it an ideal technique for the detection and quantification of microbial genes (Additional file 1: Table S1).

Bacterial identification in blood or plasma

dPCR has been used in critical care medicine for the detection of different bacterial pathogens in septic patients or patients with a suspected BSI [10, 14,15,16]. Yamamoto et al. successfully diagnosed a septic patient with a Mycobacterium tuberculosis (MTB) disseminated infection by detecting MTB complex-specific sequences in total cell-free DNA (cfDNA) in plasma. dPCR was employed after sputum, urine, and blood samples all tested negative by COBAS TaqMan MTB and MAI tests (Roche Diagnostics) and TSPOT.TB test (Oxford Immunotec) and mycobacterial culture [14]. These results indicate that dPCR is more sensitive than the other molecular and culture methods, and that dPCR could serve as a less invasive diagnostic tool for MTB infections [14]. In two other studies [15, 16] dPCR was used to detect major BSI Gram-negative pathogens in cfDNA isolated from plasma of critically ill patients. Shin et al. [15] developed a dPCR assay able to detect four major Gram-Negative pathogens and four common antimicrobial resistance (AMR) genes. Zheng et al. [16] work focused on only two of the most common MDR Gram-Negative pathogens. These assays report a time from sample collection to result of three to four hours and a detection limit of one Colony-forming Unit/ml of bacteria in the blood [15, 16]. The study by Shin et al. indicated that dPCR was also more sensible than qPCR [15]. Hu et al. [10] compared the detection of pathogens and AMR genes by dPCR with metagenomic Next-Generation Sequencing (mNGS) and with blood culture using samples from a cohort of septic patients with suspicion of BSIs. dPCR showed a great potential to identify the pathogens most commonly associated with BSIs as well as AMR genes, as it was faster and more sensible than mNGS and blood culture [10]. In these previous studies [10, 15, 16], clinical validation revealed that dPCR method was superior to blood culture in terms of specificity, sensitivity, and turnaround time, representing a promising method for the early and accurate diagnosis of BSIs [10, 15, 16]. Our group has established a ddPCR assay capable of detecting and quantifying the housekee** genes of important nosocomial bacterial species in ICUs, such as Klebsiella pneumoniae, Escherichia coli and Staphylococcus aureus [17]. These assays are compatible with duplexing, show low replication variability and very low limit of detection (less than 1 pg of DNA) (Fig. 3). Ongoing work is focused on applying the developed ddPCR assays directly to blood and designing a panel that allows testing as many samples at the same time as possible.

Fig. 3
figure 3

dPCR detection assay for E. coli, K. pneumoniae and S. aureus. A ddPCR workflow; B ddPCR results displayed as droplets of different fluorescence amplitude; C copy number of E. coli, S. aureus and K. pneumoniae in different DNA of different initial DNA concentration

Full size image

The main limitation of dPCR compared with blood culture or mNGS, but not with qPCR, is that it can only detect the pathogens included in the dPCR panels. Nevertheless, the results obtained support that early identification of MDR pathogens by dPCR can improve treatment outcomes [10, 15, 16].

Fungal identification in blood

dPCR has been also tested for the detection of candidemia [42] quantified bacterial DNA load by dPCR (through 16S rDNA), to evaluate if the bacterial density in an ICU environment influenced the establishment of the microbiome in hospitalized premature infants. The authors showed that bacterial DNA load and diversity varied between surfaces. Room-specific microbiome signatures were detected, suggesting that the microbes seeding ICU surfaces are sourced from reservoirs within the room, possible sha** hospitalized infants gut microbiome [42].

Applications of dPCR targeting host response

Results coming from high-throughput technologies such as microarrays or next-generation sequencing, which are able to analyse the entire human transcriptome, have revealed the existence of specific host response signatures potentially useful to improve diagnosis [43], severity stratification and prognosis assessment [44, 45] of severe infection. dPCR is making real the promise of translating these signatures into the clinical practice. A work from our group was pioneer in exploring the potential use of dPCR to diagnose sepsis, evidencing that the gene expression ratio between the constant region of the mu heavy chain of IgM and CD20 yielded an area under the receiver operating curve (AUROC) of 0.72 to differentiate sepsis from systemic inflammatory response syndrome (SIRS) [46]. In turn, Almansa et al. evidenced that the transcriptomic ratios between matrix metalloproteinase-8 (MMP8) or Lipocalin 2 (LCN2), which are two genes coding for the proteins contained in the neutrophil granules, with the major histocompatibility complex class II, DR alpha molecule (HLA-DRA), yielded AUROCs > 0.89 to differentiate between sepsis and SIRS [47]. The combination of gene expression profiling by dPCR with standard biomarkers commonly used in the clinical practice is also an exciting avenue of research, already explored in another work from Almansa et al., which evidenced that the combination of procalcitonin and HLA-DRA expression levels outperformed the former biomarker to detect sepsis. In this work, procalcitonin yielded an AUROC of 0.80 and the ratio Procalcitonin/HLA-DRA of 0.85 [48]. In a small study, Link et al. explored the potential of microRNA (miRNA) expression quantification by dPCR to diagnose sepsis. These authors found that miR-26b-5p yielded an AUROC of 0.80 to differentiate critically ill patients with sepsis from those with no sepsis [49]. dPCR has been also employed to evaluate the magnitude of the biological processes occurring during sepsis that are difficult to quantify with the currently available methods. For example, in a cohort of patients with infection, sepsis or septic shock, Martin-Fernandez et al. evidenced a progressive increase in the expression levels of emergency granulopoiesis related genes with severity [50]. Another signature of sepsis and sCAP is the depressed expression of those genes involved in the immunological synapse between antigen-presenting cells and T cells [51]. Menéndez et al. demonstrated that dPCR is an useful method to evidence the depressed expression of three of these genes [HLA-DRA, CD40 Ligand (CD40LG) and CD28] in patients with CAP presenting with organ failure [52], while Almansa et al. evidenced that profiling the expression levels of immunological synapse genes was also useful to identify VAP, yielding AUROCs of 0.82 for CD40LG, 0.79 for inducible T cell costimulator (ICOS), 0.78 for CD28 and 0.74 for CD3E [53]. Regarding prognosis, Busani et al. used dPCR to evidence the potential role of mitochondrial DNA as predictor of mortality in patients with septic shock due to MDR bacteria [54]. In turn, Almansa et al. showed that gene expression levels of HLA-DRA quantified by dPCR were an independent predictor of mortality in sepsis [47]. Cajander et al. had already proposed to use expression levels of this gene to identify those sepsis patients that could benefit from immunostimulatory drugs [55]. More recently, using dPCR, Bruneau et al.revealed that expression levels of a circulating ubiquitous RNA (RNase P) correlated with disease severity, invasive mechanical ventilation status and survival in patients with COVID-19 [40]. Also in severe COVID-19, Sabbatinelli et al. found that low levels in plasma of the inflamm-aging associated miRNA miR-146a were associated with no response to tocilizumab [56]. Metabolomics is a relatively new “-omic” that has demonstrated its use in the management of septic patients (e.g. lactate). The combination of this “-omics” approach with the transcriptomics markers measured by dPCR might be useful to determine patient severity, predict the need for mechanical ventilation and mortality [57].

Conclusions

This review evidences the potential of dPCR as a useful tool that could contribute to improve the diagnosis and clinical management of infection in critical care medicine. Although most of the published works consist of pilot/ exploratory studies, they show the potential of dPCR, supporting the development of further, larger studies aimed to validate the use of this technology in this field.