Point-of-Critical-Care Diagnostics (POCC-Dx) for Sepsis Enabled by Multiplexed Micro-Nano Sensing Technologies

Abstract

Sepsis is responsible for the highest economic and mortality burden in critical care settings around the world, prompting the World Health Organization in 2018 to designate it as a global health priority. Despite its high universal prevalence and mortality rate, a disproportionately low amount of sponsored research funding is directed towards diagnosis and treatment of sepsis, when early treatment has been shown to significantly improve survival. Additionally, current technologies and methods are inadequate to provide an accurate and timely diagnosis of septic patients in multiple clinical environments. For improved patient outcomes, a comprehensive immunological evaluation is critical which is comprised of both traditional testing and quantifying recently proposed biomarkers for sepsis. There is an urgent need to develop novel point-of-care, low-cost systems which can accurately stratify patients. These point-of-critical-care (POCC) sensors should adopt a multiplexed approach utilizing multi-modal sensing for heterogenous biomarker detection. For effective multiplexing, the sensors must satisfy criteria including rapid sample to result (STR) delivery, low sample volumes for clinical sample sparring, and reduced costs per test. A compendium of currently developed multiplexed micro and nano (M/N) based diagnostic technologies for potential applications towards sepsis are presented. We have also explored the various biomarkers targeted for sepsis including immune cell morphology changes, circulating proteins, small molecules, and presence of infectious pathogens. An overview of different M/N detection mechanisms are also provided, along with recent advances in related nanotechnologies which have shown improved patient outcomes and perspectives on what future successful technologies may encompass.

Graphical/Visual Abstract and Caption

To reduce mortality from common yet complicated septic infections, micro and nano technologies which target multiple sepsis biomarkers can provide faster, more accurate, and personalized diagnosis to support early treatment

1.0. INTRODUCTION

1.1. Impact and need for sepsis-related research

Sepsis is defined as an over-compensatory bipolar immune response to infection, potentially damaging multiple organs and is highly fatal for the elderly, immunocompromised, and neonates. Once diagnosed, septic individuals have days or sometimes hours to obtain proper treatment usually in the form of antibiotics, supplemental oxygen, and intravenous fluids before permanent internal injury occurs (Polat et al., 2017). This immune-system dysfunction is commonly caused by contracting an infection, with sources such as bacteria, viruses, or fungi, making sepsis stratification difficult based on the variance in origin and corresponding indicative symptoms and biomarkers to analyze (Pierrakos & Vincent, 2010). In hospital settings, many septic infections arise from contaminated surgical procedures or hospital-borne pathogens, making sepsis the highest cause of hospital-related deaths across the globe, e.g. 1 in 3 hospital deaths are sepsis-related (Fleischmann et al., 2015).

According to the Center for Disease Control (CDC), over 1.7 million people in the United States are diagnosed with sepsis each year, and 270,000 of them succumb to the disease (Rhee et al., 2019). Sepsis impacts more people than any other high-mortality disease including strokes, cancer, heart disease, and HIV as shown in Table 1 (Benjamin et al., 2017; HIV-Incidence-Fact-Sheet_508.Pdf, n.d.; Mariotto et al., 2011; Rhee et al., 2019). Sepsis contributes around $23.6US billion for healthcare burdens in the United States, due to high disease contraction in hospitals, and longer stay in hospitals intensive care units (Torio & Moore, 2016). Alarmingly, research funds directed towards understanding, diagnosing, and treating sepsis remain orders of magnitude lower compared with its high disease counterparts, with such disproportion visualized by Figure 1. For example, HIV, a disease thoroughly documented both in mechanisms and treatment, receives almost 40 times more state-funded research, yet septic individuals are 13 times more likely to succumb to the disease. The ratio for healthcare burden to research funding is therefore highest for sepsis and lowering it must be prioritized to improve survival chances in critical care settings.

Table 1:

Statistics on sepsis relative to other high mortality diseases

In the United States	New cases per year	Healthcare Burden (HB)	Government-funded research (RF)	Ratio of HB to RF	References
Sepsis	1,700,000	$23.6 B	$0.07 B	330.6	(Rhee et al., 2019), (Torio & Moore, 2016), (What Is Sepsis?, n.d.)
Stroke	795,000	$33.9 B	$1.80 B	18.8	(Benjamin et al., 2017), (NINDS 2020 Congressional Budget Justification | National Institute of Neurological Disorders and Stroke, n.d.)
Cancer	1,000,000	$173.0 B	$6.44 B	26.9	(Mariotto et al., 2011), (Edwards et al., 2014), (NCI Budget and Appropriations, 2020)
Heart Disease	1,055,000	$316.1 B	$1.32 B	239.5	(Benjamin et al., 2017), (Clemens, 2019)
HIV	38,000	$36.4 B	$2.67 B	13.6	(HIV-Incidence-Fact-Sheet_508.Pdf, n.d., p.), (CDC FACT SHEET: Today’s HIV/AIDS Epidemic, n.d.), (“U.S. Federal Funding for HIV/AIDS,” 2019)

Figure 1:

Bar graph representing incidence of common diseases (in incidence per 100,000 people per year) related to US-Dollars spent for state funded research (per 1 billion USD). As indicated, sepsis is the most common disease yet receives significantly less funding from government research projects

In clinical settings, the current workflow follows a systematic procedure based on the technology and personnel available to triage septic patients (Figure 2a). At hospitals, if patients display early signs of sepsis (i.e., SIRS criteria), antibiotics are delivered targeting a myriad of infection sources before diagnosis. This may help improve patient survival, but such antibiotic regiments have low efficiency and contributes to antibiotic resistance (Khan et al., 2019; Pradipta et al., 2013; Reddy et al., 2018; Sharma & Srivastava, 2016). During this time, blood is collected to identify the responsible pathogen(s) and refine the antibiotic prescription for the patient. These clinical tests include a bacterial culture, gram-staining, and a drug resistance test to ensure certain drugs will effectively reduce pathogen propagation (Khan et al., 2019; Reddy et al., 2018; Shrestha et al., 2007). Normally, these tests provide results days after a patient is admitted, which may result in missing a critical window for accurate patient diagnosis and strategizing effective therapeutics. This may be the most important aspect for improving patient survival.

Figure 2:

Representations of sepsis progression through function and management schemes. (a) Current workflow chart for sepsis in critical care settings. Left represents the course of action while right indicates the time expected to perform that action. (b) Sepsis evolution, from systemic inflammatory response syndrome (SIRS) to sepsis to septic shock. Each phase has different patient outcomes (blue) and use different biomarkers for determination (green). (c) Physiological mechanisms for sepsis, where the body’s inflammation response (green) is initially balanced with self immunosuppression (red). During sepsis, a cascading effect starts with excessive inflammation (SIRS), followed by compensatory anti-inflammation response (CARS), and ends with positive feedback for both conditions (septic shock). At this stage, mortality rates are significantly higher

According to Kumar et al., every hour without treatment for sepsis mortality increases by 8% and 80% of these deaths may be preventable with expedient diagnosis and treatment (Kumar et al., 2006). Current rapid detection standards include vital checks such as respiration, low blood pressure, and changes in mental status; conglomerated as the quick sequential organ failure assessment score (qSOFA). While relatively simple to evaluate, it lacks sensitivity for detecting sepsis during early stages and these parameters are subject to change with many confounding diseases (Marik & Taeb, 2017). Along with other operations for quantifying sepsis-related biomarkers, such as flow cytometry and lactate tests, a general trend is they inadequately provides diagnosis for most setting as they are frequently expensive, require large biofluid draws from patients, have limited detection range, or complicated results to interpret which only delay prognosis from trained clinicians (Hein-Kristensen et al., 2009; Mukhopadhyay et al., 2017; Umlauf et al., 2013; Venet et al., 2011). Few commercial POCC sepsis technologies are currently available (Table 2), and while some of them have shown improved performance and accuracy in clinical studies, none of them have multiplexing capabilities beyond their directed primary biomarker class (i.e., pathogens versus cell receptors versus proteins) which limits their potential to get a more comprehensive sepsis immunological profile of a patient (Reddy et al., 2018). Additionally, other limitations persist, such as higher biofluid volumes requirement (i.e., PCR-based commercial systems like LiDia® and T2Dx®) and costs for the devices are still expensive (i.e., the i-STAT 1 Wireless unit costs over $18,000 US), hindering their ability to be used in resource limited setting worldwide. There lies a need for new and improved detection systems that are inexpensive, accurate, and accessible for any individual requiring fast sepsis diagnosis. The best options for improving these strategies lies in measuring multiple septic biomarkers simultaneously using novel multiplexed micro and nano (M/N) systems.

Table 2:

Notable available POCC products for sepsis biomarker detection

Device	Detection mechanism	Biomarker measuring	Biofluid volume	Assay time
i-STAT 1 Wireless	Electrochemical	Small molecules	17–95 μL	2–10 mins
LABGEOIB10	Immunoassay	Proteins	500 μL	20 mins
LiDia®	PCR	Pathogens	10 mL	3–4 hours
Moxi GO™ II	Coulter/flow cytometry	Cell receptors/cell counts	60 μL	< 1 min
T2Dx®	PCR	Pathogens	4 mL	3–5 hours
Wolf® Cell Sorter	Fluorescence activated cell sorting (FACS)	Cell receptors/cell counts	0.15–5 mL	1 hour

1.2. Stages of sepsis and recent definition changes

The phases and definition for sepsis has gone through multiple iterations over the past decades, with the most recent rebranding occurring in 2016. Before then, four stages defined sepsis-related conditions: the systemic inflammatory response syndrome (SIRS), sepsis, severe sepsis, and septic shock. However, severe sepsis was absorbed under sepsis as it is already a severe disease and removed due to redundancy. Different classifications for SIRS were adopted as well, measured using the qSOFA method (Marik & Taeb, 2017). The progression and detailed identities for sepsis-related disease are visualized by Figure 2b (Singer et al., 2016).

During the systemic inflammatory response syndrome phase, early signs of sepsis are realized, which must include two of the following symptoms (Bone et al.): a body temperature above 38°C or below 36°C, a heart rate above 90 beats per minute (bpm), a respiratory rate (RR) above 20 breaths per minute, a white blood cell count (WBC) greater than 12,000 per μL or less than 4,000 per μL, with more than 10% of those WBC’s showing immature bands (Bone et al., 1992). Under SIRS conditions, the body generates an unnecessarily larger inflammatory response relative to the current disease-burden in the body, unbalancing the body’s natural homeostasis (Figure 2c). At this stage, symptoms related to an infection are not present, can occur without an infection, and biomarkers to determine diagnosis include expression of transmembrane proteins on certain immune cells (Bone et al. 1992). Overtime, the body overcompensates for hyper-inflammation by reducing inflammation factor levels and upregulating immunosuppression factors, an effect coined as the compensatory anti-inflammation response syndrome (CARS) (Adib-Conquy & Cavaillon, 2009). A consequence of CARS is the body is much more susceptible to pathogen proliferation and harmful infections. Sepsis occurs when SIRS leads to CARS and produces such an infection. During sepsis progression, proteins and small molecules found in the blood may yield abnormal levels, while pathogen levels that caused the infection will begin to ramp up, and the organ which the infection originated in may begin to fail. Septic shock begins once multiple organs are affected and the individual experiences low blood pressure. Now, mortality significantly increases, and treatment such as blood pressure support, pathogen-targeting antibiotics, intravenous fluids, supplemental oxygen, and vasoconstricting drugs must be administered quickly to curtail symptoms, indicating the rapid need for early diagnosis (Polat et al., 2017). For each phase, different biomarkers are studied, with a classification of common biomarkers represented by Figure 3.

Figure 3:

Overview of potential biomarkers for point-of-critical care (POCC) sepsis diagnosis

2.0. SEPSIS BIOMARKERS USED IN POINT-OF-CRITICAL-CARE SENSORS

2.1. Cellular-related biomarkers

2.1.1. Membrane Proteins

Pro/anti-inflammatory response results in over and/or under expression of several cellular membrane proteins. As detailed in Table 3, macrophage-specific surface receptors for sepsis diagnosis include C-type lectin receptors and Toll-like receptors (TLR). Both recognize pathogenic ligands, with C-type lectin receptors targeting glycan with calcium-dependent carbohydrate-recognition domains, while TLR as a class of receptors that recognize pathogen-associated molecular patterns (PAMPs) such as lipids, proteins, and nucleic acids (Kawasaki & Kawai, 2014; Tang et al., 2018). In their response, receptor-mediated intracellular cascades promote adaptive immunity recruitment and stimulate the foreign-body response. They can be measured using immunofluorescence and immunohistochemistry, and higher levels of both point towards sepsis (Chiffoleau, 2018; Tsujimoto et al., 2008).

Table 3.

Common cellular related sepsis biomarkers

	Cell Type	Primary Function	Common Detection Methods	During Sepsis	References
Membrane proteins
C-type Lectin Receptor	Macrophages	Recognize pathogenic ligands	Fluorescence microscopy	↑	(Karsten et al., 2012), (Chiffoleau, 2018)
nCD64	Neutrophils	Ig binding	Fluorescence microscopy	↑	(Zeitoun et al., 2010), (Hassan et al., 2017)
mHLA-DR	Monocytes	Identifying foreign cells	Fluorescence microscopy	↓	(Zouiouich et al., 2017), (Monneret et al., 2011)
CTLA-4	T lymphocytes	Immunosuppression	Immunoassays, fluorescence	↑	(Chang et al., 2013), (Gao et al., 2015)
Toll-like receptor (TLR)	Macrophages	Recognize pathogenic ligands	Immunohistochemistry, immunofluorescence	↑	(Kawasaki & Kawai, 2014), (Ungaro et al., 2009), (Tsujimoto et al., 2008)
Cell Motility/Chemotaxis
Neutrophil Motility	Neutrophils	Phagocytosis	Optical monitoring, cell counting	↓	(Hassan et al., 2018), (Ellett et al., 2018)
CD66b	Neutrophils	Adhesion, migration, pathogen binding	Immunoassays, fluorescence	↑	(Martins et al., 2008), (Schmidt et al., 2012)
CD11b	Neutrophils & Lymphocytes	Adhesion	Immunoassays, fluorescence	↑	(Sheneef et al., 2017)
Cell Stiffness	Leukocytes	Rigidity, deformability	Atomic force microscopy, optical tweezers	↓	(Reddy et al., 2018), (Nishino et al., 2005)

Some receptors in sepsis diagnosis have high sensitivity by appearing on atypical cells. This includes neutrophilic CD64 (nCD64) as well as monocyte expression of human leukocyte antigen (mHLA-DR). For nCD64, neutrophils uniquely express CD64 during under inflamed conditions, making it a binary indicator for sepsis and functions to bind with immunoglobulins and contribute to pathogen phagocytosis (Velásquez et al., 2013, p. 64). Contrastingly, the reduced expression for mHLA-DR occur during immunosuppression, and is used to predict infections and septic behavior (Zouiouich et al., 2017). Both are usually quantified through fluorescence microscopy or flow cytometry, and yield opposite trends during sepsis as shown by Table 3 (Chiffoleau, 2018; Monneret et al., 2011).

Another immuno-suppressive biomarker is cytotoxic T-lymphocyte antigen-4 (CTLA-4), a costimulatory receptor in T cells for the human immune system. CTLA-4 is also co-stimulated with T-cell receptor (TCR), as well as binds with CD80 and CD86 on antigen presenting cells, with higher affinity than CD28 (Alegre et al., 2001). However, CTLA-4 under normal conditions is found intracellularly and not expressed on the membrane. High surface levels of CTLA-4 also correspond to immunosuppression, and thereby acts as a sepsis promoter when found in high amounts. Chang et al. reports these findings, developing antibodies to competitively bind with both CTLA-4 and PD-1 on T cell surfaces in mice with fungal infections (Chang et al., 2013).

2.1.2. Cell motility, chemotaxis, and stiffness

Typical neutrophil receptors such as CD11b and CD66b are both elevated during sepsis but represent different mechanisms for inflammation. Explicitly, CD11b may also be found on monocytes, natural killer cells, and macrophages, participating in cell adhesion and migration in the presence of CD18 (Jämsä et al., 2012). On the other hand, CD66b, also known as carcinoembryonic antigen-related cell adhesion molecule 8 (CEACAM8), is involved in adhesion but also in altering neutrophil migration patterns as higher levels (Schmidt et al., 2012). Both have similar detection methods such as fluorescence labels or antibody-specific immunoassays, and have only recently been studied for sepsis (Schmidt et al., 2012; Sheneef et al., 2017). Directly measuring physical and mechanical properties rather than biochemical allowing measuring sepsis from multiple angles. During sepsis, neutrophils exhibit discernable changes in motility, increasing mean distance traveled over time and reversing course, pausing, and oscillating more frequently than in healthy individuals due to pathogen pursuit during inflammation (Ellett et al., 2018; Hassan et al., 2018). These changes are commonly determined using optical tracking and gated-cell counting systems (Ellett et al., 2018). Similarly, it has been observed that leukocytes are more deformable, likely an effect for increasing motility (Nishino et al., 2005). Measuring changes in stiffness require force detection and is conducted through atomic force microscopy (AFM) and using optical tweezers (Nishino et al., 2005).

2.2. Soluble biomarkers in blood plasma

2.2.1. Proteins

Circulating proteins, or cytokines, play a critical role in inflammatory response of the septic patient. IL-6, Tumor necrosis factor alpha (TNF-α), and IL-1β are all pro-inflammatory cytokines upregulated during sepsis (Table 4). In particular, IL-6 is produced by many cells at lesion sites including T cells, hepatocytes, and fibroblasts, as well as osteocytes, triggering production of C-reactive protein, fibrinogen, haptoglobin and suppressing fibronectin, transferrin, and albumin synthesis (Tanaka et al., 2014). Comparatively, TNF-α is a cytokine related to leukocyte adhesion, as well as proinflammation and neutrophil activation (Kurt et al., 2007). In septic conditions, TNF-α is found at higher concentrations during SIRS, and inhibiting mechanisms which produce the cytokine improve sepsis survival (Dellinger, 1997). Finally, IL-1β, mainly produced by macrophages, is involved with activation and proliferation to phagocytose foreign bodies, as well as block apoptosis (Kurt et al., 2007). Currently, electrochemistry, fluorescence microscopy, and immunoassays quantify cytokine levels from blood, with respective heightened septic concentrations and typical LOD values shown in Table 4 (Kurt et al., 2007; Panacek et al., 2004; Pierrakos & Vincent, 2010). Procalcitonin (PCT) and C-reactive protein (CRP) are different stimulating proteins but universally studied for predicting sepsis (Faix, 2013). The precursor to calcitonin, PCT was recently observed for having heightened levels during bacterial infections, while CRP is stimulated from IL-6 by hepatocytes and indicates phagocytic activity and inhibits proteolysis (Thompson et al., 1999; Vijayan et al., 2017). Both proteins are typically measured through electrochemistry or immunoassays, although PCT has greater sepsis sensitivity, requiring lowering concentrations for diagnosis (Faix, 2013; Vijayan et al., 2017). Soluble CD28 (sCD28) is a costimulatory receptor with T cell receptor (TCR) that is involved with activating T-lymphocyte cells in conjunction of an infectious immune response (Alegre et al., 2001). Specifically, CD28 binds to CD80 and CD86 on antigen presenting cells, which when bound activates an intracellular signaling cascade to increase cytokine production from the T cell. This thereby increases immune activity and inhibits BCL-2 production, further improving T cell survival and participation. Studies have revealed that CD28 levels are significantly reduced on T cell membranes, increasing soluble forms in blood, which highly correlates to sepsis conditions (Ramachandran et al., 2015). Comparatively, methods have been researched which show reduced CD28 stimulation may signal improved conditions following sepsis and reduced death rates from bacterial infections. With current techniques, sCD28 can be measured through immunoassays, have a 3.0 pg/mL LOD, and during sepsis values may be greater than 400 pg/mL (Jedynak et al., 2018; Li et al., 2014).

Table 4.

Common sepsis related soluble molecule biomarkers

	Primary Function	Common Detection Methods	Detection Limits	Levels During Sepsis	References
Proteins
IL-6	Pro-inflammatory cytokine	Electrochemical, fluorescence	1 pg/mL	> 1000 pg/mL	(Brown et al., 2018), (Russell et al., 2019), (Bozza et al., 2007; Panacek et al., 2004)
Tumor necrosis factor alpha (TNF-α)	Pro-inflammatory cytokine	Electrochemical, fluorescence	1.1 pg/mL	> 90 pg/mL	(de Pablo et al., 2011), (Pierrakos & Vincent, 2010)
IL-1β	Pro-inflammatory cytokine	Electrochemical, immunoassays	0.7 pg/mL	> 50 pg/mL	(de Pablo et al., 2011), (Baraket et al., 2017), (Kurt et al., 2007)
Procalcitonin (PCT)	Calcium homeostasis	Electrochemical, immunoassays	0.01 ng/mL	> 0.5 ng/mL	(Maruna et al., 2000), (Min Lim et al., 2017), (Vijayan et al., 2017)
C-reactive protein (CRP)	Phagocytic marker during inflammation	Electrochemical, immunoassays, EIS	0.01 ng/mL	> 90 ng/mL	(Thompson et al., 1999), (Ibupoto et al., 2012), (Castelli et al., 2004)
sCD28	T cell activation	Fluorescence, immunoassays	0.1 pg/mL	> 5 pg/mL	(Hui et al., 2017), (Watanabe et al., 2018), (Nolan et al., 2008)
Soluble Triggering Receptor Expressed on Myeloid cells 1 (sTREM-1)	Monocyte and neutrophil activation	Immunoassays	3 pg/mL	> 400 pg/mL	(Aeinehvand et al., 2018; Jedynak et al., 2018; Li et al., 2014)
Small Molecules
Lactate	Anaerobic metabolite	Electrochemical, EIS, fluorescence, colorimetric	0.05 mM	> 2 mM	(Brown et al., 2018; Koh et al., 2016)

Triggering Receptor Expressed on Myeloid Cells 1, or TREM-1, is a transmembrane protein found on human myeloid cells and are an integral part of the innate immune system (Jedynak et al., 2018). In many cases where it interacts with materials outside the body, it is dissolved in a fluid state in the extracellular matrix as soluble TREM-1 or sTREM-1. As sTREM-1, the protein can be found in many immune cells such as neutrophils, monocytes and their progenitor macrophages, as well as endothelial and epithelial cells (Biron et al., 2015). Under infectious conditions, sTREM-1 expression increases from all these cells significantly, making it a valuable biomarker for sepsis diagnosis or an indicator involved in sepsis survival (Jedynak et al., 2018). One of many initial studies determining the effects of sTREM-1 as an indicator of sepsis came from Gibot et al., which concluded that upregulated levels of sTREM-1 in blood plasma correlated significantly with sepsis. Specifically, they found levels higher than 60 ng/mL in blood plasma for individuals undergoing sepsis, with higher correlation and specificity compared with other biomarkers they tested (Gibot et al., 2004).

2.2.2. Small molecules

The most ubiquitously measured sepsis biomarker in clinical settings may be lactate. During anaerobic metabolism, it is consumed and converted into lactic acid and later derivates of alcohol to yield minor returns in adenosine triphosphate (ATP) when oxygen is in short supply (Lee & An, 2016). During sepsis and septic shock, sites of organ failure upregulate lactate production and is an indicator of the hyper-inflammation response, where cells are further cascading into dysfunction. In clinical settings, electrochemical sensors have been designed to measure lactate levels (Rathee et al., 2016), but various other methods are possible including EIS, fluorescence techniques, as well as colorimetric and more (Brown et al., 2018). Normally, blood lactate levels are 0.05 mM or less, but values during sepsis are determined at 2 mM or greater and may reach as high as 25 mM (Lee & An, 2016).

2.3. Pathogen Identification

Bacterial, viral, and fungal infections commonly lead to inflammatory responses and if unregulated can progress into sepsis.

2.3.1. Bacteria

Methicillin-resistant Staphylococcus aureus (MRSA) is a gram-positive bacterium which commonly infects the body through open wounds in hospitals (Table 5). This strain is related with nosocomial infections, a trait it shares closely to sepsis and is one of the most common causes (Zorzoli et al., 2016). Few treatments are available for MRSA as it is antibiotic resistant, other than resupplying body fluids and select drugs such as doxycycline. After incubating in the body for 1 to 10 days, MRSA lasts for a week to several weeks, and may lead to sepsis without the proper treatment. Clinical testing methods include immunoassays, DNA analysis through polymerase-chain reaction (PCR) linked tests, and gram-positive staining. Many methods have difficulty distinguishing results lower than 10 colony-forming units (CFU)/mL, while the pathogen may exceed 20 CFU/mL during sepsis (Klevens et al., 2007). Similarly, Mycobacterium tuberculosis is a bacterium which is mostly known for causing tuberculosis, a debilitating and commonly fatal respiratory disease that can lead to septic shock (Smith, 2003). Particle droplets from infected individuals are how the disease spreads, and while worldwide cases have declined, it is still prevalent in areas of the world with low public health (Smith, 2003). The bacterium can lay dormant in vivo for 2 to 5 years before symptoms and widespread infection develop. Neither gram-positive or gram-negative, an abundance of mycolic acid in its membrane only allows acid-directed stains for imaging and targeting (Kethireddy et al., 2013). Outside of stains, PCR and immunoassays are common identifying techniques, where CFU’s between 1 and 1000 per mL in blood samples are likely to indicate sepsis (Kethireddy et al., 2013).

Table 5.

Common sepsis related pathogens

	Common Pathogen Sources	Incubation Time	Common Detection Methods	Detection Limits	Levels During Sepsis	References
Bacteria
Staphylococcus aureus (MRSA)	Open wounds in hospitals	1 to 10 days	Immunoassays, PCR, gram-positive stain	10 CFU/mL	> 20 CFU/mL	(Zorzoli et al., 2016), (Klevens et al., 2007)
Mycobacterium tuberculosis	Air droplets	2 to 5 years	Immunoassays, PCR, acid-fast stain	0.4 CFU/mL	1 to 1000 CFU/mL	(Mishra et al., 2019), (Crump et al., 2011), (Kethireddy et al., 2013)
Klebsiella pneumoniae	Non-airborne contact	Weeks to months	Immunoassays, gram-negative stain	1 CFU/mL	> 10 CFU/mL	(Cihalova et al., 2017), (Sawano et al., 2016), (Peter et al., 2012)
Pseudomonas aeruginosa	Contaminated water, hospitals, burn sites	2 to 5 days	Gram-negative stains, triple sugar iron test	5 CFU/mL	> 10 CFU/mL	(Gang et al., 1999), (Sharma & Srivastava, 2016), (Mocan et al., 2017)
Neisseria meningitidis	Fluid droplets from infected individuals	3 to 4 days	Immunoassays, PCR, gram-negative stain	102 CFU/mL	> 103 CFU/mL	(Brigham & Sandora, 2009), (Whaley et al., 2018), (La Scolea & Dryja, 1984)
Virus
Herpes Simplex Virus (HSV)	Perinatal transmission, skin contact	several years	PCR, direct fluorescent antibody, cyto-smear	102 DNA copies/mL	> 102 DNA copies/mL	(Amel Jamehdar et al., 2014), (Berrington et al., 2009; Glas et al., 2012)
Influenza	From air droplets or surfaces	1 to 4 days	PCR, serologic assays	103 DNA copies/mL	> 105 DNA copies/mL	(Florescu & Kalil, 2014), (Stellrecht et al., 2017), (Lalueza et al., 2019)
Fungi
Candida albicans	Unsterilized catheters, across mucosal barriers	Variable (days to years)	Microscopy, PCR	10 CFU/mL	> 104 CFU/mL	(Duggan et al., 2015), (van de Groep et al., 2018), (Erdogan & Rao, 2015)
Aspergillus fumigatus	Air-borne spores	12 to 17 days	Immunoassays, bronchoscopy, PCR	3 CFU/mL	> 104 CFU/mL	(Hachem et al., 2009), (Alshareef & Robson, 2014), (Sinha et al., 2018)

Klebsiella pneumoniae is a gram-negative bacterium that is frequently linked to infections such as urinary tract infections, gastroenteritis, and hepatitis (Paczosa & Mecsas, 2016). Described as an opportunistic pathogen, symptoms from K. pneumoniae only manifest under immunosuppressed states, and thus may remain dormant for weeks or months inside the body. The bacterium occurs mainly in hospital settings or alongside other diseases, with end-stage symptoms merging into sepsis and later septic shock (Sawano et al., 2016). Most clinical diagnostic techniques include immunoassays, gram-negative stains, and a complete blood count (CBC) (Sawano et al., 2016). For these techniques, their average LOD measures 1 CFU/mL, while levels above 10 CFU/mL may point towards sepsis (Peter et al., 2012). Similar to K. pneumonia, Pseudomonas aeruginosa is an opportunistic gram-negative bacterium mostly attributed in nosocomial settings (Gang et al., 1999). However, it is predominately found in the lungs or burn sites under infectious circumstances, only increasing in load and detriment under immunosuppressed or compromised conditions. While antibiotics like β-lactam inhibitors and aminoglycosides can treat P. aeruginosa infections, the bacterium is becoming increasingly antibiotic resistant over time (Bassetti et al., 2018). Common modalities for P. aeruginosa quantification include gram-negative stains and triple sugar iron tests (TRI) (Sharma & Srivastava, 2016). While it may be common to have bacterial loads greater than 5 CFU/mL, during sepsis those numbers rise to greater than 10 CFU/mL (Mocan et al., 2017). Neisseria meningitidis is another gram-positive bacterium known as causing meningococcemia, a rare but highly fatal form of sepsis that can target the respiratory system, skin, eyes, or ears (Takada et al., 2016). People become infected through contacting fluid droplets from infected individuals or sharing water sources, and is also classified as a sexually transmitted disease (STD) (Brigham & Sandora, 2009). After becoming infected, symptoms normally appear after 3 to 4 days of incubation, while identifying N. meningitidis clinically includes PCR, gram-negative stains, or skin biopsies (Takada et al., 2016). For N. meningitidis, any detecting greaterthan 10³ CFU/mL may indicate septic conditions (La Scolea & Dryja, 1984).

2.3.2. Viruses

The herpes simplex virus (HSV) causes the STD known as herpes; a common disease, under normal conditions symptoms are chronic yet mild (Fatahzadeh & Schwartz, 2007). However, unique cases may lead to HSV-driven sepsis in the form of encephalitis, pneumonia, esophagitis, or hepatitis (Riediger et al., 2009). The virus itself may remain dormant for years before activating, or may never activate within the body (Berrington et al., 2009). HSV may be identified using PCR, direct fluorescent antibody (DFA) loading, or a cytological smear test (Fatahzadeh & Schwartz, 2007). When analyzing from PCR, more than 10² DNA copies/mL may be an indicator for sepsis (Glas et al., 2012). Another virus which causes sepsis is the influenza virus. A common infection-causing virus, life-threatening symptoms may arise for more vulnerable populations such as the immunocompromised or elderly (Florescu & Kalil, 2014). Most people become infected through contacting air droplets, surfaces, or people with the virus, and will have between 1 to 4 days post-contact to produce symptoms. For detecting influenza, PCR-based DNA sample analysis is conducted, along with serologic assays (Stellrecht et al., 2017). Under septic conditions, more than 10⁵ DNA copies/mL is required for determination (Lalueza et al., 2019).

2.3.3. Fungi

While fungal sepsis is rare compared to bacterial sepsis, Candida albicans may be the most common fungal sepsis by far. Another opportunistic pathogen, it normally produces infections after transmitting through unsterilized hospital equipment or across mucosal barriers, with infections originating in the gastrointestinal tract or mouth (Duggan et al., 2015). Incubation times have wide variance, taking days or years from initial contact to symptom progression (Erdogan & Rao, 2015). To identify and quantify C. albicans, PCR is utilized, but cellular features are also identifiable through optical microscopy, requiring a biopsy sample to locate the pathogen (van de Groep et al., 2018). C. albicans can be observed at concentration as low as 10 CFU/mL, while during sepsis the fungus can be greater than 10⁴ CFU/mL (Erdogan & Rao, 2015). A rarer yet deadly fungal-derived sepsis occurs through Aspergillus fumigatus. It is a common airborne fungus, generating a respiratory infection for people with incapacitated immune systems such as those who suffer from granulocytopenia (Nenoff et al., 1995). Typically, A. fumigatus incubates in the body for 12 to 17 days before infections transpire, and PCR, immunoassays, or bronchoscopy can identify the fungus from other cell types as low as 3 CFU/mL (Alshareef & Robson, 2014; Bénet et al., 2013; Hachem et al., 2009). However, higher fungal loads above 10⁴ CFU/mL are indicators for the infection becoming septic (Sinha et al., 2018).

3.0. MICRO AND NANO TECHNOLOGIES FOR SEPSIS BIOMARKERS

Recently, the potential for a successful sepsis diagnosis device utilizing M/N systems has increased significantly. Further improvements in lab-on-a-chip (LoC) fabrication protocols and the discovery of new measurement and signal acquisition sources which can perform multiplexed analysis of numerous biomarkers are responsible for their success, enabling a high biomarker yield which is imperative for stratifying sepsis in clinical settings (Liao et al., 2019; Reddy et al., 2018; Sinha et al., 2018). Specifically, many promising LoC devices rely on the robust yet flexible manufacturing of microfluidics and the ever-expanding diversity and modification of nanoparticles as target objects. Microfluidics provide a compartmentalized scaffold which may perform optimized and regulated sample processing as well as identify and separate biomarkers based on defined characteristics found at micro or nano-scale dimensions (Boneschansker et al., 2014; Brown et al., 2018; Ellett et al., 2018). In microfluidics, nanoparticles are the active agents performing and displaying the analytical changes in the assays, owing to their heterogenous material origins and functionalization including magneto-nanoparticles, quantum dots (QDs), electrically tuned particles, barcoded particles, and others (Ashley et al., 2020; Cihalova et al., 2017; Mok et al., 2014; Uddin et al., 2016). Our own work studies metal oxide-coated hybrid surface microparticles with unique impedance measurements within a microfluidic impedance sensing system to target leukocyte receptors related to sepsis such as CD11b, CD64, CD66b (Ashley et al., 2020; Xie et al., 2017).

Here we detail common methods for sepsis or sepsis-related biomarkers measured with micro or nanotechnologies. Relevant cases will also be explored which characterize their effectiveness and advantages versus comparable approaches, and precise translation to sepsis diagnostic applications. Such mechanisms may improve sepsis outcomes by requiring smaller biofluid volumes collected from patients, faster analysis with automated sample processing, multiplex biomarkers detection, improved diagnostic accuracy, and miniaturized systems being manufactured at low costs to deliver a POCC technological solution.

3.1. Micro and nano technologies for cell-related biomarker quantification

3.1.1. Electrochemical impedance spectroscopy

Analyzing biomarkers directly related to cells utilize many detection methods to accommodate the heterogeneity of immune cells. One option includes electrical impedance spectroscopy (EIS) of whole cells in microfluidic systems. Here, large cells with phospholipid bilayer membranes act as insulating spheres in the presence of an electric field. Dielectric media under flow and sensor electrodes then generate impedance shifts as cells pass over an applied electric field. A pulse representing the cell can be generated, and data extrapolation includes pulse amplitude, directly proportional with cell volume, and pulse width. Additionally, the applied voltage frequency modifies the detecting target. Lower frequencies (100 kHz to 5 MHz) correspond with larger structures such as the cell itself, while higher frequencies (> 10 MHz) penetrate the membrane and may provide signals on intracellular structures and organelles (Ahuja et al., 2019; Cheung et al., 2010; Hassan & Bashir, 2014). A highly quantitative tool that measures cellular properties without modification, limitations include upscaling micro/nanovolt changes, poor signal-to-noise ratios (SNR), and relatively difficult to analyze results (Cheung et al., 2010). EIS technologies for cellular biomarkers include research by Hassan et al. that screened neutrophils for nCD64 (Figure 4a). After isolating neutrophils from potentially septic whole blood, an antibody capture chamber collected cells which expressed nCD64, and were crossed-referenced with white blood cell (WBC) and platelet counts that were also determined on-chip using EIS. A versatile technique with whole blood stratification on-chip, the device could detect cell counts as low as 50 cells/μL and had a receiver operating curve (ROC) of 77% (Hassan et al., 2017). Another on-chip method includes work by Ahuja et al. which used a multifrequency EIS device and modified antineoplastic-antibody conjugates to evaluate drug efficacy on T47D cancer cells.

Figure 4:

Micro-Nano technologies for cellular antigen expression (a), mobility (b, c) and stiffness quantification (d). (a) Schematic view of nCD64 quantification neutrophils isolate from whole blood on-chip. Cells are counted electrically before and after entering the nCD64 capture region, coated with anti-CD64 antibodies. The differential cell count determines nCD64+ counts (left). Bar and whisker plot for average nCD64 counts for infected patients over time since admitted to ICU (middle). Quantitative counts of cells before and after entering capture chamber (right) (Hassan et al., 2017). (b) Optical tracking of neutrophil motility using a microfluidic biochip evaluating whole blood (left). Blood enters the loading chamber (LC) and enters the maze (M) which uses size exclusion in the channel dimensions to filter red blood cells from entering the maze (middle). Occurrences of oscillations, pauses, reverse migration, and average distance traveled in the maze are significantly different between septic and non-septic blood samples (right) (Ellett et al., 2018). (c) Quantification and visualization of cell migration towards chemokine gradients. Cells loaded in cell traps can migrate towards (chemotaxis) or away from gradient (retro-taxis) (left). Characterizing chemical gradients in the device using fluorescein labelled dextran. 50mm channels shows a shallower gradient as compared to 6mm channels. A1 represents the path from buffer channel to cell loading chamber, a2 cell loading chamber and a3 cell-loading chamber to chemokine reservoir (right) (Boneschansker et al., 2014). (d) Schematic of the stiffness dependent cell separation microfluidic device. The diagonal ridges compress cells in succession, with secondary flows in channel results in flow shift of cells proportional to their stiffness (left, middle). Fluorescent scatter plot shows the separation of Hey cells from K562 cells on a stiff outlet (right) (Wang et al., 2013)

Screening the cells at low frequencies counted total cells, while higher frequencies measured intact versus destroyed organelles for live-dead determination. Coupled with machine learning to correlate results, proper cell fates were reported as 89% accurate (Ahuja et al., 2019).

3.1.2. Fluorescence measurements

Receptor-mediated fluorescent techniques use functionalized molecules with fluorescence properties to measure values using optical differences. The technique is routinely used for organic substances and diverse protocols have been produced to couple fluorescence markers with numerous cellular receptors. Limitations in M/N technologies include low multiplexing yield as it is limited to discernable differences in visible light waves and must irreversibly modify cells to collect results (Umlauf et al., 2013; Venet et al., 2011; Zouiouich et al., 2017). Numerous studies employ fluorescence techniques to study cellular biomarkers. Research by Zouiouich et al. and later Tamulyte et al. use modified POCC flow cytometry to measure CD45 and mHLA-DR in septic individuals. Using QuantiBrite anti-HLA-DR PE antibodies to stain mHLA-DR receptors on monocytes from whole blood, both publications were able to indicate which patients had symptom outcomes based on mHLA-DR molecules per monocyte, ranging from less than 2,000 molecules per cell to greater than 8,000 molecules per cell. However, neither study was able to directly correlate results with changes in mortality (Tamulyte et al., 2019; Zouiouich et al., 2017).

3.1.3. Physical measurements

To observe and differentiate cells based on biophysical characteristics, precise device design is necessary. Here, flow and non-flow based microchannels are fine-tuned with different geometries, orientations, gradients, and stiffnesses to elicit cell-specific responses, as it is reported that motile cells like neutrophils are closely tied to the physical properties of their surroundings (Ellett et al., 2018). While advantageous for label-free detection, it is limited by translatable scope, complex device manufacturing, and underdeveloped quantification methods (Boneschansker et al., 2014; Ellett et al., 2018; Hassan et al., 2018). Interesting work by Ellet et al. observes neutrophil motility after isolating from whole blood using size-dependent microfluidic channels to compare patterns in septic versus non-septic patients. Exploiting the greater motility of septic neutrophils, significant differences in overall cell counts, pauses, oscillations, and reverse migrations taken by the cells were recorded through microchannel optical tracking. The system was sensitive enough to identify 1 μm change in position by the cells, and only required 1 μL of blood (Figure 4b) (Ellett et al., 2018). In another study, Boneschansker et al. reported a microchannel fabricated with chemokine gradients using fluorescein labelled dextran and found significantly different results of cell path directions in 50 μm channels versus 6 μm channels shown in Figure 4c (Boneschansker et al., 2014). Another study measured force and torque responses from cells flowing over cell-antigen respective antibody immobilized surfaces. Here, the antigen-antibody bond is modelled as a spring. Measured and predicted capture ratio measurements of EpCAM expressing MDA-MB-231 and BT-20 cells as a function of the applied flow rate in microchannels functionalized with EpCAM antibodies showed BT-20 cells capture ratio is much higher in rolling adhesion region (Zheng et al. 2011). Work by Jones et al., exploited changes in deformability with microstructure orientation gradients, and was able separate Hey cells from K562 with 82% purity in a flow-based device shown in Figure 4d (Wang et al., 2013). Observing changes in deformability are analyzed in research by Crawford et al., which shears granulocytes in a centrifuging microfluidic cytometer. Using a high speed camera to record changes in final cell size and elongation, they were able to separate septic versus healthy patients based on cell morphology after shearing (Crawford et al., 2018).

3.2. Micro and nano technologies for protein and small molecule quantification

3.2.1. Analytical measurements

Numerous methods or combinations thereof have been utilized in M/N systems to numerically detect proteins and small molecules. This includes modified EIS methods discussed earlier, enzymatic electrochemical detection, and surface plasmon resonance (SPR). For enzymatic detection, a protein or analyte participates in a redox reaction which generates ionic byproducts which record current changes on a three-electrode system. Successful for many analytes such as lactate, oxygen, and glucose, they are limited by electrochemical dependencies and weak sensitivity over time (Brown et al., 2018). Likewise, SPR exploits oscillations in electrons on gold or silver nanoparticle surfaces from light sources to finely measure adsorbed or functionalized materials on their surfaces based on material size. Here, differences in 1 nM may be achieved, and can render molecules as small as polynucleotides. A newer technology, many believe their application in biosensors has yet to be fully realized (Sun et al., 2020).

For one study, IL-6 and Abelson tyrosine kinase (Abi) was measured for quantity and activity, respectively (Figure 5a). With this system, streptavidin-biotin conjugated particles attached to the functionalized capture chamber, and pulses are detected from particles which did not attach, and successfully detecting IL-6 as low as 50 pg/mL (Mok et al., 2014). Interleukin-6 has also been measured by Russell et al. using EIS from antibody-functionalized gold needle-shaped electrodes on a microfabricated device, where IL-6 concentration could be identified as low as 25 pg/mL (Russell et al., 2019). In a study by Baraket et al., both IL-10 and IL-1β were evaluated on diazonium-modified antibodies electroplated on gold electrodes, measured using EIS, and detectable to 1 and 15 pg/mL, respectively (Baraket et al., 2017). For measuring PCT, Sun et al. constructed a fully integrated localized-SPR microdevice which measured PCT load on antibody-functionalized gold nanopillars and having a 0.5 ng/mL limit of detection (Figure 5b). With the detection system, it was reported that it has multiplexing capabilities up to 6 biomarkers at once (Sun et al., 2020). As shown in Figure 5c, Panneer et.al. developed a label-free EIS device for PCT, lipopolysaccharide (LPS), and lipoteichoic acid (LPA) from whole blood samples, exploiting a coulombic potential generated from antibody-coated gold electrodes encapsulated with nylon polymer membranes (Panneer Selvam & Prasad, 2017). In analyzing CRP, Ibupoto et al. produced a disposable potentiometric sensor using ZnO nanowires adsorbed with glutaraldehyde and targeting antibodies, determining a 10 pg/mL LOD for the microsystem (Ibupoto et al., 2012). Work by Ashley et al. developed a wearable multiplexed gold small molecule sensor electrochemically measuring lactate and oxygen using lactate oxidase immobilized with chitosan/single-walled carbon nanotubes and Nafion-coated electrodes, respectively (Ashley et al., 2019). Finally, in a novel system by Uddin et al. and shown in Figure 5d, a lab-on-a-disc centrifugation device targeted thrombin-coated magnetic particles and allows for detection of bead aggregates through optomagnetic concentration using a magnetic field, with imaged particle aggregates serving as a cross-reference to concentration outcomes (Uddin et al., 2016).

Figure 5:

Micro and nano technologies for proteins (a,b,d), small molecules (c). (a) A dual-layer impedance detection system for IL-6 and Abelson tyrosine kinase (Abi), where streptavidin-biotin particle conjugates functionalized with target antibodies bind with captured protein in a micro fluidic device (left). As the particles pass over and collect the protein, the particles then flow across an electric field which measures impedance pulses (middle). Total bead counts passing over the electrodes corresponds with control experiments measuring IL-6 concentration and Abi activity (right) (Mok et al., 2014). (b) A fully integrated localized surface plasmon resonance-enhanced quantum dots system measuring procalcitonin (PCT) measuring whole blood samples (left). Streptavidin-functionalized quantum dots target and bind with biotinylated PCT-targeting antibodies, subsequentially binding with antibodies functionalized on gold nanopillars (middle). PCT concentrations correlate with optical intensity from bound QD levels (right) (Sun et al., 2020). (c) Label-free impedance microfluidic biosensor from charge-linked biomarker reactions to assess PCT, lipopolysaccharide (LPS), and lipoteichoic acid (LPA) from whole blood samples (left). Here, nylon polymer membranes encapsulated three distinct antibody-coated gold electrodes, where biomarker-antibody binding generates a coulombic potential measured as impedance change for PCT, LPS, and LTA (middle and right) (Panneer Selvam & Prasad, 2017). (d) A centrifugation-based measurement system for thrombin using nano- and micro-sized magnetic beads. Detection was determined using both optomagnetic determinations from aptamer-coated nanobeads (middle channel) juxtaposed with optical imaging of aptamer-coated microbeads (outer channels) (left and middle left), where aggregation from thrombin determines sample concentration (middle right). Optomagnetic concentration measured from magnetic field pulses, and average size of imaged particle aggregates complete a two-step thrombin detection assay (right) (Uddin et al., 2016)

3.2.2. Quantum dots, colorimetric, and optical methods

Fluorescence methods are commonly employed for tracking and quantifying materials. Additionally, the use of QDs has also been utilized as pseudo-fluorescent nanoparticles. At such small length-scales, semiconductor-based nanoparticles fluoresce with distinct wavelengths based on materials chosen due to the electron movement under quantum mechanics principles. They are preferable to other fluorescent techniques with no quenching effects, and narrower spectrum bands with increased multiplexing in the same field of view (Herrera et al., 2019). Exploiting redox reactions which produce colorimetric effects are another way which microsensors can convey analyte detection (Koh et al., 2016). Considering multiplexing, Volpetti, Garcia-Cordero, and Maerkl developed a library of fluorescence-linked immunoassays on a microfluidic chip, using neutravidin-biotin antibody immobilization methods for the detection of IL-6, IL-1β, TNF-α, and 381 other biomarkers all on the same chip. With most markers distinguishable in the 10 pM range, nearly every assay along with control samples was completed in over 3 hours (Volpetti et al., 2015). Sun et al. used a localized-SPR for detecting PCT (Figure 5b) and a QD’s detection system in parallel to their localized-SPR, and produced higher optical intensities with greater PCT concentration on their nanorods (Sun et al., 2020). Additionally, Koh et al. developed a microfluidic, PDMS-based wearable multiplexing device for measuring lactate, glucose, pH, chloride, and biofluid volume from collecting perspiration. The device coupled colorimetric redox reactions for each biomarker, and used a smartphone application to quantify degrees of color change (Koh et al., 2016).

3.3. Micro and nano technologies for pathogen identification and quantification

3.3.1. Bacteria and fungi-based detection systems

Bacterial and fungal sepsis analysis use very similar technology. One common method comes from functionalized particles to capture and amplify DNA or pathogen-specific proteins. This process is similar to approaches discussed previously, using detection through QDs, EIS, or SPR. Another process separates pathogens from host cells based on density gradients and measuring the pathogen load from an isolated sample. While simplistic in design, micro-centrifuging at forces high enough for pathogen separation is technically difficult, making it problematic for POCC applications where technologies should be simplistic and streamlined (Cihalova et al., 2017; Nguyen et al., 2019; Yoo et al., 2011). Most recently, Piekarz et al. developed a 25 array microwave E. coli detection system which uses antibody-functionalized electrodes and measures capacitance change from bacterial load, estimating a 10³ CFU/mL LOD (Piekarz et al., 2020). When studying both bacteria and fungi, Lehmann et al. developed a multiplexed PCR system which can differentiate 25 pathogens from whole blood samples in under 6 hours. Here, gram-positive, gram-negative, and fungi pathogens are developed and analyzed in parallel, and has over 98% specificity between samples (Lehmann et al., 2008). As shown by Figure 6a, work by Cihalova et al. produced a multiplexed bacterial detection system using magnetic nanoparticles for bacteria separation and QD’s for measuring DNA load after cell lysis, with a LOD as low as 10² CFU/mL (Cihalova et al., 2017).

Figure 6:

Various point-of-critical care (POCC) sensors targeting pathogens, including bacteria (a, b), virus’ (c), and fungi (d). (a) Sensor employing magnetic particles to purify and culture MRSA, S. aureus, and K. pneumoniae, while attributing detection with oligonucleotide-functionalized quantum dots (QDs) targeting bacterial-specific DNA fragments (left). Representative emission spectra using the 3 QDs with their coupled bacteria (middle). Fluorescence intensity from QDs correlates with identifying bacterial concentrations from a sample (right) (Cihalova et al., 2017). (b) A solution circuit chip (SCC) which evaluates bacterial load from a sample on 20 electrochemical sensors. An array of common working, counter, and reference electrodes with the functionalized peptide nuclei acid (PNA) probes (left) which bind with specific regions on bacterial DNA (middle). Ru(NH₃)₆³⁺ targets the bound DNA phosphate backbone, reducing to Ru(NH₃)₆²⁺ which undergoes a redox reaction with Fe(CN)₆³⁻ in solution back to Ru(NH₃)₆³⁺ and an current is recorded, correlating current strength with bound DNA (right) (Lam et al., 2013). (c) Using gold nanowires and surface-enhanced Ramen spectroscopy (SERRS), exonuclease III digests complementary DNA strands from fungi, resulting in a decreased signal which is SERRS identified (left). Such decreases show identification of specific fungi from select samples, with a 100 femtomolar limit of detection (right) (Yoo et al., 2011). (d) A paper-based influenza/bacteria sensor which uses pH gradients and exploits colorimetric reactions unique with proteins and select enzymatic substrates (left). Varying the pH yields optical peak changes (ΔCIE) and can eliminate non-targeted pathogens in samples as (middle). The colorimetric assay can distinguish between different viral strains as well as their drug resistance (right) (Murdock et al., 2017)

In another recent study, a whole blood sample multiplexed PCR device is developed specifically for sepsis diagnosis, measuring 18 unique bacterial or fungal pathogens and reaching a 10 CFU/mL limit of detection (LOD) (van de Groep et al., 2018). A solution circuit chip (SCC) design has also been used by Lam et al., which employs an electrochemical redox reaction from functionalized peptide nuclei acid probes which binds to bacterial DNA, with over 24 possible bacterial samples in one chip (Figure 6b) (Lam et al., 2013).

Results by Nguyen et al. detailed a fully integrated centrifugal microdevice which measures E. coli and salmonella DNA, and correlates densities to UV-Vis absorbance characterizations to identify pathogen and relative load, producing a 10² CFU/mL LOD (Nguyen et al., 2019). Finally, Yoo et al. used a novel surface-enhanced Ramen spectroscopy (SERRS) on gold nanowires system to measure 4 isolated and purified fungal DNA following exonuclease III activity, with a very low 100 fM DNA LOD shown in Figure 6c (Loo et al., 2017).

3.3.2. Viral-based detection systems

Many virus-specific POCC detection systems directly measure DNA, either through PCR, colorimetric, or enzyme-specific substrates. While PCR is the standard in research settings for DNA extrapolation, its hands-on processing hinders it from frequent adoption in the POCC, lending more towards disposable immunoassays. For example, research by Murdock et al. have produced a paper-based chip for measuring influenza through multiple strains and antibiotic-resistant tests based on pH gradients and protein-coupled colorimetric reactions shown in Figure 6d (Murdock et al., 2017). In another complementary study, Kao et al. developed a microfluidic, fully automated PCR chip which can detect influenza FluA and the H1N1 strain in under an hour, with a 20 fg/μL LOD (Kao et al., 2011). A POCC influenza detection system has also been developed by Xu et al., fabricating highly-uniform nanobipyramidal silver-coated gold particles for a localized-SPR-based colorimetric sensor, which changes color corresponding to viral load on particle surfaces, and produced a 0.01 mU/mL LOD (Xu et al., 2017).

An assessment of commonly used multiplexing modalities are given in Table 6, which compares their technology specific requisite materials, sepsis biomarker types they can detect, and corresponding detecting limits. We also classified each technology based on their POCC multiplexing potential while considering the reported number of biomarkers recorded simultaneously, their scalability to miniaturized devices, and their ability to measure different biomarker types together (such as pathogens vs. cell receptors vs. soluble proteins etc.). Additionally, notable multiplexing M/N systems for sepsis-biomarker diagnosis are summarized by Table 7. Here, articles are displayed based on multiplexing capability, diversity of biomarkers studied, and novelty of mechanisms used.

Table 6.

Common point of critical care (POCC) micro and nano multiplexing modalities

	Requisite materials*	Detectable biomarkers	LOD	Sensitivity	References
POCC multiplexing potential Class 1: (i) can measure multiple biomarkers simultaneously (~ 2–20) in parallel, (ii) limited in measuring other biomarker types (i.e., pathogens vs. proteins vs. cell receptors)
Fluorescence	*Dyes, *imaging, antibodies, nanoparticles	Cell receptors Proteins Pathogens	N/A 10 pM 102 cells/mL	N/A 100 pM 102 cells/mL	(Mishra et al., 2019), (Crump et al., 2011), (Kethireddy et al., 2013), (Hachem et al., 2009), (Volpetti et al., 2015)
Optical/colorimetric	*Tracking system, *camera, pigments, redox agents	Cell motility Small molecules	1 μm position Δ 200 μM	N/A 0.02–1 mM	(Boneschansker et al., 2014), (Ellett et al., 2018), (Koh et al., 2016), (Sawano et al., 2016), (Peter et al., 2012)
POCC multiplexing potential Class 2: (i) can measure many biomarkers simultaneously (~ 2–100+) AND/OR, (ii) can measure other biomarker types AND/OR, (iii) can translate to POC sensors with similar sensitivity. No reports found demonstrating all three features.
Polymerase chain reaction (PCR)	*DNA polymerase, *temperature control, *DNA primers, *nucleotides, *buffer, centrifugation	Pathogens	20 fg/μL	20–30 fg/μL	(Kao et al., 2011), (van de Groep et al., 2018), (Lehmann et al., 2008)
Quantum dots (QDs)	*Semi-conducting nanoparticles, imaging/detector, antibodies	Proteins Pathogens	0.5 ng/mL 102 CFU/mL	0.4 ng/mL 102 CFU/mL	(Cihalova et al., 2017), (Sun et al., 2020)
Surface plasmon resonance (SPR/LSPR)	*Metal nanomaterials, *light source, antibodies	Proteins Pathogens	0.5 ng/mL 100 fM of DNA	0.4 ng/mL 100 fM	(Sun et al., 2020), (Yoo et al., 2011)
POCC multiplexing potential Class 3: (i) can measure multiple biomarkers simultaneously (~ 2–150+), (ii) can measure other biomarker types, and (iii) many examples of fully integrated POC sensors
Electrochemical impedance spectroscopy (EIS)	*AC current, *electrodes, nanoparticles, antibodies	Cells Cell receptors Proteins Small molecules Pathogens	50 cells/μL 2000 cells/μL 0.5 pg/mL 50 pM 103 CFU/mL	40 cells 400 cells/μL 5 pg/mL 100 pM N/A	(Hassan et al., 2017), (Ashley et al., 2019), (Lam et al., 2013), (Mok et al., 2014), (Baraket et al., 2017)

^*Material/label/agent required

Table 7.

Notable micro and nano multiplexing diagnostic systems

Article	Cell Parameter	Technology	Biomarkers	LOD	# of bio-markers	Whole biofluid (Y/N)	Assay time
(Baraket et al., 2017)	Protein quantification	Electrochemical impedance spectroscopy (EIS)	IL-6, IL-10, IL-1β	0.5 pg/mL	3	N	~3 hours
(Cihalova et al., 2017)	Bacteria loading	Quantum dots (QDs)/fluorescence	S. aureus, MRSA, K. pneumoniae	102 cells/mL	3	N	~2 days
(Crawford et al., 2018)	Granulocyte stiffness	Deformability cytometry	CD45, CD66b, deformability	Δ 1 μm in diameter	3	N	< 10 min
(Ellett et al., 2018)	Motility	Optical Tracking	Oscillations, Pausing, Reverse migration, Avg. distance	Δ 1 μm in position	5	Y (blood)	~6 hours
(Hassan et al., 2017)	Cell counting	EIS	RBCs, Platelets, WBC differential	50 cells/μL	3	Y (blood)	~2 hours
(Kao et al., 2011)	Virus loading	Microfluidic PCR	Influenza FluA and H1N1	20 fg/μL	2	N	~1 hour
(Koh et al., 2016)	Small molecule/volume quantification	Colorimetric/wearable	Lactate, pH, Cl-, glucose, Sweat volume	200 μM	5	Y (sweat)	~1 min
(Lam et al., 2013)	Bacteria loading	EIS	S. saprophyticus, S. aureus, P. aeruginosa; K. pneumonia, etc.	103 CFU/mL	20	N	~2 min
(Lehmann et al., 2008)	Bacteria/Fungi loading	DNA sequencing/PCR	E. aerogenes, P. aeruginosa, A. fumigatus, C. glabrata, etc.	10 CFU/mL	25	Y (blood)	> 6 hours
(Loo et al., 2017)	Bacteria loading	Centrifugation/DNA sequencing	M. tuberculosis, A. baumanii	103 CFU/mL	2	Y (blood)	~2 hour
(Mok et al., 2014)	Protein quantification	EIS	IL-6, Abl	50 pM	2	N	~2 hours
(Murdock et al., 2017)	Bacteria/virus loading	Colorimetric	Influenza A/B, HPV, S. pneumoniae	N/A	6	N	~10 hours
(Nguyen et al., 2019)	Bacteria loading	Centrifugation/ absorbance	E. coli, S. typhimurium, V. parahaemolyticus	102 cells/mL	3+	N	~1 hour
(Panneer Selvam & Prasad, 2017)	Protein/small molecule quantification	Electrical	PCT, LTA, LPS	0.001 to 1 μg/mL	3	Y	~2 hours
(Sun et al., 2020)	Protein quantification	Surface plasmon resonance (SPR)/QDs	PCT	0.5 ng/mL	6+	Y (blood)	~30 min
(Uddin et al., 2016)	Protein quantification	Opto-magnetic microbeads/ centrifugation	Thrombin	25 pM	8	N	~15 min
(van de Groep et al., 2018)	Bacteria/fungi loading	Real-time PCR	S. aureus, S. pneumoniae, E. coli, P. aeruginosa, C. albicans, etc.	10 CFU/mL	18	Y (blood)	~3 hours
(Volpetti et al., 2015)	Protein quantification	Fluorescence	IL-6, IL-1β, TNF-a, PSA, GFP, etc.	~10 pM	384+	N	> 3 hours
(Watkins et al., 2013)	Cell counting	EIS	CD4/CD8	2000 cells/μL	2	Y (blood)	< 30 min
(Yoo et al., 2011)	Fungi loading	SPR	A. fumigatus, C. glabrata, C. krussei, C. neoformans	100 fM of DNA	4	N	~4 hours

4.0. CONCLUSION: CHALLENGES, COMMERERICIALIZATION PATHS, AND POTENTIAL OUTCOMES

4.1. Multiplexing technology directions, machine learning, and future perspectives

Many mechanisms for biomarker detection have been presented, yet the margins for most successful multiplexing may depend on precisely how many biomarkers can be measured rapidly and economically. Currently, with more than 170 known biomarkers related to sepsis progression, certain devices may have limited multiplexing potential to adequately measure sepsis biomarkers due to technology restrictions or the inability to translate techniques to other biomarker types. For example, fluorescence quantification has been developed and characterized for decades as an accurate tool for measuring cell surface receptors, soluble proteins, and pathogens. However, multiplexing hits a plateau as detector parallelization is required for each biomarker, and unique dye wavelengths are restricted in or around the visible light spectrum. Presently, there are no fluorescence tools which can measure more than 18 dyes, and at this capacity the technology is expensive and cannot translate to miniaturized POCC metrics at scale (Jin et al., 2019). Fluorescence methods are also limited in directly quantifying protein or small molecule biomarkers without additional agents such as nanoparticles or immobilizing assays. This highlights the potential of other detection techniques which are not limited by wavelength band interference. Further, other technologies such as optical tracking (quantifies cell’s physical and mechanical properties) and PCR (quantifies pathogen loads) characteristically measure their targets but cannot translate to other biomarker types.

The technology with the highest multiplexing potential for quantifying sepsis biomarkers can be electrically based spectroscopy (EIS). EIS stands out as it has many reported examples for measuring different biomarker types using one signal input and output scheme (Baraket et al., 2017; U. Hassan et al., 2017; Lam et al., 2013; Panneer Selvam & Prasad, 2017). Further, by employing electrically sensitive nanoparticles or “barcoded” species as target agents, the ceiling for multiplexing is only limited by nanoparticle material or physical properties. Indeed, particles with different electrical properties have shown high selectivity for different types of biomarkers with the availability of an extensive library of particles far greater than other modalities (Sui et al., 2020; Xie et al., 2017). Additionally, investigative research in novel polymer particles fabricated from stop-flow lithography can produce barcoded particles and with only 5 barcoded regions the multiplexing potential may reach 120 unique biomarkers (Prakash et al., 2020). The ability to measure each biomarker accurately and impartially with the same detection source through EIS reduces materials and manufacturing costs while also facilitating sample sparring as heterogeneous biomarkers in the same sample may be measured together.

Incorporating multiplexing in diagnostic technologies will help test more biomarkers for accurate distinction between septic and non-septic individuals, however, data management and analysis may increase in complexity. Solutions to handle large multivariate data may come from machine learning, which after training on large data sets can observe and identify trends and relationships for better sepsis stratification. By building computational models which indirectly find correlations previously undiscovered, machine learning may strengthen statistical significance between different targets and improve biomarker targeting specificity. This can improve sepsis diagnosis from many perspectives, including processing a patient’s medical history from electronic medical records (EMR) to predict a patient’s likelihood of developing sepsis (Taneja et al., 2017). Similarly, many studies have already incorporated machine learning with sensors for measuring sepsis biomarkers and will continue to grow based on its automated iterative model building from substantial data sets and allows for an increase in multiplexing potential without a relative increase in processing (Ahuja et al., 2019; Ellett et al., 2018; Green et al., 2019; U. Hassan et al., 2017; Sui et al., 2020). Future POCC strategies may implement machine learning from multiple perspectives—predicting sepsis development from patient EMR data and analyzing multiple biomarkers after symptom onset—to identify sepsis earlier and treat more specifically.

4.2. Further limitations, POCC scalability/commercialization, and theranostic applications

Numerous inflammatory biomarkers, pathogens, detection methods, and sensing platforms indicate the potential for accurate and multiplexing point-of-critical-care sepsis diagnostics (POCC-Dx) technologies. However, it is important to note several limitations, notably many bacterial and fungal detection systems have limits of detection close if not similar to the densities for determining septic conditions (Table 5). This relatively low sensitivity for pathogens may produce false negative results, where they are undetected yet still possibly present at harmful loads (Glas et al., 2012). Additionally, the ever-fluid definition for sepsis prevents clinicals, researchers, or the average person from grasping its causes or implications, as the definition was changed as recently as 2016. A recent survey (2016) from the U.S. Census Bureau found that while 55% of U.S. citizens know about sepsis, only 26% were able to identify proper symptoms or effects (Singer et al., 2016). As we continue to study sepsis, a more realized and robust definition for the disease will continue to reinforce which diagnosis and treatment strategies are most effective.

One of the underlying issues with sepsis continues to be accurate diagnosis. With over 170 biomarkers linked with sepsis, most detection systems have below acceptable accuracy for clinical samples due to the complex nature sepsis has, from its origins to how it propagates through the body (Chiesa et al., 2004). Even if novel technologies are determined, a laborious approval process stifles clinical translation. Many technologies which are commercially approved in Europe still face regulatory decisions in the United States today, as lengthy evaluations of In vitro Diagnostic (IVD) devices by the Food and Drug Administration (FDA) stifle their clinical translation (“Overview of IVD Regulation,” 2019; Sinha et al., 2018). A potential solution to expedite sepsis technology translation is by developing integrated systems based on diagnostics and therapeutic technologies in a Theranostics approach (Figure 7). This attempts to integrate diagnostic technologies and delivering therapeutic agents, eliminating multi-step processes which may improve clinical impact and increase survival rates for time-sensitive diseases like sepsis.

Figure 7:

Theragnostic cycle proposed for sepsis

The cost of the POCC devices plays a significantly factor for their clinical translation, which normally remain much higher for M/N sensors during research and development phase. This is mainly due requiring expensive fabrication procedures such photolithography, electroplating, and others which need cleanroom facilities or specialized equipment, sensor characterizations using micro-scale imaging and control assays, and maintaining a supply-chain of raw materials for agents such as nanoparticles or antibodies, all of which may be tens of thousands of dollars individually. Furthermore, many testing methods for devices still use larger-scale benchtop equipment, such as syringe pumps for reagents flow in the microfluidic devices or data processing on laboratory computers, which will require further prototyping and design to integrate all processes on-chip. However, once the sensor is fully integrated, the scalability of high-throughput production will drive down the costs of the devices exponentially. For example, for MOXI GOTM II instrument, disposable cartridges are only approximately $8 US and can run two tests per chip. It is encouraging that devices designed specifically for POCC environments can be produced for accurate biomarker detection and clinical translation.

Future endeavors may explore recently conceived techniques with biosensor applications, such as QD imaging and localized-SPR for ultrasensitive detection. New breakthroughs will further improve their versatility for a diverse echelon of biomarkers. With the continuous addition of novel techniques and approaches, sepsis diagnosis will further ameliorate and concurrently improve survival from this complicated and debilitating disease.