Mononuclear phagocytes (MNPs) include circulating blood monocytes, dispersed tissue-bound macrophages, and dendritic cells (DCs) that may be either circulating or tissue-bound (36). While macrophages may live for years, blood monocytes have a circulating half-life of only 2 to 3 days (36, 37). Also, while circulating monocytes can replace tissue-resident macrophages, turnover rate varies by organ system. Turnover is higher in barrier organs—for example, gut and dermis—than in other organs, such as heart, pancreas, liver, and central nervous system. Replacement may be hastened in any organ by a local inflammatory process that leads to monocyte influx.

Innate immune responses act rapidly as a first line of defense against invasive, infectious pathogens (1). Initially, neutrophils and monocytes recognize pathogen-associated molecular patterns (PAMPs) and damage-associated molecular patterns (DAMPs). These interactions induce MNPs to release multiple cytokines, such as TNF-α, interleukin (IL)-1, and IL-6, that attract and activate other immune cells (1, 4). While neutrophils primarily kill microbes, MNPs kill microbes and, in addition, present their unique antigenic content to the adaptive immune system (4, 38).

MNPs link innate and adaptive immune systems by their ability to adopt either pro- or anti-inflammatory functions (4, 39). Pro-inflammatory functions eliminate infectious or injurious stimuli and activate antigen-specific helper T lymphocytes, whereas anti-inflammatory functions maintain homeostasis, conduct efferocytosis, and thereby control autoimmunity. Critically, MNPs express class II major histocompatibility complex (MHC) proteins that activate antigen-specific helper T lymphocytes and secrete cytokines to nourish and/or activate diverse cell types ( Figure 1 ). More numerous neutrophils by contrast are primarily pro-inflammatory, live only for days after a 6- to 12-hour circulating half-life, and do not characteristically present foreign antigens to adaptive immune lymphocytes (43).

The adaptive immune system comprises antigen-specific T lymphocytes that are cytotoxic, or are responsible for self-tolerance (T regulatory cells) as well as antibody-producing plasma cells that result from B cell differentiation (1, 3). Although initially slower to respond than the innate immune system, the adaptive immune system drives antigen-specific recognition and generates immunologic memory. Immunologic memory generates faster, stronger repeat immune responses against previously encountered antigens.

In sepsis, both innate and adaptive arms of the human immune system are altered (1). In addition, multiple cytokine levels are elevated, including GM-CSF. These cytokines drive proliferation of circulating innate immune cells, including neutrophils, monocytes, and eosinophils, by signaling through specific cell surface receptors (44, 45). For example, high affinity GM-CSF receptors are found principally on myeloid cells, including neutrophils, MNPs, and eosinophils.

Numerous cytokines, including GM-CSF, have pleiotropic effects that vary depending on local cytokine concentrations in the vicinity of specific cell surface receptors (44, 46–48). GM-CSF pleiotropism relies on higher order extracellular assembly of heterodimeric receptor chains, as well as 4 distinct intracellular signaling pathways, including: mitogen activated protein kinase (MAPK); nuclear factor kappa-B cells (NFĸB); phosphoinositide 3-kinase (PI3K); and signal transducer and activator of transcription 5 (STAT5). Such diversity explains GM-CSF’s capacity to generate survival, differentiation, activation, and/or proliferation signals, depending on cytokine concentration at the receptor level, as well as other local stimuli. At low GM-CSF concentrations, PI3K signaling leads to survival, whereas at high concentrations, PI3K, MAPK, and STAT5 signaling lead to survival and cell proliferation (47, 48). Correspondingly, both ligand and dose-specific effects on NFĸB signaling have been described in primary macrophages (49, 50), and such effects have been observed to influence epigenomic programming (51).

Recently, GM-CSF effects on MNP metabolism were revealed in mouse models with disrupted GM-CSF signaling (52). These models demonstrated a critical role of GM-CSF in maintaining mitochondrial structure and function, as well as fatty acid beta oxidation, tricarboxylic acid cycle activity, oxidative phosphorylation, and adenosine triphosphate (ATP) generation. These effects of GM-CSF on metabolic capacity enable MNPs to fulfill energy-intensive innate immune functions, including: respiratory burst generation, phagocytosis, antigen presentation, cytokine secretion, and efferocytosis (52–54). All these functions rely on metabolic energy and fail in its absence. By extension, metabolic capacity in tissue-bound macrophages throughout the body may be maintained by ongoing low-level and/or pulsatile GM-CSF expression. This activity aligns with known ongoing low-level yet plastic GM-CSF expression by diverse cell types, including endothelial, epithelial, and immune cells, as well as fibroblasts (2, 55).

We hypothesize that myeloid proliferation driven by high cytokine levels in sepsis leads to cell division that outpaces time and/or GM-CSF stimulation needed for maturation of cellular metabolic capacity. Thus, sustained high inflammatory cytokine secretion may counterintuitively result in degradation of metabolic capacity of newly formed MNPs to fulfill immune functions. Consequently, immature MNPs with insufficient metabolic capacity to support normal innate immune functions appear “immunosuppressive.” In support of this model, GM-CSF reverses monocyte hyporesponsiveness in multiple in-vitro systems (56–60). Multiple reports support that GM-CSF increases blood monocyte levels, upregulates monocyte responsiveness, and increases HLA-DR expression, which is known to enhance antigen presentation and adaptive immune responses (48, 54, 61, 62).

The polymorphic MHC gene family in humans is on chromosome 6 and encodes multiple class II MHC proteins, including HLA-DP, HLA-DQ, and HLA-DR (68). Historically, these proteins were recognized as transplantation antigens, serving as targets for immune rejection of transplanted tissue. During infection, MNPs phagocytose pathogens that are then digested to yield foreign peptides that combine intracellularly with class II MHC proteins, such as HLA-DR (4, 69). Normally, monocytes and macrophages express HLA-DR levels ranging from 15,000 antibodies bound per cell (Ab/c) to as high as 60,000 Ab/c (70, 71); and a commonly used lower limit of HLA-DR in healthy subjects is 15,000 Ab/c (33, 72, 73). The large spread in the reported HLA-DR levels is most likely explained by biologic variability, as well as differences in assay reagents and flow cytometry methods used over years to quantitate HLA-DR expression levels (74–77). Peptide-MHC complexes are transported to the cell surface where they mediate antigen-specific recognition by CD4+ helper T lymphocytes. Once activated by peptide-MHC recognition, CD4+ T lymphocytes boost adaptive immune responses by activating other T and B lymphocytes that can recognize and target the invading pathogen (3, 4, 26). Because HLA-DR functions as the bridge between innate MNPs and antigen-specific T lymphocytes, low HLA-DR levels lead to diminished antigen presentation and reduced adaptive immune activation (4, 78). When HLA-DR is low, CD4+ T lymphocytes are not activated, hence cannot augment either B-cell stimulation to produce specific antibodies or CD8+ cytotoxic T lymphocyte generation to target infected cells directly (3, 4).

Despite HLA protein diversity, common determinants recognized by monoclonal antibodies enable flow cytometric quantitation of surface class II MHC expression level on blood cells (4, 40, 76). Although flow cytometry enables monocyte HLA-DR (mHLA-DR) quantitation, other cells expressing HLA-DR are also detected, including DCs, macrophages, B cells, and T cells (4, 30, 76, 79–82). Thus, to generate mHLA-DR specificity, cells are also stained for CD14 (also known as the LPS receptor), of which, expression is restricted to monocytes. Combined CD14 and HLA-DR staining enables quantitation of CD14+ classical and intermediate monocytes, the most abundant and rapidly replenished populations in blood. Results are typically reported either as percent of CD14+ monocytes expressing HLA-DR or as mean fluorescence intensity (MFI) of antibody against HLA-DR on CD14+ monocytes (77, 83).

HLA-DR downregulation and reduced monocyte responsiveness are described features of sepsis IP (4, 26). As detailed in Table 1 , low HLA-DR correlates with adverse clinical outcomes, including increased risk for nosocomial infections, end-organ failure, longer ICU hospitalizations, and mortality (30, 33, 75, 84–89, 91–93).

Inter-laboratory variability initially posed a challenge to analytic validity of HLA-DR testing to identify sepsis IP (4). Now, a system offering standardized quantitative measurement of cell surface HLA-DR proteins (Quantibrite™; Becton, Dickinson and Company [BD]) is available. Developed in 2001, Quantibrite™ beads allow estimation of Ab/c, enabling monocyte cell surface HLA-DR protein quantitation to stratify patients based on mHLA-DR levels (79, 94–96). This assay uses phycoerythrin (PE)-labeled anti-HLA-DR monoclonal antibodies for estimating Ab/c (97). Geometric MFI values can be analyzed further to calculate numbers of Ab/c, which represents numbers of HLA-DR proteins on the monocyte surface (96, 97). Using standard instrument settings, flow cytometry data are converted into number of PE molecules per cell. Based on a known ratio of PE to antibodies against HLA-DR, Ab/c can be calculated, hence quantitating HLA-DR protein on CD14+ monocytes. With Quantibrite™, moderate immunosuppression is defined as about 10,000–15,000 Ab/c (74, 79). In several studies, a cut-off value of 8000 Ab/c was used to indicate IP. HLA-DR levels below 8000 Ab/c indicate more severe sepsis IP (4, 79). In some studies, 30% CD14+/HLA-DR+ cells corresponded to 5000 Ab/c for severe IP, whereas 45% CD14+/HLA-DR+ cells corresponded to about 8000 Ab/c for moderate IP (4, 79). Numerous studies have employed Quantibrite™ to measure HLA-DR-defined IP (26, 33, 70, 73–75, 77, 88, 90, 98–101).

Multiple literature analyses support mHLA-DR expression by flow cytometry as a sepsis IP biomarker and mortality predictor (80). One such review evaluated mHLA-DR in patients with complicated intra-abdominal infections and sepsis from 12 studies (n=761) (102). Results from 10 of these studies showed strong associations between low mHLA-DR expression and mortality. By contrast, 2 studies showed no prognostic value of mHLA-DR expression level. Proposed factors contributing to nonsignificant results in these 2 studies include: homogeneity of enrolled patients, young age, small sample sizes, and heterogeneity among experimental protocols (77, 87, 100, 103). Another review assessed mHLA-DR in critically ill patients with coronavirus disease of 2019 (COVID-19), sepsis, or bacterial infections from 15 studies (n=1160) (104). Of these studies, 4 monitored mHLA-DR expression with flow cytometry by a standardized protocol that reported results as Ab/c. Initial mHLA-DR expression was lower for COVID-19 patients than for controls (10,000 Ab/c vs 15,000 Ab/c) yet higher for COVID-19 patients than for septic shock patients (10,000 Ab/c vs 5000 Ab/c). Lower mHLA-DR expression was associated with higher ICU mortality and greater disease severity at hospital admission. A meta-analysis evaluated 8 prospective cohort studies to evaluate HLA-DR as a biomarker for sepsis in patients after trauma (n=639) (105). Results from 7 studies showed that HLA-DR by flow cytometry for detecting sepsis IP had a pooled sensitivity of 81% and a pooled specificity of 67%.

While various thresholds for detecting IP have been proposed, a minimum threshold for raising secondary infection and mortality risks has to date been neither standardized nor adopted (3, 4, 77, 84, 106). Hence, HLA-DR testing by flow cytometry can now be implemented with analytic validity, and multiple studies support its clinical validity. Yet, both a definitive threshold for sepsis IP and clinical utility for therapeutic response prediction remain, for now, unconfirmed.

Notably, 3 additional approaches to mHLA-DR measurement have been investigated. First, measurement of HLA-DR expression levels by polymerase chain reaction (PCR) was explored in several clinical studies (4, 70, 80). In 1 such study, quantitative real-time PCR (qRT-PCR) and mHLA-DR flow cytometry were used to assess HLA-DR and class II transactivator (CIITA) in patients with bacteremic sepsis (n=60) (70). TaqMan gene qRT-PCR expression assays were used to measure HLA-DR-α subunit (HLA-DRA) and CIITA, whereas Quantibrite™ was used to measure mHLA-DR by flow cytometry. Similar patterns for initial reductions in HLA-DRA, mHLA-DR, and CIITA were all followed by subsequent increases over time (p<0.001). Hence, qRT-PCR yields results somewhat similar to flow cytometry with low variability and reproducibility. While qRT-PCR may be robust for detecting HLA-DR expression in patients with sepsis, qRT-PCR results are non-specific for monocytes since circulating DCs, B cells, and activated T cells also express HLA-DR (70, 80). As such, it may not reliably reflect mHLA-DR expression in monocytes that drives sepsis IP (4, 80, 107).

Second, myeloid-derived suppressor cells (MDSCs) have been described in patients with sepsis (4, 108–110). Although not standardized, all MDSC descriptions include “low HLA-DR expression.” Hence, MDSCs are invariably monocytes with low HLA-DR. In sepsis, MDSCs associate with: prolonged immunosuppression, diminished T cell functions, development of nosocomial infections, higher reinfection rates, and hospital readmissions (4, 109, 111, 112).

Finally, several studies support that dynamic changes by serial mHLA-DR monitoring might predict mortality better than static mHLA-DR monitoring (80, 113, 114). Correspondingly, persistence of low mHLA-DR levels suggests slow or no recovery from sepsis IP (4, 12, 13, 113, 115, 116). Given inter-individual variability of mHLA-DR in sepsis, dynamic change or HLA-DR slope might increase prognostic significance of low mHLA-DR expression for mortality prediction (4, 13, 93, 98, 117). Thus far, no standardized approaches to serial mHLA-DR monitoring have been either developed or tested prospectively.

As in adults, low mHLA-DR in children associates with nosocomial infections and mortality (92, 118–121). Nonetheless, patient age affects monocyte subtypes and function, so direct comparison of adults vs children may be confounding (121). While adult monocytes are predominantly classical (CD14+/CD16-), neonatal monocytes are mostly intermediate (CD14+/CD16+) or nonclassical (CD14-lo/CD16+) subtypes that express lower levels of HLA-DR (121–123). These differences result in reduced T cell activation in neonates compared with adults (121). Also, neonates have proportionally more regulatory T cells than adults, and that difference may also limit immune responses in children with sepsis (124, 125).

One study compared mHLA-DR expression among critically ill children with sepsis, trauma-related hospital acquired infection, or recent surgery (n=37; median age, 9 years) vs healthy control children (n=37; median age, 3 years) (92). Results showed lower mHLA-DR expression (67% vs 95%; p<0.001) and lower mHLA-DR MFI (3219 vs 6545; p<0.001) for critically ill children vs healthy controls at all examined time points, in particular on classical monocytes and in children admitted for sepsis. Another study evaluated blood samples in hospitalized children with sepsis (n=30) vs healthy controls (n=21) for mHLA-DR expression using Quantibrite™ technology (98). As with adults, mHLA-DR expression in pediatric patients with sepsis was lower than that in controls (p=0.0001). Finally, a prospective, single-center, observational study evaluated mHLA-DR levels using Quantibrite™ in children with septic shock admitted to a pediatric ICU (n=26; median age, 2 years) with healthy controls (n=30) (90). As seen elsewhere, mHLA-DR levels were lower for patients with septic shock than for healthy controls (p<0.001).

While HLA-DR is well-documented for sepsis IP detection, other potential biomarkers are also being explored. LPS-induced TNF-α production from peripheral blood cells reflects innate immune system function via myeloid cell capacity to respond to an inflammatory stimulus (3, 30, 126). Although both ex-vivo TNF-α secretion and HLA-DR expression assess monocyte dysfunction via metabolic capacity to fulfill basic immune functions, ex-vivo TNF-α secretion is less specific for monocytes as responding myeloid cells include both neutrophils and monocytes. Independent of sepsis IP, TNF-α levels may also be influenced by a variety of other factors, such as: type of LPS used, blood volume, incubation conditions, and LPS concentration (3).

In contrast to substantial literature examining mHLA-DR prognostic significance in adult sepsis, there are fewer reports on TNF-α, and most are in small groups of children (3, 74, 91, 127, 128). Overall, these studies support clinical validity of measuring TNF-α by ex-vivo LPS stimulation. Although a few studies describe standardized protocols for measuring LPS-induced TNF-α production for sepsis, scalable analytic validity may remain challenging (29, 129). As seen for HLA-DR quantitation, no receiver operating characteristic (ROC) curve analysis has been performed to define a TNF-α threshold for sepsis IP.

ALC is another laboratory parameter that reflects immune system function (29, 130). The reference range for ALC varies with age. Normal for adults varies between 1000 and 4800 cells/µL, and for children, between 3000 and 9500 cells/µL (131, 132). Lymphopenia occurs when a patient’s ALC is below normal and can increase risk for infection (133).

In sepsis IP, circulating lymphocyte populations (e.g., CD4+ T cells, CD8+ T cells, B cells) are characteristically reduced due to tissue sequestration and apoptosis (26). Reductions at sepsis onset typically persist for up to 28 days. Increased apoptosis of both innate immune cells and adaptive immune cells in sepsis results in leukopenia, which associates with higher risks of secondary infections and death (134–136).

A retrospective, single-center cohort study monitored blood parameters in patients with bacteremia and sepsis (n=335) for secondary infection risk and mortality (130). Results showed higher ALC at Day 4 for survivors vs non-survivors (1100 cells/µL vs 700 cells/µL; p<0.0001). Also, 28-day and 1-year mortality were higher in severe (40% vs 10% and 58% vs 29%; p<0.001) and moderate (25% vs 10%; p=0.003, and 40% vs 29%; p=0.025) lymphopenia vs those without persistent lymphopenia. Multivariable analysis showed that Day 4 ALC was associated with both 28-day (odds ratio [OR], 0.68; p=0.009) and 1-year mortality (OR, 0.74; p=0.008). Severe persistent lymphopenia (< 0.6 x 10³ cells/μL) was also associated with development of secondary infections (OR, 2.11; 95% confidence interval [CI], 1.02–4.39; p=0.04) (26, 130). Thus, persistent lymphopenia on the fourth day after a sepsis diagnosis predicted mortality and may be a valid marker of sepsis-induced immunosuppression.

In another single-center study, cross-sectional analysis was performed of ALC as an outcome predictor in patients with sepsis presenting to an emergency department (n=124) (137). Results showed a higher need for ICU admission (51.9% vs 14%; p<0.001) and higher rates of 28-day mortality (88.1% vs 11.9%; p<0.001) for patients with lymphopenia vs those without lymphopenia. In addition, age and sequential organ failure assessment (SOFA) scores were higher for patients with lymphopenia vs without.

Lower monocyte counts are also seen in sepsis and can impact health outcomes (64). A retrospective, single-center database analysis of patients with sepsis (n=2012) showed higher 28-day mortality rates, higher bacteremia rates, and higher incidence of organ dysfunction for patients with initial monocyte counts < 250 cells/μL.