The contribution of the monocyte-macrophage lineage to immunotherapy outcomes

Visual Abstract

Abstract

Macrophages execute core functions in maintaining tissue homeostasis, in which their extensive plasticity permits a spectrum of functions from tissue remodeling to immune defense. However, perturbations to tissue-resident macrophages during disease, and the subsequent emergence of monocyte-derived macrophages, can hinder tissue recovery and promote further damage through inflammatory and fibrotic programs. Gaining a fundamental understanding of the critical pathways defining pathogenic macrophage populations enables the development of targeted therapeutic approaches to improve disease outcomes. In the setting of chronic graft-versus-host disease (cGVHD), which remains the major complication of allogeneic hematopoietic stem cell transplantation, colony-stimulating factor 1 (CSF1)–dependent donor-derived macrophages have been identified as key pathogenic mediators of fibrotic skin and lung disease. Antibody blockade of the CSF1 receptor (CSF1R) to induce macrophage depletion showed remarkable capacity to prevent fibrosis in preclinical models and has subsequently demonstrated impressive efficacy for improving cGVHD in ongoing clinical trials. Similarly, macrophage depletion approaches are currently under investigation for their potential to augment responses to immune checkpoint inhibition. Moreover, both monocyte and tissue-resident macrophage populations have recently been implicated as mediators of the numerous toxicities associated with chimeric antigen receptor T-cell therapy, further highlighting potential avenues of macrophage-based interventions to improve clinical outcomes. Herein, we examine the current literature on basic macrophage biology and contextualize this in the setting of cellular and immunotherapy. Additionally, we highlight mechanisms by which macrophages can be targeted, largely by interfering with the CSF1/CSF1R signaling axis, for therapeutic benefit in the context of both cellular and immunotherapy.

While T cells are central to the activity of allogeneic stem cell transplant, chimeric antigen receptor T-cell therapy, and many bispecific antibodies against hematological malignancies, other hematopoietic cells play important roles in modulating responses. Adams and colleagues review understanding of the ontogeny, function, and contributions of monocytes and macrophages to clinical outcomes with these immunotherapies. The authors further discuss how newly available agents targeting these cells can be used for therapeutic advantage.

Introduction

Monocytes and macrophages represent key cell types for tissue protection and preservation of homeostasis. Beyond their capacity as innate immune defenders, tissue-resident macrophages (TRMs) also contribute to tissue development and remodeling, with specific features and functions depending on their ontogeny and local environment.^1,2 However, the dynamics of the monocyte-macrophage lineage are largely perturbed during disease, and contribute to maladaptive processes with severe symptomatic repercussions.³ Notably, the clinical outcomes and complications that arise after adoptive cell and immunotherapies for the treatment of hematological malignancies are associated with significant changes in monocyte and TRM populations, leading to disease features that have proven difficult to treat adequately.⁴ The critical reliance of macrophages on different signaling pathways, such as colony-stimulating factor 1 (CSF1), can be used to devise specific therapeutic strategies to target this cell type. As such, approaches to functionally modulate or deplete macrophages have recently been shown to significantly improve clinical outcomes.⁵

Chronic graft-versus-host disease (cGVHD) remains the primary late complication of allogeneic hematopoietic stem cell transplantation (allo-HSCT), in which fibrotic manifestations dependent on donor-derived macrophages cause significant morbidity and impair quality of life.⁴ Chimeric antigen receptor (CAR) T-cells as therapy for B-cell malignancies and multiple myeloma are associated with complications including cytokine release syndrome (CRS), hemophagocytic lymphohistiocytosis (HLH)/macrophage activation syndrome (MAS), and immune effector cell–associated neurotoxicity syndrome (ICANS). Recent studies have implicated both monocytes and TRMs, respectively, in mediating these complications.^6,7 Tumor-associated macrophages (TAMs) have been directly implicated in promoting tumorigenesis,⁸ and can hinder the capacity for immune checkpoint inhibitors (ICIs) to reinvigorate T-cell antitumor responses.⁹ Moreover, TRMs have recently been described to contribute to immune-related adverse events (iRAEs) associated with ICIs.¹⁰ As such, the modulation and/or depletion of macrophages is of particular interest in each of these settings. In this review, we summarize the current state of knowledge on the origins and functions of the monocyte-macrophage lineage, and how this may be perturbed in each of these settings. Moreover, we highlight recent studies demonstrating the efficacy of targeting macrophages for improving outcomes after therapy for hematological malignancies.

Origins and functions of the monocyte-macrophage lineage

Monocytes and macrophages belong to the mononuclear phagocyte family but arise from distinct developmental origins and exhibit specific functions (Figure 1).¹ Both embryonic and adult hematopoiesis in the fetal liver and bone marrow (BM), respectively, give rise to monocytes throughout the lifespan. In the adult BM, common myeloid progenitor cells produce lineage-restricted granulocyte-monocyte, monocyte-dendritic cell, and common monocyte-committed progenitors, beyond which 2 broad and functionally distinct categories of “classical/inflammatory” (Ly6C^hiCCR2⁺ in mice, CD14⁺CD16^– in humans) and “nonclassical/patrolling” (Ly6C^loCCR2^– in mice, CD14^loCD16⁺ in humans) monocytes arise.^11,12 Ly6C^hi monocytes are the obligatory precursors to Ly6C^lo monocytes in circulation, differentiating via an intermediate population in a CSF1-dependent manner and regulated by the transcription factors Nr4a1 and C/EBPβ.^13,14 The former are rapidly recruited to sites of infection and/or tissue damage with concomitant inflammation, relying heavily on the C-C chemokine receptor 2 (CCR2)/C-C motif chemokine ligand 2 (CCL2) axis, among other chemokines (eg, CCL5, CCL7, and CX3CL1).¹⁴ Upon entry to an inflamed tissue, Ly6C^hi monocytes may give rise to monocyte-derived inflammatory Ly6C^hi macrophages, which subsequently release cytokines (eg, tumor necrosis factor [TNF], interleukin-1β [IL-1β], and nitric oxide), and process and present antigen,¹⁴ or to Ly6C^lo macrophages.¹⁵ Typically, nonclassical Ly6C^lo monocytes will instead actively patrol the vascular lumen, maintain endothelial integrity, and differentiate into macrophages within tissues,¹⁶ in which Ly6C^lo macrophages demonstrate prolonged CSF1 dependence.¹⁷ Ly6C^hi macrophages can phenotypically switch to Ly6C^lo macrophages within tissues depending on the provision of CSF1, IL-4, IL-10, and other signaling cues including the CX3CL1/CX3CR1 axis.^14,18, 19, 20 As such, the trajectory to Ly6C^lo macrophages is highly plastic and environmentally dependent.

Figure 1.

Overview of differentiation pathways of monocytes and TRMs. TRM populations arise via distinct differentiation pathways, derived from either erythromyeloid progenitors or monocytes generated from the BM. Common myeloid progenitors in the BM generate lineage-restricted granulocyte-monocyte and monocyte-dendritic cell progenitors, which subsequently produce monocyte-committed progenitors. Classical/inflammatory circulating monocytes differentiate in a CSF1-dependent manner into nonclassical/patrolling monocytes. Nonclassical monocytes may give rise to monocyte-derived macrophages in distinct tissue niches, which require continual input from the BM to replenish the macrophage population. Alternatively, embryonically derived macrophages seed tissue prenatally, are sustained via self-renewal, and remain critically dependent on CSF1R signaling for survival.

Within their new tissue niche, monocyte-derived macrophages acquire functions similar to the local macrophage population and will contribute to the induction and/or resolution of inflammation depending on the experimental system and milieu in which they differentiate.²¹ Recent work using single-cell RNA sequencing highlighted that during inflammation, tissue macrophages, and most notably the infiltrating monocyte-derived constituents, will commit to 1 of 4 conserved activation states, ascribing functions of phagocytosis, oxidative stress, inflammation, or remodeling, in which the commitment is both stimulus and tissue dependent.²² Notably, embryonically seeded macrophages generally exhibit more transcriptional stability during an inflammatory response, whereas the infiltrating monocyte-derived macrophages tend to contribute the most functional heterogeneity.²² Such pioneering studies are critical in efforts to redefine the spectrum of monocyte/macrophage functionality, which has historically been limited to the M1 “inflammatory”/M2 “anti-inflammatory” dichotomy. For the purpose of this review, we refer to specific markers and functions of monocytes/macrophages to capture their nuances in each described setting.

Previous dogma held that this monocyte-to-macrophage differentiation pathway was also responsible for generating and constantly renewing organ-specific TRM populations in homeostasis.²¹ However, fate mapping and sequencing efforts have elucidated an embryonic origin for most TRM subsets, excluding gut and certain brain border–associated macrophages, which require ongoing BM input.^23,24 Erythromyeloid progenitors seed developing organs prenatally, acquire specific transcriptional programs, and are sustained during homeostasis by self-renewal, giving rise to notable populations including microglia in the brain parenchyma, epidermal Langerhans cells, and Kupffer cells of the liver.²⁵ The local milieu shapes signature TRM characteristics, exemplified by the role of transforming growth factor β (TGF-β) in defining unique molecular and functional signals in microglia, whereas other central nervous system (CNS)-associated macrophages differentiate independently of this cytokine.²⁶ TRM populations subsequently contribute to homeostatic tissue-specific development and remodeling, maintenance of vascular integrity, debris clearance, and immune surveillance.¹ However, disease-related shifts in local cytokine and growth factor availability skews macrophage function to become either harmful or beneficial, which may also fluctuate temporally. For example, disease-associated microglia emerge during neurodegeneration with a distinct protective signature²⁷; however, chronic microglia stimulation induces dysfunctional subpopulations that subsequently promote pathology progression.²⁸

CSF1R inhibition as a strategy to target macrophages

The development and homeostasis of macrophage populations is regulated via signaling of CSF1 and IL-34 through the class 3 receptor tyrosine kinase CSF1 receptor (CSF1R; reviewed elsewhere).^29,30 Although ligand availability varies across tissues, provision of CSF1 is a major determinant of the number of resident macrophages within a given organ.³¹ Human monocytes differentiated in vivo with CSF1 adopt an immunosuppressive phenotype.³² Similarly in mice, CSF1 elicits macrophages supportive of tissue remodeling and fibrosis, highlighted by expression of CD206 and TGF-β. In contrast, proinflammatory macrophages express TNF, major histocompatibility complex class 2, and inducible nitric oxide synthase (iNOS).³³ In both steady state and disease, CSF1-neutralizing antibodies, and CSF1R-blocking antibodies and small-molecule tyrosine kinase inhibitors represent viable tools for examining CSF1R dependence, and as valuable therapeutic strategies for macrophage depletion both experimentally and clinically (Figure 2). However, circulating CSF1 levels are regulated by the rate of release relative to the rate of clearance, predominantly via receptor-mediated endocytosis and intracellular degradation by CSF1R-expressing cells.³⁴ Therefore, rapid systemic increases in CSF1 levels are common after CSF1R antibody blockade, which should be considered in the setting of ongoing inflammation.³⁵ Notably, although the differentiation of Ly6C^hi to Ly6C^lo monocytes is a CSF1R-dependent phenomenon, the Ly6C^hi population can persist in the absence of CSF1R signaling, and can potentially accumulate during blockade and contribute to exacerbated inflammation.³³ Alternative approaches to targeting Ly6C^hi inflammatory monocytes have been established, such as immune-modifying particles targeting the MARCO scavenging receptor (reviewed elsewhere).³⁶ Although CSF1R inhibitor withdrawal readily reverses macrophage depletion, the duration of therapy must also be considered as to the potential long-term effects of sustained TRM loss and absence of their tissue-supporting functions. A summary of clinically and experimentally available agents with documented uses in the setting of hematological malignancies is provided in Table 1.

Figure 2.

Sites of activity of inhibitors of the CSF1 signaling pathway. CSF1 and IL-34 signal through the class 3 receptor tyrosine kinase CSF1R to regulate monocyte and macrophage differentiation and survival. CSF1-neutralizing antibodies bind to the circulating ligand to prevent receptor binding. CSF1R-blocking antibodies prevent signaling induced by either CSF1 or IL-34. Small-molecule tyrosine kinase inhibitors act intracellularly to block phosphorylation, thus preventing receptor activation.

Table 1.

CSF1R inhibiting agents evaluated clinically and experimentally for hematological malignancies

Agent	Use parameters	Specifications and target involvement	Remarks and outcomes
Axatilimab (commercially Niktimvo, Incyte Corporation)	FDA approved for cGVHD after failure of at least 2 prior lines of systemic therapy	Humanized IgG4 monoclonal antibody targeting human CSF1R	Efficacy, safety, and tolerability assessed in phase 1/2 open-label study (NCT03604692)37 Randomized phase 2 AGAVE-201 study (Syndax, Incyte; NCT04710576) met primary efficacy at 3 different doses. Superior efficacy at lowest dose (0.3 mg/kg every 2 weeks)5
Emactuzumab (RG-7155)	Experimental	Humanized IgG1 monoclonal antibody targeting human CSF1R	Used to examine impact of macrophage depletion on chronic lymphocytic leukemia38
M279 (Amgen)	Experimental	Monoclonal IgG1 antibody against mouse CSF1R	Used to examine and confirm the CSF1R dependency of donor macrophages in cutaneous and pulmonary cGVHD33 Confirmed role of TAM in impairing myeloma control39
CS4 (human), CS7 (mouse) (Eli Lilly & Co)	Experimental	Human and mouse CSF1R inhibitors	Used to explore the potential of macrophage depletion for multiple myeloma treatment40
Pexidartinib (PLX3397, commercially Turalio)	FDA approved for treatment of adults with symptomatic tenosynovial giant cell tumor	Selective small-molecule inhibitor of the human CSF1R, KIT, and FLT3-ITD kinases	Evaluated in phase 1/2 study for activity in relapsed/refractory FLT3-ITD–mutant AML (NCT01349049)41 Evaluated for maximum tolerable dose in phase 1 trial for pediatric refractory leukemias (NCT02390752)42
Edicotinib (JNJ-40346527)	Experimental	Selective inhibitor of the human CSF1R tyrosine kinase	Evaluated in phase 1/2 study for treatment of relapsed or refractory classical Hodgkin lymphoma (NCT01572519). Well tolerated but limited efficacy as monotherapy43 Phase 2 study for treatment of patients with relapsed or refractory AML terminated for insufficient enrollment (NCT03557970)
GW-2580	Experimental	Selective inhibitor of human c-FMS kinase	Used to examine role of CSF1R signaling in AML cell viability44 Used to study the role of CSF1R signaling in monocyte-dependent survival of mantle cell lymphoma45

c-FMS, encodes the CSF1R; FDA, US Food and Drug Administration; FLT3-ITD, FMS-like tyrosine kinase-3 internal tandem duplication; IgG, immunoglobulin G.

Monocytes and macrophages as mediators of deleterious therapy outcomes

Macrophages as a novel therapeutic target in cGVHD

Allo-HSCT offers a curative therapy for hematological malignancies, among other disorders of the blood.⁴⁶ Despite prophylaxis using calcineurin inhibitors (cyclosporine and tacrolimus), methotrexate, and posttransplant cyclophosphamide, GVHD still occurs in 30% to 70% of allo-HSCT recipients.^47,48 GVHD can be temporally categorized based on the involvement of different immune cell populations and disease mechanisms, resulting in characteristic pathologies. Acute GVHD (aGVHD) is underlain by interactions between donor T cells and host antigen presenting cells in the midst of heightened inflammation induced by pretransplant conditioning,⁴⁹ and is a major risk factor for developing cGVHD. Presentations include skin rash, diarrhea, liver dysfunction, and neurological deficits.^50,51 cGVHD presents with heterogeneous multiorgan clinical manifestations,⁵² including sclerotic skin, lichen planus lesions of the skin and mucous membranes, bronchiolitis obliterans of the lungs, and CNS abnormalities, often culminating in severe disability.^53,54 cGVHD remains a major contributor to impaired quality of life⁵⁵ and is the leading cause of nonrelapse morbidity and mortality in patients surviving >2 years after transplant.⁵⁶ Moreover, the preference for peripheral blood stem cell transplantation (PBSCT) over conventional BM transplantation in current clinical practice is associated with an increased incidence of cGVHD, which is typically less responsive to historical treatments aimed at aGVHD.57, 58, 59

Monocytes in aGVHD

The involvement of monocytes and macrophages in aGVHD and cGVHD pathology differs both temporally and spatially. Early studies described a marked elevation of circulating serum CSF1 levels during aGVHD, which has the capacity to expand splenic macrophages as a response to ongoing donor T-cell proliferation.³⁴ Preconditioning regimens also induce the release of damage-associated molecular patterns from the gastrointestinal tract early after transplant.⁶⁰ Residual host inflammatory monocytes bind damage-associated molecular patterns such as adenosine triphosphate from damaged cells via expression of the P2Y2 purinergic receptor, leading to their recruitment to the gut and subsequent production of reactive oxygen species, inducing tissue damage.⁶⁰ Similarly, elevated levels of S100 proteins in the serum and gastrointestinal tract bind to toll-like receptor 4 on monocytes, resulting in their activation and subsequent induction of T helper 17 cell responses.⁶¹ These data suggest that early monocytic events help to shape inflammatory responses during aGVHD development. In human studies, elevated numbers of circulating CD14⁺ monocytes before conditioning is associated with a greater risk of developing grade 2 to 4 aGVHD,⁶² potentially explained by the capacity for monocytes from both patients with aGVHD and cGVHD to induce more IL-17⁺IFNγ⁺ CD4 T cells than monocytes from healthy donors or donors without GVHD.⁶¹ Comparison of classical, intermediate, and nonclassical human monocytes revealed that each shows a distinct profile of cytokine expression (interferon gamma [IFN-γ], TNF, IL-6, and IL-10), highlighting the potential for each subset to contribute distinctly to disease.⁶³ Moreover, monocytes from aged donor grafts were shown to skew toward a proinflammatory transcriptional signature,⁶⁴ and harbored more nonclassical monocytes when compared with young donor grafts, which was associated with increased incidence of grade 2 to 4 aGVHD.⁶⁵ Collectively, these findings demonstrate important and varied contributions of monocyte subsets to GVHD development. Alternatively, PBSCT grafts sourced from granulocyte colony-stimulating factor-mobilized donors contain a regulatory population of CD34⁺ monocytes with the capacity to suppress alloreactive T-cell responses,⁶⁶ highlighting the importance of the local microenvironment and signaling cues in dictating monocyte-related outcomes.

Macrophages in aGVHD

Studies examining the specific contribution of antigen presenting cells in GVHD induction have revealed a protective role for conditioning-resistant host TRMs, which control alloreactive T-cell responses in the periphery early after transplant, putatively via the CD47 pathway.⁶⁷ Although, a notable exception to this generalization is the requirement for host microglia in mediating T-cell infiltration and inflammation in the brain during aGVHD.⁶⁸ However, peripheral expansion of recipient splenic macrophages by CSF1 administration is protective against aGVHD, whereas pretransplant macrophage depletion via the administration of anti-CSF1R antibody has consistently proven to exacerbate aGVHD severity and mortality.^35,67 Similarly, CSF1R blockade early after transplant augmented aGVHD in both the gastrointestinal tract and CNS,⁶⁹ and thus, collectively, these findings highlight the importance of host macrophages in restraining donor T-cell responses, and the potential detriment of CSF1R blockade during active inflammation. In contrast, donor monocyte-derived cells respond to T-cell–derived granulocyte macrophage colony-stimulating factor (GM-CSF) and promote tissue damage by direct IL-1β and reactive oxygen species production.⁷⁰ Moreover, donor monocyte-derived macrophages accumulate in both the skin and gut during aGVHD, in which skin-engrafting macrophages rapidly adopt the phenotype of the resident population.71, 72, 73 Transcriptionally, GVHD-related macrophages upregulate genes related to proinflammatory cytokine production and antigen presentation, and can induce allogeneic T-cell proliferation, highlighting their capacity to contribute to disease progression.⁷¹ As such, the accumulation of donor macrophages in the skin is reported as a negative prognostic factor for steroid-refractory GVHD.^73,74

Macrophages in cGVHD

Considering cGVHD, only a few human studies have highlighted perturbations to the circulating monocyte compartment, whereby CD14⁺ monocytes tend to accumulate in the BM and show increased expression of costimulatory molecules,⁷⁵ whereas intermediate monocytes expand in the blood.^76,77 However, although such studies have the capacity to measure temporal fluctuations in monocyte proportions and surface marker expression, they have been limited by an inability to correlate blood monocyte changes with that of macrophages in target organs, and therefore the implications for altered monocyte populations in cGVHD are largely unknown.⁷⁸ Rather, the mechanisms by which donor macrophages contribute to target organ pathology in cGVHD have been extensively detailed in preclinical studies, paving the way for testing novel therapeutic strategies (Figure 3). Transfer of granulocyte colony-stimulating factor–mobilized grafts in both major and minor histocompatibility mismatch mouse models induced GVHD, whereby development of cutaneous pathology was found to be IL-17A dependent.⁷⁹ Subsequent studies in both PBSCT and BM transplantation models revealed that transfer of Il-17a^−/− grafts could prevent skin pathology.⁷⁹ Notably, IL-17 acted as a macrophage chemoattractant, whereby F4/80⁺ skin-infiltrating macrophages were largely absent in recipients of Il-17a^−/− grafts.⁷⁹ Studies in the same PBSCT model proposed that the F4/80⁺CD11b⁺Ly6C^lo donor BM-derived monocyte population is the precursor to pathogenic macrophages that sequester in the skin.³³ These macrophages polarize toward a profibrotic phenotype in the skin, in which they produce TGF-β, which is known to be a potent fibroblast activator, resulting in collagen deposition and development of scleroderma.^33,80 Notably, skin-infiltrating macrophages were disease promoting and critically dependent on CSF1/CSF1R signaling for differentiation and survival, demonstrated by their widespread depletion upon transfer of Csf1r^−/− allografts or administration of an anti-CSF1R monoclonal antibody, which subsequently reduced pathology.³³ Similar IL-17A– and CSF1R-dependent pathogenicity was described for bronchiolitis obliterans of the lungs, indicating that targeting donor-derived macrophages during cGVHD may have therapeutic activity across multiple peripheral organs. Of note, the infiltration of donor-derived macrophages to the brain is a key feature of CNS cGVHD.⁸¹ However, treatment with an anti-CSF1R antibody failed to induce widespread CNS macrophage depletion, nor could it attenuate features of neuroinflammation.⁶⁹ These findings highlight divergent outcomes between the CNS and periphery in response to systemic therapies during cGVHD, and identifies the need for including monitoring of CNS presentations during ongoing clinical trials, and the development of alternative brain-targeted therapies for patients exhibiting CNS cGVHD.

Figure 3.

Contribution of macrophages to cGVHD pathology and therapeutic benefit of CSF1R antibody blockade. IL-17 production from T helper 17 (Th17)/Tc17 cells functions as a chemoattractant for Ly6C^lo monocytes, which subsequently differentiate into CD206⁺ macrophages in the presence of CSF1 within tissue. Production of profibrotic factors by these donor-derived macrophages leads to fibroblast activation and collagen deposition, resulting in sclerosis. Antibody blockade of the CSF1R with axatilimab reduces the frequency of nonclassical monocytes and tissue-resident CSF1R-expressing macrophages, subsequently attenuating fibrotic pathology and cGVHD severity.

Clinical studies have also confirmed the presence of pathogenic macrophage populations in the skin during cGVHD, associated primarily with aberrant tissue repair mechanisms and fibrosis.⁷³ RNA sequencing of lichen planus and morphea-like (sclerotic) skin biopsies from patients with cGVHD compared with healthy controls revealed a robust interferon-response signature, with upregulation of genes associated with antigen processing and presentation.⁸² Most recently, a phase 1/2 clinical trial examined the safety, tolerability, and efficacy of axatilimab, an high affinity, humanized anti-CSF1R monoclonal antibody, after failure of at least 2 prior systemic therapies.⁴ Treatment with 1 mg/kg every 2 weeks resulted in a median time to response of 4 weeks with benefit observed in all involved organs, including the joints/fascia, skin, and lungs, with 33% of patients experiencing sustained responses for at least 20 weeks.⁴ Notable improvements were observed in skin-tightening symptoms and/or severity, thus meeting the primary efficacy end point, and more than half (52%) of responders subsequently tolerated glucocorticoid dose reductions.⁴ Correlative studies also revealed reductions in circulating nonclassical monocytes, a reduction in the number of CSF1R⁺ macrophages in paired skin biopsies, and a significant decline in plasma levels of TGF-β in responding patients.⁴ This work has recently been extended in the AGAVE-201 trial, which intriguingly demonstrated equivalent or improved efficacy and minimized toxicity of axatilimab at the lowest tested dose (0.3 mg/kg).⁸³ Interestingly, this dose did not result in elevated systemic CSF1 levels, suggesting incomplete receptor blockade. This dosing ambiguity, such that the lowest possible dose produced the most favorable response, is yet to be fully understood with respect to monocyte/macrophage differentiation, degree of depletion, and levels of receptor inhibition. Furthermore, the precise mechanism via which axatilimab functions (eg, Fc-mediated effects vs cytokine starvation) are yet to be elucidated. Hence, other modes of inhibiting CSF1R signaling, including small-molecule inhibitors, should not be assumed to induce the same therapeutic benefit until rigorously evaluated in preclinical systems.

Cumulatively, the evidence presented here indicates a key role for monocytes and macrophages in pathological manifestations of aGVHD and cGVHD. Notably, the current literature suggests a strong dichotomy between the protective role of host macrophages during peripheral aGVHD, and the pathogenicity of donor macrophages during cGVHD. Consequently, CSF1R blockade is a valuable and efficacious strategy for the depletion of donor macrophages late after transplantation.

Monocytes and macrophages: key in CAR T-cell therapy outcomes

CAR T-cells targeting CD19 or B-cell maturation antigen, although revolutionary in cancer treatment, induce severe toxicities including CRS, HLH/MAS, and ICANS.⁸⁴ CRS is reported to occur in 50% to 90% of patients receiving CD19 CAR T cells,⁸⁵ with CAR T-cell trafficking, expansion, killing, and activation of bystander immune cells (predominantly monocytes/macrophages) causing dramatic increases in systemic cytokine levels (eg, interferon-γ, GM-CSF, IL-6, and TNF).⁸⁶ Subsequent endothelial injury and vascular damage contribute to hypoxia and hypotension, often presenting in conjunction with fever, fatigue, and headache.^86,87 CRS typically precedes ICANS, although in rarer instances the 2 may co-occur.⁸⁸ Clinically, ICANS manifestations present on a spectrum from confusion, attention deficits, and impaired fine motor skills, escalating to cerebral oedema, coma, seizures, and death.⁸⁹ At present, CRS is managed by administration of the IL-6 receptor antagonist tocilizumab or corticosteroids, whereas ICANS management strategies are limited to corticosteroids.⁹⁰ Understanding mechanisms driving these toxicities are critical for developing targeted therapeutic strategies and managing complications refractory to standard-of-care treatment.

CRS

Recent studies in mice demonstrate that CD19 CAR T cells interacting with their tumor target leads to their release of cytokines including GM-CSF and IFN-γ, and robust activation of Ly6C^hiF4/80^int/lo macrophages, which express high levels of iNOS and produce IL-6.⁹¹ IL-6 can induce iNOS production and, as such, blockade of the IL-6R (tocilizumab) could alleviate CRS mortality, which was associated with reduced proportions of iNOS⁺ macrophages, indicating that modulating macrophage function can attenuate CAR T-cell–related complications.⁹¹ However, whether tocilizumab functions via effects mediated on macrophages, or further upstream by modulating CAR T-cell activation, or both, requires further investigation. Interestingly, however, although monocytes can also produce IL-1 and IL-6 during CRS, monocyte/macrophage depletion via liposomal clodronate reduced CAR T-cell expansion and leukemia clearance.⁶ Therefore, blocking downstream mediators, such as iNOS, may be a more appropriate strategy during CRS than directly ablating myeloid populations.

HLH/MAS

HLH/MAS, recognized in American Society of Transplantation and Cellular Therapy guidelines as immune effector cell–associated HLH-like syndrome, poses a significant mortality risk after CAR T-cell transfer, owing to the development of pancytopenia with BM failure, coagulopathy, and multiorgan dysfunction.^84,92 Initial reports suggested these toxicities to be severe CRS manifestations; however, recent studies indicate a variable time to onset with the potential for delayed presentations coinciding with, or after, CRS resolution.^93,94 T-cell–derived cytokines (eg, IFN-γ) activate macrophages, which subsequently release cytokines (TNF, IL-6, and IL-1β), and ferritin and plasminogen activator, correlating with clinical hyperferritinemia and hypofibrinogenemia, respectively.^84,92 Ferritin production in macrophages is augmented by uptake of hemoglobin-haptoglobin complexes by CD163.⁹² Macrophage activation of STAT1 by IFN-γ receptor signaling drives hemophagocytosis of erythrocytes, contributing to multiorgan failure and potential lethality.⁹⁵ Reports on management of HLH are heterogeneous but are treated in the frontline setting with corticosteroids, with inclusion of etoposide, anakinra, and ruxolitinib across different studies.⁸⁴ A recent report demonstrated utility of IFN-γ blockade (emapalumab) in managing HLH/MAS after CD19 CAR T-cell therapy, however this approach is yet to be validated in larger prospective studies.⁹⁶

ICANS

Most recently, treatment of leukemia- or lymphoma-bearing mice with syngeneic CD19 CAR T cells in an immunocompetent mouse model revealed blood-brain barrier damage followed by extensive microglia expansion, activation, and production of TNF, CCL2, and GM-CSF, which was associated with neurological deficits.⁷ Microglia activation was also evident in autopsied brain tissue of patients who received CD19 CAR T cells for B-cell non-Hodgkin lymphoma and subsequently developed ICANS, indicating microglia activation is conserved across mice and humans in this setting. Microglia depletion via a CSF1R small-molecule inhibitor could alleviate the neurological deficits observed in mice, highlighting microglia as a critical mediator of ICANS. Moreover, CD19 CAR T-cell transfer induced phosphorylation of multiple kinases in the transforming growth factor-β-activated kinase 1 (TAK1)-NF-κB-p38 mitogen-activated protein kinase (MAPK) pathway in microglia. Subsequent genetic and pharmacological blockade (Takinib) of TAK1 could reduce microglia activation sufficient to improve neurological function, and concomitantly augmented CAR T-cell antilymphoma activity.⁷ Notably, although clinical studies indicate that although tocilizumab is beneficial for CRS management, it does not share the same efficacy for neurological events, and may even be detrimental.⁹⁷ These recent findings regarding microglia activation would support this conclusion, highlighting divergent mechanisms between CRS and ICANS that need to be appropriately targeted (Figure 4). Collectively, these seminal studies on CRS and ICANS highlight the utility of modulating host monocyte and/or TRM responses to manage CAR T-cell complications.

Figure 4.

Functions of monocytes and macrophages in CAR T-cell–related complications. In the periphery, CAR T-cell engagement with target antigen leads to release of cytokines including GM-CSF and IFN-γ causing the robust activation of proinflammatory macrophages. Macrophages contribute to the release of IL-6, which aids in amplifying and sustaining CAR T-cell activation. Treatment with tocilizumab (IL-6R antagonist) is the current treatment used for the management of CRS. In the brain, CAR T-cell therapy is associated with blood-brain barrier damage followed by activation of tissue-resident microglia, which release numerous cytokines including GM-CSF, TNF, and CCL2. Inhibition of the TAK1 kinase with Takinib can reduce microglia activation, which was associated with improved cognitive function in mice. p38-MAPK, p38-mitogen-activated protein kinase.

Beyond clear roles in mediating immune toxicity, monocytes in the initial leukapheresis product can also impair T-cell activation before transduction.⁹⁸ Consequently, monocyte depletion from the initial product can augment transduction efficiency and CAR T-cell expansion.^98,99 Transcriptional profiling of leukapheresis products identified a monocyte signature defined by MS4A4A, CD86, CD163, and SIGLEC5, with expression of this gene set found to be associated with lower circulating CAR T-cell numbers after infusion, and characterizing patients as having significantly shorter progression-free survival.¹⁰⁰ Moreover, monocyte dysregulation (based on HLA-DR downregulation) is associated with CAR T-cell failure,¹⁰¹ and patients with elevated frequencies of circulating monocyte-myeloid–derived suppressor cells fail to show durable responses after CD19-directed CAR T-cell therapy.¹⁰² In summary, monocytes influence CAR T-cell production, therapeutic responses, and toxicity outcomes.

Macrophage-mediated complications of cancer and immunotherapy

Although monocytes and macrophages contribute significantly to cell therapy complications, a plethora of reports have also described the association between macrophages within the tumor microenvironment (TME) and disease progression. TAMs have largely been described as protumor, given their capacity to promote regulatory T-cell infiltration to the TME, expression of immune checkpoint ligands, provision of growth factors, and remodeling of the extracellular matrix to aid tumor expansion and dissemination.¹⁰³ Intratumor and intertumor heterogeneity is extensive, in part attributable to the dual ontogeny of TAM populations, whereby those arising from local TRMs are often involved in tissue remodeling, whereas BM monocyte-derived TAMs coordinate immunosuppressive functions and influence the function of other immune cells.^9,104 These data further highlight phenotypic and functional complexity dependent on cellular ontogeny, which plays a critical role in tumor immunity. In the context of hematological malignancies, a number of studies have revealed features of TAMs that position them as appealing therapeutic targets by examining the TME in the BM, lymph nodes, and spleen.

Coculture of BM-derived macrophages isolated from acute myeloid leukemia (AML)-bearing mice with AML tumor cells permitted greater tumor growth than macrophages from nonleukemic mice, indicating that TAMs have a propensity to support AML expansion.¹⁰⁵ Similarly, macrophages were required for tumor engraftment in a mouse model of chronic lymphocytic leukemia, because macrophage depletion attenuated tumor growth and improved survival.¹⁰⁶ Additional studies examining the therapeutic potential of CSF1R blockade for multiple myeloma revealed that in vivo treatment could prevent TAM accumulation and improve myeloma control.³⁹ In vitro mechanistic studies revealed that low doses of anti-CSF1R antibody can repolarize macrophages toward more proinflammatory/antitumor phenotypes, evidenced by transcriptional cytokine upregulation (eg, Tnf) and increased expression of both genes and proteins related to antigen processing/presentation.⁴⁰ Moreover, anti-CSF1R treatment augmented myeloma sensitivity to chemotherapy, highlighting the potential benefit of targeting TAMs in complementary therapeutic strategies.⁴⁰

In line with these preclinical studies, it is unsurprising that increased TAM frequencies are associated with poor prognosis in humans.^105,107, 108, 109 Increased macrophage accumulation in the lymph node has been associated with increased risk of relapse and poorer prognosis after both autologous¹⁰⁷ and allogeneic transplant.¹⁰⁸ Cumulatively, this suggests that CSF1R inhibition, and/or other approaches to deplete TAMs, may be beneficial for removing protumor macrophages. Conceptually, this was supported by treatment with a selective CSF1R inhibitor of samples from patients with AML, which reduced tumor cell viability due to loss of trophic support by TAM depletion. Notably, mononuclear cells isolated from primary AML samples demonstrate a range of dose-dependent responses to CSF1R small-molecule inhibition. Furthermore, CSF1R inhibitor resistance was associated with poor prognosis cytogenetics and reduced overall survival.⁴⁴ Although trials of single-agent CSF1R inhibition (PLX3397) for modifying the TME in refractory leukemias and lymphoma have demonstrated favorable safety profiles,⁴² efficacy of monotherapy has been limited, thus warranting further exploration into combination therapy approaches.¹¹⁰ However, although preclinical studies have produced evidence supporting a synergistic effect of combining CSF1R inhibition with ICIs,^111,112 a number of clinical trials examining this partnership have been terminated or failed to produce clinical efficacy (eg, ClinicalTrials.gov identifiers: NCT02452424 and NCT02777710). Thus, although the addition of CSF1R inhibitors to augment immunotherapy is conceptually attractive, this theory requires further development to devise strategies for targeting TAMs that generate clear clinical benefit.

Moreover, as the use of ICIs continues to grow, the potential for iRAEs to develop is becoming increasingly recognized. Recently, neurological iRAEs after anti–programmed cell death protein 1 (anti–PD-1) treatment were shown to be mediated by direct activation of Syk kinase signaling in microglia in tumor-bearing mice, with microglia activation also evident in cortical brain samples from patients treated with anti–PD-1.¹⁰ Microglia depletion in mice via CSF1R inhibition, or Syk inhibition as a more targeted approach to attenuate microglia reactivity, could improve behavioral responses in anti–PD-1–treated mice.¹⁰³ These findings indicate that TRM populations can contribute to complications after immunotherapy, and that targeting macrophages may have dual effects on both antitumor responses and immune toxicity.

Conclusions and outlook

Our understanding of the ontogeny, function, and disease contributions of monocytes and macrophages has improved dramatically over recent years. The monocyte-macrophage lineage has proven to be a critical dictator of therapeutic outcomes in the setting of hematological malignancies. Importantly, the translation of CSF1R inhibition from preclinical models to successful clinical trials and now US Food and Drug Administration approval represents a significant breakthrough in the treatment of cGVHD. Future investigations would benefit from devising strategies to phenotypically modulate monocytes/macrophages in the absence of ablation. Moreover, understanding the heterogeneity amongst disease-driving macrophages will help to identify critical subpopulations that may be more directly targeted, thus sparing potentially beneficial populations. Combining axatilimab with other approved cGVHD therapies is also of clinical interest for managing specific disease manifestations.¹¹³ Collectively, understanding biological mechanisms is critical to devising targeted therapeutic strategies, paving the way for future macrophage-directed approaches in the clinic to potentially improve outcomes of treatment for hematological malignancies.

Conflict-of-interest disclosure: K.P.A.M. received honoraria from Incyte and Deciphera. G.R.H. has consulted for Generon Corporation, NapaJen Pharma, iTeos Therapeutics, Commonwealth Serum Laboratories, Cynata Therapeutics, Neoleukin Therapeutics, and Incyte Pharma and has received research funding from Compass Therapeutics, Syndax Pharmaceuticals, Applied Molecular Transport, Serplus Technology, Heat Biologics, Laevoroc Oncology, iTeos Therapeutics, Genentech, Incyte Pharma, and Commonwealth Serum Laboratories. R.C.A. declares no competing financial interests.