Animal models of anemia of inflammation: Methodological advances, limitations, and translational applications

Abstract

Anemia of inflammation (AI) represents one of the most prevalent secondary anemias in clinical practice, frequently complicating a wide spectrum of chronic inflammatory diseases. Given its highly heterogeneous etiology and complex pathophysiological mechanisms, the establishment of rational and high‐fidelity animal models is vital for elucidating the core mechanisms of AI and optimizing clinical diagnostic and therapeutic strategies. This narrative review provides a structured categorization of current AI animal models into three major paradigms: sterile inflammation models, disease‐specific models, and genetically engineered models. We provide an in‐depth comparative analysis of their underlying modeling principles, methodological protocols, and characteristic phenotypic profiles, while comprehensively evaluating their respective translational advantages and intrinsic limitations. Furthermore, we discuss the prospective applications of these models in deciphering pathomechanisms, validating therapeutic targets, and developing interventional strategies. To enhance reproducibility and translational relevance, we propose a streamlined phenotyping panel and a model selection framework. The precise selection and rational application of these tailored models will contribute to improved reliability of fundamental AI research and accelerate its translational progression from bench to bedside.

Anemia of inflammation is a highly prevalent secondary anemia complicating chronic inflammatory diseases. Current animal models are categorized into sterile inflammation, disease‐specific, and genetically engineered models, with comparative analysis of their modeling principles, characteristic phenotypes, and translational advantages to elucidate core mechanisms and accelerate bench‐to‐bedside translation.

1. INTRODUCTION

Anemia of inflammation (AI) is the hypoproliferative anemia that arises during chronic infection, autoimmune disease, cancer, and other inflammatory conditions. ¹ Long known as anemia of chronic disease (ACD), ² the condition is now widely called AI, reflecting the central role of inflammation in its pathogenesis. AI is the second most common anemia worldwide, surpassed only by iron deficiency anemia (IDA). ¹ , ³ , ⁴ As the predominant anemic entity observed in hospitalized and chronically ill cohorts, ¹ AI was identified as the cause of anemia in 70% of a clinical cohort of 191 hospitalized elderly patients, ⁵ and anemia itself affected an estimated 1.92 billion people worldwide in 2021. ⁴ , ⁶ Together, these figures highlight the substantial clinical burden of AI among geriatric patients and patients with chronic comorbidities. ⁵ , ⁷ , ⁸ Furthermore, patients afflicted with AI frequently exhibit debilitating clinical symptoms, including persistent fatigue, progressive cachexia, vertigo, and cognitive impairment, which severely diminish their quality of life. ⁹ Despite advancements in clinical diagnostics, human biospecimens offer limited capacity for capturing the dynamic pathological progression of the disease and are frequently confounded by concurrent iron deficiency. Consequently, the development of robust animal models that faithfully recapitulate these clinical hallmarks is crucial for elucidating the complex inflammation–iron crosstalk and providing a foundational platform for the preclinical evaluation of novel therapeutics, such as hepcidin antagonists.

The disruption of iron homeostasis constitutes the primary pathogenic mechanism of AI. ¹⁰ Hepcidin, a 25‐amino acid peptide hormone predominantly synthesized by hepatocytes and encoded by the Hamp gene, ¹¹ serves as the central regulatory hub of systemic iron metabolism. ¹² Hepcidin expression is governed by the complex interplay of multiple signaling pathways: the canonical bone morphogenetic protein/mothers against decapentaplegic homolog (BMP/SMAD) pathway senses circulating iron levels via the hemochromatosis protein‐transferrin receptor ½‐hemojuvelin (HFE‐TFR1/2‐HJV) complex, facilitating SMAD phosphorylation and Hamp transcription. Transmembrane serine protease 6 (Tmprss6) mediates negative regulation, and the nuclear factor erythroid 2‐related factor 2‐bone morphogenetic protein 6 (Nrf2‐BMP6) axis mediates iron‐overload feedback. ¹³ , ¹⁴ , ¹⁵ Under inflammatory conditions, the interleukin 6/janus kinase/signal transducer and activator of transcription 3 (IL‐6/JAK/STAT3) and interleukin 1β (IL‐1β)/nuclear factor kappa‐B (NF‐κB) signaling cascades are markedly activated in macrophages and T cells, prompting phosphorylated STAT3 and CCAAT/enhancer‐binding protein delta (C/EBPδ) to synergistically bind the Hamp promoter and drive a pathological surge in hepcidin levels. ¹⁶ , ¹⁷

This elevated hepcidin exerts its biological effects by targeting its sole functional receptor, ferroportin (FPN). ¹⁸ By inducing the internalization and ubiquitin‐mediated degradation of FPN, hepcidin disrupts the central axis of systemic iron regulation. ¹⁹ , ²⁰ , ²¹ This mechanism not only suppresses intestinal iron absorption mediated by divalent metal transporter 1 (DMT1) and duodenal cytochrome b (DCYTB) but also concurrently obstructs macrophage iron recycling, leading to significant intracellular iron retention within the liver and spleen. ²² In experimental animal models, this cellular transport blockade manifests as a characteristic profile of functional iron deficiency—marked by hypoferremia and diminished transferrin saturation alongside normal or paradoxically elevated ferritin levels. ²³

Beyond dysregulated iron metabolism, impaired erythropoiesis and truncated erythrocyte longevity constitute another pivotal axis of AI pathogenesis. Pro‐inflammatory cytokines impede erythropoietic output by suppressing renal erythropoietin (EPO) synthesis and erythroid erythroferrone (ERFE) expression, while concurrently activating macrophages to enhance erythrophagocytosis, thereby curtailing the circulatory lifespan of red blood cells. This dual defect of “suppressed production and accelerated depletion” is highly coupled with functional iron deficiency, collectively culminating in mild‐to‐moderate normocytic normochromic anemia. Ultimately, these interconnected mechanisms establish a pathological closed loop of “inflammation–iron restriction–anemia” ²⁴ , ²⁵ , ²⁶ (Figure 1).

FIGURE 1.

The schematic diagram for the pathophysiological mechanisms of anemia of inflammation (AI). Chronic diseases trigger inflammation and release cytokines that drive AI through three interconnected axes, including hepcidin‐mediated iron dysregulation, impaired erythropoiesis, and shortened RBC survival, ultimately leading to normocytic normochromic anemia. BMP, bone morphogenetic protein; DCYTB, duodenal cytochrome b; DMT1, divalent metal transporter 1; EPO, erythropoietin; ERFE, erythroferrone; FPN, ferroportin; Hamp, hepcidin antimicrobial peptide; HFE, hemochromatosis protein; HJV/Hjv, hemojuvelin; IL‐1β, interleukin‐1β; IL‐6, interleukin‐6; JAK, janus kinase; RBC, red blood cell; SMAD, mothers against decapentaplegic homolog; STAT3, signal transducer and activator of transcription 3; TFR1/2, transferrin receptor 1/2; TMPRSS6, transmembrane serine protease 6; TNF‐α, tumor necrosis factor‐α. Created with BioRender.com.

Current experimental models of AI have been categorized into three main classes: sterile inflammation models, disease‐specific models, and genetically engineered models. All of these models exhibit the systemic phenotypic characteristics of inflammation‐associated anemia. Hematologically, these include mild‐to‐moderate normocytic normochromic anemia, decreased serum iron concentration and transferrin saturation, normal or elevated ferritin levels, reduced EPO and ERFE expression, and a decline in reticulocyte counts. From the iron metabolism perspective, the core features of AI include significant upregulation of hepatic hepcidin expression, downregulation of FPN levels, and iron sequestration within the liver and spleen. Inflammatory hallmarks in these models are reflected in elevated levels of inflammatory cytokines such as IL‐6, IL‐1β, and tumor necrosis factor α (TNF‐α). These factors converge to form the pathological closed loop of “inflammation–iron restriction–impaired erythropoiesis–anemia.” These experimental models have played a pivotal role in elucidating AI pathogenesis and accelerating the development of targeted therapeutics, such as hepcidin antagonists. ⁷ , ²³ To improve reproducibility and cross‐study comparability, we propose a streamlined phenotyping panel for AI models (Table 1).

TABLE 1.

Recommended streamlined phenotyping panel for anemia of inflammation models.

Parameter category	Parameter	Reference range (mouse, C57BL/6J)	Reference range (rat)
Hematological	Hemoglobin (Hb)	6–15 g/dL 27 , 28 , 29	Female: 7–15 g/dL 30 , 31
Hematological	RBC count	2–9 × 109/mL 29 , 32 , 33	Female: 6–8 × 109/mL 31
Hematological	MCV	39–58 fL 34 , 35 , 36	Female: 37–45 fL 30 , 37
Hematological	MCH	10–15 pg 29 , 32 , 38	Not specified
Hematological	MCHC	29–32 g/dL 26 , 36 , 39	Female: 30–33 g/dL 30 , 37
Hematological	Reticulocyte count	2–5 × 1011/L 33 , 39 , 40	Female: 2.5%–7.5% 30
Iron metabolism	Serum iron	30–220 μg/dL 34 , 41 , 42	Female: 10–250 μg/dL 30 , 31
Iron metabolism	Transferrin saturation	18%–30% 29 , 34 , 43	Female: 10%–20% 37
Iron metabolism	Ferritin	100–200 μg/L 28 , 43	Not specified
Iron metabolism	Hepatic Hamp mRNA	2–6‐fold increase vs. control 27 , 40 , 44	Female: 1.5–2‐fold increase vs. control 30
Iron metabolism	Tissue iron (liver)	40–120 μg/g 27 , 32 , 45	Female: 100–300 μg/g 30 , 31
Iron metabolism	Tissue iron (spleen)	20–100 μg/g 27 , 32 , 33	Female: 400–700 μg/g 30 , 31
Inflammatory	Serum IL‐6	1–11 ng/mL 27 , 46 , 47	Female: 140–210 pg/mL 30 , 37 , 48
Inflammatory	Serum TNF‐α	25–1000 pg/mL 36 , 49 , 50	Female: 3.5 pg/mL; male: 55–130 pg/mL 48 , 51 , 52
Inflammatory	Serum IL‐1β	3–100 pg/mL 32 , 50	Female: 30 pg/mL 48 , 52
Disease‐specific	CKD: creatinine	25–150 μmol/L 49 , 53 , 54	Male: 2.3–2.8 mg/dL 45 , 55
Disease‐specific	CKD: BUN	5–40 mmol/L 54 , 56	Male: 42–100 mg/dL 55
Disease‐specific	IBD: colon length	3.5–8.5 cm 36 , 57	Not specified
Disease‐specific	SLE: anti‐dsDNA antibody	Elevated vs. control 58	Not specified
Disease‐specific	RA: arthritis score	2–10 (per paw) 59 , 60	4–12 (per paw) 48

Abbreviations: BUN, blood urea nitrogen; CKD, chronic kidney disease; dsDNA, double‐stranded DNA; IBD, inflammatory bowel disease; IL‐1β, interleukin‐1β; IL‐6, interleukin‐6; MCH, mean corpuscular hemoglobin; MCHC, mean corpuscular hemoglobin concentration; MCV, mean corpuscular volume; RA, rheumatoid arthritis; RBC, red blood cell; SLE, systemic lupus erythematosus; TNF‐α, tumor necrosis factor‐α.

Previous reviews have provided important foundations for understanding animal models of AI. Rivera and Ganz summarized infectious and noninfectious mouse models and emphasized their distinct strengths and limitations, ⁶¹ whereas Fraenkel highlighted mechanistic insights obtained mainly from the heat‐killed Brucella abortus (HKBA) model. ⁷ However, the subsequent development of disease‐specific models, genetically engineered strains, and mechanism‐based therapeutic studies warrants an updated assessment of the field. To ensure methodological transparency, this narrative review was based on a structured literature search conducted in PubMed up to August 2026. Search terms included combinations of “anemia of inflammation,” “anemia of chronic disease,” “animal model,” “hepcidin,” “ferroportin,” “iron metabolism,” and the names of individual modeling approaches. The reference lists of relevant primary studies and review articles were additionally screened to identify eligible publications. Priority was given to in vivo studies that described model‐establishment procedures or reported hematological, inflammatory, iron‐metabolism, erythropoietic, mechanistic, or therapeutic outcomes relevant to AI. In vitro‐only studies, studies of anemia without a defined inflammatory component, and publications lacking sufficient methodological or phenotypic information were generally excluded. Accordingly, this review classifies current models into sterile inflammation, disease‐specific and clinically relevant, and genetically engineered models. We compare their construction protocols, characteristic phenotypes, pathophysiological fidelity, translational relevance, therapeutic applications, and limitations. By integrating classical models with recent mechanistic and translational evidence, this review aims to provide a practical framework for selecting appropriate models based on specific experimental objectives.

2. STERILE INFLAMMATION MODELS

2.1. Chemical stimulation induction model: Sterile abscess model induced by turpentine oil

The sterile abscess model induced by turpentine oil (TO) (primarily comprising α/β‐pinene) has a long‐standing history. ⁶² Because its initial application in canine iron metabolism studies in 1949, ⁶³ , ⁶⁴ this chemical irritant has become a common experimental tool for establishing acute (16–48 h) and chronic (2–6 weeks) murine models of AI ⁴⁶ (Figure 2A).

FIGURE 2.

Sterile inflammation models of anemia of inflammation in C57BL/6 mice. (A) Turpentine oil (TO), (B) heat‐killed Brucella abortus (HKBA), (C) lipopolysaccharide (LPS) and (D) Zymosan. ip, intraperitoneal injection; sc, subcutaneous injection. Created with BioRender.com.

Injecting turpentine into the interscapular fat pad in mice or the subplantar region of the hind paw in rats triggers a localized inflammatory cascade via chemical irritation, thereby simulating the pathological state of AI. Mechanistically, the localized injection‐induced tissue necrosis releases massive amounts of damage‐associated molecular patterns (DAMPs). These DAMPs synergistically provoke the robust release of pro‐inflammatory cytokines, such as IL‐6 and TNF‐α, ⁶⁵ , ⁶⁶ , ⁶⁷ from immune cells via toll‐like receptor (TLR) ⁴⁷ , ⁶⁸ /NF‐κB/mitogen‐activated protein kinase (MAPK) cascades, ⁶⁶ , ⁶⁹ NLR family pyrin domain containing 3 (NLRP3) inflammasome activation, ⁶⁸ and hypothalamic–pituitary–adrenal (HPA) axis neuroendocrine responses. ⁷⁰ These surging cytokines then orchestrate profound crosstalk between the IL‐6/JAK/STAT3 ³² , ⁶⁹ , ⁷¹ and BMP/SMAD pathways ³⁴ to drive pathological hepcidin overexpression in the liver ³² , ⁶⁹ , ⁷¹ ; the subsequent degradation of FPN fundamentally obstructs iron efflux, ultimately resulting in systemic iron utilization restriction and the AI phenotype. ³⁵

This model robustly replicates AI features with well‐defined kinetics: IL‐6 surges rapidly postmodeling and persists into the chronic phase ³² , ³⁴ , ³⁵ ; TNF‐α peaks at 16 h and gradually declines after 3 weeks, ³² whereas IL‐1β exhibits sustained upregulation during the intermediate stage of inflammatory progression. ⁶⁷ Overall, this model is valued for its relatively high success rate and robust reproducibility, and it consistently and reliably replicates the iron‐restricted erythropoiesis observed in AI patients. It has been widely applied in pharmaceutical research and has been used to explore the efficacy and molecular mechanisms of therapeutic candidates such as prolyl hydroxylase inhibitors, ⁷² black pepper extract, ⁷¹ low‐anticoagulant heparin–iron complex, ⁶⁹ Caralluma tuberculata, ⁶⁶ gadolinium (III) complex‐derived nonsteroidal anti‐inflammatory drug, ⁶⁵ intestinal HIF2α activators, ²⁷ and other related therapeutic substances.

However, this model has recognized limitations: intense local irritation leads to abscess formation and tissue necrosis at the injection site. The resulting nonspecific nociception and excessive local inflammation may confound the interpretation of systemic iron metabolism data. ⁷ From an animal welfare perspective, tissue damage and associated pain should be minimized through appropriate anesthesia and analgesia protocols.

2.2. Pathogen‐associated molecular pattern induced models

Although active pathogen infection models were employed in early studies, their strict reliance on high‐level biosafety laboratories and the introduction of confounding factors—such as direct iron sequestration by pathogens and pathogen‐induced hemolysis—severely compromise the mechanistic purity of AI. Consequently, these models have been largely superseded by sterile inflammation models utilizing heat‐killed pathogens or lipopolysaccharide (LPS).

2.2.1. HKBA‐induced model

B. abortus is intrinsically a gram‐negative intracellular bacterium. ⁷³ In early studies, researchers established that a single intraperitoneal injection of HKBA elicits inflammatory responses and iron‐restrictive anemia lasting for weeks in vivo. ³⁸ This breakthrough effectively resolved the stability issues associated with traditional sterile inflammation models in maintaining long‐term anemia, and this model is one of the most extensively validated and widely used platforms in AI research. ¹ , ⁷

Mechanistically, the cell wall of HKBA contains abundant degradation‐resistant atypical LPS and other structural components. Acting as persistent antigenic stimuli, these components continuously activate macrophages via the TLR2/4 pathways, ⁷⁴ eliciting a robust Th1‐skewed immune response and the massive release of pro‐inflammatory cytokines such as IL‐6 and TNF‐α. ³⁸ , ⁷⁵ Studies extensively utilizing the HKBA model indicate that elevated IL‐6 directly upregulates hepcidin expression via the JAK2/STAT3 pathway ⁷⁶ , ⁷⁷ ; concurrently, this process profoundly crosstalks with the BMP/SMAD signaling axis ²⁸ , ⁷⁸ and the ERFE‐hepcidin regulatory axis. ³³ Together, they synergistically drive the internalization and degradation of FPN, ultimately precipitating severe functional iron restriction. ²⁷ , ⁷⁹

The HKBA model has now established a highly standardized protocol: typically, 8–12‐week‐old male C57BL/6J mice are induced via a single intraperitoneal injection of 2–5 × 10⁸ HKBA particles (e.g., the classical strain 1119‐3) ⁷⁶ (Figure 2B). This model exhibits well‐defined dynamic progression. Serum IL‐6 rises sharply within 6 to 24 h after injection, triggering the acute‐phase response. The model then transitions into a 3–4‐week chronic inflammatory phase characterized by typical and stable normocytic normochromic anemia, with hemoglobin concentrations generally reaching their nadir around day 14. ³⁸ By day 28, as HKBA components are gradually cleared and inflammation resolves, hemoglobin levels begin to recover spontaneously.

The core advantage of this model lies in its operational simplicity and highly stable phenotype, whereby a single administration effectively recapitulates the prolonged iron dyshomeostasis observed in clinical chronic AI. Numerous studies on the pathogenesis of AI have been conducted based on this model, ⁷ , ²⁹ , ³⁹ and it is also suitable for large‐scale screening of targeted agents, including hepcidin antagonists. ⁷⁸ However, the model has inherent drawbacks. Batch differences in immunogenicity tend to cause phenotypic variations, so commercial standardized batches are recommended. Furthermore, fundamentally driven by the macrophage “foreign body clearance” response, this model retains certain pathological disparities from the complex microenvironments of genuine progressive chronic diseases, such as tumors or autoimmune disorders.

2.2.2. LPS‐induced endotoxemia model

LPS, a core structural component of the outer membrane of gram‐negative bacteria, is classically employed to induce endotoxemia for acute AI modeling. This model was first established in 1962 by intraperitoneally injecting rats with LPS extracted from Salmonella typhosa and Escherichia coli. Longitudinal monitoring shows high sensitivity of mammals to endotoxin. Generally, intraperitoneal injection at 1 mg/kg leads to a sharp drop in plasma iron levels within 2 h, hitting the lowest point at 8–12 h ⁸⁰ (Figure 2C).

Mechanistically, in vivo LPS stimulation provokes immune cells to secrete massive amounts of IL‐6, thereby triggering a systemic inflammatory response and a precipitous elevation in hepcidin levels, ⁴⁰ which typically peaks 4 h postadministration. ⁴³ Notably, the regulatory efficacy of LPS is highly dependent on the basal BMP6/SMAD signaling axis, ⁸¹ wherein Hjv, a critical coreceptor in the BMP pathway, plays an important role. The study revealed that Hjv ablation not only blunts SMAD5 activation ⁸² but also completely abrogates the positive transcriptional regulation mediated by the IL‐6/STAT3 cascade. ⁴³ , ⁸² Beyond canonical hepcidin‐centered mechanisms, LPS‐induced inflammation can independently exacerbate anemia by activating the hepatic NF‐κB pathway, which downregulates microRNA‐122 (miR‐122) and EPO expression, thereby contributing to impaired erythropoiesis. ⁸³ LPS‐induced anemia is fundamentally an innate antimicrobial defense mechanism orchestrated by the host to restrict iron availability to invading pathogens. ⁸⁴ , ⁸⁵ Its core pathophysiological features universally converge on the aberrant expression of hepcidin and the consequent systemic redistribution of iron, typified by macrophage iron retention. ⁸² , ⁸⁶

In recent years, the applications of the LPS model have expanded to encompass highly translational and innovative modeling strategies. For instance, a 2021 study simulated inflammatory complications by subcutaneously injecting LPS (0.5 μg/g) into iron‐overloaded pregnant mice during early and late gestation. The findings demonstrated that the synergistic “two‐hit” combination of inflammation and iron overload elicited severe embryotoxicity, including gross malformations and lethality, whereas neither factor alone produced such severe consequences. ⁸⁷ Another pivotal study in 2021 targeted central nervous system (CNS) inflammation via intracerebroventricular LPS administration, revealing that localized CNS inflammation can distally regulate peripheral hepatic hepcidin expression and systemic iron metabolism through the JAK2/STAT3 pathway. ⁸⁸

In summary, the LPS‐induced endotoxemia model possesses significant advantages, including a short modeling cycle, high efficiency, and robust reproducibility. Furthermore, it offers remarkable flexibility, allowing researchers to simulate varying intensities of inflammation by titrating the dosage and administration frequency. Nevertheless, its limitations are equally apparent. The model struggles to faithfully recapitulate the long‐term pathological progression of chronic AI, and its relatively singular trigger mechanism fails to fully encompass the complex, multifactorial microenvironments typically observed in clinical patients.

2.2.3. Zymosan‐induced model

Zymosan primarily comprises β‐1,3‐glucan and mannan derived from the Saccharomyces cerevisiae cell wall. ⁸⁹ These components synergistically activate innate immunity via the Dectin‐1/Syk‐CARD9‐NF‐κB ⁹⁰ , ⁹¹ and TLR2‐MyD88‐NF‐κB/MAPK ⁹² , ⁹³ signaling axes, respectively, triggering an explosive release of pro‐inflammatory cytokines such as IL‐6, TNF‐α, and IL‐1β. ⁸⁹ , ⁹⁴ , ⁹⁵ These surging cytokines induce anemia through a three‐pronged mechanism. First, the IL‐6/JAK2/STAT3 axis drives aberrant hepcidin expression, causing systemic iron restriction ⁴⁰ ; second, TNF‐α/IFN‐γ/IL‐1β directly suppresses bone marrow erythroid differentiation via SOCS proteins and FasL pathways ²⁶ , ⁹⁶ , ⁹⁷ ; third, hyperactive erythrophagocytosis and oxidative hemolysis synergistically accelerate peripheral erythrocyte depletion. ⁹⁸ , ⁹⁹ , ¹⁰⁰

Experimentally, intraperitoneal injection of sterile lyophilized Zymosan A is utilized to establish acute (≤120 h) or chronic (2 weeks) models ¹⁰⁰ , ¹⁰¹ (Figure 2D). This approach rapidly provokes systemic inflammatory response syndrome (SIRS) and multiple organ dysfunctions, ¹⁰² , ¹⁰³ , ¹⁰⁴ but it is seldom applied to explore iron‐metabolism mechanisms and cannot simulate chronic autoimmune anemia mediated by adaptive immunity.

2.3. Direct pro‐inflammatory cytokine injection models

IL‐6, a central cytokine driving systemic inflammation, is not only widely utilized in vitro to stimulate cellular inflammatory responses but is also highly effective at inducing AI‐related phenotypes in vivo. Studies have demonstrated that a single intraperitoneal injection of recombinant IL‐6 (25 μg/kg) in mice rapidly upregulates hepcidin expression and induces marked hypoferremia ¹⁰⁵ (Figure 3A). In addition to direct protein administration, tail vein injection of recombinant adenoviral vectors expressing IL‐6 (1 × 10¹¹ particles/mouse) can achieve sustained in vivo transgene expression, thereby stably reproducing the aforementioned iron dysregulation ⁴¹ (Figure 3A). Notably, this exogenous cytokine induction strategy has been successfully validated in nonhuman primates (NHPs), such as cynomolgus monkeys; both acute and subchronic models exhibit highly consistent hepcidin surges and precipitous declines in serum iron, ¹⁰⁶ demonstrating excellent translational value.

FIGURE 3.

Direct pro‐inflammatory cytokine models of anemia of inflammation. (A) Interleukin‐6 (IL‐6), (B) interleukin‐22 (IL‐22), (C) interleukin‐1β (IL‐1β), (D) tumor necrosis factor‐α (TNF‐α), and (E) interferon‐gamma (IFN‐γ). ip, intraperitoneal injection. Created with BioRender.com.

The principal advantage of this model lies in its exquisite pathway specificity. It enables the precise targeting of the IL‐6/STAT3 signaling axis while effectively precluding interference from other inflammatory factors. Furthermore, the model is characterized by minimal tissue trauma, low mortality rates, and robust phenotypic reproducibility. However, its limitations are notable. Recombinant proteins possess a short in vivo half‐life and thus cannot sustain long‐term erythroid suppression for chronic AI modeling. Consequently, establishing a chronic model frequently relies on high‐frequency repeated dosing or gene transduction, which significantly escalates both operational complexity and experimental costs.

Beyond IL‐6, other pro‐inflammatory cytokines, including IL‐22, IL‐1β, TNF‐α, and interferon‐γ (IFN‐γ), can also be directly administered to construct AI animal models with distinct mechanistic emphases. For example, IL‐22 induces hepcidin expression via the STAT3 pathway independently of IL‐6, offering high specificity but limited translational relevance due to supraphysiological dosing ¹⁰⁷ (Figure 3B). IL‐1β rapidly triggers hypoferremia through BMP/SMAD signaling, providing a simple and efficient acute model but also failing to recapitulate chronic disease progression ¹⁰⁸ (Figure 3C). TNF‐α drives iron dysregulation in a hepcidin‐independent manner, enabling the investigation of its isolated effects, though it cannot replicate sustained endogenous cytokine elevation ¹⁰⁹ (Figure 3D). IFN‐γ promotes anemia via macrophage hyperactivation and direct erythroid suppression, allowing the precise dissection of its specific contributions but with inherent limitations in mimicking the full complexity of clinical inflammatory microenvironments ²⁶ (Figure 3E).

3. DISEASE‐SPECIFIC AND CLINICALLY RELEVANT MODELS

Although sterile inflammation models excel in dissecting acute and pure cytokine‐driven mechanisms, they inherently lack the complex, multifactorial clinical comorbidities. Therefore, disease‐specific models are important.

3.1. Anemia of chronic kidney disease models

Chronic kidney disease (CKD) is a clinical syndrome characterized by persistent abnormalities in renal function or structure, ¹¹⁰ and anemia represents one of its most critical complications. ¹¹¹ Its pathogenesis is highly complex, involving a relative deficiency of EPO, ¹¹² hepcidin‐mediated iron dysmetabolism, ¹¹³ , ¹¹⁴ the direct suppression of erythropoiesis by pro‐inflammatory cytokines, and the detrimental effects of uremic toxins. ²³ Compared to acute inflammation models, CKD models can more authentically recapitulate the clinical pathological milieu where chronic micro‐inflammation and renal failure are intertwined. ⁴⁵

Currently, mainstream CKD animal models are broadly categorized into two major classes: chemical/drug‐induced models and surgical/physical intervention models.

3.1.1. Chemical/drug‐induced models

These modeling approaches inflict direct damage to the renal parenchyma via the administration of nephrotoxic agents, representing the most frequently utilized strategy for CKD modeling. Representative models encompass those induced by adenine, Adriamycin, ⁵³ , ¹¹⁵ cisplatin, ¹¹⁶ gentamicin, ⁵⁵ , ¹¹⁷ , ¹¹⁸ and angiotensin II. ¹¹⁹

Among these, the adenine‐induced model stands as the most widely applied pharmacological tool for investigating CKD‐associated anemia and iron dysmetabolism. After 3 to 8 weeks of a high‐adenine diet (0.2%–0.75%) in rodents (Figure 4A), the excess adenine is metabolized by hepatic xanthine oxidase into highly insoluble 2,8‐dihydroxyadenine (2,8‐DHA). ¹²⁰ Subsequently, 2,8‐DHA rapidly crystallizes and precipitates within the renal tubules, causing mechanical obstruction ¹²¹ , ¹²² that consequently activates signaling cascades, including the NLRP3 inflammasome, TLRs, and NF‐κB pathways. ¹²⁰ , ¹²³ , ¹²⁴ This robust inflammatory response provokes a surge in pro‐inflammatory cytokines (e.g., IL‐6, TNF‐α), driving the aberrant upregulation of hepcidin. These events occur concomitantly with progressive tubular atrophy and interstitial fibrosis, ultimately culminating in an irreversible loss of renal function.

FIGURE 4.

Disease‐specific and clinically relevant models of anemia of inflammation. (A) chronic kidney disease (CKD), (B) inflammatory bowel disease (IBD), (C) systemic lupus erythematosus (SLE), (D) PG‐PS, (E) collagen‐induced arthritis (CIA), and (F) adjuvant‐induced arthritis (AA). 5/6 Nx, 5/6 nephrectomy; AIDs, autoimmune diseases; CFA, complete Freund's adjuvant; DSS, dextran sulfate sodium; ip, intraperitoneal injection; PG‐PS, peptidoglycan–polysaccharide; RA, rheumatoid arthritis. Created with BioRender.com.

This model is characterized by its operational simplicity and highly stable pathological phenotype, allowing researchers to precisely focus on the dynamic accumulation of inflammatory cytokines and the evolution of iron dysmetabolism—particularly governed by the BMP/SMAD regulatory axis—and anemia. ⁴⁵ , ¹²⁵ However, traditional high‐dose administration exerts overt systemic toxicity, frequently leading to severe animal wasting and high mortality rates. To address this, existing studies have confirmed that modified low‐dose intermittent regimens effectively enhance survival rates while largely preserving the core pathological hallmarks of CKD‐induced renal injury and anemia. ¹²⁶

3.1.2. Surgical/physical intervention models

These models utilize surgical interventions to directly alter renal anatomy or hemodynamics, aiming to closely simulate the irreversible loss of renal parenchyma or obstructive nephropathy observed in clinical settings. Representative models encompass 5/6 nephrectomy (5/6 Nx), unilateral ureteral obstruction (UUO), ¹²⁷ , ¹²⁸ , ¹²⁹ and ischemia–reperfusion injury (IRI). ¹³⁰

Among these, the 5/6 Nx model is recognized as a standard model for fundamental CKD research (Figure 4A). After 5/6 Nx in rodents, ¹³¹ the critically reduced number of remnant nephrons is forced into compensatory hypertrophy. The chronic hyperfiltration and high perfusion pressure sustained by these nephrons persistently induce microvascular damage and glomerulosclerosis, ¹³² , ¹³³ gradually progressing to CKD. The core advantage of this model lies in its ability to stably recapitulate the hallmarks of chronic systemic inflammation ¹³⁴ and absolute/relative deficiency of endogenous EPO production. ¹³⁵ This is invariably accompanied by abnormal hepcidin elevation and iron dysmetabolism‐driven anemia. ¹³⁶ , ¹³⁷

Recent mechanistic investigations further elucidate that the massive accumulation of reactive oxygen species (ROS) in the remnant kidney profoundly activates inflammatory cascades, such as NF‐κB and MAPK, which exacerbate renal parenchymal injury while directly suppressing bone marrow erythropoiesis. ¹³⁸ Furthermore, LONP1‐SOD2 redox imbalance causes mitochondrial dysfunction, exacerbating renal fibrosis and anemia together with disrupted iron balance. These findings collectively support targeted therapy for CKD anemia. ⁵⁴

Nevertheless, this surgical approach entails several practical drawbacks, including surgical trauma, the need for microsurgical expertise, and the risk of postoperative infection and acute renal failure. These concerns underscore the importance of appropriate perioperative care, analgesia, and humane endpoints in accordance with the 3Rs principles.

3.2. Inflammatory bowel disease–associated anemia models

Inflammatory bowel disease (IBD), primarily encompassing ulcerative colitis and Crohn's disease, ¹³⁹ is frequently accompanied by secondary anemia centered on iron deficiency, representing its most prevalent extraintestinal complication. ¹⁴⁰ In foundational research, administering 2%–5% (w/v) dextran sulfate sodium (DSS) in drinking water to mice for seven consecutive days remains the most classic protocol for establishing IBD models (Figure 4B), characterized macroscopically by progressive weight loss, hematochezia, and colon shortening. ⁴⁴ , ¹⁴¹ Microscopically, this model is accompanied by mucosal epithelial shedding, crypt architectural disintegration, and tight junction protein downregulation. ³⁶ , ⁴⁴ , ¹⁴¹ The damaged localized tissue exhibits not only massive inflammatory cell infiltration ³⁶ but also an explosive release of pro‐inflammatory cytokines, including IL‐6, TNF‐α, IL‐17, and IL‐1β, manifesting as a severe inflammatory cascade. ¹⁴¹

Anemia in this model arises from combined local tissue injury and systemic immunity. Initially, high‐concentration DSS induces gastrointestinal hemorrhage and drastic iron depletion. ¹⁴² Furthermore, aberrant inflammatory signals facilitate myeloid cell proliferation and suppress erythropoiesis (with the proportion of bone marrow erythroid cells plummeting from 16.3% to 9.8%). ¹⁴³ Then, proliferating myeloid cells migrate to intestinal lesions, aggravating tissue damage, which triggers a vicious cycle of excessive immune response and hematopoietic dysfunction.

Most uniquely, this model challenges the classical paradigm where hepcidin overexpression plays the central, dominant role in anemia of inflammation. Studies have verified that hepcidin exhibits only transient high expression during the initial 48 h of onset; as the disease progresses, the massive accumulation of TNF‐α in the colonic microenvironment potently suppresses the SMAD1 pathway, markedly counteracting the transcriptional activation of hepcidin by IL‐6. ⁵⁷ Consequently, systemic hepcidin cannot serve as the sole mechanistic explanation for chronic‐phase anemia; instead, locally secreted hepcidin produced in situ by type 2 conventional dendritic cells (cDC2s) within the intestinal lamina propria appears to exert a far more critical regulatory role in restricting intestinal iron absorption. ¹⁴⁴ Given the extreme complexity of the pathophysiological networks in IBD, factors such as intestinal dysbiosis have also been proven to be non‐negligible synergistic variables that drive inflammation and exacerbate secondary anemia. ³⁶ , ¹⁴⁵ , ¹⁴⁶ To address the mechanistic limitations of the DSS model, the scientific community has also developed alternative colitis models—such as the spontaneous IL‐10 knockout (KO) model ¹⁴⁷ or the adoptive T‐cell transfer model ¹⁴⁸ —to comprehensively investigate the multifaceted landscape of this disease.

3.3. Autoimmune diseases models

Autoimmune diseases (AIDs) arise from a breakdown in immune tolerance, leading to hyperactive immune responses against self‐antigens. ¹⁴⁹ To date, over 100 distinct AIDs have been identified. ¹⁵⁰ , ¹⁵¹ Consequently, establishing relevant animal models is of paramount importance for elucidating underlying pathogenic mechanisms and developing targeted therapeutics.

3.3.1. Systemic lupus erythematosus models

Systemic lupus erythematosus (SLE) is a complex, multifactorial systemic autoimmune disease ¹⁵² that can involve multiple organ systems, including the skin, ¹⁵³ hematological system, ¹⁵⁴ and kidneys. ¹⁵⁵ Given the complexity of its pathological processes, this review mainly introduces one classic SLE experimental animal model that is most closely associated with AI.

Murphy Roths Large (MRL)/lpr mice are a murine strain established on the MRL genetic background, which carries a recessive mutation in the Fas gene on chromosome 19 (designated as lymphoproliferation, or lpr mutation). This genetic defect blocks the Fas/FasL apoptotic pathway, leading to aberrant proliferation of autoreactive lymphocytes and persistent attack against host tissues. As early as the 1980s, the MRL/lpr mouse model was well recognized as a classic experimental tool for exploring the core pathological mechanisms of SLE. ¹⁵⁶

Under standard housing conditions, these mice spontaneously develop typical SLE pathological phenotypes at 20 weeks of age ⁵⁸ (Figure 4C). At this stage, marked splenomegaly is consistently observed, indicative of aberrant extramedullary hematopoiesis and potential increased erythrocyte destruction. Furthermore, the simultaneous involvement of multiple organ systems, including arthritis, cutaneous rashes, and vasculitis, reflects a substantial systemic inflammatory burden, closely approximating the pathological background of chronic disease‐related anemia driven by widespread tissue inflammation in clinical settings. ¹⁵⁷ Notably, pathological iron accumulation is frequently detected in the kidneys and spleen of these mice, providing an ideal in vivo platform for investigating SLE‐associated iron dysmetabolism. ⁵⁸ However, the limitations of this model are equally evident. Its inflammation is dominated by IFN‐γ instead of IFN‐α, making it incapable of fully recapitulating pathological changes in anemia in human SLE. ¹⁵⁷ Nevertheless, the robust reproducibility of this model in simulating sustained lymphocyte activation and multisystem chronic inflammation renders it a valuable tool for exploring hematopoietic microenvironmental alterations under chronic inflammatory conditions.

3.3.2. Rheumatoid arthritis models

Rheumatoid arthritis (RA) is a chronic, systemic autoimmune inflammatory disease that primarily affects the joints, exhibiting a higher prevalence among individuals over 40 years of age. ¹⁵⁸ , ¹⁵⁹ Although its exact pathogenesis remains to be fully elucidated, it critically involves aberrant immune responses triggered by protein citrullination, with genetic susceptibility playing a pivotal role. ¹⁶⁰ Typical clinical manifestations encompass joint swelling, severe pain, ankylosis, and eventual joint deformity; severe cases are frequently accompanied by constitutional symptoms such as anorexia and low‐grade fever. ¹⁵⁸ , ¹⁶¹ This section focuses on several classical and widely utilized animal models of RA.

Peptidoglycan–polysaccharide–induced arthritis models

Intraperitoneal administration of peptidoglycan–polysaccharide (PG‐PS) at 15 mg/kg or intra‐articular delivery at 0.17 mg/mL activates the NF‐κB/MAPK signaling cascade in Lewis rats (Figure 4D). The elevated TNF‐α and IL‐1β levels not only directly inhibit renal EPO expression ³⁷ and trigger apoptosis of erythroid progenitors but also facilitate hepcidin upregulation. ⁴⁰ , ¹⁶² , ¹⁶³ A progressive anemic phenotype synchronized with chronic inflammation is typically observed 2–4 weeks postinduction. ³⁰ It should be noted that PG‐PS is a biologically derived preparation, and interbatch variability in immunogenicity may influence model reproducibility; the use of commercially standardized preparations is recommended where available.

Collagen‐induced arthritis models

Type II collagen combined with adjuvant is injected subcutaneously into the tail (approximately 2 cm from the base of the tail, avoiding the tail veins) of 6–8‐week‐old DBA mice to induce arthritis and elicit immune cross‐reaction targeting autologous joint tissues ¹⁶⁴ (Figure 4E). Following a secondary booster, it produces stable joint swelling, bone marrow erythroid suppression, and elevated serum levels of key pro‐inflammatory factors, including TNF‐α, IFN‐γ, and IL‐17A. It is widely considered the model most closely resembling human RA.

Adjuvant‐induced arthritis models

Primarily established in rats, the adjuvant‐induced arthritis (AA) model is induced via subcutaneous injection of 100 μL complete Freund’s adjuvant (CFA) containing 10 mg/mL heat‐killed Mycobacterium tuberculosis into the plantar tissue, which triggers cross‐reactive immune injury to joint tissues (Figure 4F). It presents biphasic joint swelling from 3 days to approximately 3 weeks after injection, accompanied by sustained hemoglobin reduction driven by elevated systemic pro‐inflammatory cytokines, including IL‐1β, TNF‐α, and IL‐6. ⁵¹

4. GENETICALLY ENGINEERED MODELS

Although traditional induced models can partially recapitulate complex clinical pathology, genetically engineered strains precisely target the inflammation–hepcidin axis to dissect molecule‐specific AI effects. Hamp‐targeted genetically engineered models imply that Hamp may act as a key mediator, which could help support preliminary therapeutic screening for hepcidin antagonists.

Hepcidin overexpression in Hamp transgenic (Tg) mice is commonly achieved via liver‐specific albumin promoters or doxycycline‐inducible systems. ¹⁶⁵ This model accurately recapitulates the classical hematological and pathological features of AI patients, including profound hypoferremia and macrophage iron sequestration. Its principal advantage lies in constructing a pathological microenvironment without systemic inflammation, which facilitates the independent assessment of the damage caused by hyperhepcidinemia to iron homeostasis and erythropoiesis. ¹⁶⁶ On the contrary, Hamp KO mice subjected to chronic inflammation induced by HKBA or TO represent a classic standard model for verifying hepcidin‐dependent pathogenesis of AI. ²⁹

Beyond direct Hamp gene modification, targeting upstream regulatory proteins facilitates finer dissection of AI signaling networks. In Tmprss6‐deficient mice, impaired Hjv cleavage constitutively activates the BMP/SMAD pathway, driving sustained Hamp expression and systemic iron restriction. ¹⁶⁷ , ¹⁶⁸ This hepcidin‐driven iron blockade closely mirrors the functional iron deficiency of AI, providing a robust research model for testing therapeutic strategies to restore iron availability.

5. MODEL COMPARISON AND SELECTION FRAMEWORK

The selection of an appropriate AI animal model should be guided by the specific research question, the mechanism of interest, and practical considerations, including model duration, technical complexity, and animal welfare. To facilitate this process, we propose a comparative framework organized around key dimensions of model selection, together with a decision‐support flowchart (Figure 5) and detailed comparative information (Tables 2 and S1).

FIGURE 5.

The model selection flowchart for animal models of anemia of inflammation. Selection starts from the research focus and proceeds through three decision axes: mechanism of interest, disease context, and genetic tractability, and candidate models are then filtered by practical considerations (biosafety, surgical expertise, model duration, cost and throughput, and animal welfare) to reach the final model choice. 5/6 Nx, 5/6 nephrectomy; AA, adjuvant‐induced arthritis; AI, anemia of inflammation; BMP, bone morphogenetic protein; BSL‐2, biosafety level 2; CIA, collagen‐induced arthritis; CKD, chronic kidney disease; DSS, dextran sulfate sodium; Hamp, hepcidin antimicrobial peptide; HKBA, heat‐killed Brucella abortus; IBD, inflammatory bowel disease; IDA, iron deficiency anemia; IL‐1β, interleukin‐1β; IL‐6, interleukin‐6; JAK2, Janus kinase 2; KO, knockout; LPS, lipopolysaccharide; MRL/lpr, Murphy Roths Large/lymphoproliferation; RA, rheumatoid arthritis; SLE, systemic lupus erythematosus; SMAD, suppressor of mothers against decapentaplegic; STAT3, signal transducer and activator of transcription 3; Tg, transgenic; Tmprss6, transmembrane serine protease 6; TO, turpentine oil. Created with BioRender.com.

TABLE 2.

Overview and recommended use of animal models of anemia of inflammation.

Model category	Representative models	Modeling logic and typical course	Defining AI features	Best suited for/principal caveat
Sterile inflammation (chemical injury)	Turpentine oil (TO) 32 , 35 , 63	Turpentine under the skin → sterile abscess and inflammation (16–48 h acute; 2–6 weeks chronic)	Inflammation raises hepcidin → iron trapped → anemia	Acute‐to‐chronic studies, drug testing; abscess and pain may confound
Sterile inflammation (persistent bacterial stimulus)	Heat‐killed Brucella abortus (HKBA) 29 , 38 , 39	One injection of heat‐killed bacteria → sustained inflammation for 3–4 weeks	Persistent hepcidin → iron trapping → mild anemia (recovery by 4 weeks)	Chronic AI, hepcidin–antagonist screening; batch variability
Sterile inflammation (acute endotoxemia)	Bacterial endotoxin (LPS) 40 , 43 , 80	One dose of endotoxin → acute systemic inflammation within hours	Hepcidin peak at 4 h; serum iron nadir at 8–12 h → acute iron restriction	Acute pathway studies; not for chronic AI
Sterile inflammation (complex innate activation)	Zymosan 101 , 104 , 169	Yeast cell wall particles → strong innate immune activation; severe inflammation (up to 5 days or 2 weeks)	Iron restriction + impaired red cell production + red cell destruction	Multipathway innate inflammation; too severe for routine iron studies
Direct cytokine‐driven	IL‐6 protein 105 ; IL‐6 adenoviral vector 41 ; IL‐22 107 ; IL‐1β 108 ; TNF‐α 109 ; IFN‐γ 26	Cytokines given directly (protein or gene transfer) → activates one pathway at a time (hours to weeks)	Hepcidin up (IL‐6, IL‐22, IL‐1β) or red cell suppression (TNF‐α, IFN‐γ)	Single‐cytokine mechanisms; above‐physiological dosing, short‐lived
Disease‐specific (chronic kidney disease)	Adenine diet 114 , 120 , 122 ; 5/6 nephrectomy (5/6 Nx) 131 , 133 , 135	Adenine diet (3–8 weeks) or surgical kidney removal (4–12 weeks) → kidney failure	Inflammation + low erythropoietin + iron retention → CKD anemia	CKD anemia, treatment testing; toxicity or surgery limit throughput
Disease‐specific (inflammatory bowel disease)	DSS‐induced colitis 57 , 139 , 141	DSS in drinking water → gut injury and colitis (7 days)	Bleeding (iron loss) + inflammation (red cell suppression) → mixed anemia	Mixed iron deficiency/inflammatory anemia; complex mechanisms
Disease‐specific (autoimmune disease)	MRL/lpr SLE, 58 , 156 , 157 PG‐PS, 30 , 40 , 163 CIA, 164 and AA arthritis 51	Lupus‐prone mice (spontaneous) or arthritis immunization → chronic autoimmune inflammation (weeks)	Immune activation + inflammation → red cell suppression and anemia	Autoimmune‐associated anemia; slow onset, strain‐dependent
Genetically engineered	Hepcidin (Hamp) transgenic 165 , 166 ; Hamp knockout 29 ; Tmprss6‐deficient 167 , 168	Hepcidin gene overexpressed or deleted → hepcidin on/off without inflammation	Directly tests whether hepcidin causes iron restriction and anemia	Hepcidin causality, target validation; no real inflammation

Abbreviations: AA, adjuvant‐induced arthritis; AI, anemia of inflammation; CIA, collagen‐induced arthritis; CKD, chronic kidney disease; DSS, dextran sulfate sodium; IFN‐γ, interferon‐gamma; IL‐6, interleukin‐6; IL‐1β, interleukin‐1β; LPS, lipopolysaccharide; MRL/lpr, Murphy Roths Large/lymphoproliferation; PG‐PS, peptidoglycan–polysaccharide; SLE, systemic lupus erythematosus; Tmprss6, transmembrane serine protease 6; TNF‐α, tumor necrosis factor‐α.

5.1. Model selection based on AI mechanisms

For studies focused on acute hepcidin‐dependent AI, the LPS‐induced endotoxemia model offers the most rapid and reproducible approach, with hepcidin elevation detectable within hours of administration. ⁴⁰ , ⁴³ The HKBA model provides a more sustained inflammatory stimulus, with hepcidin elevation and anemia persisting for several weeks, making it suitable for studies of chronic hepcidin‐driven iron dysregulation and for preclinical evaluation of hepcidin‐targeted interventions. ⁷⁶ , ⁷⁹ , ¹⁷⁰ The TO model generates a biphasic inflammatory response combining acute and chronic features and is appropriate for longitudinal studies of sterile inflammation‐driven AI. ³² , ³⁴ , ³⁵ The Zymosan‐induced model provides acute or chronic innate immune activation but is less frequently used for iron metabolism studies. ¹⁰⁰ , ¹⁰⁴ , ¹⁶⁹ Regarding the basis for pathway selection, the TO, HKBA, LPS, Zymosan, and IL‐6 injection models are suitable for interrogating the IL‐6/JAK2/STAT3–hepcidin axis, whereas the LPS, Tmprss6‐deficient, and IL‐1β injection models are informative for studying BMP/SMAD‐mediated regulation.

For studies of hepcidin‐independent or partially hepcidin‐independent mechanisms, the DSS‐induced colitis model and the adenine‐induced CKD model are informative, as these conditions involve additional pathogenic factors, including gastrointestinal blood loss (in IBD) and EPO deficiency (in CKD) that modify the anemia phenotype beyond hepcidin‐mediated iron restriction alone. ¹⁴² , ¹⁴⁵ , ¹⁴⁶

5.2. Model selection by disease context

When investigating AI in the context of specific clinical conditions, disease‐specific models are preferable. The adenine‐induced CKD model reproduces the combined effects of reduced EPO production and inflammation‐driven hepcidin elevation. ⁴⁵ , ¹²⁰ , ¹²⁵ The 5/6 Nx model, although surgically more demanding, provides a model of progressive renal failure with anemia reflecting both EPO deficiency and uremic inflammation. ¹³⁴ , ¹³⁵ , ¹³⁶ The DSS colitis model captures the mixed iron deficiency–inflammatory anemia characteristic of IBD. ¹⁴⁰ , ¹⁴¹ For AID‐associated AI, MRL/lpr mice (SLE) and collagen‐induced arthritis (CIA) or AA models (RA) provide platforms for studying the chronic autoimmune–inflammation–iron dysregulation axis. ⁵¹ , ⁵⁸ , ¹⁶⁴

5.3. Model selection by genetic tractability

For studies requiring precise genetic dissection of AI mechanisms, genetically engineered models offer distinct advantages. Hamp Tg mice enable the investigation of hepcidin‐specific effects in the absence of systemic inflammation. ¹⁶⁵ , ¹⁶⁶ Hamp KO mice subjected to inflammatory challenge allow definitive assessment of the hepcidin dependence of AI phenotypes. ²⁹ Tmprss6‐deficient mice provide a model of constitutive hepcidin overexpression for studying chronic iron restriction. ¹⁶⁷ , ¹⁶⁸

5.4. Practical considerations

Practical factors influencing model selection include biosafety requirements (HKBA requires BSL‐2 facilities), surgical expertise (5/6 Nx requires microsurgical training), model duration (LPS: hours to days; HKBA: weeks; TO: hours or weeks; disease‐specific models: weeks), cost and throughput, and animal welfare. From a refinement perspective, investigators should plan anesthesia and analgesia, humane endpoints, and severity monitoring for models associated with abscess formation, surgical injury, systemic toxicity, or wasting, in accordance with the 3Rs principles.

6. BIOLOGICAL HETEROGENEITY AND MODEL VALIDITY

Most current AI models are established in young, male, and otherwise healthy animals. This represents a significant limitation given that clinical cohort studies consistently identify elderly and multimorbid patients as the populations most affected by AI. ¹⁷¹ , ¹⁷² This discrepancy arises because biological heterogeneity can interfere with the validity of existing models, yet these variables remain largely untested in current preclinical systems.

Sex‐dependent differences in iron metabolism have been documented across both in vitro and in vivo studies. ¹⁷³ Female animals generally exhibit lower baseline iron stores, higher hepcidin levels relative to iron stores, and distinct inflammatory response profiles compared to males. ¹⁷⁴ , ¹⁷⁵ , ¹⁷⁶ These differences may confound results when AI models are established exclusively in male animals.

Age‐associated alterations in erythropoiesis and iron metabolism are also relevant. Aging is accompanied by diminished erythropoietic reserve, altered BMP/SMAD signaling in the bone marrow microenvironment, and immunosenescence, all of which may affect AI model outcomes. ¹⁷⁷ , ¹⁷⁸

Common comorbidities such as obesity and diabetes can independently alter iron metabolism and inflammatory responses. ¹⁷⁹ , ¹⁸⁰ , ¹⁸¹ Accordingly, when protocols optimized for healthy animals are extrapolated to experimental animals with multiple comorbidities, key variables during model induction remain difficult to control.

Collectively, these considerations support including both sexes, considering age as a biological variable, and introducing clinically relevant comorbidities in AI studies when feasible, thereby enriching biological samples and enhancing the translational validity of the resulting models.

7. CONCLUSION

In conclusion, this review delineates the modeling strategies, pathological phenotypes, underlying molecular mechanisms, and model selection schemes of anemia of inflammation across various animal models. Based on the available evidence, preclinical animal models are valuable for deciphering AI‐driven pathogenesis, refining clinical diagnostic approaches, and facilitating the development of targeted therapeutic agents.

However, it must be acknowledged that existing single animal models still exhibit non‐negligible limitations in comprehensively reconstructing the full pathophysiological landscape of human AI. These models still demonstrate significant discrepancies from the genuine clinical microenvironment, particularly regarding the fidelity of simulating long‐term chronic progression, the capacity to integrate complex clinical comorbidities, and the precise replication of highly heterogeneous inflammatory triggers. These translational barriers may impede the translation of fundamental AI research findings into clinical applications.

Therefore, the optimization and innovation of AI animal models still hold immense developmental potential. First, at the methodological level, it is necessary to develop novel composite modeling strategies to address the limitations of existing single‐target models. Second, given the exceptionally high incidence and mortality rates of AI among the elderly, hospitalized patients, and those with underlying chronic diseases, ¹ future modeling subjects should not be restricted solely to young, healthy wild‐type animals. Introducing aged animals with natural senescence or metabolic comorbidities, as well as NHPs that are evolutionarily closer to the human immune system, will greatly increase the clinical relevance of these models, thereby providing higher‐quality evidence for the development of novel therapeutic regimens. Finally, the adoption of standardized phenotyping panels and reporting guidelines across laboratories would enhance cross‐study comparability. Although substantial progress has been achieved in elucidating the mechanisms of AI, ⁷⁶ bridging the “bench‐to‐bedside” chasm between preclinical animal models and truly efficacious clinical interventions still requires prolonged and rigorous interdisciplinary scientific endeavors.

AUTHOR CONTRIBUTIONS

Jiayi Cheng: Visualization; writing – original draft; writing – review and editing; investigation. Xiaodi Zhou: Writing – original draft; visualization; investigation. Qian Liu: Conceptualization; visualization; writing – review and editing. Zixian Liu: Writing – review and editing. Hao Wu: Conceptualization; funding acquisition; writing – review and editing.

FUNDING INFORMATION

This work was supported by the National Natural Science Foundation of China (32522101 and 32370808), the Natural Science Foundation for Distinguished Young Scholars of Hubei province, China (2024AFA056), and the Fundamental Research Funds for the Central Universities (2662025DKPY005 and 2662026DKPY007).

CONFLICT OF INTEREST STATEMENT

Hao Wu is an editorial board member of Animal Models and Experimental Medicine (AMEM) and a corresponding author of this article. To minimize bias, she was excluded from all editorial decision making related to the acceptance of this article for publication.

ETHICS STATEMENT

This article is a literature review based on existing publications and does not involve any new studies with human participants or animals conducted by the authors. Therefore, ethical approval and informed consent are not applicable.

Supporting information

Table S1. Detailed methodological and translational comparison corresponding to Table  2. 2,8‐DHA, 2,8‐dihydroxyadenine; AA, adjuvant‐induced arthritis; AI, anemia of inflammation; BMP, bone morphogenetic protein; cDC2, type 2 conventional dendritic cell; CIA, collagen‐induced arthritis; CKD, chronic kidney disease; DAMPs, damage‐associated molecular patterns; DSS, dextran sulfate sodium; EPO, erythropoietin; ERFE, erythroferrone; FasL, Fas ligand; FPN, ferroportin; Hamp, hepcidin antimicrobial peptide; HJV/Hjv, hemojuvelin; HKBA, heat‐killed Brucella abortus; IBD, inflammatory bowel disease; IFN‐γ, interferon‐gamma; IgG1, immunoglobulin G1; IL‐1β, interleukin‐1 beta; IL‐6, interleukin‐6; IL‐17, interleukin‐17; IL‐22, interleukin‐22; JAK, janus kinase; LPS, lipopolysaccharide; MAPK, mitogen‐activated protein kinase; MRL/lpr, Murphy Roths Large/lymphoproliferation; NF‐κB, nuclear factor kappa B; NLRP3, NLR family pyrin domain containing 3; PAMP, pathogen‐associated molecular pattern; PG‐PS, peptidoglycan–polysaccharide; RA, rheumatoid arthritis; RBC, red blood cell; ROS, reactive oxygen species; SD, Sprague–Dawley; SIRS, systemic inflammatory response syndrome; SLE, systemic lupus erythematosus; SMAD, suppressor of mothers against decapentaplegic; SOCS, suppressor of cytokine signaling; STAT3, signal transducer and activator of transcription 3; TLR, toll‐like receptor; Tmprss6, transmembrane serine protease 6; TNF‐α, tumor necrosis factor‐alpha; TO, turpentine oil.

DATA AVAILABILITY STATEMENT

Data sharing not applicable to this article as no datasets were generated or analysed during the current study.