Recurrent pregnancy loss (RPL) is a major clinical challenge in reproductive medicine, inflicting not only physiological harm but also profound psychological trauma on patients. Despite decades of research, up to 50% of cases remain idiopathic and are classified as unexplained RPL (URPL) (1). It is crucial to recognize that URPL is likely not a single disease entity but rather a common clinical endpoint for a heterogeneous group of underlying pathologies. The immuno-metabolic dysregulation model proposed here may represent a significant, yet not exclusive, pathway contributing to a substantial subset of these cases. Among the many potential etiologies, immune dysregulation is widely considered a key factor strongly associated with URPL. The rationale for this hypothesis lies in the exquisite immunological paradox of a successful pregnancy: the maternal immune system must establish and maintain tolerance to the semi-allogeneic fetus, which carries paternal antigens, while simultaneously preserving its capacity to defend against pathogens (2, 3).

Traditionally, immunological research into RPL has focused on local imbalances in the proportions or functions of immune cells at the maternal-fetal interface, such as the Treg/Th17 cell axis (4). However, these studies often treat the uterus as an isolated local phenomenon, an approach that largely fails to explain the upstream drivers of this local immune dysfunction. Why does the program for establishing immune tolerance repeatedly fail in certain women? What are the root pathophysiological mechanisms?

In recent years, groundbreaking advances in the field of immunometabolism have offered a novel perspective. This field has revealed that the functional state of an immune cell is intricately linked to its intrinsic metabolic programming (5, 6). For instance, tolerogenic regulatory T (Treg) cells rely on oxidative phosphorylation (OXPHOS) to sustain their suppressive function, whereas pro-inflammatory T helper 17 (Th17) cells depend on aerobic glycolysis to support their rapid proliferation and effector functions (7, 8). This discovery implies that the immune cell imbalances observed in RPL may, at their core, represent a profound dysregulation of metabolic programming.

Building on this premise, this review aims to move beyond the confines of traditional local-centric immune studies by proposing an integrative ‘gut-systemic-decidual’ model of immunometabolic dysregulation. We hypothesize that URPL is not a problem confined to the endometrium but rather a systemic disease that originates with a distal homeostatic imbalance (e.g., in the gut), progresses through systemic immunometabolic reprogramming, and ultimately manifests at the maternal-fetal interface. This article will systematically elucidate how gut dysbiosis, as the initial ‘distal disturbance’, triggers a cascade that reshapes the maternal systemic immunometabolic landscape; how these functionally pre-conditioned immune cells, upon migrating into the decidua, encounter a ‘hostile microenvironment’ that is itself metabolically compromised; and how this dual systemic and local pressure ultimately drives the functional failure of key immune cells (notably Treg and dNK cells), thereby dismantling maternal-fetal immune tolerance. By integrating multi-dimensional, recent evidence, we aim to provide a more complete and in-depth explanation for the pathophysiology of URPL and to illuminate new directions for the development of novel diagnostic markers and precision preventive and therapeutic strategies.

A growing body of evidence suggests that the immune dysregulation in RPL is not a localized uterine event but that its pathophysiological roots can be traced to distant organ systems. a concept well-established in the context of the gut-skin and gut-brain axes (9, 10)**. Among these, the gut, as the body’s largest immune organ and microbial reservoir, is emerging as a focal point of investigation. A healthy gut microbiota and its metabolites, such as short-chain fatty acids (SCFAs), are critical regulators for maintaining systemic immune homeostasis (11). SCFAs not only provide energy to colonocytes and maintain intestinal barrier integrity but also directly regulate the differentiation and function of distant immune cells, such as T cells. Classic animal model studies have long established that SCFAs like butyrate, produced by commensal bacteria, are key factors in inducing the differentiation of colonic Treg cells (12, 13), thereby establishing a systemic environment of immune tolerance. A recent landmark study provided direct and compelling evidence for this “gut-immune” axis: in a clinical cohort of patients with unexplained RPL (URPL), gut dysbiosis and the associated reduction in microbial metabolites (such as SCFAs and secondary bile acids) were directly and significantly correlated with a reduction in protective peripheral Treg cells and an increase in pro-inflammatory Th1/Th17 cells (14). Deeper mechanistic investigations further revealed that SCFAs (particularly propionate and butyrate) can directly act on G protein-coupled receptors (GPR43) on the surface of T cells in distant immune organs (such as the spleen), potently promoting their differentiation into immunosuppressive Treg cells and consequently ameliorating a mouse model of fetal loss (15). This provides a clear mechanistic link for the concept that a healthy gut ecosystem actively cultivates a systemic immune tolerance environment favorable for pregnancy through its metabolic output.

However, when the gut microbiota homeostasis is disrupted (gut dysbiosis), this delicate balance collapses. The overgrowth of pathogenic bacteria and the decline of beneficial bacteria lead to a decrease in the production of beneficial metabolites like SCFAs, coupled with an increase in inflammatory molecules such as lipopolysaccharide (LPS) (4). This concept is supported by early evidence: two decades ago, prospective studies began to explore the association between host genetic susceptibility to LPS (such as CD14 gene polymorphisms) and the risk of RPL, laying the early groundwork for an “endotoxin hypothesis” (16). More importantly, dysbiosis is often accompanied by impaired intestinal barrier function, a condition clinically referred to as “leaky gut” (17, 18). This allows microbial products like LPS, which are normally confined to the intestinal lumen, to cross the barrier and enter the bloodstream, triggering a sustained, low-grade systemic inflammatory response (1, 18, 19) that defines “metabolic endotoxemia”. Notably, this pathological cascade may be more complex; for instance, some evidence suggests the existence of a “gut-lung-uterus” axis, where inflammatory signals originating from the gut may first affect the lungs and then extend to the uterus via the circulatory system, forming a multi-organ cascade (20). Furthermore, this gut-decidual axis may not be a one-way street. The profound hormonal shifts during pregnancy, including rising levels of progesterone and estrogen, can themselves modulate the composition and function of the gut microbiota. This suggests a potential bidirectional feedback loop, where pregnancy-induced changes in the gut could either reinforce a healthy state or, in susceptible individuals, exacerbate pre-existing dysbiosis, further amplifying the systemic inflammatory pressure on the maternal-fetal interface.

Circulating LPS, as a potent pathogen-associated molecular pattern (PAMP), exerts widespread “pre-sensitizing” and “metabolic reprogramming” effects on the maternal immune system. The detrimental impact of this systemic inflammatory state is multi-faceted. First, in addition to inducing the production of large quantities of pro-inflammatory cytokines (such as TNF-α and IL-6) from monocytes/macrophages via the classic Toll-like receptor 4 (TLR4) signaling pathway (17), LPS can also activate the complement system. This leads to the generation of the potent inflammatory mediator C5a, which directly mediates placental injury and pregnancy loss through its receptor, C5aR1 (21). This chronic, low-grade inflammatory environment constitutes a “pre-sensitization” process for the maternal immune system, systematically altering the differentiation potential of T cells by inhibiting Treg cell differentiation while promoting the expansion of Th17 cells, thereby disrupting systemic immune balance (22). Second, at a deeper level, LPS fundamentally re-wires the metabolic programs of immune cells. Persistent LPS stimulation induces a shift toward glycolytic metabolism in myeloid cells like monocytes, a pro-inflammatory “war-ready” phenotype that renders them more easily activated to release inflammatory mediators (23). Within the cell, continuous LPS stimulation inhibits the activity of core metabolic sensors like AMPK and SIRT1. This not only drives cells toward a pro-inflammatory phenotype dependent on glycolysis but can also directly trigger a form of metabolism-dependent cell death termed “ferroptosis,” causing direct damage to tissues such as the endometrium (24). Furthermore, LPS can amplify inflammation at a post-transcriptional level through a newly discovered and elegant mechanism. It promotes the sequestration of RC3H1, a protein responsible for degrading TNF-α mRNA, within intracellular stress granules by a protein named MNSFβ. This sequestration prevents the degradation of TNF-α mRNA, leading to its uncontrolled expression and creating a self-amplifying inflammatory loop (25) (Figure 1).

If gut dysbiosis is the “distant storm” that initiates the pathological process, then the decidual microenvironment is the local battlefield where this conflict ultimately culminates. Under physiological conditions, the early-pregnancy decidua is a unique immune-privileged microenvironment that actively suppresses immune attacks to protect the embryo through multiple mechanisms. Among these, decidual stromal cells (DSCs) play the role of central coordinators. They not only possess intrinsic immunosuppressive functions but also secrete various cytokines and chemokines to recruit and regulate the phenotype and function of immune cells (especially Treg cells and specialized dNK cells), collectively building a robust niche of immune tolerance (25) (Figure 2). This microenvironment is also metabolically unique: it is a relatively hypoxic yet nutrient-rich area, a distinctive metabolic feature crucial for maintaining immune tolerance (26, 27). Importantly, this privileged status is not a passive “isolation” but a dynamic equilibrium actively constructed and maintained by a sophisticated signaling network between maternal and fetal cells. For example, normal dNK cells secrete pigment epithelium-derived factor (PEDF) to protect DSCs from inflammatory and apoptotic damage, forming a positive protective feedback loop (28).

In patients with RPL, however, this “immune sanctuary” transforms into a metabolically hostile environment (29). This transformation is particularly evident in its spatial dimension. The latest spatial transcriptomics studies have clearly depicted that the implantation site (IZ) of a normal decidua is a protective zone enriched with dNK1 cells and M2-like macrophages. In RPL, however, this protective structure is dismantled; the protective cells are sharply reduced and replaced by an infiltration of cytotoxic CD8+ T cells. In essence, the spatial distribution pattern of cells undergoes a fundamental reversal (30). The root of this spatial collapse lies in the profound dysregulation of molecular and metabolic programs within the microenvironment. Multiple integrative transcriptomic and metabolomic studies have consistently revealed significant metabolic disturbances in the decidual tissue of RPL patients (14, 31, 32). Even at the upstream epigenetic level, RPL-associated villi and decidua exhibit a state of widespread DNA hypermethylation, affecting the gene expression of numerous key signaling pathways (33). This metabolic dysregulation is not merely an intrinsic cellular malfunction but is also driven by aberrant signaling molecules in the microenvironment. For instance, DSCs from RPL patients excessively secrete the extracellular matrix protein Decorin, which acts as a detrimental signaling molecule that directly targets decidual macrophages, inducing their mitochondrial dysfunction (34). Concurrently, systemic lipid imbalance further exacerbates the local pathology. It leads to a sharp increase in downstream inflammatory metabolites of arachidonic acid (AA) in the decidua—oxylipins such as PGE2 and PGF2α—and these inflammatory lipid molecules further shape a pro-inflammatory microenvironment (35).

Amid these chaotic metabolic changes, a hallmark feature is an aberrant upregulation of the glycolytic pathway and an impaired tricarboxylic acid (TCA) cycle, leading to the excessive accumulation of acidic metabolites like lactate (31, 32). It is noteworthy, however, that the dysregulation of glucose metabolism is not a simple “upregulation” or “downregulation” but rather a failure in the control of key metabolic nodes. A pioneering study discovered that the glycolytic intermediate fructose-1,6-bisphosphate (FBP) is a critical signal for inducing tolerogenic interactions between DSCs and macrophages, and its deficiency in RPL may be a significant cause of immune dysregulation (36). Furthermore, autophagy, a core mechanism for maintaining cellular metabolic homeostasis, is severely defective, preventing the timely clearance of damaged organelles and proteins and thus exacerbating cellular stress and dysfunction (37). This dysregulation can even trigger “ferroptosis,” a form of metabolism-dependent cell death, inflicting direct physical damage on decidual and trophoblast cells (23).

This metabolically disordered microenvironment not only directly harms maternal immune cells but also poses a grave threat to the survival and function of trophoblast cells, which are semi-allogeneic. At the same time, the disturbed microenvironment disrupts the intricate metabolic programs of the trophoblasts themselves. For example, the expression of HSD3B1, a key enzyme for progesterone synthesis in invasive extravillous trophoblasts (EVTs), is downregulated, leading to impaired local synthesis of progesterone, an endogenous immunosuppressant (38). The secretion of the critical growth factor GDF15 by trophoblasts may also be reduced, further weakening their invasive capacity (39). Even at a more refined molecular level, trophoblast function is severely disrupted by epigenetic and post-transcriptional dysregulation. For example, the abnormal upregulation of long non-coding RNAs (such as Lnc-HZ05) disrupts cytoskeletal rearrangement, thereby inhibiting the formation of migrasomes, a key structure for trophoblast migration (40).

More importantly, this hostile battlefield is not silent but is filled with erroneous signals and disrupted interactions. This “acidified” and “dysregulated” metabolic microenvironment, shaped by a multitude of factors (41), ultimately exerts a direct and decisive influence on the function of immune cells.

For example, high concentrations of lactate and a hypoxic environment can stabilize the expression of HIF-1α, which plays a dual role in different cells: on one hand, it inhibits the proliferation and function of Treg cells (42); on the other, it promotes the survival and differentiation of pro-inflammatory Th17 cells, thus directly driving the Treg/Th17 imbalance from a metabolic level. Meanwhile, other dysregulated metabolite profiles (such as abnormal tryptophan metabolism) can directly impair the cytotoxic regulation and vascular remodeling capabilities of dNK cells via pathways like the aryl hydrocarbon receptor (AHR) pathway (5). Single-cell sequencing has revealed with unprecedented precision that the decidua in RPL does not suffer from a uniform dysfunction but is instead dominated by pathogenic cell subpopulations. For instance, the massive emergence of “inflammatory” iDSCs and “glycolytic” glyDSCs actively secrete inflammatory factors, remodeling the entire communication network of the microenvironment (43). This breakdown in communication is also reflected between mother and fetus. Normally, fetal villi deliver an inhibitory signal, miR-29a-3p, to maternal dNK cells via exosomes to suppress their aggression. In RPL, however, this critical tolerance signal is significantly weakened, leading to a communication breakdown and the eruption of conflict (44). Therefore, when immune cells that have already been “pre-sensitized” in the systemic circulation arrive at a decidua filled with erroneous cell subpopulations, incorrect signaling molecules, and aberrant metabolic products, not only do they fail to be reprogrammed into a tolerogenic phenotype, but their inherent inflammatory propensity is further amplified and solidified, ultimately leading to an immune attack on the embryo.

If the deterioration of the decidual microenvironment sets the local stage for the conflict, then the functional collapse of key immune cells within this environment is the direct cause of pregnancy failure. Here, we will focus on two core cell types essential for maintaining maternal-fetal tolerance—regulatory T cells (Tregs) and decidual natural killer (dNK) cells—to dissect how profound errors in their metabolic programming lead to their functional collapse in RPL.

This review has systematically challenged the traditional view of unexplained recurrent pregnancy loss (URPL) as an isolated uterine event. By integrating the latest evidence from immuno-metabolism, microbiology, and maternal-fetal medicine, we have constructed and substantiated a novel “gut-systemic-decidual” immuno-metabolic dysregulation model. We propose that URPL is not merely a local immune imbalance but a systemic disease initiated by distant gut dysbiosis, propagated through the metabolic reprogramming of systemic immune cells and inflammatory signaling, and ultimately erupting in a metabolically dysregulated decidual microenvironment. From the impaired systemic differentiation of Treg cells due to reduced gut SCFAs, to the “metabolic endotoxemia” triggered by LPS leakage; from the “metabolic hostility” of the decidual microenvironment, to the “functional collapse” of key immune cells (Tregs and dNK cells) due to errors in metabolic programming—this series of interconnected pathological events collectively dismantles maternal-fetal immune tolerance. This integrated model not only provides a deeper, more logical pathophysiological explanation for up to 50% of “unexplained” RPL cases but also opens up entirely new horizons for future research and clinical practice. It is important to acknowledge, however, the limitations of the current evidence. Much of the human data supporting this model is correlational, and future longitudinal studies and intervention trials are essential to firmly establish causality. Moreover, given the heterogeneity of URPL, this model likely explains a significant subset of cases, and the potential for bidirectional interactions within the gut-decidual axis warrants further investigation.

Looking ahead, based on this new paradigm, research in the RPL field will enter a new era of multidimensional, dynamic, and precision-based approaches.

In diagnostics, the future goal is to build a dynamic warning system capable of “full-temporal, full-spatial” risk assessment. This will require moving beyond the current static multi-omics snapshots to develop artificial intelligence (AI) predictive models that can integrate gut microbiota, serum metabolites, the epigenetic status of immune cells, and their real-time functions (such as Treg plasticity and memory status). This approach will be crucial for patient stratification, distinguishing subtypes of URPL based on their primary driver (e.g., ‘gut-driven inflammatory type’ vs. ‘local decidual metabolic type’). We envision that the management of high-risk pregnant women in the future may involve creating their personalized “Digital Twin” models, enabling pre-emptive warnings and guiding targeted preventive approaches through dynamic monitoring and algorithmic prediction before the cascade of immune imbalance is triggered. Furthermore, incorporating paternal factors (such as sperm epigenetic and metabolic defects) as standard parameters into these models will be a critical step toward achieving a truly comprehensive diagnosis.

In therapeutics, intervention strategies will shift from symptomatic treatment to multi-target, mechanism-based precision interventions. Future treatment regimens will be a multidimensional combination, potentially including: (1) “Engineered probiotics”: not just simple strain supplementation, but “Live Biotherapeutic Products (LBPs)” that are engineered to produce specific SCFAs or immunomodulatory molecules; (2) “Treg memory remodeling therapy”: using targeted drugs or cell therapies to repair the potentially depleted “ex-Treg” memory cell pool in RPL patients, restoring their rapid response capability in subsequent pregnancies; and (3) “Decidual targeted delivery systems”: using nanoparticles and other carriers to precisely deliver PD-1 agonists, metabolic regulators, or anti-inflammatory drugs to the maternal-fetal interface, allowing for highly selective intervention in the local microenvironment without affecting systemic immune function. These interventions represent promising preventive approaches that can be tailored to patient subgroups identified through advanced diagnostics.

In basic research, several core scientific questions remain to be answered. We need higher-resolution spatiotemporal multi-omics technologies to track the migration trajectory and metabolic state evolution of individual immune cells from the gut to the decidua in real time. We need more sophisticated humanized mouse models and decidual organoids to validate the causal relationships of key molecules in a setting that more closely resembles the human body. Most importantly, we need to answer an ultimate question: How do we define and maintain an immuno-metabolic homeostasis that is “most favorable for pregnancy”? This is not only for a successful pregnancy but also because the maternal immune-metabolic state during gestation profoundly influences the long-term health of the offspring through “metabolic programming”.

In conclusion, by expanding the etiological understanding of RPL from the local uterine environment to a systemic “gut-systemic-decidual” framework, we not only provide a critical new perspective for understanding this stubborn clinical challenge but also point toward a clearer research direction for safeguarding the long-term health of both mother and child.