Experimental animal models and patient-derived platforms to bridge preclinical discovery and translational therapeutics in pulmonary arterial hypertension

Abstract

Pulmonary arterial hypertension (PAH) is a progressive vascular disease characterized by sustained elevation of pulmonary arterial pressure, endothelial cell dysfunction, and right ventricular failure. A wide range of experimental animal models, including the monocrotaline model, Sugen combined with hypoxia, and pulmonary artery banding in large animals, have been pivotal in uncovering disease mechanisms such as vascular remodeling, metabolic dysregulation, and hypoxia-inducible signaling. More recently, human-based platforms, including induced pluripotent stem cell-derived vascular cells, organ-on-chip systems, and precision-cut lung slices, have emerged as powerful tools to model patient-specific pathophysiology and study pharmacological responses. These systems enable the interrogation of BMPR2 mutations, mitochondrial dysfunction, and sex-specific responses, factors often overlooked in traditional preclinical models. Moreover, integrating these platforms with omics technologies and comorbidity-driven experimental systems addresses key translational gaps. This review provides an overview of animal and human-based models used in PAH research and highlights emerging strategies to enhance their translational relevance. We advocate for a multi-platform and precision medicine-oriented approach that bridges preclinical insights with clinical outcomes to accelerate therapeutic development in PAH.

Introduction

Pulmonary hypertension (PH) is a progressive and multifactorial disorder characterized by a mean pulmonary arterial pressure (PAP) greater than 20 mmHg at rest, as defined by the updated international hemodynamic criteria [1, 2]. This revised threshold, lowered from the historical cutoff of 25 mmHg, reflects the clinical need for early detection and intervention [2]. The World Health Organization (WHO) classifies PH into five etiological groups: Group 1 (pulmonary arterial hypertension, or PAH), Group 2 (PH due to left heart disease), Group 3 (PH associated with lung disease and/or hypoxia), Group 4 (chronic thromboembolic PH), and Group 5 (PH with unclear or multifactorial mechanisms) [3]. Group 1 PAH remains the primary focus of mechanistic investigations and therapeutic development.

PAH is characterized by progressive pulmonary vascular remodeling, sustained vasoconstriction, in situ thrombosis, and elevated pulmonary vascular resistance (PVR), which together impose chronic pressure overload on the right ventricle (RV) [4]. The RV initially adapts through hypertrophy but eventually undergoes maladaptive remodeling, including capillary rarefaction, metabolic reprogramming, and fibrotic remodeling. This leads to RV failure, which is the ultimate cause of death in PAH patients [5, 6]. At the cellular level, PAH involves coordinated dysfunction across multiple pulmonary vascular cell types, including endothelial cells (PAECs), smooth muscle cells (PASMCs), fibroblasts, and immune cells. Endothelial dysfunction is an early and pivotal event marked by reduced nitric oxide (NO) bioavailability, impaired prostacyclin signaling, and increased production of vasoconstrictive and mitogenic mediators, such as endothelin-1 (ET-1) and thromboxane A2 [7–9]. These imbalances promote hyperproliferation of PASMCs, adventitial fibrosis, and the narrowing of the vascular lumen [10]. These pathogenic mechanisms, including endothelial dysfunction, inflammation, metabolic dysregulation, and extracellular matrix remodeling, collectively drive progressive pulmonary vascular remodeling and RV maladaptation (Fig. 1). Understanding the interplay among these processes is critical for selecting appropriate experimental models that both recapitulate specific disease features and identify therapeutic targets.

Fig. 1.

Pathobiological hallmarks of pulmonary hypertension (PH). Chronic pulmonary vascular remodeling, characterized by intimal fibrosis, smooth muscle hyperplasia, adventitial thickening, and plexiform lesions, narrows the arterial lumen, disrupts perfusion, and increases pulmonary vascular resistance. Key mechanisms contributing to PH include perivascular inflammation with cytokine release and immune activation; endothelial dysfunction marked by reduced nitric oxide and prostacyclin bioavailability, BMPR2 loss, and EndoMT; metabolic dysregulation involving aerobic glycolysis, glutaminolysis, mitochondrial impairment, and PDK activation; and extracellular matrix remodeling with collagen accumulation, elastin degradation, and altered mechanotransduction. Created with BioRender.com.

Clinically, accurate diagnostic and prognostic biomarkers are crucial for the timely diagnosis, effective monitoring, and personalized therapeutic strategies for PAH [11]. The evolution and integration of clinical laboratory science have facilitated the standardized implementation and interpretation of laboratory biomarkers, significantly enhancing patient stratification and clinical decision-making [12]. For example, biomarkers such as N-terminal pro-brain natriuretic peptide (NT-proBNP), initially validated in heart failure patients, have proven essential in assessing cardiac dysfunction and severity of disease in patients with PAH, demonstrating the value of translating biomarker research across cardiovascular disorders [13, 14]. Inflammatory and immunologic markers, such as serum IgE and eosinophil counts, are commonly used in the context of allergic and respiratory diseases [15]. Their relevance in PAH may help identify inflammatory subtypes and support precision medicine approaches.

Despite decades of mechanistic advances, the translation of preclinical discoveries into effective therapies for PAH remains limited. This translational gap stems from several factors, including species-specific differences in vascular biology, incomplete modeling of human pathophysiology, and a lack of standardized endpoints [16–18]. Nevertheless, well-characterized experimental models continue to serve as essential tools for dissecting the molecular mechanisms of diseases and evaluating the potential therapies. Among these, rodent models, such as monocrotaline (MCT), Sugen-hypoxia (SuHx), and chronic hypoxia, have been instrumental in reproducing the key features of human PAH, including pulmonary vascular remodeling, inflammation, and RV hypertrophy [19–23]. While no single model fully replicates the complex vascular, immune, and cardiac features of human PAH [24], each provides valuable insights into specific aspects of the disease. Therefore, model selection should be tailored to the biological process under investigation, whether that is endothelial injury, vascular occlusion, or RV adaptation [25].

In this review, we provide an in-depth and updated synthesis of the experimental models used in PAH, emphasizing both classical and advanced platforms. To ensure comprehensive coverage, we conducted a targeted literature review using combinations of keywords such as “pulmonary arterial hypertension,” “animal models,” “preclinical models,” “experimental models,” “alternative models,” “organ-on-chip,” “iPSC,” and “translational research” across PubMed and Web of Science databases until March 2025. Here, we highlight recent technological innovations, including human-based systems and omics-integrated approaches, that enhance mechanistic resolution. Finally, we propose a multi-platform strategy that integrates in vivo, ex vivo, and in vitro models to improve translational fidelity and better connect preclinical insights with clinical applications.

Classical rodent models in PAH: foundations of preclinical discovery

Preclinical drug development for PAH relies heavily on animal models to elucidate disease mechanisms and evaluate therapeutic efficacy. Among the most widely used models are monocrotaline, Sugen 5416 plus hypoxia, and chronic hypoxia models, each recapitulating key pathological features such as endothelial dysfunction, pulmonary vascular remodeling, and RV hypertrophy to varying degrees of clinical relevance [19, 26]. More advanced models, including left pneumonectomy followed by MCT or SU5416 administration and pulmonary artery ligation in rodents and large animals (e.g., pigs and sheep), provide a more severe phenotype that mimics the advanced stages of human PAH [23, 27]. Model selection is influenced by the specific pathophysiological process and therapeutic pathway under investigation [19, 28]. Figure 2 contextualizes the severity and spectrum of pulmonary vascular and RV remodeling in commonly used preclinical models. Given the heterogeneity of PAH, no single model fully captures the spectrum of the disease. Therefore, a multi-model approach remains essential for gaining comprehensive insights into disease progression and validating candidate interventions across different biological contexts. The strategic integration of diverse experimental systems has significantly advanced the understanding of PAH pathogenesis and facilitated the identification of novel therapeutic targets.

Fig. 2.

Progression of pulmonary vascular and right ventricular (RV) remodeling in PAH. The pulmonary artery (PA) undergoes structural changes due to chronic vascular stress, leading to a progressive development from normal to intimal thickening and medial hypertrophy, ultimately resulting in severe luminal occlusion. The RV initially adapts through concentric hypertrophy to maintain cardiac output, but eventually develops maladaptive remodeling, leading to right ventricular dilation and dysfunction due to persistent overload. The schematic diagram illustrates the experimental PAH models, categorized by the severity of pulmonary vascular remodeling and right ventricular dysfunction. Vascular remodeling in chronic hypoxia models remains mild, whereas RV function remains intact. The Sugen/hypoxia (SuHx), monocrotaline, and pulmonary artery banding models produce more severe vascular and RV remodeling, which differs between species and the induction methods. Created with BioRender.com.

Monocrotaline model

The MCT model remains a cornerstone of preclinical PAH research because of its reproducibility, simplicity, and robust vascular phenotype [29]. Derived from Crotalaria spectabilis, MCT is administered subcutaneously (commonly 60 mg/kg) and undergoes hepatic bioactivation by cytochrome P450 enzymes, primarily CYP3A4, generating a toxic pyrrole metabolite, dihydro-monocrotaline (MCTP) [30–32]. This bioactive intermediate selectively injures PAECs, initiating a cascade of events that culminate in pulmonary vascular remodeling and RV hypertrophy [33]. Mechanistically, MCT-induced endothelial injury is characterized by marked megalocytosis, suppression of endothelial nitric oxide synthase, and disruption of intracellular membrane trafficking [34]. These alterations impair caveolar nitric oxide signaling and promote pro-proliferative, anti-apoptotic responses [34]. Critically, MCT exposure downregulates bone morphogenetic protein receptor 2 (BMPR2) expression and activates BMP antagonists, leading to dysfunctional BMP signaling, a defining feature of heritable and idiopathic PAH [35, 36]. Histologically, the MCT model induces medial hypertrophy, perivascular inflammation, and RV fibrosis, but lacks the complex plexiform and neointimal lesions observed in end-stage human PAH [37]. While the vascular changes are robust, the disease phenotype is variably progressive and partially reversible, with limitations attributed to the aqueous instability of MCTP and species-specific metabolism. Notably, mice metabolize MCT inefficiently and develop PH inconsistently, likely due to the differential expression of hepatic CYP isoforms [38, 39].

Overall, the MCT rat model offers unique insights into the early interplay between endothelial cell dysfunction and smooth muscle cell proliferation [40–43]. It is particularly well-suited for dissecting inflammation-mediated mechanisms and pharmacologic testing of anti-inflammatory, anti-proliferative, and RV-targeted therapies [19, 44]. Combinatorial strategies, such as MCT with left pneumonectomy, can intensify vascular injury and yield neointimal and angio-obliterative lesions, partially recapitulating advanced human PAH [23]. However, the systemic toxicity of MCT, including hepatic veno-occlusive disease and RV myocarditis, must be carefully considered when interpreting its translational relevance. In summary, although the MCT model does not fully replicate the complex vascular pathology of human PAH, it remains an indispensable model for studying the key molecular drivers of pulmonary vascular injury and RV adaptation.

Chronic hypoxia model

The chronic hypoxia model is one of the most established and widely used preclinical models of PH [45]. It simulates sustained low oxygen exposure, typically 10% inspired O₂, yielding a PaO₂ of ~ 40 mmHg over 2–4 weeks, thereby reproducing the pulmonary vascular adaptations observed in high-altitude PH and chronic lung diseases [46, 47]. Hypoxic vasoconstriction of pulmonary resistance vessels is followed by concentric medial hypertrophy and muscularization of distal pulmonary arterioles, leading to elevated PAP and increased RV afterload [45, 47, 48]. This model recapitulates the progressive onset and sustained progression of vascular remodeling, distinguishing it from more acute or toxic models, such as MCT. Notably, chronic hypoxia directly targets PAECs, altering their production of vasoactive mediators, growth factors, and inflammatory signals that modulate the behavior of SMCs and adventitial fibroblasts [28]. These cell-to-cell interactions drive persistent vasoconstriction and maladaptive remodeling. A key strength of the chronic hypoxia model lies in its ability to interrogate conserved molecular responses to oxygen deprivation [47]. Central to this is the activation of hypoxia-inducible factor 1-alpha (HIF-1α), which stabilizes under low oxygen tension and triggers transcriptional programs that promote cell proliferation, metabolic reprogramming, and vascular remodeling [49, 50]. Mitochondrial dysfunction emerges as a primary upstream signal. In hypoxia, reduced reactive oxygen species production by mitochondrial complexes I and III leads to membrane depolarization, calcium influx, and inhibition of voltage-gated potassium (Kv) channels, pathways that are similarly dysregulated in human PAH [51].

Genetic predisposition to mitochondrial dysfunction further amplifies disease severity. For example, fawn-hooded rats spontaneously develop PH due to defective electron transport chain components linked to a chromosome 1 mutation [52]. This defect leads to insufficient reactive oxygen species generation under normoxic conditions, mimicking a “pseudohypoxic” state with constitutive HIF-1α activation and Kv channel downregulation [53]. These animals serve as a compelling platform for studying the metabolic contributions to disease onset. Although the model exhibits moderate severity and lacks plexiform lesions, features that limit translational fidelity, its pathophysiological relevance is reinforced by its responsiveness to metabolic therapy. Pyruvate dehydrogenase kinase inhibition with dichloroacetate restores mitochondrial glucose oxidation, reverses glycolytic reprogramming in PASMCs, and has been shown to regress established PH in both hypoxia- and MCT-induced models [54]. Nonetheless, interspecies variability (e.g., greater susceptibility in Sprague–Dawley rats than in C57BL/6 mice), along with differences in age, sex, and hypoxia tolerance, influences disease penetrance and progression [55–57]. Moreover, the reversibility of pulmonary vascular changes upon reoxygenation challenges its utility in modeling irreversible human PAH. Importantly, RV failure, a key determinant of patient prognosis, was not reliably reproduced in this model. Despite its limitations in capturing late-stage human pathology, the chronic hypoxia model remains a critical experimental system for uncovering hypoxia-sensing mechanisms, mitochondrial dysfunction, and metabolic plasticity, which all contribute to early PAH pathogenesis.

Sugen-hypoxia model

The SU5416/hypoxia (SuHx) model represents a pivotal advancement in preclinical PAH research [25]. This two-hit model combines a single administration of SU5416, a selective vascular endothelial growth factor receptor 2 (VEGFR2) inhibitor, with three weeks of chronic hypoxia (10% O₂), followed by reoxygenation [46]. Combined VEGF signaling inhibition with hypoxic stress induces selective pulmonary endothelial cell apoptosis, clonal expansion of apoptosis-resistant endothelial populations, and angioproliferative remodeling, which recapitulates the hallmark features of severe human PAH [58]. Pathologically, the SuHx model induces intimal thickening, neointimal occlusion, and plexiform-like lesions in rats, along with elevated pulmonary arterial pressure, RV hypertrophy, and progressive RV dysfunction [59]. One of the key strengths of the SuHx model is its durability, as PH persists for weeks beyond the hypoxic phase, enabling the study of long-term vascular and RV remodeling [46]. It also provides a high-fidelity platform for evaluating anti-remodeling therapies and dissecting the molecular mechanisms involving BMPR2, HIF signaling, and metabolic dysregulation [60–62]. However, species-specific responses limit generalizability. While Sprague–Dawley rats robustly develop angio-obliterative lesions and RV failure, mice, particularly C57BL/6J mice, demonstrate a less severe phenotype [25]. In murine models, SuHx induces PH without fully replicating the plexiform architecture observed in rats or humans. These differences may arise from species-specific VEGF signaling, bronchial circulation, or regenerative endothelial capacity. Additionally, technical challenges associated with echocardiographic and hemodynamic assessments in small rodents reduce throughput and reproducibility. The SuHx model remains a critical tool in translational PAH research, particularly for interrogating angioproliferative vascular remodeling and testing the efficacy of targeted interventions. To maximize the translational value, researchers must carefully consider the species, genetic background, and experimental design when employing this model.

Pneumonectomy combined with a second hit: a model of severe angio-obliterative PAH

Pneumonectomy, combined with a “second hit” model, represents a powerful strategy to induce severe PAH with high histopathological fidelity to human disease [27]. In this model, left lung resection serves as the initial insult, followed by the administration of either MCT or the VEGFR2 inhibitor SU5416 as a secondary pathogenic stimulus [27]. This dual-hit approach leads to profound pulmonary vascular remodeling, neointimal proliferation, and RV hypertrophy, closely mimicking the advanced angioproliferative lesions characteristic of human PAH. Surgically, the procedure involves ligation and excision of the left pulmonary artery, vein, and bronchus in anesthetized Sprague-Dawley rats, thereby redirecting all pulmonary blood flow to the right lung [23, 27]. This unilateral pneumonectomy increases shear stress, turbulent flow, and compensatory vascular remodeling in the remaining lung. One week post-surgery, animals receive a single subcutaneous dose of MCT (60 mg/kg) or SU5416 (20 mg/kg), initiating progressive pulmonary vascular injury and endothelial cell dysfunction [23, 27]. Over the following 6–8 weeks, this model evolves into a severe PAH phenotype, marked by sustained pulmonary hypertension, RV pressure overload, and complex neointimal and angio-obliterative lesions [23]. Hemodynamic assessments typically reveal elevated pulmonary artery and RV systolic pressures, paralleling the progression of late-stage PAH [23]. The model’s ability to recapitulate plexiform-like vascular pathology makes it an excellent platform for dissecting the chronic molecular mechanisms of endothelial proliferation and therapeutic resistance.

However, several limitations must be considered. The surgical component introduces confounding physiological stress, which may independently alter vascular tone and remodeling, complicating the attribution of effects solely to MCT or SU5416 administration. Furthermore, the procedure requires significant technical expertise, is associated with increased mortality and inflammatory response, and exhibits high inter-animal variability, which limits its widespread use and reproducibility across laboratories. Overall, the pneumonectomy “second-hit” model is valuable but technically challenging. It is particularly useful for studying advanced lesion formation and testing high-fidelity interventions. Nonetheless, careful experimental design and consideration of surgical variables are essential to ensure the translational relevance and interpretability of the findings. These limitations underscore the urgent need for standardizing model selection, rigorous experimental design, and alignment with clinical phenotypes to enhance translational reliability, as emphasized in recent expert consensus reports [63, 64]. To improve the translational reliability of preclinical PAH research, studies must adhere to rigorous methodology, including clearly defined inclusion and exclusion criteria, sufficient sample sizes to ensure statistical power, randomization, blinded outcome assessment, and standardized reporting practices [63]. Strengthening the rigor of preclinical studies can reduce bias, improve reproducibility, and mitigate the limitations inherent to animal models used in these studies. Incorporating advanced imaging techniques, employing multiple complementary PAH models, and validating findings across rodent and human systems are critical steps toward clinical applicability. Moreover, greater emphasis should be placed on evaluating the right ventricular function, which is a major prognostic determinant in patients with PAH. Furthermore, multicenter preclinical studies, which mirror the structure of clinical trials across multiple research institutions, should be actively encouraged to strengthen reproducibility, increase statistical robustness, and facilitate broader validation of promising therapeutic candidates.

Pig pulmonary artery banding model: a large-animal approach to RV failure in PAH

Large animal models are critical for bridging the translational gap between rodent and human clinical trials. Pigs offer unique advantages in cardiovascular and pulmonary research because of their anatomical and physiological similarities to humans.

The pulmonary artery (PA) banding model in pigs replicates pressure overload-induced RV remodeling and failure, providing a high-fidelity platform for studying the natural progression of RV dysfunction in PAH [65, 66]. This model involves surgical constriction of the main pulmonary artery, which increases the PAP and imposes a chronic afterload on the RV [66]. The degree of constriction can be titrated to induce either gradual, adaptive hypertrophy or, more severe, maladaptive remodeling [67]. Over a period of approximately five weeks, animals develop pathophysiological changes consistent with pressure overload, including RV hypertrophy, reduced ejection fraction, and fibrotic remodeling. Importantly, this model isolates RV remodeling from primary pulmonary vascular injury, allowing for the study of RV-specific pathophysiological mechanisms [68]. In PAH, patient outcomes are more closely associated with RV function than with pulmonary vascular resistance. The PA banding model thus enables a precise investigation of the transition from compensated to decompensated RV remodeling, a central yet poorly understood process in PAH progression [69]. In the compensated phase, the RV undergoes concentric hypertrophy, with preserved systolic function and minimal fibrosis. Over time, or with tighter constriction, the RV transitions to a decompensated state, characterized by extracellular matrix accumulation, inflammatory cell infiltration, capillary rarefaction, and cardiomyocyte apoptosis [70]. These changes culminate in decreased cardiac output, impaired exercise tolerance, and heart failure. This staged remodeling progression provides an ideal framework for investigating the molecular drivers of RV adaptation and failure. The large size of pigs facilitates real-time hemodynamic monitoring, cardiac MRI, pressure-volume loop analysis, and interventional procedures that are not feasible in rodents. However, this model presents significant challenges. The surgical procedure is invasive and requires specialized expertise, postoperative care, and ethical considerations. Altered hemodynamics and tissue hypoxia may also introduce systemic effects that confound RV-specific interpretation. The PA banding model in pigs is a robust and clinically relevant system for studying RV remodeling in patients with PAH. Although not a model of pulmonary vascular disease per se, this experimental model is indispensable for understanding RV pathophysiology and testing RV-targeted interventions, which is a critical frontier in PAH therapeutics.

Advanced and specialized animal models in PAH research

Classical rodent models, such as monocrotaline and SuHx, have significantly advanced our understanding of PAH pathobiology. However, these models do not fully capture the heterogeneity, genetic, and molecular complexity observed in patients. These limitations have prompted the development of more refined animal models that incorporate additional physiological, genetic, or environmental perturbations to better replicate human-relevant pathophysiological features, ranging from plexiform lesion formation and right ventricular maladaptation to sex- and genotype-specific disease features (Table 1). Several advanced models that leverage multi-hit paradigms, transgenic technologies, and metabolic perturbations have been developed to recapitulate specific aspects of PAH pathogenesis with greater fidelity.

Table 1.

Comparative summary of pulmonary hypertension animal models: vascular remodeling, RVSP, and RV dysfunction. This table summarizes commonly used animal models of pulmonary hypertension (PH), categorized by species and group classification. Each model is evaluated for its degree of pulmonary vascular remodeling, elevation in right ventricular systolic pressure (RVSP), and severity of right ventricular (RV) dysfunction using a semi-quantitative scale (1 = mild, 5 = severe). Key Pathobiological features and technical considerations are described, including model-specific mechanisms such as inflammation, hypoxia, mitochondrial dysfunction, and metabolic stress. Classical models (e.g., monocrotaline, chronic hypoxia, SuHx) and advanced two-hit or Transgenic systems (e.g., MCT + pneumonectomy, Sirt3/Ucp2 double knockout) are compared to highlight their utility in replicating distinct aspects of PAH pathogenesis and progression. This comparative framework supports informed model selection for mechanistic and translational studies

Model	Species	PH group	Vascular remodeling	RVSP	RV dysfunction	Key characteristics
Monocrotaline (60 mg/kg)	Rat	Group 1	3	4	4	Reproducible, inflammation-driven endothelial injury; induces medial hypertrophy and fibrosis; lacks plexiform lesions; risk of systemic toxicity, including myocarditis.
MCT + Pneumonectomy	Rat	Group 1	4	4	5	Two-hit model combining endothelial insult and mechanical stress; induces angio-obliterative remodeling and RV failure; technically challenging.
MCT + SuHx	Rat	Group 1	4	4	5	Combined inflammatory and hypoxic insult; exacerbated remodeling and RV decompensation; simulates aggressive disease with complex vascular pathology.
Sugen 5416 + Hypoxia (SuHx)	Rat	Group 1	3	4	4	Induces neointimal proliferation and plexiform-like lesions; sustained PH post-hypoxia; high translational fidelity; strain-dependent variability.
Chronic Hypoxia (10% O2, 2–4 weeks)	Mouse	Group 3	1	1	1	Mild, reversible medial thickening; simple, low-cost implementation; commonly used to test genetic susceptibility.
Chronic Hypoxia (10% O2, 2–4 weeks)	Rat	Group 3	2	3	1	Moderate vascular remodeling and RVSP elevation; reversible with reoxygenation; mimics high-altitude and COPD-related PH.
Fawn-Hooded Rat	Rat	Group 1/3	2	2	1	Mitochondrial dysfunction drives spontaneous PH under normoxia; exacerbated by hypoxia; models metabolic contribution to pulmonary vascular disease.
Mitomycin C	Rat	PVOD	2	1	2	Induces pulmonary veno-occlusive disease with occlusion of small veins and capillaries; remodeling is age-dependent and associated with endothelial-to-mesenchymal transition (EndMT).
Ucp2−/−	Mouse	-	1	1	1	Endothelial Ucp2 deficiency leads to increased mitophagy and apoptosis; mild PH phenotype with limited RV involvement; implicates mitochondrial quality control in vascular integrity.
Sirt3−/−	Mouse	-	2	2	1	Sirt3 deficiency promotes vascular remodeling and elevated RVSP; associated with LOXL2 upregulation; muscle-specific deletion suggests peripheral metabolic crosstalk in PH.
Ucp2−/−; Sirt3−/−	Mouse	-	4	3	2	Double knockout model with spontaneous severe PH and plexiform-like lesions; demonstrates mitochondrial dysfunction, insulin resistance, and robust vascular pathology.

Two-hit models to recapitulate advanced angio-obliterative PAH

The addition of a secondary pathogenic insult to conventional models significantly increases disease severity and histopathological relevance of the model. For example, combining MCT with left pneumonectomy intensifies hemodynamic stress and induces severe neointimal proliferation, medial thickening, and RV failure [23, 27, 32]. Left pneumonectomy augments hemodynamic stress by redistributing pulmonary blood flow and creating turbulent shear forces in the remaining lung, which synergize with MCT to drive angio-obliterative changes [23, 27]. Similarly, MCT paired with an aortocaval shunt results in high-flow pulmonary circulation, leading to marked RV dysfunction and mortality. These multi-hit models closely mimic late-stage human PAH and are particularly useful for evaluating anti-remodeling or RV-protective therapies. However, these models are technically demanding, often requiring surgical expertise and meticulous execution, and are associated with greater inter-animal variability, necessitating careful experimental design and validation of the results.

Among the classical models, MCT and SuHx offer complementary mechanistic insights. MCT induces acute inflammation-mediated endothelial injury through hepatic bioactivation, making it well-suited for studying immune-mediated remodeling and early vascular dysfunction. In contrast, the SuHx model combines VEGFR2 inhibition with hypoxic stress to drive chronic angio-proliferative remodeling and progressive RV failure, closely resembling advanced human PAH. While MCT is valuable for investigating early disease processes and short-term interventions, SuHx is preferable for studying irreversible vascular pathology and long-term anti-remodeling therapy. Ultimately, model selection should be aligned with the specific biological pathways and translational endpoints under investigation.

Although the MCT and SuHx models have provided foundational knowledge, they represent only a subset of the available tools. Ongoing progress highlights the need for more physiologically relevant models that better capture the complexity and heterogeneity of human PAH [71]. This includes systems that account for genetic predisposition, sex-based differences, metabolic dysfunction, and comorbidities. Moreover, the integration of high-resolution omics, advanced imaging, and refined phenotyping platforms has enhanced the capacity of animal models to align with human pathophysiology [71]. Building upon these classical models, a new generation of genetically engineered and disease-primed systems has emerged. These models were developed to investigate specific pathogenic mechanisms with greater granularity while offering new opportunities to evaluate precision-targeted therapies. The following examples highlight key specialized models that reflect the distinct aspects of PAH biology.

Female-specific plexiform lesions in Mts1 transgenic mice

The Mts1 (S100A4)-overexpressing transgenic mouse model demonstrates sex-specific susceptibility to severe vascular remodeling, with female mice uniquely developing plexiform-like lesions without having elevated RVSP [72]. This finding implicates estrogen signaling and cytoskeletal remodeling pathways in the development of severe pulmonary vascular lesions in patients with PH. It also highlights the modulatory role of sex hormones on vascular cell behavior and supports the hypothesis that estrogen metabolites may drive pathological remodeling in genetically susceptible individuals [72]. Collectively, this model emphasizes the importance of considering sex as a biological variable in the design and interpretation of preclinical studies on PAH.

BMPR2 mutation models with a second-hit requirement

Rodents bearing heterozygous BMPR2 mutations (e.g., Bmpr2^+/−, Bmpr2^+/R899X) do not always develop spontaneous PAH [73–75]. However, when combined with inflammatory triggers such as IL-6 overexpression, lipopolysaccharide, or 5-lipoxygenase activation, these animals exhibit pronounced vascular remodeling and elevated RVSP. West et al. developed a smooth muscle-specific, doxycycline-inducible BMPR2 mutant mouse (SM22-rtTA × TetO7-BMPR2^R899X) carrying a truncating R899X mutation [75]. After 9 weeks of transgene induction, starting at 4 weeks of age, all mice showed pulmonary vascular pruning, as demonstrated by fluorescent microangiography. Approximately one-third of the patients developed elevated RVSP, accompanied by distal vessel muscularization and large structural vascular lesions. In a separate study, rats harboring a monoallelic 71-bp deletion in exon 1 of Bmpr2 (Δ71 rats) exhibited reduced BMPRII expression and diminished SMAD1/5/9 phosphorylation [76]. These animals developed age-dependent spontaneous PAH with low penetrance (16–27%), mirroring the incomplete penetrance observed in human cases. Compared to wild-type controls, Δ71 rats demonstrated heightened susceptibility to hypoxia-induced pulmonary hypertension, progressive pulmonary vascular remodeling with a pro-proliferative phenotype, and a reduced pulmonary microvascular density. This reflects the low penetrance observed in human mutation carriers and emphasizes the importance of environmental or epigenetic “second hits” in disease manifestation. These models allow mechanistic dissection of gene-environment interactions and are particularly valuable for preclinical testing of BMPR2-targeted therapies in a genetically predisposed context.

Sirt3/Ucp2 double knockout mouse: a metabolically primed plexogenic model

A rare but pathologically rich model is the Sirt3/Ucp2 double knockout mouse, which recapitulates several hallmarks of severe human PAH, including plexiform-like lesions, metabolic syndrome features, and progressive RV failure [77]. These mice exhibit spontaneous vascular remodeling, mitochondrial dysfunction, and insulin resistance, reflecting a syndromic form of PAH driven by oxidative stress and impaired mitochondrial bioenergetics. Importantly, histological analysis revealed inflammatory plexiform lesions prior to the increase in RVSP, indicating that vascular remodeling precedes and potentially drives hemodynamic deterioration. This model offers a powerful platform for investigating metabolic-epigenetic coupling and evaluating mitochondria-targeted interventions. In 2024, Jheng et al. investigated the impact of skeletal muscle SIRT3 deficiency on pulmonary vascular health using skeletal muscle-specific Sirt3 knockout mice (Sirt3^skm–/–) [78]. These mice exhibited reduced pulmonary vascular density, proliferative vascular remodeling, and elevated pulmonary pressure. Mass spectrometry-based secretome analysis of SIRT3-deficient myocytes identified increased secretion of lysyl oxidase-like 2 (LOXL2). Elevated LOXL2 protein levels were confirmed in the plasma and skeletal muscle of Sirt3^skm–/– mice, a rat model of PH-HFpEF, and in patients with PH-HFpEF, implicating skeletal muscle-derived LOXL2 as a potential mediator of pulmonary vascular dysfunction in this disease.

Mitomycin-C and cyclophosphamide models of PVOD

To model pulmonary veno-occlusive disease (PVOD), which primarily affects the pulmonary venous and capillary compartments, researchers have employed mitomycin-C and cyclophosphamide [79–81]. These alkylating agents induce dose-dependent venular remodeling and modest increases in pulmonary artery pressure, with females exhibiting greater susceptibility. Mitomycin C, in particular, has been shown to induce capillary congestion and venular occlusion, capturing the key histopathological features of PVOD. These models are vital for studying post-capillary forms of pulmonary hypertension and developing therapies that differentiate between the arterial and venous remodeling processes.

Recent therapeutic advances and contributions of animal models

Refined preclinical models play an essential role in validating emerging therapies. For instance, activin-targeting fusion proteins, which have recently been shown to improve pulmonary vascular resistance and exercise capacity in phase III trials, were initially evaluated in the SuHx model [82, 83]. MCT and hypoxia models have similarly supported the development of mitochondria-targeted therapies, such as dichloroacetate, which reverses metabolic remodeling in patients with PAH. Nevertheless, many therapies that show efficacy in animal models fail once in clinical translation, highlighting the ongoing “valley of death” between the preclinical promise and patient outcome. A recent study by Kelly and Chan highlighted how advances in molecular biology, combined with improvements in patient stratification and diagnostic tools, are driving the emergence of precision medicine approaches in PAH [84]. Despite advances in preclinical modeling, translating therapies from animal models to humans with PAH remains challenging. Risk stratification systems, such as the COMPERA and REVEAL registries, have improved clinical management and highlighted discrepancies between experimental endpoints and patient-centered outcomes [85]. Novel biomarkers, including actigraphy-based step counts, are being evaluated as sensitive correlates of disease progression and treatment response. Future research should prioritize the development of models and endpoints that better reflect the multifactorial nature of human PAH and its clinical trajectory.

Advancing comorbidity-integrated experimental systems

The growing recognition of PAH as a systemic disorder, often presenting with metabolic and cardiovascular comorbidities, necessitates the development of more clinically representative experimental models. Contemporary studies indicate that up to 60% of patients with PAH present with coexisting conditions, such as obesity, diabetes, systemic hypertension, or heart failure, each of which alters disease progression, hemodynamics, treatment responses, and survival outcomes [86–88]. The presence of these comorbidities not only complicates diagnosis and patient stratification but also contributes to pulmonary vascular remodeling and right ventricular dysfunction [89, 90].

Recent advances in animal modeling have sought to bridge this translational gap by employing multi-hit strategies that incorporate comorbid metabolic conditions into classical PAH models. For example, combining high-fat diet-induced obesity with MCT or SuHx exposure in rodents recapitulates key clinical features, such as systemic inflammation, insulin resistance, and altered drug pharmacokinetics, offering a more realistic representation of the obese PAH patient phenotype [87, 91]. Similarly, diabetic PAH models using streptozotocin-induced hyperglycemia followed by hypoxic or pharmacological triggers have revealed metabolic vulnerabilities, including enhanced glycolytic dependency and mitochondrial dysfunction, which are not observed in non-diabetic contexts [88]. Furthermore, emerging clinical meta-analyses have shown that obesity may paradoxically confer a survival advantage in PAH, despite being associated with worse baseline functional status; this phenomenon is known as the “obesity paradox” [92, 93]. However, this paradox appears to influence treatment responsiveness and outcome heterogeneity, underscoring the need for preclinical models that can mechanistically dissect these interactions. Incorporating such comorbidity-driven heterogeneity into experimental design not only enhances translational fidelity but also provides critical platforms for evaluating personalized and stratified therapeutic strategies that reflect real-world patient populations.

Seven observational studies have collectively demonstrated that patients with PAH and cardiopulmonary comorbidities share a distinct clinical phenotype [89]. These individuals tend to be older and have a more balanced sex distribution than idiopathic PAH cohorts. Functionally, they present with diminished exercise tolerance, reflected by shorter six-minute walk distances, elevated N-terminal pro-brain natriuretic peptide (NT-proBNP) levels, and impaired pulmonary gas exchange, as indicated by reduced diffusion capacity for carbon monoxide. Hemodynamically, this subgroup exhibits elevated right atrial pressure and pulmonary artery wedge pressure, yet paradoxically displays lower mean PAP and pulmonary vascular resistance. Arterial oxygen saturation is also reduced in these patients, suggesting a compromised cardiopulmonary reserve. Overall, a meta-analysis of six studies revealed that the presence of comorbidities confers an 86% increased risk of mortality, with pulmonary comorbidities imparting the highest prognostic burden [89]. Therapeutically, patients with comorbidities were less frequently prescribed combination PAH-specific therapies. Importantly, this study suggests that comorbid conditions substantially modify the clinical trajectory and management of PAH [89]. These findings support the need for comprehensive phenotyping and tailored therapeutic strategies to address the elevated risk profile in this heterogeneous patient population [89].

In this context, identifying robust biomarkers and understanding the underlying molecular mechanisms are critical for refining patient stratification and therapeutic targeting in comorbid PAH populations. Galectin-3, for example, has emerged as an informative biomarker, which is associated with systemic inflammation and fibrosis, and has been extensively studied in conditions such as gestational diabetes and heart failure [94]. In PAH, measuring galectin-3 could potentially help identify early manifestations of cardiovascular or inflammatory comorbidities, thereby guiding earlier and more tailored therapeutic interventions. Additionally, molecular pathways such as DNA damage repair mechanisms, which are crucially implicated in megakaryopoiesis and endothelial cell function [95], represent another avenue for elucidating novel pathogenic processes involved in PAH-associated vascular remodeling and thrombosis, especially in the context of metabolic or cardiovascular comorbidities. Ultimately, developing comorbidity-integrated experimental models that leverage such biomarkers and mechanistic insights is essential for achieving precision medicine [96].

In parallel with animal modeling, engineered in vitro platforms have emerged as powerful tools for better replicating human pulmonary vascular physiology and pathology. Traditional animal models often fail to capture the full spectrum of human cellular interactions, mechanical forces, and microenvironmental signals. To address these limitations, recent efforts have focused on biomimetic technologies, including organ-on-chip systems, microfluidic devices, and 3D vascular constructs. These platforms allow for precise manipulation of cell types, flow conditions, and tissue architecture, supporting real-time functional readouts and human-specific drug testing [97]. When integrated with in vivo models, such systems have the potential to enhance translational accuracy and accelerate the discovery of targeted therapies for PAH. Nevertheless, animal models remain indispensable and have significantly advanced our understanding of the pathobiology of PAH. Their limitations, including species-specific differences in vascular responses, gene regulation, and right ventricular adaptation, continue to challenge direct translation (Table 2). Therefore, leveraging both animal and humanized in vitro systems in parallel may represent the most robust strategy for future translational research.

Table 2.

Comparative summary of classical, advanced, and human-based models used in PAH. This table outlines key features of commonly used preclinical and translational PAH models, including animal-based systems (e.g., monocrotaline, hypoxia, suhx, genetic, and large animal models) and emerging human-derived platforms (e.g., iPSC-derived cells, 3D organoids, organ-on-chip systems, and precision-cut lung slices). For each model type, we highlight advantages, limitations, relevance to human PAH pathophysiology, and therapeutic applications. The table aims to guide the selection of appropriate models for disease mechanism studies, therapeutic screening, and translational validation

Model type	Advantages	Limitations	Similarity to human pathophysiology	Therapeutic applications
Monocrotaline (MCT) Rat Model	Easy to induce; replicates vascular remodeling and inflammation	Lacks plexiform lesions; liver toxicity confounds interpretation	Moderate - inflammatory PAH features, endothelial dysfunction	Testing anti-inflammatory, endothelial-targeted drugs
Chronic Hypoxia (CH) Mouse/Rat Model	Reproducible and well-characterized; mimics hypoxia-driven vasoconstriction	Mild phenotype; does not induce occlusive vascular disease	Low to moderate - vasoconstriction without severe remodeling	Exploring hypoxia signaling and vasodilators
SU5416 + Hypoxia (SuHx) Rat Model	Best recapitulates plexiform-like lesions and severe angio-obliterative PH	Species-limited (works mainly in rats); technical and cost constraints	High - vascular obliteration, neointimal and plexiform lesions	Preclinical drug screening for angioproliferative PAH
Pneumonectomy + Vascular Injury	Models severe angioproliferative PAH with surgical precision	High variability; technically challenging	High - complex vascular remodeling and occlusion	Late-stage therapy evaluation and interventional studies
Genetic Models (e.g., BMPR2, S100A4)	Mimics heritable PAH; allows mechanistic genetic dissection	Low penetrance of disease; may require second-hit strategies	Moderate to high - mimics familial PAH pathobiology	Testing gene therapy and pathway-specific modulators
Large Animal Models (Pig, Dog, Sheep)	Closest anatomical and physiological match to humans	Costly, ethically complex, require infrastructure	High - RV function and hemodynamic parameters mimicked	Validating RV therapies and pressure-overload responses
iPSC-derived Pulmonary Cells	Patient-specific modeling; retains the genetic context of the disease	Immature phenotype; variability in differentiation	High - retains genetic background and disease signatures	Precision medicine and patient-specific drug testing
3D Organoids	3D architecture with multicellular complexity	Limited vascular structure; lacks full tissue organization	Moderate - partial disease architecture representation	Studying fibrosis, inflammation, and limited drug screens
Organ-on-Chip Systems	High physiological relevance under controlled microfluidic flow	Requires engineering expertise; low throughput	High - dynamic biomechanical cues and drug response	Modeling vascular response under flow and shear stress
Precision-Cut Lung Slices (PCLS)	Maintains lung tissue architecture; responsive to drugs ex vivo	Short culture lifespan; dependent on the availability of human donor tissue	High - near-native response to pharmacological agents	Ex vivo testing of PAH drugs and immunomodulators

Human-based in vitro and ex vivo platforms: patient-specific translational tools

As alternatives to traditional animal models, human-based experimental systems, such as precision-cut lung slices (PCLS), induced pluripotent stem cell (iPSC)-derived pulmonary vascular cells, organoids, and organ-on-chip platforms, have gained prominence for their ability to recapitulate human-specific cellular interactions and genetic backgrounds. PCLS allow for the study of acute vasoreactivity and drug responses in intact human lung tissues. In contrast, iPSC-derived endothelial and smooth muscle cells facilitate the modeling of patient-specific genetic variants, including BMPR2 mutations, as well as the testing of personalized therapeutic strategies. Organoid and organ-on-chip technologies offer advanced microphysiological environments that mimic the complex hemodynamics and multicellular architecture of the pulmonary vasculature, providing complementary insights into in vivo models and bridging critical translational research gaps.

Primary human pulmonary artery cells have long served as foundational in vitro models for investigating the cellular and molecular mechanisms of PAH. These cells, isolated directly from patient lungs or commercially available sources, retain disease-specific phenotypes, including hyperproliferation, reduced apoptosis, and abnormal inflammatory signaling. Notably, studies using hPAECs from patients with idiopathic PAH have demonstrated altered BMPR2 signaling, increased expression of endothelin-1, and dysregulated metabolic pathways consistent with the Warburg phenotype. While primary cells offer a direct insight into human diseases, their utility is often limited by donor variability, a short lifespan in culture, and loss of native vascular architecture. Nevertheless, they are widely employed for drug screening, mechanistic dissection of signaling pathways (e.g., TGF-β, PDGF, and Notch), and transcriptomic profiling, specifically in studies aimed at validating findings derived from animal models or high-throughput screens.

Induced pluripotent stem cell-derived vascular cells and lung organoids provide a patient-specific platform that overcomes many limitations of primary cells. iPSCs can be reprogrammed from the somatic cells of patients with PAH and differentiated into pulmonary endothelial cells, smooth muscle cells, or fibroblasts, allowing researchers to model genotype-phenotype relationships in vitro. A 2023 study demonstrated that BMPR2-mutant iPSC-PAECs exhibit impaired angiogenesis, which is reversible by BMP9 [98]. Patient-derived iPSCs harboring BMPR2 mutations recapitulate PAH phenotypes (e.g., glycolytic shift and apoptosis resistance) [98]. Furthermore, three-dimensional lung organoids derived from iPSCs enable the partial reconstruction of pulmonary microvascular networks and are being explored for modeling vascular remodeling and inflammation in the lungs. These systems are amenable to CRISPR/Cas9 gene editing for mechanistic investigations and drug testing, making them especially valuable for personalized medical approaches to PAH. However, the limitations of this approach include the use of immature cell phenotypes and the absence of hemodynamic stress, which can be addressed by integrating organ-on-chip technology.

Organ-on-chip systems have emerged as cutting-edge technologies that mimic the structural, mechanical, and fluidic microenvironments of the pulmonary vasculature with high fidelity [99]. These microfluidic devices incorporate human vascular cells into perfusable tissue-engineered channels, allowing them to recapitulate key biological pathomechanisms, such as shear stress, cyclic stretch, and barrier function. Recent applications in PAH research have included models of endothelial dysfunction and permeability, neutrophil transmigration, and pulmonary artery stiffness [100]. Microfluidic systems incorporating PAH patient ECs/SMCs under shear stress replicate the in vivo hemodynamics. Previous studies have reported that an organ-on-chip model of vascular endothelial and smooth muscle cell interactions can facilitate the study of dynamic changes in molecular and functional cell phenotypes under physiological flow conditions in response to key factors linked to the development of PAH, such as BMPR2 silencing and hypoxia. The model can accommodate blood-derived endothelial cells from patients, offering the prospect of tailoring medicine to individual PAH patients [100]. These platforms enable high-throughput drug screening but require validation using in vivo models to assess systemic effects.

Precision-cut lung slices represent a powerful ex vivo method that retains the native lung tissue architecture and cell diversity, allowing for real-time analysis of the pulmonary vascular tone, immune responses, and remodeling [101]. PCLS generated from human or animal lungs can be maintained in culture for several days, enabling short-term pharmacological testing and imaging [102]. Importantly, PCLS derived from explants from patients with PAH have been used to study acute vasoreactivity to agents such as nitric oxide donors and Rho-kinase inhibitors [103]. In a 2020 study, Alsafadi et al. applied PCLS to evaluate fibrotic responses in the pulmonary arteries and observed cell-type-specific transcriptomic changes following TGF-β stimulation [104, 105]. The integration of live imaging with immunofluorescence and RNA profiling further enhances the utility of PCLS as a translational bridge between the in vitro and in vivo systems. Although limited by the short culture duration and logistical challenges in acquiring fresh human lung tissue, PCLS remains a valuable platform for the rapid evaluation of therapeutic targets under near-physiological conditions.

Although in vitro and ex vivo human platforms, including primary cells, iPSC-derived organoids, organ-on-chip devices, and PCLS, offer unparalleled access to human-specific gene expression, signaling dynamics, and cellular responses relevant to PAH, they inherently lack the systemic complexity that is crucial for fully recapitulating disease progression. Critical features of PAH, such as the bidirectional communication between the pulmonary vasculature and right ventricle, systemic involvement of the immune system and hormonal regulation, and the chronic impact of hemodynamic stress on multiple organ systems, are challenging to model in isolation within these reductionist systems. Consequently, the true strength of these human-based models lies in their strategic integration with in vivo animal models, which provide temporal and organism-level insights into disease initiation and remodeling processes. Integrated approaches, such as validating omics-derived targets from rodent models in human iPSC-derived cells or using PCLS to assess acute drug responses observed in vivo, are essential for developing therapies with high translational potential. However, it is becoming increasingly clear that neither approach alone can fully capture the complexities of human diseases. Ultimately, a multi-platform strategy that synergistically combines the mechanistic precision of human-derived models with the physiological relevance of whole-animal systems represents the most promising path forward in PAH research and drug discovery, allowing for a more comprehensive understanding of this complex disease than what can currently be achieved.

Translational gaps and future directions

Despite significant advances in PAH modeling, translational efficacy remains limited. Therapeutic candidates demonstrating preclinical efficacy often fail in clinical trials, a phenomenon attributed to interspecies differences in vascular biology, the use of non-representative disease endpoints, and poor alignment with human pathophysiological complexity.

Future studies should prioritize the development of integrative multiscale platforms that combine mechanistic fidelity with clinical relevance. Hybrid model systems that combine traditional rodent frameworks with human iPSC-ECs, PCLS, and computational disease modeling can contextualize findings and validate conserved molecular signatures. For instance, transcriptomic signatures identified in SU5416/hypoxic rat lungs should be cross-validated in lung tissues from patients with PAH, enabling a robust cross-species validation. Large animal models, such as CRISPR-engineered pigs harboring PAH-associated mutations, provide a unique platform for studying pulmonary vascular remodeling and RV adaptation under human-like hemodynamic conditions. Therefore, increasing the genetic and sex-based diversity of preclinical models is essential. Addressing the unresolved “estrogen paradox” requires systematic comparisons of sex-specific metabolic responses in monocrotaline-exposed rodents, complemented by analyses of patient-derived iPSC-vascular cells to uncover cell-autonomous mechanisms. Models such as female Mts1 transgenic mice uniquely develop plexiform-like lesions, supporting the hypothesis that estrogen signaling influences vascular pathology in a sex-specific manner, which is a critical consideration for translational research. Translational advancement also depends on refining experimental endpoints. Conventional metrics, such as RV systolic pressure, lack predictive validity for human outcomes. Incorporating non-invasive methods of monitoring human rest/activity cycles in rodent models to approximate functional capacity, analogous to the six-minute walk test, can improve clinical congruence. Furthermore, echocardiographic indices, such as the TAPSE/PASP ratio, should be routinely integrated into the preclinical assessments to parallel right ventricular-pulmonary arterial coupling metrics used in patient care.

Computational and artificial intelligence platforms

The integration of computational modeling and artificial intelligence (AI) has emerged as a powerful frontier in PAH research, enabling an enhanced mechanistic understanding and accelerating predictive therapeutic discovery. Traditional mathematical models, incorporating hemodynamic principles, cellular signaling networks, and multiscale physiological interactions, have long contributed to elucidating complex disease dynamics that are beyond the reach of conventional experimental systems. Recent machine learning (ML) applications in PAH research have demonstrated improved diagnostic performance across a range of clinical, imaging, and biomarker data sets. These approaches outperform conventional methods in identifying at-risk patients, refining phenotypes, and supporting non-invasive screening, thereby enhancing early detection and informed clinical decision-making [106, 107]. AI models trained on chest radiography images have demonstrated exceptional diagnostic accuracy in detecting PAH, surpassing the performance of experienced clinicians and providing a non-invasive adjunct to current screening strategies [108]. Similarly, recent ML algorithms have shown great potential for identifying at-risk patients up to six months before diagnosis, enabling earlier referral and intervention [106]. Computational frameworks also extend to the development of explainable and transparent AI tools for characterizing the lung disease burden in PH using 3D Computed Tomography-based models, which improve disease phenotyping and prognostic accuracy [109, 110]. Notably, machine learning–guided decision tools are being developed to optimize right heart catheterization referrals, thereby bridging the diagnostic gap for patients with ambiguous clinical presentations [111]. Furthermore, meta-analyses have confirmed the high diagnostic performance of ML-enhanced tools applied to electrocardiography, echocardiography, and radiologic data, further supporting their integration into non-invasive diagnostic workflows [107]. As these computational platforms continue to mature by incorporating comorbid conditions, drug interactions, and genetic variability, they offer the promise of “virtual laboratories” that reduce dependence on animal models, enhance clinical trial design, and facilitate precision medicine strategies tailored to individual patients. Ultimately, a unified translational framework is needed that integrates in vivo and in vitro models with computational tools, accounts for sex and genetic heterogeneity, and aligns surrogate endpoints with patient-centric outcomes. This approach holds promise for accelerating the development of precision therapeutics tailored to individual pathobiological profiles.

Conclusion

Research on PAH continues to advance through animal modeling, molecular biology, and human-based systems, which have improved our understanding of the disease mechanisms and the speed of drug discovery. A comprehensive research strategy that integrates in vivo, ex vivo, and in vitro models with omics technologies and translational endpoints will facilitate the connection between laboratory research and clinical practice, ultimately enhancing outcomes for patients with PAH. The combination of animal systems for in vivo pathophysiology with human-based tools for molecular precision provides a method for developing therapies that match specific PAH subtypes.

Author contributions

ES and MB contributed to the writing and editing of the manuscript. All the co-authors approved the final version of the manuscript.

Funding

This study was supported by the National Institutes of Health/National Heart, Lung, and Blood Institute (NIH/NHLBI) under grants K01HL159038-01A1 and R25HL146166 (to MB), the American Heart Association Career Development Award (24CDA1269532 to MB), and the American Thoracic Society Research Program (Grant Nos. 23-24U1 to MB).

Data availability

Not applicable, as no dataset was generated or analyzed in this study.

Declarations

Ethics approval and consent to participate

Not applicable, as it did not involve human participants or animal subjects.

Consent for publication

This study was exempt from ethical review as it did not involve individual data requiring consent.

Competing interests

The authors declare no competing interests.