Proteomics is the large-scale study of proteins, the molecules that perform much of the work inside cells and tissues. While DNA provides a relatively stable set of instructions, proteins reflect what cells are actively doing in a particular environment. They can reveal inflammation, tissue repair, immune-cell activity, altered metabolism and structural damage. In atherosclerosis, this distinction is especially important because an arterial plaque is built from multiple cell types and extracellular materials. It may include immune cells, smooth-muscle cells, connective-tissue components, lipids, calcium deposits and regions of tissue breakdown. Each component can produce or modify proteins, creating a molecular record of the processes occurring within the lesion. By applying proteomic analysis to different areas of advanced carotid plaques, the study by Sinha and colleagues examines that record at a level of detail that conventional anatomical descriptions cannot provide.

The carotid arteries supply blood to the brain, and advanced disease in these vessels can become dangerous when a plaque narrows the artery or sheds material that travels downstream. Yet the size of a plaque alone does not fully describe its biological behavior. Two plaques with similar dimensions may differ in composition and in the processes taking place inside them. One may contain relatively organized tissue, while another may include active inflammatory signaling, weakened structural regions or extensive remodeling. The study’s emphasis on spatial heterogeneity addresses this problem directly. “Spatial” in this context means that the molecular profile depends on where a sample is taken within the plaque. A protein associated with immune activation may be abundant in one region but scarce in another, while proteins involved in tissue structure or repair may follow a different distribution. Such variation can disappear when an entire plaque is ground together and analyzed as a single sample.

This is one of the central challenges in tissue biology. An average measurement can be accurate for the sample as a whole while still concealing the most biologically important location. If a small, high-activity region represents only a fraction of a plaque, its molecular signals may be diluted by less active material. Spatially resolved approaches attempt to preserve the relationship between molecular measurements and tissue anatomy. In practice, this requires careful sampling or methods that connect protein signals to defined areas of a tissue section. The resulting data can be used to compare regions within the same plaque and to identify patterns that recur across plaques. The title of the study indicates that this strategy revealed both spatial and molecular heterogeneities, suggesting that advanced carotid lesions differ internally as well as from one patient to another. That distinction is crucial: a plaque may be heterogeneous because it contains multiple microenvironments, while different plaques may also arrive at advanced disease through different biological routes.

The molecular diversity of atherosclerotic plaques reflects the fact that atherosclerosis is not simply a storage problem caused by excess cholesterol. It is a chronic disease involving the artery wall, circulating lipoproteins, immune responses, cellular stress and tissue remodeling. Lipoproteins can enter the vessel wall, where they may be chemically modified and taken up by immune cells. These cells can accumulate lipid and alter the local inflammatory environment. Smooth-muscle cells, meanwhile, can change their behavior and contribute to structural tissue, but under some conditions they may also adopt different cellular states. The extracellular matrix—the network of proteins that gives tissue strength and organization—can be produced, rearranged or degraded. Proteomic measurements are capable of capturing many of these processes because they survey the molecules that mediate them. A complex protein pattern may therefore act as a molecular fingerprint of the local balance between injury, inflammation, repair and structural failure.

The study is also significant because carotid plaques can be examined in clinical contexts that are directly relevant to human disease. Advanced lesions removed during treatment provide an opportunity to investigate tissue that has reached a stage associated with substantial arterial pathology. Such samples can preserve evidence of the processes that shaped the plaque over time, although they represent a particular stage of disease rather than the entire history of its development. Proteomic analysis cannot, by itself, prove that a specific protein caused a plaque to progress or become dangerous. It can identify associations and molecular signatures that point to mechanisms requiring further testing. This distinction matters when translating discovery science into medical claims. A protein may be abundant because it drives disease, because it is produced in response to damage, or because it accumulates as a by-product of another process. Spatial information helps narrow those possibilities by showing whether the protein is concentrated in regions with corresponding structural or cellular features.

The work may ultimately help researchers move toward more precise descriptions of cardiovascular risk. Current clinical evaluation relies on a combination of symptoms, imaging, medical history and established risk factors. These tools are indispensable, but they do not measure every molecular process occurring within an individual plaque. A future diagnostic framework could potentially combine imaging with molecular markers that indicate particular plaque states. For example, researchers might seek signatures associated with active inflammation, matrix breakdown, calcification or organized repair. The present study does not establish such a clinical test, and its findings should not be interpreted as a ready-made predictor of stroke. Its contribution is more fundamental: it maps the complexity that any successful predictor will need to account for. If dangerous biology is localized rather than evenly distributed, sampling and measurement strategies must be designed accordingly.

The findings also raise questions about how therapies interact with plaque biology. Treatments that reduce circulating cholesterol, lower blood pressure or dampen inflammation can alter the environment in which plaques develop, but their effects may not be identical in every region of every lesion. A molecular atlas of plaque heterogeneity could help researchers determine whether particular cellular states respond differently to treatment or whether some regions remain biologically active despite improvement elsewhere. Proteomics may also support the discovery of candidate drug targets by identifying pathways shared by high-risk regions across multiple plaques. Any such application will require validation in larger patient groups and independent experimental systems. Researchers will need to establish whether the proteins detected in tissue can also be measured reliably in blood, whether they change before clinical events and whether they improve prediction beyond existing methods.

For now, the study’s most immediate message is conceptual but powerful: advanced atherosclerotic plaques should not be regarded as chemically or biologically uniform. Their internal geography matters. By linking protein composition to location, the research provides a framework for asking more precise questions about how plaques grow, remodel and potentially become harmful. It also illustrates why modern cardiovascular research increasingly combines pathology with high-throughput molecular measurement. A microscope can show the architecture of a lesion, while proteomics can reveal the active molecular programs embedded within that architecture. Together, these approaches may produce a more realistic account of disease than either can provide alone. The next challenge will be to determine which of the molecular patterns identified in advanced carotid plaques are causes, consequences or warning signals—and whether those patterns can be translated into earlier detection and safer, more individualized care.

The study employs the SHAP (SHapley Additive exPlanations) methodology to elucidate complex relationships between various biological markers and health outcomes. By unraveling these connections, the research emphasizes the potential of machine learning in transforming raw health data into actionable insights. Such methodologies are particularly advantageous in complex scenarios where established health indicators may fall short in delivering precise assessments of health risks, especially in populations that have traditionally been underserved in scientific research.

At the core of the study lies the TyG-ABSI measure, which integrates elements of biochemical markers—like triglycerides and waist circumference—to provide a holistic view of an individual’s metabolic condition. This composite measure provides deeper insights compared to traditional metrics, potentially enabling healthcare providers to identify at-risk individuals more effectively. The TyG-ABSI metric stands out because it transcends mere body mass index calculations, recognizing that obesity’s metabolic implications are far more nuanced than simple numeric representations suggest.

Participants in the research included individuals from low-income backgrounds, who often face greater barriers to healthcare access and preventive measures. By focusing on this demographic, the study shines a critical light on health disparities and the pressing need for tailored strategies that address the unique challenges faced by these communities. The results underscore how socioeconomic factors significantly influence health outcomes, particularly regarding obesity and its associated complications.

Moreover, the application of machine learning algorithms offers a myriad of advantages in predictive analytics. Through the deployment of these advanced modeling techniques, the research team is capable of assessing vast amounts of health data quickly and efficiently. This not only aids in identifying correlations that may not be overtly apparent through traditional statistical methods but also supports a more personalized approach to health management. The advent of explainable AI through SHAP further enhances this capacity, allowing for interpretations that can be understood and communicated more effectively to patients.

The researchers meticulously evaluated the predictive power of TyG-ABSI in comparison with other established indicators, such as BMI and waist-to-hip ratio measurements. The analysis yielded significant findings, with TyG-ABSI presenting a higher correlation with the incidence of carotid plaques. This compelling evidence positions TyG-ABSI as a potentially transformative tool in the realm of preventive cardiology, highlighting its ability to provide an early warning system for cardiovascular diseases.

Additionally, the research emphasizes the importance of machine learning’s explainability. With the growing reliance on AI in healthcare, ensuring that algorithms can be understood by clinicians is paramount. The SHAP analysis offers a unique advantage in this regard, enabling healthcare providers to gain insights into the decision-making process of machine learning models. This feature not only enhances trust in AI-assisted diagnoses but also fosters collaborative decision-making between patients and healthcare providers.

The impact of this research extends beyond academic intrigue; it holds profound implications for public health policy. As nations grapple with increasing rates of obesity and related diseases, integrating innovative measures like TyG-ABSI into standard health assessments could significantly bolster preventative strategies. Policymakers are urged to consider how such advanced methodologies can revolutionize health screenings, particularly in resource-limited settings where rapid and reliable health assessments are critically needed.

Furthermore, the study illuminates the intersection of technology and healthcare, advocating for sustained investments in research that harnesses the potential of digital health innovations. By embracing machine learning and data-driven approaches, healthcare systems can evolve to become more proactive rather than reactive, focusing on prevention rather than intervention. This paradigm shift could enhance overall health outcomes and reduce the burden of chronic diseases, particularly in vulnerable populations.

The research team calls upon fellow scientists and healthcare practitioners to replicate their findings and explore the applicability of TyG-ABSI within different demographic groups. Future studies could aim to refine the measurement further, expanding its utility across diverse populations and health conditions. The collaborative effort of the global scientific community is vital in validating and promoting interventions that address public health challenges effectively.

Moreover, leveraging community engagement in research initiatives can help bridge gaps in health literacy, empowering individuals with the knowledge and tools they need to manage their health proactively. It is essential to foster a culture of awareness around metabolic obesity and its risks, ensuring that communities recognize the importance of early detection and intervention.

In conclusion, Hao et al.’s research represents a significant leap forward in our understanding of metabolic obesity and its implications for cardiovascular health. The integration of the TyG-ABSI index, underpinned by machine learning methodologies, heralds a new era of health assessments that could empower low-income populations and mitigate health disparities. As the scientific community continues to explore these innovative approaches, the ultimate goal remains clear: to develop sustainable solutions that enhance health outcomes for all individuals, regardless of their socioeconomic background.

This study not only highlights the challenges faced by low-income populations but also showcases the transformative potential of technology in addressing these challenges. By embracing innovative health metrics and advanced analytical tools, we can move closer to a future where effective healthcare is accessible to everyone.

Subject of Research: Metabolic obesity indicator and its relationship with carotid plaque in low-income populations.

Article Title: TyG-ABSI as a novel metabolic obesity indicator for carotid plaque: an explainable machine learning study using SHAP in low-income population.

Article References:
Hao, J., Chen, R., Abudukeremu, D. et al. TyG-ABSI as a novel metabolic obesity indicator for carotid plaque: an explainable machine learning study using SHAP in low-income population.
BMC Endocr Disord 25, 281 (2025). https://doi.org/10.1186/s12902-025-02099-5

Image Credits: AI Generated

DOI: https://doi.org/10.1186/s12902-025-02099-5

Keywords: TyG-ABSI, metabolic obesity, carotid plaque, machine learning, SHAP, low-income population, health disparities, cardiovascular risk, preventive cardiology, public health.