Researchers led by Yue Wang and colleagues at the China-Japan Friendship Hospital analyzed data from 706 patients with type 2 diabetes, integrating eleven behavioral and metabolic variables including smoking status, physical activity, body mass index, waist circumference, blood pressure, and lipid profiles. Using latent class analysis, a statistical method that identifies hidden subgroups within a population, the team uncovered three distinct behavioral-metabolic phenotypes.

The first phenotype, characterized by relatively favorable metabolic profiles and healthier behaviors, was labeled the low-risk class. The second, marked by central obesity, elevated blood pressure, and dyslipidemia, was named the metabolic-risk class. The third, distinguished by smoking and relatively preserved metabolic measures, was called the behavioral-risk class.

Crucially, these phenotypes carried different risks of complications. Over a median follow-up of 3.9 years, patients in the metabolic-risk class faced approximately twice the risk of diabetic kidney disease compared with the low-risk class, while those in the behavioral-risk class showed roughly 1.7 times the risk of peripheral neuropathy.

The findings suggest that the metabolic-risk class, with its constellation of central adiposity and hypertension, primarily drives renal hemodynamic abnormalities and fibrosis, whereas the behavioral-risk class, dominated by smoking, promotes oxidative stress, mitochondrial dysfunction, and Schwann cell injury that underlie nerve damage.

Notably, the study also revealed that socioeconomic status shaped phenotype membership. Patients with higher educational attainment were more likely to belong to the low-risk class, while those with lower education were overrepresented in the metabolic-risk and behavioral-risk classes.

The researchers emphasize that their approach moves beyond treating type 2 diabetes as a uniform condition and instead recognizes hidden behavioral-metabolic patterns that can guide precision prevention strategies.

However, the authors caution that the study was conducted at a single center in China, and the phenotypes may differ across populations. They also note that the observational design cannot establish causation, and that longer follow-up is needed to fully capture progressive microvascular complications.

Nevertheless, the findings provide a compelling framework for identifying patients who face the greatest risks of kidney and nerve damage, offering a path toward more targeted interventions for the millions affected by type 2 diabetes worldwide.

The distinction between the two higher-risk phenotypes identified in the study deserves closer attention, because it illustrates why single-factor risk assessment often falls short in clinical practice. The Obesity-Related Metabolic Risk phenotype, which accounted for nearly forty percent of the cohort, combined obesity, physical inactivity, insulin resistance, and suboptimal blood pressure control into a single coherent profile. This clustering matters physiologically: excess adipose tissue, particularly visceral depots, secretes pro-inflammatory adipokines and free fatty acids that worsen insulin resistance, while sustained hypertension raises glomerular capillary pressure and accelerates structural kidney damage. When these abnormalities coexist, their effects on the renal microvasculature are not merely additive but potentially synergistic, which helps explain why this phenotype carried roughly double the risk of incident diabetic kidney disease compared with the low-risk reference group.

The Behavioral-Visceral Adiposity phenotype presents a more counterintuitive picture. Although members of this group had lower overall body mass index, they exhibited higher visceral fat area and engaged in unhealthy lifestyle behaviors such as smoking, excessive alcohol consumption, or poor dietary patterns. This dissociation between general adiposity and visceral fat accumulation is well recognized in metabolic research, sometimes described as the phenomenon of normal-weight individuals with metabolically adverse fat distribution. Visceral adipose tissue is more metabolically active and more lipolytic than subcutaneous fat, releasing fatty acids directly into the portal circulation and promoting hepatic insulin resistance, systemic inflammation, and endothelial dysfunction. For peripheral nerves, which depend on endoneurial microvessels that are vulnerable to oxidative injury, this combination of toxic exposures and hidden visceral fat appears particularly damaging, and indeed this phenotype showed the strongest association with incident diabetic peripheral neuropathy in the study.

The use of latent class analysis rather than conventional regression on individual risk factors represents a methodological choice worth understanding. Traditional approaches typically estimate the independent effect of each variable while statistically adjusting for the others, which can obscure the reality that patients do not experience risk factors in isolation. Latent class analysis instead treats the observed pattern of behaviors and metabolic measurements as expressions of an underlying, unobserved typology, allowing the data themselves to reveal how risk factors naturally group together in real populations. This probabilistic classification also assigns each participant a membership probability across all classes rather than forcing a rigid categorization, which better reflects the gradations seen in clinical practice.

Several technical details of the study design strengthen its internal validity. The investigators established separate outcome-specific risk sets, excluding participants who already had diabetic kidney disease when analyzing kidney outcomes and those with prevalent neuropathy when analyzing nerve outcomes. This approach reduces the possibility of reverse causation, in which subclinical complications influence the behavioral and metabolic characteristics being measured. The use of a C-peptide-based index of insulin resistance, rather than the conventional calculation based on insulin itself, was a pragmatic adaptation for a cohort in which many participants received exogenous insulin therapy, since administered insulin would otherwise distort the measurement of endogenous insulin secretion.

The socioeconomic findings add an important dimension to the interpretation. Socioeconomic status was independently associated with phenotype membership, meaning that social circumstances partly determine which behavioral-metabolic pattern a patient develops. More striking was the effect modification observed in the stratified analyses: the elevated risks associated with the Obesity-Related Metabolic Risk phenotype were concentrated among patients with lower socioeconomic status, while the neuropathy risk linked to the Behavioral-Visceral Adiposity phenotype appeared mainly among those with higher socioeconomic status. This pattern suggests that material deprivation amplifies the metabolic pathway to kidney disease, perhaps through limited access to care, poorer medication adherence, or dietary constraints, whereas the behavioral phenotype may exert its neuropathic effects more strongly in higher-status patients whose metabolic risk is otherwise comparatively contained. Such heterogeneity implies that identical clinical profiles may warrant different intervention priorities depending on a patient’s social context.

From a mechanistic standpoint, the differential organ vulnerability observed across phenotypes is consistent with established pathophysiology. Diabetic kidney disease is driven largely by hemodynamic stress, glomerular hyperfiltration, and progressive fibrosis, processes that are tightly linked to blood pressure control and adiposity-related insulin resistance. Diabetic peripheral neuropathy, by contrast, involves axonal degeneration and Schwann cell injury mediated by oxidative stress, mitochondrial dysfunction, and microvascular rarefaction, processes to which smoking and other toxic behavioral exposures contribute directly. The study’s finding that the metabolic phenotype predicted both complications while the behavioral phenotype selectively predicted neuropathy aligns with this mechanistic divergence and offers a biological rationale for phenotype-specific screening strategies.

For clinicians, the practical implication is that risk stratification in type 2 diabetes could become more informative by considering how risk factors cluster rather than evaluating each in isolation. A patient with modest body mass index but elevated visceral fat and unhealthy habits may be overlooked by conventional screening focused on weight and glycemia, yet the study suggests such patients face substantial neuropathy risk. Conversely, patients with overt obesity and hypertension, particularly those facing socioeconomic disadvantage, warrant intensified surveillance for early kidney involvement, including regular assessment of urinary albumin excretion and estimated glomerular filtration rate.

It should also be acknowledged that the cohort was drawn from a single metabolic management center in Beijing, and the specific phenotype prevalences and hazard ratios may not generalize directly to other populations with different dietary customs, smoking prevalence, or healthcare access. The observational design further means that phenotype membership itself may shift over time, and longitudinal reclassification studies would help clarify whether patients transition between phenotypes and how such transitions modify complication risk. Nonetheless, the demonstration that behavioral and metabolic characteristics combine into recognizable, clinically meaningful patterns, and that socioeconomic context conditions their consequences, provides a foundation for the kind of multidimensional risk assessment that precision prevention in diabetes increasingly demands.

Subject of Research: Behavioral-metabolic phenotypes and latent class analysis in type 2 diabetes

Article Title: Behavioral-metabolic phenotypes and differential risks of diabetic kidney disease and peripheral neuropathy in type 2 diabetes: a latent class analysis

Article References: Wang, Y., Liu, Z., Zhang, Z., Hu, A., Liu, L., Zeng, S., Liu, P., Wang, Y., & Sun, L. (2026). Behavioral-metabolic phenotypes and differential risks of diabetic kidney disease and peripheral neuropathy in type 2 diabetes: a latent class analysis. Frontiers in Endocrinology, 17, Article 1962966. https://doi.org/10.3389/fendo.2026.1962966

Image Credits: AI Generated

DOI: 10.3389/fendo.2026.1962966

Keywords: type 2 diabetes, latent class analysis, diabetic kidney disease, diabetic neuropathy, behavioral-metabolic phenotypes, socioeconomic status, visceral adiposity, Behavioral-metabolic, phenotypes, differential, risks, diabetic