In the spectrum of microvascular complications associated with type 2 diabetes, concurrent chronic kidney disease (CKD) has emerged as a significant issue in contemporary nephrology. Studies show that CKD affects 20% to 40% of individuals with diabetes, making it the leading cause of end-stage renal disease (ESRD) globally (1–3). The impact of kidney disease extends beyond the decline in renal function, establishing a bidirectional relationship between kidney damage progression and increased cardiovascular risk. The progression of diabetic kidney disease severely affects patients’ quality of life. As renal function gradually declines, patients face an increased risk of end-stage renal disease (ESRD), and the associated social medical costs continue to rise. The pathogenesis of diabetic nephropathy is complex, and metabolic factors play a key role in its progression. Although the granular mechanisms linking metabolic dysregulation to chronic kidney disease remain incompletely elucidated, robust biological frameworks support this association (4). Therefore, the timely identification and therapeutic targeting of metabolism-related risk factors represents a cornerstone strategy. Early preventive measures can effectively lower chronic kidney disease (CKD) with T2DM incidence and optimize patient outcomes throughout the disease process.

Previous investigative studies have illuminated the intricate relationship between Metabolic Syndrome, the trajectory of chronic kidney disease (CKD) progression, and the manifestation of albuminuria and total proteinuria development (2, 4, 5). A comprehensive meta-analysis encompassing 24 studies and involving 6,573,911 participants revealed sophisticated revelations into this relationship. Specifically, 5 cohort studies assessed the risk of albuminuria or proteinuria in MetS patients, demonstrating a significant association (OR 1.43; 95% CI: 1.10-1.86). Stratified subgroup analyses further elucidated these findings, revealing albuminuria risk at OR 1.15 (95% CI: 1.11-1.18) and proteinuria risk at OR 1.76 (95% CI: 1.57-1.97). Additionally, 6 cohort studies investigated the risk of rapid kidney function decline among MetS patients, revealing a statistically significant correlation (OR 1.25; 95% CI: 1.07-1.47) (6).

Remarkably, disparate investigations exhibited substantial heterogeneity in risk assessment. This inherent variability in research outcomes potentially stems from multiple critical determinants: population demographic divergences, variance in follow-up durations, and region-specific and ethnic-specific contextual factors. Furthermore, the multifaceted definitional landscape of Metabolic Syndrome (MetS) may substantially influence research interpretations. Furthermore, the multifaceted definitional landscape of Metabolic Syndrome (MetS) can substantially influence research interpretations. The World Health Organization (WHO) first conceptualized insulin resistance syndrome in 1998 (7). Subsequent frameworks, such as the National Cholesterol Education Program Adult Treatment Panel III (NCEP-ATP III) (8). and the International Diabetes Federation (IDF), introduced evolving definitions, with the IDF in 2005 (9) emphasizing abdominal adiposity as a key component. By 2009, a collaborative consortium established a unified MetS standard (10). The MetS diagnostic framework encompasses critical metabolic parameters: blood pressure, glucose, metabolism, lipid profiles, and obesity, all these factors are linked to chronic kidney disease progression. Notably, while serum uric acid remains excluded from primary MetS diagnostic criteria, it is increasingly recognized as a significant metabolic biomarker. Although serum uric acid is not included in the classic definition of metabolic syndrome, numerous studies have shown that it plays an important role in metabolic syndrome. For example, in the genetic study by Yang (11), metabolic factors such as BMI, high-density lipoprotein (HDL), and systolic blood pressure (SBP) were found to have a significant causal relationship with serum urate levels and gout risk. Furthermore, there is a bidirectional causal effect between serum urate and metabolic traits, with hyperuricemia being closely associated with obesity, insulin resistance, and other metabolic issues. Therefore, in this study, we have included serum urate as a key metabolic indicator in the metabolic grouping of patients with diabetic nephropathy to more comprehensively assess its relationship with metabolic risk and kidney disease progression.

The distinctive contribution of our investigation lies in innovatively employing Latent Class Analysis (LCA). Unlike traditional regression techniques, LCA represents a sophisticated mixed modeling approach designed to identify the most optimal data representation through the premise of unobserved heterogeneous subgroups within the dataset. Currently no studies have used latent class analysis to identify a subgroup of T2DM associated CKD patients at high risk for progression of renal deterioration due to metabolic disorders. Therefore, by integrating metabolic markers blood pressure, glycemic profiles, lipid metabolism, body mass index, and serum uric acid, our study aimed to identify a potentially high-risk subgroup of metabolic disorders in patients with T2DM associated CKD patients and to assess their risk of renal function progression during short-term, intermediate, and long-term follow-up periods.

Metabolic disorders, including glucose, lipid, blood pressure, and uric acid abnormalities, contribute to CKD progression. Given that metabolic risk factors comprise multiple indicators rather than a single factor, our study innovatively employed Latent Class Analysis (LCA) to stratify diabetic nephropathy patients from our center into distinct risk subgroups based on metabolic risk factors. These two subgroups demonstrated significant differences in baseline characteristics, with Class 2 patients exhibiting more severe metabolic disorders compared to Class 1 patients, and subsequently showing significantly different risks for renal composite endpoint events during follow-up. Class 2 patients were characterized by higher BMI, systolic blood pressure, diastolic blood pressure, glycated hemoglobin, total cholesterol, triglycerides, and low-density lipoprotein. We also found that Class 2 patients had higher levels of Hb, PCR and ACR, also a higher proportion of female patients, but lower albumin (Alb) levels compared to Class 1 patients.

Additionally, patients in Class 2 with poorer metabolic profiles are younger on average, yet have higher urinary protein excretion. Existing studies have shown that patients with early-onset type 2 diabetes typically face greater challenges in metabolic regulation. One study found that the rate of β-cell function decline in adolescents with type 2 diabetes is much faster than in adults, with an annual loss rate of 25-30% (34), compared to only 7% (35) in adults. This accelerated β-cell function decline may be one of the key drivers of metabolic dysregulation in younger patients, and the loss of β-cell function is closely associated with the occurrence and progression of diabetes complications, including diabetic nephropathy (36). Furthermore, adolescents with type 2 diabetes not only face greater challenges in terms of β-cell function but also exhibit significantly lower insulin sensitivity compared to adults, indicating more severe insulin resistance. Studies have also shown that younger-onset patients often have unfavorable fat distribution characteristics, which further exacerbates insulin resistance and leads to elevated metabolic indicators such as blood glucose, lipids, and blood pressure. Such metabolic dysregulation causes younger patients to experience more severe metabolic issues in a shorter period, thereby increasing the risk of kidney damage (37). In addition to physiological factors, patients with type 2 diabetic nephropathy typically face poorer lifestyle and management controls, especially among younger patients who fail to adopt timely and effective treatment and interventions, exacerbating their metabolic dysregulation. Therefore, despite the younger age of the patients in the second group, their more severe metabolic dysregulation might lead to a higher risk of progression to diabetic nephropathy and renal function deterioration.

During the 3-year, 5-year, and 10-year follow-up periods, Class 2 patients demonstrated higher incidence rates of composite renal endpoint events compared to Class 1 patients. Specifically, after adjusting for confounding factors, the risk of renal composite endpoint events in Class 2 was 1.718-fold (HR: 1.718, 95% CI: 1.202- 2.455, P < 0.05), 1.662-fold (HR: 1.662, 95% CI: 1.178- 2.345, P < 0.05), and 1.826-fold (HR: 1.826, 95% CI: 1.305- 2.555, P < 0.05) higher than Class 1 at 3-year, 5-year, and 10-year follow-up, respectively. Given the significantly increased risk of renal composite endpoint events in Class 2 patients across all follow-up periods, we should pay more attention to patients with severe metabolic disorders and implement more stringent lifestyle modifications or intensified medical interventions to reduce their risk of renal composite endpoint events. Our study demonstrates that Latent Class Analysis based on metabolic disorder factors is an effective method for risk stratification of renal composite endpoint events in CKD patients with type 2 diabetes.

Latent Class Analysis (LCA) is a distinctive statistical method that explains response patterns across various indicators by identifying latent (unobserved) classes (38). Its most notable characteristic is its “person-centered” research perspective, which effectively categorizes individuals with similar characteristics into different subgroups and calculates the probability of each individual belonging to a specific subgroup. Unlike traditional statistical methods (such as regression analysis) that focus on relationships between variables, LCA emphasizes the heterogeneous characteristics within populations by identifying groups with similar behavioral or characteristic patterns, thereby revealing the underlying heterogeneity within populations (39). Technically, LCA employs Full Information Maximum Likelihood (FIML) to handle missing data, which offers advantages over traditional deletion or imputation methods. In medical research, LCA has been widely applied to disease subtype classification. For instance, Calfee et al. successfully used LCA to identify two biologically distinct subtypes of Acute Respiratory Distress Syndrome (ARDS) (40), those were subtype 2 (hyperinflammatory) and subtype 1 (hypoinflammatory). Their findings revealed significant differences in responses to PEEP (Positive End-Expiratory Pressure) strategies and fluid management approaches between subtypes. This enables early identification of high-risk patients to guide individualized treatment plans, improve prognostic accuracy, support precision medicine strategies in treatment, and potentially enhance disease outcomes through optimized treatment protocols. However, the application of LCA in predicting renal outcomes among CKD patients with T2DM remains relatively limited. Our study innovatively applies LCA to identify risk stratification for renal outcomes in T2DM associated CKD. Unlike traditional studies that focus on single metabolic risk factors, we integrated a series of metabolism-related risk factors to identify high-risk subgroups, including blood pressure, blood glucose, blood lipids, body mass index, and uric acid. Importantly, we established our model without considering outcome variables, an approach that enhances result stability and reproducibility. By examining follow-up results at 3, 5, and 10 years, our study provides evidence for short-term, medium-term, and long-term risk prediction. This analysis method, based on full sample information, maximizes statistical power and reduces bias.

The latent variables in our model were associated with kidney function deterioration, consistent with previous literature reports. Our research found that metabolic factors, including systolic blood pressure, diastolic blood pressure, total cholesterol, low-density lipoprotein, triglycerides, BMI, and glycated hemoglobin, were associated with kidney function decline. Among these, hypertension is a crucial risk factor for CKD progression, with poor blood pressure control accelerating kidney function decline. Studies have shown that elevations in both systolic and diastolic blood pressure are significantly associated with the rate of kidney function decline. A cohort study of 2,769 CKD patients found that every 10mmHg increase in systolic blood pressure was associated with a 17% increased risk of kidney function deterioration (41). A cohort study of CKD patients conducted from 2003 to 2015, including 2,588 patients from a region in East Asia, demonstrated a non-linear relationship (U-shaped curve) between systolic blood pressure and CKD progression risk. Persistent abnormalities carried greater risk than intermittent ones, and blood pressure variability was also identified as a significant factor. Subgroup analysis indicated that diabetic patients had a higher risk of kidney function progression (42). Furthermore, dyslipidemia is closely associated with CKD progression. Research has demonstrated that elevated levels of total cholesterol, low-density lipoprotein, and triglycerides are significantly associated with increased risk of kidney function decline. In a prospective study involving 11,847 participants, high triglyceride levels were associated with a 26% increased risk of CKD occurrence (43). A meta-analysis including 21,411 diabetic nephropathy patients found that TC and LDL-C levels were higher in DN patients who progressed to ESRD. Controlling lipid levels, particularly TC and LDL-C, may slow the progression of DN to ESRD (44). Glycated hemoglobin (HbA1c) levels are closely associated with the progression of diabetic nephropathy. Studies have shown that each 1% increase in HbA1c is associated with a 23% higher risk of kidney function decline, and good glycemic control can significantly slow the deterioration of kidney function (45). Similarly, the progression of BMI is also closely related to the progression of kidney function (46). The relationship between metabolic factors and kidney function progression involves complex mechanisms, including RAAS system activation due to insulin resistance, promotion of oxidative stress, induction of endothelial dysfunction, and increased expression of inflammatory factors. Additionally, hemodynamic changes, adipose tissue factor-mediated injury, oxidative stress and mitochondrial dysfunction, and systemic inflammation all play roles in this process (47–49).

Evidence suggests a close association between hyperuricemia and metabolic syndrome (50–53). Therefore, in this study, we used hyperuricemia as an important marker of metabolic status in patients with diabetic nephropathy, alongside blood glucose, lipids, blood pressure, and BMI, as criteria for metabolic grouping. However, in our study, after grouping based on metabolic-related indicators through Latent Class Analysis (LCA), we found no statistically significant difference in serum uric acid levels between the two groups. Possible reasons for this include: First, serum uric acid levels are significantly influenced by renal excretory function. In a CKD cohort where baseline eGFR is similar across latent classes, uric acid levels may be uniformly elevated, reducing its ability to distinguish between different groups. In addition, uric acid is affected by various confounding factors such as diuretic use, urate-lowering medications, dietary habits, and genetic background. These factors may be unevenly distributed or inadequately measured in our cohort, potentially masking the relationship between uric acid and other metabolic factors. In our study, LCA classified participants by identifying indicators with greater between-group differences (blood pressure, lipids, glucose, and BMI), so even if uric acid is biologically relevant, its contribution to the classification may have been limited. Lastly, the single baseline measurement of uric acid did not reflect its changes over time, limiting our ability to further explore its role in the progression of renal decline. Therefore, the lack of significant differences in serum uric acid between the groups in our study does not contradict prior findings but more likely reflects cohort characteristics, confounding factors, and measurement limitations. Future sensitivity analyses, such as stratification by CKD stage, exclusion of patients on urate-lowering medications or diuretics, or using eGFR-adjusted uric acid residuals, may help further clarify the relationship between uric acid and renal function progression.

Our study has several notable strengths: 1) This is the first study to employ Latent Class Analysis (LCA) based on baseline variables to identify high-risk subpopulations among patients with T2DM associated CKD. Unlike previous literature that analyzed individual metabolic indicators’ relationships with prognosis, our approach integrated multiple metabolic indicators to classify patients into distinct subgroups. This method better reflects patients’ overall metabolic status and provides more comprehensive risk assessment, potentially revealing patterns that traditional analytical methods might miss. Traditional risk stratification methods using single or few indicators may overlook patients’ overall characteristics and fail to identify complex relationships between metabolic abnormalities and CKD progression. By identifying different patient subtypes, this approach provides a foundation for personalized treatment and helps understand specific risk factors and prognostic differences among subgroups. 2) Based on model fitting indices and clinical judgment, we selected a two-class model, which represents the most parsimonious model while maintaining accuracy. 3) We utilized line graphs to illustrate differences in continuous variables between classes and bar charts to demonstrate differences in categorical variables, enhancing the visualization of our findings. 4) Our study promotes the application of Latent Class Analysis in risk prediction, demonstrating its utility in clinical research. 5) As a retrospective cohort study, selection and observation biases were minimized through appropriate study design and methodology. 6) Our extended follow-up period allowed us to evaluate short-term, medium-term, and long-term renal outcomes, maximizing our risk prediction capabilities and providing a comprehensive assessment of disease progression.

Despite the strengths of this study, several limitations should be acknowledged: First, our findings are derived from a single-center cohort, and as such, the generalizability of the results may be limited. The metabolic characteristics and progression trajectories of chronic kidney disease (CKD) may vary across different populations and regions. Therefore, our conclusions warrant external validation through multi-center, prospective studies to assess their applicability to broader populations. Second, although we adjusted for multiple metabolic-related covariates in our analysis, the retrospective design of the study may leave room for unmeasured confounders. These unmeasured factors could still influence the observed associations, introducing potential bias into our findings. Third, the metabolic indicators in our study were only measured at baseline, which limits our ability to capture their dynamic changes over time. This static approach may not reflect the natural fluctuations of these indicators, and the absence of key variables, such as waist circumference and fasting blood glucose, may have affected the accuracy of our risk stratification. Additionally, without tracking these indicators during follow-up, we were unable to assess their evolving role in CKD progression. Future studies incorporating dynamic measures or time-dependent variables could provide a more comprehensive understanding of metabolic changes and renal outcomes.

Using Latent Class Analysis, we identified two subpopulations of T2DM associated CKD with different risks of renal outcome events during 3-year, 5-year, and 10-year follow-up periods, based on metabolism-related factors including blood pressure, blood glucose, blood lipids, and body mass index. During the 3-year, 5-year, and 10-year follow-up periods, Class 2 subgroup demonstrated 1.718-fold, 1.662-fold, and 1.826-fold higher risks of composite renal endpoint events compared to Class 1 subgroup, respectively. By integrating multiple metabolic indicators to classify patients into distinct subgroups, our approach better reflects patients’ overall metabolic status and provides more comprehensive risk assessment, potentially revealing patterns that traditional analytical methods might miss. Our study, through identifying different patient subtypes, helps understand specific risk factors and prognostic differences among subgroups, thereby providing a foundation for personalized treatment approaches.