Skip to main content
Download PDF

Circulating CCN6/WISP3 in type 2 diabetes mellitus patients and its correlation with insulin resistance and inflammation: statistical and machine learning analyses

BMC Medical Informatics and Decision Making volume 25, Article number: 114 (2025) Cite this article

Abstract

Introduction

Cellular Communication Network Factor 6 (CCN6) is an adipokine whose production undergoes significant alterations in metabolic disorders. Given the well-established link between obesity-induced adipokine dysfunction and the development of insulin resistance and type 2 diabetes mellitus (T2DM), this study investigates the potential role of CCN6 as a biomarker for T2DM. The present study aimed to investigate the association between serum CCN6 levels and T2DM, as well as its risk factors, for the first time.

Methods

In this case-control study, a total of 80 individuals diagnosed with T2DM and 80 healthy control individuals, who referred to Shariati hospital (Tehran, Iran), were included in the study. Biochemical parameters including fasting blood glucose (FBG), aspartate transaminase (AST), alanine transaminase (ALT), triglycerides (TG), total cholesterol (TC), high-density lipoprotein (HDL), and low-density lipoprotein (LDL) were determined using the AutoAnalyzer instrument. The circulating levels of CCN6, adiponectin, Tumor necrosis factor-α (TNF)-α, Interleukin 6 (IL-6), and insulin were quantified using ELISA. The Student t-test was applied to data that presented as mean ± standard deviations (SD). Moreover, the Gini Index was utilized to determine the weight of each factor in T2DM classification. Additionally, various machine learning models were employed to develop classifiers for predicting T2DM.

Results

T2DM patients demonstrated significantly lower levels of CCN6 (1259.76 ± 395.02 pg/ml) compared to controls (1979.17 ± 471.99 pg/ml, P < 0.001), as well as lower levels of adiponectin (P < 0.001) and higher levels of TNF-α and IL-6 (P < 0.001) compared to non-T2DM individuals. In the T2DM group, CCN6 exhibited negative correlations with insulin, Homeostatic Model Assessment for Insulin Resistance (HOMA-IR), body mass index (BMI), IL-6, and TNF-α. Logistic regression analysis indicated an increased risk of T2DM, with a CCN6 cutoff value of 1527.95 pg/mL distinguishing T2DM patients with 86.3% sensitivity and 73.8% specificity. The Gini Index highlighted that HOMA-IR, IL6, and CCN6 had the highest weighting on T2DM.

Conclusion

Our research identified a significant and negative association between serum CCN6 levels and the likelihood of T2DM, as well as inflammation biomarkers (IL-6 and TNF-α). CCN6 shows promise as a potential biomarker for T2DM; however, further investigations are necessary to validate this finding and assess its clinical utility.

Peer Review reports

Introduction

Diabetes Mellitus is a medical condition characterized by challenges in the body’s efficient storage and utilization of glucose [1]. The most prevalent form of this condition is type 2 diabetes mellitus (T2DM), characterized by persistent elevation of blood glucose and insulin levels. This elevation is primarily triggered by insulin resistance and impaired insulin function in vital tissues such as the liver, adipose tissue, and skeletal muscle [1, 2]. In our current era, obesity represents a substantial threat to human well-being [1] this has resulted in a surge of obesity-related diseases, including insulin resistance and T2DM. Diet stands out as one of the significant factors influencing the onset and progression of these conditions [1, 3].

Obesity carries significant health implications, triggering inflammation and insulin resistance, ultimately culminating in diabetes. Furthermore, insulin resistance can be associated with various factors such as oxidative stress, endoplasmic reticulum stress, hypoxia, aging, and lipodystrophy, all contributing to the development of T2DM [4, 5]. In cases of obesity, M1 macrophages in adipose tissue release numerous pro-inflammatory cytokines, disrupting insulin signaling and promoting insulin resistance [1]. Moreover, other immune cells in adipose tissue influence inflammation and insulin resistance, including mast cells, dendritic cells, neutrophils, and B and T lymphocytes. Beyond fat storage, adipose tissue also plays a crucial role in releasing adipokines, a group of secretory cytokines [6]. These adipokines play a crucial role in the pathogenesis, prognosis, diagnosis, and treatment of metabolic disorders triggered by obesity [3, 7, 8]. Due to the association of obesity with T2DM, the expression of a group of adipokines is increased and reduced in the induction of insulin resistance associated with obesity and T2DM [9].

Adipokines are crucial in regulating carbohydrate and lipid metabolism, maintaining body energy balance, promoting homeostasis, and modulating inflammation. Notably, specific adipokines and cytokines, such as vaspin, resistin, apelin, ghrelin, omentin, TNF-α, and interleukin-6 (IL-6), are implicated in the development of insulin resistance associated with obesity and T2DM [9]. Another significant adipocytokine in this context is cellular communication network factor 6 (CCN6/WISP3), which belongs to the CCN matrix protein family and is synthesized and secreted from adipose tissue [10].

WISP3 is one of the six members of the CCN protein family, and it holds significant importance in the mitochondrial electron transport system and tumor suppression, particularly in certain metastatic cancers [10, 11]. Significantly, earlier research has shown that CCN6 acts as a tumor suppressor and inhibits the action of insulin-like growth factor 1 (IGF-1) when breast cancer advances. Epithelial cells could become invasive and migrate if the function of this adipokine is reduced because it may alter the IGF-1 growth factor receptor [12, 13]. Additionally, a 2022 study found that in NASH (Non-Alcoholic Steatohepatitis) mice, elevated CCN6 expression decreased fibrosis, hepatic steatosis, and the inflammatory response [14]. However, there is currently little knowledge of the function of CCN6 in T2DM patients. Further investigation is required to comprehend its potential significance in the context of T2DM patients.

Furthermore, machine learning techniques have demonstrated promising outcomes in identifying biomarkers for various diseases, including coronary artery disease, breast cancer [15], and Alzheimer’s disease [16]. In metabolic diseases such as T2DM, machine learning has emerged as a powerful method for complex data analysis and pattern recognition regarding disease progression, insulin resistance, and obesity-related complications. Machine learning algorithms, for example, have been applied to predict T2DM risk from clinical, biochemical, and genetic variables with the prospect of personalized medicine approaches [17,18,19]. These techniques permit the integration of diverse variables, such as adipokine levels, inflammatory markers, and anthropometric parameters, to improve diagnostic sensitivity and identify novel biomarkers. Machine learning techniques are employed in the current research to assess the utility of CCN6 as a biomarker in diagnosing patients with T2DM.

Through the application of sophisticated algorithms, we aim to contrast the diagnostic potential of CCN6 and its relationship with other conditions such as inflammation, insulin resistance, and obesity. This type of analysis not only gives greater insight into CCN6 in T2DM but also offers an understanding of how machine learning assists in the speeding up of biomarker discovery for metabolic disorders. Considering the association between CCN6/WISP3 serum levels and obesity, inflammation, insulin resistance, and other physiological processes that could influence T2DM development, there is potential for CCN6 to be used as a biomarker for the early diagnosis and treatment of T2DM. As far as we know, the relationship between CCN6 serum levels and type 2 diabetes has not been addressed in any previous research.

Early detection of T2DM is crucial in order to intervene early and prevent complications such as cardiovascular disease, neuropathy, and nephropathy. Current tests for diagnosis, including HbA1c and fasting plasma glucose, typically are not successful at identifying those individuals who are pre-diabetic or at high risk of T2DM. New biomarkers, including CCN6, can enhance early diagnosis and enable tailored therapeutic strategies to ultimately benefit patient outcomes.

Hence, the current study will be the first to investigate the serum levels of CCN6 in T2DM patients and explore its potential correlation with TNF-α, IL-6, and other biochemical parameters.

Methods

Participants

The current study comprised 80 patients in the T2DM group (52 males and 28 females) and 80 individuals in the control group (53 males and 27 females), aged between 45 and 75 years, who referred to Shariati hospital (Tehran, Iran) [20]. The T2DM group was diagnosed according to the American Diabetes Association (ADA) compliant clinical tests [21]. The medical histories of all individuals were checked for the absence of renal diseases, autoimmune disorders, chronic inflammations, cancer, steroid medication, a family history of thyrotoxicosis, immune system suppression, and anti-inflammatory drugs in the past six months, as well as individuals who had smoked or used tobacco products during the previous three months, were not allowed to participate in the study. Comprehensive medical records of both participant groups were recorded. The current study was conducted with adherence to ethical principles and received approval (IR.TUMS.SPH.REC.1401.280) from the Ethics Committee of the University of Medical Sciences, with written consent obtained from all participants. Ethical guidelines in accordance with the Helsinki Declaration were followed.

Assessment of anthropometric parameters

The Body Mass Index (BMI) of the participants was calculated using the standard formula of weight (kg)/height (m2). Furthermore, a standard aneroid sphygmomanometer was used to monitor their resting systolic blood pressure (SBP) and diastolic blood pressure (DBP).

Assessment of biochemical parameters

A volume of 7 ml of fasting venous blood was collected from each participant in both groups. Subsequently, the serum levels of fasting blood glucose (FBG), aspartate transaminase (AST), alanine transaminase (ALT), triglycerides (TG), total cholesterol (TC), high-density lipoprotein (HDL), and low-density lipoprotein (LDL) were examined using an autoanalyzer and commercially available test kits. The serum insulin concentration in both groups was assessed using the enzyme-linked immunosorbent assay (ELISA) method and the commercial Monobind kit. Additionally, the Homeostatic Model Assessment for Insulin Resistance (HOMA-IR ) index in this study was determined using the standard formula: [FBG (mg/dl)] × [fasting blood insulin (µU/mL)/405].

Assessment of cytokines and adipokines

The serum concentrations of CCN6, TNF-α, IL-6, and adiponectin were measured using the ELISA method. The CCN6 kit (KEH04333, Aviva system biology; USA) had intra-assay and inter-assay coefficients of variation (CV) of 8% and 10%, respectively, and a minimum detectable concentration of less than 50 pg/mL. The levels of adiponectin were determined using the Adipogen kit (South Korea) with intra-assay and inter-assay CV values of 3.4% and 4.3%, respectively, and a minimum detectable range of 0.1 nanograms per milliliter.

The R&D Systems kit, which has a minimum detectable range of 5.5 pg/mL for TNF-α and 0.11 pg/mL for IL-6, was used to measure the serum levels of these two chemicals. The intra-assay and inter-assay CV values for TNF-α were 7.4% and 5.2%, respectively, while for IL-6, they were 9.6% and 6.9%, respectively.

Statistical analysis

All statistical analyses were performed using SPSS version 27, and a significance level of p < 0.05 was considered statistically significant. Categorized data were tested using the chi-square test and presented in terms of frequency and percentage. The evaluation of data normality was carried out by employing the Kolmogorov-Smirnov test. Following this assessment, the Student t-test was applied to data that demonstrated a normal distribution, and the resulting results were presented as mean values alongside their corresponding standard deviations (SD). In cases where the data did not exhibit a normal distribution, the U Mann-Whitney test was employed, and the results were reported as median values along with their respective interquartile ranges (IQR). Pearson correlation analysis was utilized to evaluate the relationship between CCN6 and continuous variables, while binary logistic regression was employed to investigate the connection between CCN6 and the probability of developing T2DM. To assess the diagnostic performance of the CCN6 ELISA test in distinguishing between the patient and control groups, the ROC curve was utilized. Moreover, the cutoff value, sensitivity, and specificity of the CCN6 ELISA test were determined based on the ROC curve.

Machine learning modeling and evaluation

RapidMiner 10.1 was used for the modeling and assessment procedure [22]. To determine the significance of each factor about T2DM, we employed the Gini Index [23]. Following that, we utilized five different machine learning algorithms: Naïve Bayes [24], Decision Tree [25], Gradient Boosted Trees [26], K-Nearest Neighbor (KNN) [27], Random Forest [26], and, to develop predictive models for T2DM. The performance of these models was assessed using various metrics such as accuracy, area under the curve (AUC), f-measure, kappa, sensitivity, and specificity [28]. To ensure the reliability of our results, we implemented a 10-fold cross-validation technique [29].

Results

Anthropometric parameters

The data presented in Table 1 reveals that there were no statistically noteworthy differences observed in BMI (P = 0.77), SBP (P = 0.75), or DBP (P = 0.62) when comparing the control group with the T2DM group. However, it’s worth noting that the mean age of the T2DM group shows a marginal elevation compared to the control group (P = 0.13).

Table 1 Basic anthropometric, biochemical and immunological characteristics of 80 control people and 80 T2DM patients. The student t-test was applied to data that presented as mean ± standard deviations (SD)
Full size table

Biochemical parameters

In contrast to the control group, the T2DM group showed considerably higher levels of FBG, HOMA-IR, and insulin (P < 0.001), as shown in Table 1. Additionally, TG, TC, and LDL-C values were higher in the T2DM group than in the control group (P < 0.001, P = 0.002, and P = 0.002, respectively). On the other hand, it was discovered that the patient group’s serum HDL-C concentration was lower than the control group’s (P = 0.005).

Cytokines and adipokines

There was a considerable reduction in the average serum levels of CCN6 among individuals with T2DM, declining from 1979.17 ± 471.99 pg/ml to 1259.76 ± 395.02 pg/ml compared to the control group (P < 0.001). As well as, the serum concentration of adiponectin decreased from 13.56 ± 3.52 µg/ml to 9.12 ± 2.74 µg/ml (P < 0.001) (Fig. 1 (a, b) and Table 1). Moreover, the serum levels of inflammatory cytokines IL-6 and TNF-α were notably elevated in the T2DM group compared to the control group (P < 0.001 for both). These findings are graphically represented in Fig. 1 (c, d) and detailed in Table 1.

Fig. 1
figure 1

a, b Serum levels of CCN6 and adiponectin were lower in the T2DM group than the control group, respectively. c, d Serum levels of TNF-α and IL-6 were higher in T2DM group than the control group, respectively. Note: *<0.05, **<0.01.

Full size image

Correlation of CCN6 serum level with clinical parameters

Within both the non-T2DM and T2DM groups, a negative notable correlation was identified between serum CCN6 levels and BMI [(r = -0.298, P = 0.007) and (r = -0.251, P = 0.025), respectively], as showed in Tables 2 and 3.

Table 2 The correlation analysis of various variables with CCN6 in the control group. Pearson correlation analysis was utilized to evaluate the relationship between CCN6 and continuous variables
Full size table
Table 3 The correlation analysis of various variables with CCN6 in the T2DM group. Pearson correlation analysis was utilized to evaluate the relationship between CCN6 and continuous variables
Full size table

As demonstrated in Tables 2 and 3, a clear negative and inverse relationship was evident between CCN6 levels and insulin within both the non-T2DM and T2DM groups, as indicated by the correlation coefficients [(r = -0.231, P = 0.040) and (r = -0.334, P = 0.002), respectively]. Additionally, in control and patient groups a similar negative and significant correlation was observed between CCN6 and HOMA-IR [(r = -0.240, P = 0.032) and (r = -0.349, P = 0.002), respectively].

In the T2DM group, a negative and robust correlation was identified between CCN6 and TNF-α (r = -0.257, P = 0.022) and IL-6 (r = -0.403, P < 0.001), as detailed in Table 3. These findings highlight the potential role of CCN6 in the onset of type 2 diabetes by suggesting that the relationships between CCN6 and several variables varied between the control and T2DM groups.

Association of CCN6 serum level with the risk of T2DM

The effect of a 100-unit change in serum CCN6 concentration on the risk of acquiring type 2 diabetes was assessed using a binary logistic regression analysis. The findings consistently indicated a significant relationship between both the unadjusted model (OR [95% CI] = 0.67 [0.60–0.76]) and the adjusted models, which accounted for factors like age, gender, and BMI (OR [95% CI] = 0.62 [0.54–0.72]) as demonstrated in Table 4.

Table 4 Binary logistic regression for odd ratio of T2DM status according to 100-unit change in CCN6
Full size table

Figure 2 displays the ROC curve analysis, revealing a CCN6 threshold value of 1527.95 pg/mL for discriminating T2DM patients from the control group. The sensitivity (86.3%) and specificity (73.8%) of the study were excellent. At a 95% confidence level, the computed area under the curve was 0.88 (0.83–0.93, p < 0.001). These results suggest that CCN6 shows potential as a prospective diagnostic indicator for T2DM, demonstrating favorable precision in distinguishing between T2DM patients and individuals without the condition.

Fig. 2
figure 2

ROC curve for diagnosis of T2DM according to CCN6 serum levels. A cut-off value of CCN6 (1527.95 pg/mL) was identified to distinguish between T2DM patients and the control group, demonstrating good sensitivity and specificity (86.3% and 73.8%, respectively). The area under the curve was calculated as 0.88 [0.83–0.93 (95% Cl), p < 0.001]

Full size image

Machine learning modeling and evaluation

Figure 3 illustrates the relative weight of factors associated with T2DM. Among these factors, HOMAIR, IL6, and CCN6 exhibit the highest weighting on T2DM. Also, Fig. 4 shows the Decision Tree Model.

Fig. 3
figure 3

Factors weight on T2DM calculated by gini index

Full size image
Fig. 4
figure 4

Decision tree model for classifying T2DM cases

Full size image

The performance comparison of the models is depicted in Table 5. The Random Forest model demonstrates the highest performance across multiple metrics, making it the best model for T2DM classification in this comparison. The Random Forest model has the highest overall correctness in predicting cases of type 2 diabetes, with an accuracy of 92.50% +/- 3.95%. It also demonstrates a high degree of agreement between its predicted classifications and the actual T2DM labels, with a kappa value of 0.850 +/- 0.079 when accounting for the likelihood of agreement by chance. This implies that it successfully differentiates between positive (T2DM) and negative cases and that its predictions are highly reliable. The Random Forest model also demonstrates excellent performance in correctly identifying positive instances (T2DM) while minimizing false positives and false negatives. Its F-Measure of 92.39% +/- 4.27% shows that precision and recall are well balanced. The Random Forest model exhibits remarkable performance in terms of sensitivity and specificity.It exhibits a sensitivity of 92.50% +/- 8.74%, reflecting its ability to accurately identify T2DM cases. Simultaneously, it achieves a specificity of 92.50% +/- 6.45%, indicating its capability to correctly identify non-T2DM cases.

Table 5 The T2DM classification models performance comparison
Full size table

The Random Forest model performs well on several criteria, such as sensitivity, specificity, accuracy, kappa, AUC, and F-Measure. Its strong performance places it at the top of this comparison of T2DM classifications, indicating that it can be used to forecast T2DM instances with high accuracy and dependability.

Discussion

It has been established that adipokines, like hepatokines and myokines [30,31,32], play critical roles in the regulation of metabolic parameters [3, 7]. Accordingly, the objective of our study is to investigate, for the first time, the association between serum levels of CCN6 and T2DM disease. Previous studies have explored the relationship of CCN6 with breast cancer, skeletal disorders like rheumatoid arthritis and pulmonary fibrosis [33,34,35]. Additionally, the association of other members of this family, such as CCN1, CCN2, and CCN3, has been studied in conditions like obesity, insulin resistance, and diabetic wound healing [36,37,38,39,40]. Nevertheless, the relationship between serum CCN6 levels and type 2 diabetes has not yet been investigated. Consequently, our research offers a fresh look at serum CCN6 levels in T2DM patients and explores how they relate to TNF-α, IL-6, and other biochemical markers.

Our study revealed that, in contrast with the control group, those with type 2 diabetes had significantly lower serum levels of CCN6. This discovery suggests a potential link between lower CCN6 levels and the onset of T2DM. Notably, the established cutoff value has proven effective, with sensitivity and specificity exceeding 70%, in accurately distinguishing T2DM patients from the control group. These results are consistent with previous studies that have emphasized the role of CCN6 in metabolic disorders.

For instance, research findings have indicated that WISP1/CCN4 may play a role in the development of diabetes and obesity. WISP1 has thus been established as a viable clinical marker for the assessment of metabolic diseases and diabetes as a result of these findings [41]. As well as, in an animal-based investigation, it was demonstrated that the expression levels of CCN1 and CCN2 exhibited a substantial increase within the retinal tissues of diabetic mice as compared to their non-diabetic counterparts [42].

In addition, in a study conducted in 2023 by Song and colleagues, they demonstrated that serum levels of CCN6 have decreased in patients and mice models afflicted with NASH. They also found that an increase in this adipokine leads to reduced inflammation, steatosis, and hepatic fibrosis in NASH-affected mice [14].

Consistent with our study, in 2019, Yang Li and fellow researchers demonstrated that individuals with T2DM exhibited elevated levels of CCN3 in their serum when contrasted with the control group. Furthermore, they established a positive association between the serum levels of this adipokine and inflammatory markers, including CRP, TNF-α, and IL-6 [38].

In our study, we observed a significant negative correlation between CCN6 and inflammatory factors such as IL-6, TNF-α. Also, we noticed a negative and noteworthy correlation between CCN6 and indicators of insulin resistance, specifically insulin levels and HOMA-IR. This discovery implies that CCN6 may have the potential to serve as a therapeutic target for enhancing insulin sensitivity in individuals diagnosed with T2DM. Previous researches also demonstrated a correlation between serum levels of CCN6 and insulin resistance. They indicated that increased CCN6 expression in NAFLD-afflicted mice inhibits the Ask1-p38MAPK/JNK signaling pathway in adipose tissue, leading to reduced inflammation and insulin resistance [14]. Moreover, in animal hepatic cells and human muscle cells, CCN4 acts as an inhibitor of phosphorylation of Akt and its glycogen synthase kinase 3β substrates, insulin receptor, FOXO1, p70S6 kinase, suppression of gluconeogenic gene expression, and inhibition of insulin-stimulated glycogen synthesis. This has led to disruptions in insulin functionality [43, 44].

As evidenced, in 2019, Klimontov and colleagues identified CCN4 as a potential biomarker for obesity in individuals with Type 2 Diabetes. Their research revealed a connection between CCN4 serum levels, central abdominal fat mass, and dysfunction of adipose tissue [43]. In line with these findings, our study also demonstrated a significant negative correlation between CCN6 serum levels and BMI within both the patient and control groups.

Furthermore, inhibition of WISP3 expression was shown to enhance the effects of the IGF-1 factor and neoplastic progression in breast epithelial cells [45]. In a separate study in 2018, the impact of CCN6 inhibition on apoptosis of pulmonary cells under hypoxic conditions was investigated, and their results indicated that CCN6 can inhibit the extrinsic apoptosis pathway by suppressing caspase 8, thereby safeguarding pulmonary cells [46].

It’s worth highlighting that our research did not uncover any substantial or statistically significant link between CCN6 serum levels and the serum lipid profile or FBG in individuals diagnosed with T2DM. However, a different research investigation documented a notable and positive correlation between CCN3 levels and TG as well as FBG [38].

On the other hand, the Gini Index indicated that the levels of CCN6 in the serum play a significant role in identifying patients with T2DM. The random forest model demonstrated the highest performance in this study, which is consistent with a previous study that also showed the Random Forest model outperforms the Decision Tree model for T2DM prediction. In a study, a random forest model was proposed with an accuracy of 71.1%, while our study reported a model with 92.50% accuracy. This improvement in accuracy indicates that the current study successfully selected suitable features for the T2DM prediction.

In another recent study [47] the random forest model achieved an accuracy of 94.4% and outperformed both the decision tree and multiple logistic regression models for T2DM classification. The improved performance in this study may be attributed to the implementation of the SMOTE (Synthetic Minority Over-sampling Technique) technique, which aims to balance the number of records in the minor class with those in the majority class [48]. By applying the SMOTE technique and conducting tests on the synthetic records generated by SMOTE, the models achieved higher performance compared to their real performance on the original dataset.

The random forest model could serve as a viable approach for predicting the risk of T2DM by considering various factors such as, IL6, CCN6, Adiponectin, TNF, HDL, TG, LDL, TC, BMI, SBP, and DBP. Moreover, this model might be employed in decision support systems to assist clinical professionals.

Conclusion

Our results reinforce the idea that CCN6 serum concentrations was decreased in individuals with T2DM, and also, exhibited correlations with TNF-α, IL-6, BMI, insulin and HOMA-IR measurements. This implies that CCN6 may serve as a prospective biomarker for anticipating the onset of T2DM. Nonetheless, additional research is imperative to establish definitive conclusions.

Data availability

The datasets utilized and/or scrutinized throughout the study are obtainable upon reasonable request from the corresponding author.

Abbreviations

T2DM:

Type 2 diabetes mellitus

CCN6:

Cellular Communication Network Factor 6

IGF-1:

Insulin-like growth factor 1

NASH:

Non-Alcoholic Steatohepatitis

IGT:

Impaired glucose tolerance

ELISA:

Enzyme-linked immunosorbent assay

BMI:

Body mass index

SBP:

Systolic blood pressure

DBP:

Diastolic blood pressure

FBG:

Fasting blood glucose

IL-6:

Interleukin-6

TNF-alpha:

Tumor necrosis factorα

HOMA-IR:

Homeostatic Model Assessment for Insulin Resistance

AST:

Aspartate transaminase

ALT:

Alanine transaminase

TG:

Triglycerides

TC:

Total cholesterol

LDL-C:

Low-density lipoprotein cholesterol

HDL-C:

High-density lipoprotein cholesterol

CCN6:

Cellular communication network factor 6

FOXO1:

Forkhead box protein O1

References

  1. Zatterale F, Longo M, Naderi J, Raciti GA, Desiderio A, Miele C, Beguinot F. Chronic adipose tissue inflammation linking obesity to insulin resistance and type 2 diabetes. Front Physiol. 2020;10:1607.

    Article  PubMed  PubMed Central  Google Scholar 

  2. Westman EC. Type 2 diabetes mellitus: a pathophysiologic perspective. Front Nutr. 2021;8:707371.

    Article  PubMed  PubMed Central  Google Scholar 

  3. Cheng J-X, Yu K. New discovered adipokines associated with the pathogenesis of obesity and type 2 diabetes. Diabetes, Metabolic Syndrome and Obesity: Targets and Therapy. 2022:2381–9.

  4. Wondmkun YT. Obesity, insulin resistance, and type 2 diabetes: associations and therapeutic implications. Diabetes Metabolic Syndrome Obes. 2020:3611–6. https://doiorg.publicaciones.saludcastillayleon.es/10.2147/DMSO.S275898

  5. Miao X, Davoudi M, Alitotonchi Z, Ahmadi ES, Amraee F, Alemi A, Afrisham R. Managing cardiovascular events, hyperglycemia, and obesity in type 2 diabetes through MicroRNA regulation linked to glucagon-like peptide-1 receptor agonists. Diabetol Metab Syndr. 2025;17(1):13.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  6. Wu W, Hu X, Zhou X, Klenotic PA, Zhou Q, Lin Z. Myeloid deficiency of CCN3 exacerbates liver injury in a mouse model of nonalcoholic fatty liver disease. J Cell Communication Signal. 2018;12:389–99.

    Article  Google Scholar 

  7. Catalina MO-S, Redondo PC, Granados MP, Cantonero C, Sanchez-Collado J, Albarran L, Lopez JJ. New insights into adipokines as potential biomarkers for type-2 diabetes mellitus. Curr Med Chem. 2019;26(22):4119–44.

    Article  PubMed  Google Scholar 

  8. Afrisham R, Farrokhi V, Ayyoubzadeh SM, Vatannejad A, Fadaei R, Moradi N, et al. CCN5/WISP2 serum levels in patients with coronary artery disease and type 2 diabetes and its correlation with inflammation and insulin resistance; a machine learning approach. Biochem Biophys Rep. 2024;40:101857.

    CAS  PubMed  PubMed Central  Google Scholar 

  9. Jaganathan R, Ravindran R, Dhanasekaran S. Emerging role of adipocytokines in type 2 diabetes as mediators of insulin resistance and cardiovascular disease. Can J Diabetes. 2018;42(4):446–56. e1.

    Article  PubMed  Google Scholar 

  10. Ahmed KA, Hasib TA, Paul SK, Saddam M, Mimi A, Saikat ASM, et al. Potential role of CCN proteins in breast cancer: therapeutic advances and perspectives. Curr Oncol. 2021;28(6):4972–85.

    Article  PubMed  PubMed Central  Google Scholar 

  11. Yu X, Mao R, Feng W, Zhao Y, Qin J, Yang Y, et al. WISP3 suppresses ESCC progression by inhibiting the IGF-2-IGF1R-AKT signaling cascade. Exp Cell Res. 2021;409(1):112871.

    Article  CAS  PubMed  Google Scholar 

  12. Lorenzatti G, Huang W, Pal A, Cabanillas AM, Kleer CG. CCN6 (WISP3) decreases ZEB1-mediated EMT and invasion by Attenuation of IGF-1 receptor signaling in breast cancer. J Cell Sci. 2011;124(10):1752–8.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  13. Huang W, Pal A, Kleer CG. On how CCN6 suppresses breast cancer growth and invasion. J Cell Communication Signal. 2012;6:5–10.

    Article  Google Scholar 

  14. Song Y, Li C, Luo Y, Guo J, Kang Y, Yin F, et al. CCN6 improves hepatic steatosis, inflammation, and fibrosis in non-alcoholic steatohepatitis. Liver Int. 2023;43(2):357–69.

    Article  CAS  PubMed  Google Scholar 

  15. Thalor A, Joon HK, Singh G, Roy S, Gupta D. Machine learning assisted analysis of breast cancer gene expression profiles reveals novel potential prognostic biomarkers for triple-negative breast cancer. Comput Struct Biotechnol J. 2022;20:1618–31.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  16. Chang C-H, Lin C-H, Lane H-Y. Machine learning and novel biomarkers for the diagnosis of Alzheimer’s disease. Int J Mol Sci. 2021;22(5):2761.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  17. Wu G, Wu S, Xiong T, Yao Y, Qiu Y, Meng L, et al. Identification of biomarkers for the diagnosis of type 2 diabetes mellitus with metabolic associated fatty liver disease by bioinformatics analysis and experimental validation. Front Endocrinol (Lausanne). 2025;16:1512503.

    Article  PubMed  Google Scholar 

  18. Kim S, Park J, Son Y, Lee H, Woo S, Lee M, et al. Development and validation of a machine learning algorithm for predicting diabetes retinopathy in patients with type 2 diabetes: algorithm development study. JMIR Med Inf. 2025;13:e58107.

    Article  Google Scholar 

  19. Hosseini N, Tanzadehpanah H, Mansoori A, Sabzekar M, Ferns GA, Esmaily H, Ghayour-Mobarhan M. Using a robust model to detect the association between anthropometric factors and T2DM: machine learning approaches. BMC Med Inf Decis Mak. 2025;25(1):49.

    Article  Google Scholar 

  20. Alhilfi ASN, Afrisham R, Sefidan AM, Fadaei R, Moradi N, Saed L, Einollahi N. A positive correlation of serum SFRP1 levels with the risk of developing type 2 diabetes mellitus: a case-control study. Lab Med. 2024;55(6):739–44.

  21. Care D. 2. Classification and diagnosis of diabetes: standards of care in. Diabetes Care. 2023;46:S19.

    Article  Google Scholar 

  22. Tabatabaei Sa, Fadaei R, Moradi N, Farrokhi V, Vatannejad A, Afrisham, R et al. Circulating levels of C1q/TNF-α-related protein 6 (CTRP6) in coronary artery disease and its correlation with inflammatory markers. J Diabetes Metabolic Disorders. 2024:23(1):1233–41.

  23. Daniya T, Geetha M, Kumar KS. Classification and regression trees with Gini index. Adv Mathematics: Sci J. 2020;9(10):8237–47.

    Google Scholar 

  24. Wickramasinghe I, Kalutarage H. Naive Bayes: applications, variations and vulnerabilities: a review of literature with code snippets for implementation. Soft Comput. 2021;25(3):2277–93.

    Article  Google Scholar 

  25. Priyanka KD. Decision tree classifier: a detailed survey. Int J Inform Decis Sci. 2020;12(3):246–69.

    Google Scholar 

  26. Bentéjac C, Csörgő A, Martínez-Muñoz G. A comparative analysis of gradient boosting algorithms. Artif Intell Rev. 2021;54:1937–67.

    Article  Google Scholar 

  27. Cunningham P, Delany SJ. K-nearest neighbour classifiers-a tutorial. ACM Comput Surv (CSUR). 2021;54(6):1–25.

    Article  Google Scholar 

  28. Sasikala B, Biju VG, Prashanth C, editors. Kappa and accuracy evaluations of machine learning classifiers. 2017 2nd IEEE International Conference on Recent Trends in Electronics, Information & Communication Technology (RTEICT); 2017: IEEE.

  29. Zhang X, Liu C-A. Model averaging prediction by K-fold cross-validation. J Econ. 2023;235(1):280–301.

    Article  Google Scholar 

  30. Hassan EA, Al–Zuhairi WS. The impact of gender on the relation of serum angiopoietin-like protein 8 (ANGPTL8) level with metabolic parameters in adults. Obes Med. 2023;40:100494.

    Article  Google Scholar 

  31. Al-Zuhairi WS, Sadeghi L, Hassan EA. Evaluation of cardiac myosin binding Protein-C3 (cMyBP-C3) as potential risk factor of acute coronary syndrome in diabetic patients. Indian J Pharm Educ Res. 2024;58(4s):s1234–41.

    Article  Google Scholar 

  32. Miao X, Alidadipour A, Saed V, Sayyadi F, Jadidi Y, Davoudi M, et al. Hepatokines: unveiling the molecular and cellular mechanisms connecting hepatic tissue to insulin resistance and inflammation. Acta Diabetol. 2024;61(11):1339–61.

    Article  CAS  PubMed  Google Scholar 

  33. Batmunkh R, Nishioka Y, Aono Y, Azuma M, Kinoshita K, Kishi J, et al. CCN6 as a profibrotic mediator that stimulates the proliferation of lung fibroblasts via the integrin Β1/focal adhesion kinase pathway. J Med Invest. 2011;58(3, 4):188–96.

    Article  PubMed  Google Scholar 

  34. Huang W, Gonzalez ME, Toy KA, Banerjee M, Kleer CG. Blockade of CCN6 (WISP3) activates growth factor–independent survival and resistance to Anoikis in human mammary epithelial cells. Cancer Res. 2010;70(8):3340–50.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  35. Miller DS, Sen M. Potential role of WISP3 (CCN6) in regulating the accumulation of reactive oxygen species. Biochem Biophys Res Commun. 2007;355(1):156–61.

    Article  CAS  PubMed  Google Scholar 

  36. Twigg SM. Regulation and bioactivity of the CCN family of genes and proteins in obesity and diabetes. J Cell Communication Signal. 2018;12:359–68.

    Article  Google Scholar 

  37. Brigstock D. The CCN family: a new stimulus package. J Endocrinol. 2003;178(2):169–75.

    Article  CAS  PubMed  Google Scholar 

  38. Li J-Y, Wang Y-D, Qi X-Y, Ran L, Hong T, Yang J, et al. Serum CCN3 levels are increased in type 2 diabetes mellitus and associated with obesity, insulin resistance and inflammation. Clin Chim Acta. 2019;494:52–7.

    Article  CAS  PubMed  Google Scholar 

  39. Fadhil Jaafar A, Afrisham R, Fadaei R, Farrokhi V, Moradi N, Abbasi A, Einollahi N. CCN3/NOV serum levels in coronary artery disease (CAD) patients and its correlation with TNF-α and IL-6. BMC Res Notes. 2023;16(1):306.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  40. Afrisham R, Alasvand G, Jadidi Y, Farrokhi V, Fadaei R, Alizadeh S, Moradi N. CCN3/NOV serum levels in Non-Alcoholic fatty liver disease (NAFLD) patients in comparison with the healthy group and their correlation with TNF-α and IL-6. Curr Mol Med. 2024. https://doiorg.publicaciones.saludcastillayleon.es/10.2174/0115665240285539240214111947

  41. Maiese K. Prospects and perspectives for WISP1 (CCN4) in diabetes mellitus. Curr Neurovasc Res. 2020;17(3):327–31.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  42. Hughes J, Kuiper E, Klaassen I, Canning P, Stitt A, Van Bezu J, et al. Advanced glycation end products cause increased CCN family and extracellular matrix gene expression in the diabetic rodent retina. Diabetologia. 2007;50:1089–98.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  43. Klimontov VV, Bulumbaeva DM, Fazullina ON, Lykov AP, Bgatova NP, Orlov NB, et al. Circulating Wnt1-inducible signaling pathway protein-1 (WISP-1/CCN4) is a novel biomarker of adiposity in subjects with type 2 diabetes. J Cell Communication Signal. 2020;14:101–9.

    Article  Google Scholar 

  44. Hörbelt T, Tacke C, Markova M, Herzfeld de Wiza D, Van de Velde F, Bekaert M, et al. The novel adipokine WISP1 associates with insulin resistance and impairs insulin action in human myotubes and mouse hepatocytes. Diabetologia. 2018;61:2054–65.

    Article  PubMed  Google Scholar 

  45. Zhang Y, Pan Q, Zhong H, Merajver SD, Kleer CG. Inhibition of CCN6 (WISP3) expression promotes neoplastic progression and enhances the effects of insulin-like growth factor-1 on breast epithelial cells. Breast Cancer Res. 2005;7(6):1–10.

    Article  Google Scholar 

  46. Wei S, Wang K, Zhao Z, Huang X, Tang W, Zhao Z. WISP-3/CCN6 inhibits apoptosis by regulating caspase pathway after hyperoxia in lung epithelial cells. Gene. 2018;673:82–7.

    Article  CAS  PubMed  Google Scholar 

  47. Sahebhonar M, Dehaki MG, Kazemi-Galougahi MH, Soleiman-Meigooni S. A comparison of three research methods: logistic regression, decision tree, and random forest to reveal association of type 2 diabetes with risk factors and classify subjects in a military population. J Archives Military Med. 2022;10(2).

  48. Chawla NV, Bowyer KW, Hall LO, Kegelmeyer WP. SMOTE: synthetic minority over-sampling technique. J Artif Intell Res. 2002;16:321–57.

    Article  Google Scholar 

Download references

Acknowledgements

The current investigation is derived from research project with no. 1402-1-102-63794; hence, we would like to acknowledge the Tehran University of Medical Sciences for their financial support.

Funding

The current investigation was supported by the Tehran University of Medical Sciences (grant no. 1402-1-102-63794).

Author information

Authors and Affiliations

  1. Department of Medical Laboratory Sciences, School of Allied Medical Sciences, Tehran University of Medical Sciences, Tehran, Iran

    Reza Afrisham & Yasaman Jadidi

  2. Liver and Digestive Research Center, Research Institute for Health Development, Kurdistan University of Medical Sciences, Sanandaj, Iran

    Nariman Moradi

  3. Department of Health Information Management, School of Allied Medical Sciences, Tehran University of Medical Sciences, Tehran, Iran

    Seyed Mohammad Ayyoubzadeh

  4. Sleep Disorders Research Center, Kermanshah University of Medical Sciences, Kermanshah, Iran

    Reza Fadaei

  5. Department of Pharmacology, Vanderbilt University, Nashville, TN, USA

    Reza Fadaei

  6. Department of Hematology and Blood Banking, School of Allied Medicine, Iran University of Medical Sciences, Tehran, Iran

    Omid Kiani Ghalesardi

  7. Department of Hematology and Transfusion Sciences, School of Allied Medical Sciences, Tehran University of Medical Sciences, Tehran, Iran

    Vida Farrokhi & Shaban Alizadeh

Authors
  1. Reza Afrisham
    View author publications

    You can also search for this author inPubMed Google Scholar

  2. Yasaman Jadidi
    View author publications

    You can also search for this author inPubMed Google Scholar

  3. Nariman Moradi
    View author publications

    You can also search for this author inPubMed Google Scholar

  4. Seyed Mohammad Ayyoubzadeh
    View author publications

    You can also search for this author inPubMed Google Scholar

  5. Reza Fadaei
    View author publications

    You can also search for this author inPubMed Google Scholar

  6. Omid Kiani Ghalesardi
    View author publications

    You can also search for this author inPubMed Google Scholar

  7. Vida Farrokhi
    View author publications

    You can also search for this author inPubMed Google Scholar

  8. Shaban Alizadeh
    View author publications

    You can also search for this author inPubMed Google Scholar

Contributions

RA and SHA participated in study design. NM, RF and SMA carried out experimental tests and interpreted results. YJ, OKGH and VF wrote the draft of the manuscript. RF and RA edited the manuscript. All authors read and approved the final manuscript.

Corresponding authors

Correspondence to Reza Afrisham or Shaban Alizadeh.

Ethics declarations

Ethical approval

The protocols were performed in compliance with the Declaration of Helsinki and approved by the Ethics Committee of Tehran University of Medical Sciences (ID number: IR.TUMS.SPH.REC.1401.280). As well as, informed consent was obtained from all participants.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Afrisham, R., Jadidi, Y., Moradi, N. et al. Circulating CCN6/WISP3 in type 2 diabetes mellitus patients and its correlation with insulin resistance and inflammation: statistical and machine learning analyses. BMC Med Inform Decis Mak 25, 114 (2025). https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s12911-025-02957-1

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s12911-025-02957-1

Keywords

BMC Medical Informatics and Decision Making

ISSN: 1472-6947

Contact us