Introduction

Type 2 diabetes mellitus (T2DM) stands as one of the most widespread metabolic disorders worldwide, with its incidence steadily escalating, posing a formidable challenge to global public health [1]. Data from the International Diabetes Federation (IDF) reveals that over half a billion individuals globally are afflicted by diabetes, with T2DM accounting for approximately 90% of these cases [2]. The core pathophysiological characteristics of T2DM include insulin resistance and/or dysfunction of β-cells. These disruptions upset the delicate equilibrium crucial for blood glucose regulation, subsequently triggering a series of complications that significantly impact patients’ quality of life and longevity [3].

With the progress in metabolomics and lipidomics research, the critical role of lipid metabolism dysregulation in the onset and progression of T2DM has gained increasing recognition [4,5]. Within the diverse lipid metabolic pathways associated with T2DM, the regulation of phospholipids, especially those involving lysophosphatidylcholines, has emerged as a focal point of interest [6]. Lysophosphatidylcholine acyltransferase 3 (LPCAT3), a pivotal enzyme in phospholipid biosynthesis, stands out as the most abundant member of the LPCAT family in the liver. It significantly affects membrane phospholipid composition, cellular signal transduction, and lipid metabolic pathways [7,8]. Recent investigations have further clarified that a deficiency of LPCAT3 in metabolic tissues, including the liver, skeletal muscle, and adipose tissue, can augment insulin signaling, enhance insulin sensitivity, and result in improved glucose tolerance [9–11].

However, most previous studies have primarily focused on the role of LPCAT3 within tissue-specific contexts, and there remains a significant gap in research regarding the function of serum LPCAT3. Given the potential importance of serum LPCAT3 in glucose and lipid metabolism, coupled with the escalating global burden of T2DM, it is worthwhile to explore the relationship between serum LPCAT3 levels and metabolic variables, as well as its potential predictive value for T2DM risk. Delving into these associations may offer valuable insights into the pathogenesis of T2DM and potentially pave the way for the identification of novel biomarkers that may aid in early diagnosis and intervention.

In this study, the primary objective is to explore the association between serum LPCAT3 levels and the risk of T2DM. The secondary objectives include examining the correlations between serum LPCAT3 levels and various metabolic variables, such as blood glucose indicators (e.g., fasting blood glucose (FBG), glycated hemoglobin (HbA1c)), blood lipid indicators (e.g., high-density lipoprotein cholesterol (HDL)), and obesity-related anthropometric indicators (e.g., body mass index (BMI), waist circumference (WC)).

Materials and methods

Results

Comparison of clinical parameters and serum LPCAT3 levels between patients with T2DM and individuals with NGT

As presented in Table 1, individuals with T2DM exhibited higher levels of WC, WHR, BMI, SBP, ALT, eGFR, TG, FBG, 2hPG, HbA1c, HOMA-IR, and hs-CRP compared to those with NGT. In contrast, the T2DM group had lower levels of serum Cr and HDL. No significant differences were observed in gender distribution, age, DBP, AST, UA, TC, or LDL between the two groups. The median serum LPCAT3 level in the NGT group was 24.43 ng/ml, with an interquartile range of 14.41 to 49.37 ng/ml, which was higher than the median level of 21.51 ng/ml (interquartile range: 8.47 to 35.63 ng/ml) observed in the T2DM group.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Basic characteristics of participants.

https://doi.org/10.1371/journal.pone.0329301.t001

Assessment of the associations between LPCAT3 levels and clinical parameters in the NGT and T2DM groups using Spearman correlation analysis

Table 2 reveals that in individuals with NGT, negative correlations were observed between LPCAT3 levels and HDL, FBG, and HbA1c. Additionally, among patients with T2DM, LPCAT3 levels showed negative associations with BMI and WC.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Exploring the Spearman correlation between LPCAT3 and clinical parameters.

https://doi.org/10.1371/journal.pone.0329301.t002

Stepwise linear regression analysis identifies independent factors influencing LPCAT3 levels

A stepwise linear regression analysis was performed, utilizing LPCAT3 as the dependent variable and the remaining variables as independent variables. As detailed in Table 3, the findings reveal that BMI, HDL, and FBG were identified as independent negative predictors of serum LPCAT3 levels. The model yielded an R-squared value of 0.049, indicating that it could only explain 4.9% of the variation in serum LPCAT3 levels. This suggests the existence of numerous unidentified factors that may contribute to the regulation of serum LPCAT3.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Independent predictors of serum LPCAT3 identified by stepwise linear regression.

https://doi.org/10.1371/journal.pone.0329301.t003

When assessing the impact of confounding factors on our regression models, we found that variables like age, gender, blood pressure, and liver/kidney function indicators generally had minimal effects on the outcomes (S2 Table). Notably, significant collinearity was observed among eGFR, Cr, and age. After evaluating each variable’s contribution, we excluded eGFR from the model, as it provided no additional information beyond what was already captured by the other correlated variables. In contrast, obesity-related anthropometric measures, along with lipid and glucose metabolism indicators, substantially impacted the model’s stability. This highlights the importance of these factors in our analysis.

Our analysis of the relationships between LPCAT3 and obesity-related anthropometric measures revealed a complex interplay. Specifically, although WHR alone showed statistical significance as an independent variable, its association was attenuated when considered alongside BMI or WC. When BMI and WC were separately included in the model, appropriate models were successfully fitted. Optimal model-fitting analysis revealed that the model incorporating BMI outperformed the one with WC. However, when both variables (BMI and WC) were simultaneously included in the model, despite an acceptable level of collinearity, neither achieved statistical significance (S3–S7 Tables). This underscores the delicate nature and context-dependency of these relationships, implying that the choice of anthropometric measures could potentially have a substantial impact on the outcomes.

In the context of lipid-related factors, TC and LDL exhibited significant collinearity. As conventional methods such as centering and standardization failed to produce satisfactory results, we applied PCA to TC and LDL. This strategy led to the development of a model with a relatively good fit. Nevertheless, our findings revealed that HDL lost its statistical significance when all lipid-related factors—including TC, LDL, and TG—were taken into account in the model (S8 Table). This suggests that the relationship between LPCAT3 and lipid-related factors is significantly more complex than initially hypothesized, transcending a mere negative correlation with HDL. Instead, it involves an intricate web of interactions among various lipid components, which collectively modulate LPCAT3 levels.

The relationship between glycometabolic indicators and LPCAT3 levels is also intricate. The linear regression model incorporating FBG outperforms others, as evidenced by a higher R² value, indicating a superior fit. However, when 2hPG, HbA1c, or HOMA-IR is substituted for FBG, each variable still leads to a statistically valid model, albeit with a reduced R² value. Notably, once FBG is included in the model, adding either 2hPG or HbA1c causes FBG to lose its statistical significance (S9–S14 Tables). This implies that while LPCAT3 may exhibit an association with various glycometabolic indicators, the relationships among them are intricate and interdependent. It is likely that the relatively stronger correlation observed between FBG and LPCAT3 may obscure the more subtle associations of the other indicators.

Overall, our analysis of collinearity among the independent variables has revealed a complex and unstable network of relationships which are highly dependent on the specific correlated variables incorporated into the model.

PLS analysis of correlations between LPCAT3 and obesity-related anthropometric, lipid, and glucose indicators

By conducting collinearity and confounding factor analyses, we identified complex associations between LPCAT3 and various indicators. The primary focus was on obesity-related anthropometric indicators (e.g., BMI, WC, WHR), blood lipid indicators (e.g., TC, TG, HDL, LDL), and blood glucose indicators (e.g., FBG, 2hPG, HbA1c, HOMA-IR). To gain a deeper insight into these relationships, we performed partial least squares (PLS) analysis. In this analysis, LPCAT3 was designated as the dependent variable (Y), while the obesity-related anthropometric indicators, blood lipid indicators, and blood glucose indicators were treated as distinct sets of independent variables (X).

Fig 1A and 1B illustrate the PLS analysis results concerning the correlations between LPCAT3 and obesity-related anthropometric indicators. Fig 1A, a PLS bi-plot, shows that principal component 1 (PC1) explains approximately 78.6% of the total variance, while principal component 2 (PC2) accounts for about 18.7%. BMI, WC, and WHR are projected onto this bi-plot, contributing to the variation in the principal components. Notably, LPCAT3 exhibits a relatively strong negative correlation with WC and BMI. Fig 1B, a Coefficient Plot, displays coefficients of approximately −0.064 (SE = 0.186) for BMI, −0.067 (SE = 0.346) for WC, and −0.040 (SE = 0.207) for WHR. These coefficients suggest negative correlations between all three indicators and LPCAT3, with BMI and WC potentially having a stronger impact. Nevertheless, the relatively large standard errors indicate low stability and reliability of these estimates, reflecting considerable uncertainty in the observed correlations.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. PLS analysis of correlations between LPCAT3 and obesity-related anthropometric indicators.

In Panel A (PLS bi-plot), principal component 1 accounts for 78.6% and principal component 2 for 18.7% of total variance. BMI, WC, and WHR are projected in the plot, with LPCAT3 showing a relatively strong negative correlation with WC and BMI. Panel B (Coefficient Plot) presents correlation coefficients between LPCAT3 and obesity-related indicators: approximately −0.064 (SE = 0.186) for BMI, −0.067 (SE = 0.346) for WC, and −0.040 (SE = 0.207) for WHR. All variables were standardized prior to PLS analysis. Abbreviations: PLS: partial least squares; LPCAT3: lysophosphatidylcholine acyltransferase 3; BMI: body mass index; WC: waist circumference; WHR: waist-to-hip ratio.

https://doi.org/10.1371/journal.pone.0329301.g001

Fig 2A and 2B present the PLS analysis results concerning the correlations between LPCAT3 and blood lipid indicators. Fig 2A, a PLS bi-plot, shows R² values of R²X[1] = 0.558 and R²X[2] = 0.203 for the first two components, indicating the proportion of variance in the X-variables explained by each component. TC, HDL, and LDL demonstrate certain correlations with the principal components, with HDL exhibiting a relatively strong negative correlation with LPCAT3. In contrast, TG contributes less to the variation in the principal components and shows a weaker association with LPCAT3. Fig 2B, a Coefficient Plot, displays coefficients of approximately −0.001 (SE = 0.087) for TC, −0.017 (SE = 0.185) for TG, −0.071 (SE = 0.219) for HDL, and −0.009 (SE = 0.106) for LDL. HDL appears to have a slightly stronger negative correlation with LPCAT3 compared to the other lipid indicators. However, the relatively large standard errors suggest imprecise model estimates, thereby limiting the reliability of the findings.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. PLS analysis of correlations between LPCAT3 and blood glucose indicators.

Panel A (PLS bi-plot) displays R² values of R²X[1] = 0.558 and R²X[2] = 0.203, indicating the proportion of variance in X-variables explained by each component. TC, TG, HDL, and LDL correlate with LPCAT3, with HDL showing a relatively strong negative correlation and TG a weaker one. Panel B (Coefficient Plot) shows correlation coefficients between LPCAT3 and blood lipid indicators: about −0.001 (SE = 0.087) for TC, −0.017 (SE = 0.185) for TG, −0.071 (SE = 0.219) for HDL, and −0.009 (SE = 0.106) for LDL. All variables were standardized prior to PLS analysis. Abbreviations: PLS: partial least squares; LPCAT3: lysophosphatidylcholine acyltransferase 3; TC: total cholesterol; TG: triglyceride; HDL: high-density lipoprotein cholesterol; LDL: low density lipoprotein cholesterol.

https://doi.org/10.1371/journal.pone.0329301.g002

Fig 3A and 3B depict the PLS analysis results regarding the correlations between LPCAT3 and blood glucose indicators. Fig 3A, a PLS bi-plot, shows R² values of R²X[1] = 0.783 and R²X[2] = 0.098 for the first two components. FBG, 2hPG, HbA1c, and HOMA-IR are projected onto this bi-plot, contributing to the variation in the principal components. FBG shows the strongest negative correlation with LPCAT3, followed by HOMA-IR. Fig 3B, a Coefficient Plot, presents coefficients of approximately −0.115 (SE = 0.213) for FBG, −0.023 (SE = 0.148) for 2hPG, 0.020 (SE = 0.225) for HbA1c, and −0.069 (SE = 0.365) for HOMA-IR. FBG and HOMA-IR appear to have relatively stronger negative correlations with LPCAT3. Still, the relatively large standard errors indicate low stability and reliability of these coefficient estimates, introducing a degree of uncertainty into the observed correlations.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. PLS analysis of correlations between LPCAT3 and blood glucose indicators.

Panel A (PLS bi-plot) presents R² values of R²X[1] = 0.783 and R²X[2] = 0.098. FBG, 2hPG, HbA1c, and HOMA-IR are projected, with FBG showing the strongest negative correlation with LPCAT3, followed by HOMA-IR. Panel B (Coefficient Plot) presents correlation coefficients between LPCAT3 and blood glucose indicators: roughly −0.115 (SE = 0.213) for FBG, −0.023 (SE = 0.148) for 2hPG, 0.020 (SE = 0.225) for HbA1c, and −0.069 (SE = 0.365) for HOMA-IR. All variables were standardized prior to PLS analysis. Abbreviations: PLS: partial least squares; LPCAT3: lysophosphatidylcholine acyltransferase 3; FBG: fasting blood glucose; 2hPG: 2-hour post-oral glucose tolerance test blood glucose level; HbA1c: glycated hemoglobin A1c; HOMA-IR: homeostasis model assessment of insulin resistance.

https://doi.org/10.1371/journal.pone.0329301.g003

In summary, the PLS analysis highlights the intricate and multifaceted associations among these variables. These associations collectively influence LPCAT3 levels. Although correlations between LPCAT3 and the examined indicators (obesity-related anthropometric, blood lipid, and blood glucose) are observed, these correlations are notably weak and exhibit considerable uncertainty, as indicated by the relatively large standard errors. This underscores the complexity of these relationships and the need for cautious interpretation of the results.

S2 Fig displays the prediction versus observation plots for the three models. Generally, the Root Mean Squared Error of Estimation (RMSEE) and Root Mean Squared Error of Cross-Validation (RMSECV) values of these three models are quite similar, suggesting that all three models exhibit stability and good generalizability. Nevertheless, the comparatively high values of RMSEE and RMSECV indicate that the predictive performance of these models is unsatisfactory.

Logistic regression analysis of T2DM incidence across serum LPCAT3 tertiles, using the lowest tertile as reference

The 506 participants were categorized into three groups based on their serum LPCAT3 concentrations: T1 (< 15.965 ng/mL), T2 (15.965–32.614 ng/mL), and T3 (> 32.614 ng/mL). The incidence of T2DM showed a progressive decline with increasing LPCAT3 levels, with rates of 55.88% in T1, 51.48% in T2, and 41.42% in T3. As detailed in Table 4, the initial analysis revealed a significantly lower incidence of T2DM in the T3 group compared to the T1 group. However, after sequentially adjusting for obesity-related anthropometric indicators including BMI, WC and WHR, the difference between these two groups became insignificant. No statistical difference in T2DM incidence was observed between the T1 and T2 groups.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Logistic regression analysis of serum LPCAT3 level and T2DM prevalence.

https://doi.org/10.1371/journal.pone.0329301.t004

ROC curve analysis: evaluating serum LPCAT3 as a biomarker for predicting T2DM incidence via AUC

Fig 4 demonstrates that serum LPCAT3 levels have a statistically significant yet relatively weak predictive capability for the incidence of T2DM. S15 Table presents the results of an ROC curve analysis for predicting T2DM incidence based on serum LPCAT3 levels. The optimal cut-off point for LPCAT3 was determined to be 30.133 ng/ml, with a sensitivity of 42.19% (95% CI: 36.1–48.5) and a specificity of 71.83% (95% CI: 65.8–77.3). The Youden index at this cut-off point, calculated as Sensitivity + Specificity – 1, would be approximately 0.140. The area under the curve (AUC) was 0.580 (95%CI: 0.531–0.630, p < 0.01), indicating the overall discriminative ability of serum LPCAT3 levels in predicting T2DM incidence.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. ROC curve for predicting T2DM incidence using serum LPCAT3 levels.

The red dot in the figure represents the optimal cut-off point for LPCAT3, which is 30.133 ng/ml, with a sensitivity of 42.2% and a specificity of 71.8%. The area under the curve (AUC) was 0.580, and the 95% confidence interval for the AUC ranged from 0.531 to 0.630 (p < 0.01).

https://doi.org/10.1371/journal.pone.0329301.g004

Discussion

In the present study, we found that participants newly diagnosed with T2DM had lower serum levels of LPCAT3 compared to individuals with NGT. Within the NGT cohort, we observed a negative correlation between serum LPCAT3 concentrations and both FBG and HbA1c levels. Importantly, FBG was identified as a potential independent negative predictor of LPCAT3 levels. These findings collectively suggest a complex and multifaceted involvement of LPCAT3 in glucose metabolic pathways, warranting further investigation to elucidate the underlying mechanisms and potential therapeutic implications. In 2016, Cash JG et al. were the pioneers in reporting that the overexpression of LPCAT3 in the liver alleviated postprandial hyperglycemia and improved glucose tolerance. They elucidated that this beneficial effect stemmed from the reduction of intracellular lysophosphatidylcholine levels, consequently attenuating its inhibitory influence on mitochondrial fatty acid β-oxidation [17]. However, in 2023, Tian Y et al. discovered that liver-specific deletion of LPCAT3 reduced polyunsaturated phosphatidylcholine (PUFA-PC) content within liver cell membranes. This alteration in phospholipid composition may enhance insulin receptor (INSR) endocytosis and signaling by modifying the physical properties of the membrane. Notably, mice with suppressed LPCAT3 expression exhibited upregulated hepatic fibroblast growth factor 21 (FGF21) expression, increased glucose uptake in brown adipose tissue (BAT), and improved systemic insulin sensitivity and glucose tolerance [9]. Apart from the liver, targeted inhibition of LPCAT3 in skeletal muscle and adipose tissue has also been shown to enhance insulin signaling and improve glucose metabolism, consistent with the findings presented by Tian Y et al. In skeletal muscle, the suppression of LPCAT3 led to an enhancement in INSR phosphorylation, thereby augmenting insulin signaling and sensitivity. This effect was attributed to alterations in the lipid arrangement within the cell membrane, particularly through the promotion of Monosialotetrahexosylganglioside (GM-1) microdomain clustering. Mice with a skeletal muscle-specific deletion of LPCAT3 demonstrated increased insulin sensitivity and improved blood glucose regulation [10]. Conversely, the overexpression of LPCAT3 resulted in diminished insulin sensitivity [18]. Likewise, adipose-specific suppression of LPCAT3 decreased tissue PUFA-PC levels, concurrently increasing phosphorylation of INSR and AKT, thereby indicating the activation of the insulin signaling pathway. This activation subsequently enhanced insulin sensitivity and improved glucose tolerance [11]. In a murine model where systemic suppression of LPCAT3 expression was achieved through the administration of antisense oligonucleotides, significant improvements were observed in both glucose tolerance and insulin sensitivity [19]. In general, most studies support the idea that inhibiting LPCAT3 offers benefits in strengthening insulin signaling, alleviating insulin resistance, and improving glucose tolerance. Elevated activity and expression of LPCAT3 have also been noted in animal models exhibiting insulin resistance [9,11]. However, contrary to initial expectations, our research indicates that serum LPCAT3 levels are decreased in patients with T2DM and show a negative correlation with FBG. The underlying causes of this discrepancy remain elusive, prompting us to explore potential contributing factors. Notably, previous in vivo studies have primarily focused on elucidating how LPCAT3 expression in animal tissues affects glucose metabolism. In contrast, our study delves into the levels of LPCAT3 present in human serum. Therefore, it is of great significance to explore the source and function of serum LPCAT3. As we known, LPCAT3 is an endoplasmic reticulum (ER) membrane protein highly expressed in various metabolic tissues, including the liver, small intestine, and adipose tissue, as well as in other tissues such as the testis and kidney [20]. ER proteins are typically secreted via mechanisms such as co-translational translocation, post-translational translocation, and vesicular transport [21]. Although no former studies have clarified the source of serum LPCAT3, it is inferred that it may be secreted through these ER pathways or released due to cell damage. Endoplasmic reticulum stress, inflammatory responses, lipid metabolism disorders, and cellular damage induced by T2DM [3] may contribute to these processes. It is important to note that serum LPCAT3 may not originate from a single tissue but rather from the cumulative contributions of multiple tissues expressing LPCAT3. The release of LPCAT3 from tissues into the serum is influenced not only by cellular secretory mechanisms but also by other factors, such as the efficiency of blood circulation and the metabolic clearance by the liver and kidneys. These factors may lead to serum LPCAT3 levels not accurately reflecting the actual tissue levels. Furthermore, systemic LPCAT3 is collectively modulated by multiple tissues and organs. This collective modulation potentially hinders a direct reflection of the biological functions of individual tissues through systemic LPCAT3. Additionally, the expression of LPCAT3 in specific tissues exhibits distinct regulatory patterns, and its biological functions vary across different tissues. For instance, in the liver, LPCAT3 is involved in lipogenesis and lipid transport and is regulated by the liver X receptor (LXR) [7]. In contrast, in macrophages, it is modulated by inflammatory signals and plays a role in affecting inflammation [22]. Given these tissue-specific regulatory patterns and functional differences, the functions of systemic LPCAT3 are unlikely to be a simple summation of those in individual tissues. In conclusion, LPCAT3 has a complex origin, modulation, and tissue-specific functions. Considering these characteristics, both the serum levels and functions of LPCAT3 may not fully represent its role in any single specific tissue. Therefore, further investigation is needed. Apart from considering the source of serum LPCAT3, it is imperative to take into account other possible factors that may affect serum LPCAT3. For instance, in patients with T2DM, persistent hyperglycemia and insulin resistance may trigger certain compensatory mechanisms, ultimately affecting the expression or activity of LPCAT3. Moreover, T2DM is a complex and multifaceted disorder, presenting with chronic inflammation and dysregulation of lipid metabolism, among other pathological features. The potential involvement of LPCAT3 in these processes will be discussed in subsequent sections. It is noteworthy that in our study, the association between LPCAT3 and T2DM, as well as glucose metabolism indicators, appears relatively weak. This is evidenced by the small area under the ROC curve, low R-squared values in the predictive models, and substantial standard errors from PLS analysis, which collectively highlight the tenuous and complex nature of this association. To gain a deeper understanding, future research should adopt a more comprehensive approach. Studies with larger sample sizes are essential to enhance statistical reliability. Additionally, examining LPCAT3 expression across a broader range of tissues may elucidate its overall role in glucose metabolism. Moreover, conducting long-term observations of the interaction between LPCAT3 and glucose metabolism over time will provide a clearer understanding of their relationship, rather than relying solely on one-time data.

The pivotal role of LPCAT3 in lipid metabolism has garnered substantial attention, emerging as a prominent topic of discussion. Its deficiency initiates a cascade of events that exert a profound influence on various lipid-related processes. Primarily, LPCAT3 deficiency activates β-catenin-dependent Wnt signaling pathway, which subsequently suppresses the expression of genes and transcription factors crucial for lipid synthesis, including fatty acid synthase (FAS), diacylglycerol O-acyltransferase homolog 1 (DGAT1), sterol regulatory element-binding protein (SREBP)-1c, SREBP-2, CCAAT/enhancer-binding protein alpha (C/EBPα), and peroxisome proliferator-activated receptor γ (PPARγ). Consequently, adipocyte differentiation and lipid accumulation are impaired [23]. Notably, among these factors, the downregulation of the SREBP-1c pathway due to LPCAT3 deficiency—a well-documented phenomenon in research—slows down fat production and accumulation [8,9,19,24–26]. Secondly, LPCAT3 is indispensable for lipid absorption, transport, secretion, and storage. As a key enzyme involved in phosphatidylcholine remodeling, LPCAT3 modulates the composition and fluidity of cell membrane phospholipids, thereby influencing lipid absorption and secretion. Studies on LPCAT3-knockout mice have shown a significant reduction in intestinal lipid absorption capacity, accompanied by decreased plasma levels of cholesterol, phospholipids, and triglycerides [27–29]. Furthermore, LPCAT3 promotes the formation of a phospholipid membrane microenvironment that is enriched in polyunsaturated fatty acids (PUFAs), particularly arachidonic acid. This microenvironment facilitates triglyceride accumulation within the membrane, enhancing triglyceride transport efficiency. In the absence of LPCAT3, triglycerides cannot be effectively incorporated into lipoproteins, leading to their accumulation in liver and intestinal cells and a decrease in plasma triglyceride levels [30,31]. Moreover, the enzymatic activity of LPCAT3 on cellular membranes increases the production of ω-6 arachidonic acid-containing phosphatidylethanolamine, which promotes triglyceride storage within adipocytes [32]. Additionally, LPCAT3 affects the abundance of membrane proteins, including niemann-pick C1-like 1 (NPC1L1), ATP-binding cassette transporter A1 (ABCA1), and ATP-binding cassette transporter G8 (ABCG8), which are vital for lipid absorption and cholesterol secretion. LPCAT3 deficiency disrupts the intestinal absorption of cholesterol, triglycerides, and phospholipids and reduces the capacity of small intestinal epithelial cells to secrete cholesterol via chylomicron and HDL pathways [27]. Finally, LPCAT3 deficiency disrupts the assembly of very low-density lipoprotein (VLDL) particles, resulting in the formation of smaller VLDL particles with reduced lipid content. This underscores the critical role of LPCAT3 in VLDL biogenesis [31]. In conclusion, LPCAT3 is of paramount importance in lipid metabolism, and reduced LPCAT3 expression may be associated with lower levels of serum cholesterol and triglycerides. In this study, we investigated the associations between serum LPCAT3 levels and various lipid-related parameters in individuals with NGT and those with T2DM, uncovering several intriguing discoveries. Specifically, within the NGT group, we observed a negative correlation between serum LPCAT3 levels and HDL concentrations. Stepwise linear regression analysis subsequently identified HDL as an independent negative predictor of serum LPCAT3 levels. However, it is crucial to acknowledge that even within the NGT population, the correlation between LPCAT3 and HDL is relatively weak, as evidenced by the small correlation coefficient. In linear regression models, the attenuated association between HDL and LPCAT3 when adjusting for other lipid factors highlights the intricate regulatory network of lipid metabolism. PLS analysis further reveals the instability and complexity of the relationship between LPCAT3 and the overall lipid profile. Given that the current study does not fully elucidate these intricate connections, further research is warranted. Future research should aim to decipher the precise mechanisms underlying LPCAT3’s interactions with lipid components within this complex network, potentially through large-scale longitudinal monitoring of LPCAT3 and lipid levels, as well as in vitro and in vivo manipulation of LPCAT3 activity.

Our study has uncovered a notable negative correlation between serum LPCAT3 levels and obesity-related body measurements, including BMI and WC. This finding appears to contradict previous researches that have focused on the tissue-specific expression or functional roles of LPCAT3. Previous studies have reported elevated LPCAT3 expression in the skeletal muscle of obese individuals [10,18,19,24]. Conversely, mice with liver-specific [9,19] or small intestine-specific [27] LPCAT3 deficiency exhibit reduced subcutaneous white adipose tissue, weight loss, and resistance to obesity. In addition to its roles in lipid biosynthesis, absorption, and transport (as previously discussed), LPCAT3 may also modulate energy metabolism, thereby influencing body weight regulation. In mice with reduced LPCAT3 activity, adipose tissue displays disrupted lipid metabolism, characterized by increased lipid hydrolysis and fatty acid cycling, which ultimately contributes to sustained energy expenditure and limited weight gain [32]. Moreover, a global knockout of LPCAT3 results in a decrease in the respiratory exchange ratio, indicating a shift towards using fat as the primary energy source. This shift may be associated with weight loss and reduced adipose tissue mass [19]. Additionally, inhibiting LPCAT3 may enhance the secretion of FGF21, which in turn increases energy expenditure [19]. While tissue-specific suppression of LPCAT3 expression has been shown to confer resistance to weight gain in preclinical models, our research reveals a potential negative trend between serum LPCAT3 levels and obesity-related anthropometric measures. This inconsistency implies a unique role of serum LPCAT3 distinct from its tissue-based function or suggests the presence of previously undiscovered mechanisms. In our study of individuals with T2DM, serum LPCAT3 levels exhibit a negative correlation with BMI and WC. Stepwise linear regression analysis indicates that BMI is an independent negative predictor of LPCAT3 levels. However, the strength of this association is relatively weak, as evidenced by the correlation coefficient. Subsequent multiple linear regression models and PLS analysis further reveal an unstable yet complex relationship between the three obesity-related anthropometric parameters included in this study (BMI, WC, and WHR) and serum LPCAT3 levels. To enhance our understanding of these relationships, future research should utilize advanced body composition assessment techniques, such as dual-energy X-ray absorptiometry (DXA) scans, which provide detailed information on body fat distribution, to distinguish the contributions of visceral and subcutaneous adipose tissue depots. Furthermore, longitudinal studies tracking concurrent changes in anthropometric measures and serum LPCAT3 levels over time will advance our understanding of their interplay in metabolic health.

Notably, when serum LPCAT3 is used as a predictor for the onset of T2DM, its predictive power diminishes after adjusting for obesity-related anthropometric measures. A further stratified analysis based on body mass index (BMI) has shown that the negative correlation between serum LPCAT3 levels and T2DM is more pronounced in overweight/obese individuals (BMI ≥ 24 kg/m²). In contrast, this correlation did not reach statistical significance among those with a BMI < 24 kg/m². These observations highlight the intricate interplay among serum LPCAT3 levels, T2DM, and obesity. It is plausible that the association between LPCAT3 and T2DM is indirect, potentially mediated by obesity and adipocyte-related biological mechanisms. This hypothesis holds some merit. Firstly, obesity, a well-established risk factor for T2DM, manifests as insulin resistance, chronic low-grade inflammation, and dyslipidemia in affected individuals [33]. These metabolic disturbances may affect the physiological interaction between LPCAT3 and T2DM. Moreover, previous studies have shown that the regulatory effects of LPCAT3 on insulin signaling in various tissues, including the liver [9], skeletal muscle [10,18], and adipose tissue [11,17], are mediated by changes in membrane lipid composition and fluidity. Notably, obesity not only directly affects membrane lipid composition, leading to decreased fluidity [34], but also modulates LPCAT3 expression in the liver [24] and skeletal muscle [8,18], thereby influencing glycerophospholipid remodeling in cell membranes. Recognizing the importance of understanding the intricate associations among LPCAT3, obesity, and T2DM is paramount. Nevertheless, the present study falls short of further elucidating the intricate underlying mechanisms. Large-scale population-based cohort studies are needed to validate the LPCAT3-T2DM correlation and explore its variation across different obesity levels. Incorporating more confounding factors and using more specific animal or cellular models can also enhance our understanding of their relationship. These efforts may lead to novel therapeutic approaches for T2DM, especially in obese individuals.

Given that gender and age are crucial factors in the pathogenesis of metabolic diseases, we conducted stratified analyses to identify potential effect modifiers. Our findings indicate that serum LPCAT3 levels exhibit a more robust negative correlation with BMI, HDL, and FBG in males. To the best of our knowledge, no prior studies have reported gender-based disparities in LPCAT3 expression or function concerning metabolic disorders. However, it is widely acknowledged that males and females demonstrate significant differences in metabolic parameters and hormonal profiles [35]. We hypothesize that these gender-related differences may elucidate the observed variations in the associations between LPCAT3 levels and metabolic indicators. Interestingly, our study observed substantial disparities in multiple baseline parameters between males and females. For instance, males had higher BMI and lower HDL levels compared to females (S23 Table). Nevertheless, these baseline differences alone cannot fully elucidate the gender-specific variations in the associations between LPCAT3 levels and metabolic parameters. For example, in a further BMI-stratified analysis, we found no evidence to suggest that these associations could be solely attributed to BMI differences (S24 Table). This implies that the observed differences may arise from multiple confounding factors or other unidentified variables. Gender-stratified analysis of the relationship between LPCAT3 and the risk of T2DM revealed a more pronounced negative correlation in males. Notably, in the BMI-stratified analysis, a negative correlation between LPCAT3 and T2DM risk was observed in the obese/overweight population but not in those with normal weight. Given that males have a higher BMI than females in our study, this finding aligns with the closer association between LPCAT3 and T2DM risk in males. However, this explanation does not hold when considering the results of an age-stratified analysis. Among participants, those aged <40 years had the highest BMI; however, no correlation was found between LPCAT3 and T2DM risk. In contrast, in the 40–60 age group, with a relatively lower BMI, a significant negative correlation between LPCAT3 and T2DM risk was observed, suggesting the presence of other underlying factors. The regression analysis stratified by age further complicates our understanding of the relationships between LPCAT3 and metabolic parameters (BMI, HDL and FBG). Correlations between LPCAT3 and some variables were detected in the 40–60 age group and the ≥ 60 age group. Conversely, no correlations were found between LPCAT3 and BMI, HDL, or FBG in the < 40 age group. However, we must interpret the results of the age-stratified analysis with caution. To minimize selection bias, we consecutively recruited eligible newly diagnosed T2DM patients. Consequently, the 40–60 age group comprised the largest number of participants, which may suggest that this age range represents a high-incidence period for T2DM. In contrast, the relatively elderly population (≥60 years) had a smaller sample size, which may be attributed to strict screening criteria, such as excluding those who had taken medication within one month before enrollment. Additionally, elderly individuals frequently present with more complex comorbidities. These comorbidities have the potential to confound the observed correlation between LPCAT3 levels, metabolic parameters, and T2DM in this population. The age group under 40 had a relatively small participant count and presented with higher BMI values. These findings might suggest a comparatively lower prevalence of T2DM in the population under 40. Nevertheless, obesity remains a potential risk factor for T2DM. It is imperative to acknowledge that, due to the small sample size and the presence of complex health conditions, the contingency and complexity of the correlation results are likely to be elevated. To further address these issues, future studies ought to employ larger sample sizes, adopt well-conceived age-stratified research strategies, and integrate more all-encompassing observational indicators. These efforts will facilitate a more in-depth elucidation of the potential contributions of age and gender to the associations among LPCAT3, metabolic parameters, and the risk of T2DM. Comparisons of demographic and clinical parameters across different age groups are presented in S25 Table.

It is noteworthy that LPCAT3, beyond its involvement in metabolic pathways, plays a pivotal role in various physiological and pathological processes. Its complex dual effects on inflammatory responses and ferroptosis regulation are particularly striking. On one hand, it enhances membrane unsaturation, thereby alleviating endoplasmic reticulum stress induced by saturated free fatty acids (FFAs) and mitigating inflammatory responses and cellular damage from lipid peroxidation [36,37]. On the other hand, LPCAT3 increases PUFA-PC levels in the cell membrane, rendering the membrane more vulnerable to free radical attacks. This heightens the cell’s susceptibility to ferroptosis and facilitates ferroptosis initiation [38,39]. These processes are intricately linked to diabetes [40] and lipid metabolism [41]. The question of whether alterations in LPCAT3 influence glucose and lipid metabolism via the inflammation and ferroptosis pathways remains an area of great interest and warrants further investigation.

It is remarkable that LPCAT3 is widely expressed in various tissues, including the liver and kidneys, where it plays a pivotal role in maintaining cell membrane integrity, supporting cellular function, and facilitating signal transduction. Its involvement in inflammatory responses and ferroptosis regulation further highlights its crucial role in modulating organ function, as supported by numerous studies [42,43]. In the liver, LPCAT3 exerts multifaceted effects. Some research underscores its role in promoting lipid synthesis, thereby influencing hepatic lipid metabolism [28,31], while other studies suggest it mitigates cellular damage caused by elevated lipid peroxidation [36,37]. A liver injury model involving liver-specific LPCAT3 knockout mice demonstrated reduced hepatic necrosis areas and attenuated elevations in serum transaminases (ALT and AST), indicating a potential protective role for LPCAT3 in liver health [44]. However, our current study found no significant correlations between serum LPCAT3 levels and hepatic function indicators such as AST and ALT, nor discernible differences in LPCAT3 levels between individuals with and without fatty liver disease, as detailed in S2 Table. Similarly, the role of LPCAT3 in the kidneys is complex and closely linked to ferroptosis. Overexpression of LPCAT3 in renal tubular epithelial cells leads to decreased cell viability and ferroptosis induction [45], and elevated LPCAT3 expression in the blood during acute kidney injury models underscores its potential involvement in kidney injury progression [46]. As a member of the lysophosphatidylcholine acyltransferase family, LPCAT3 may also influence renal hemodynamics by regulating the renal free arachidonic acid pool, thus affecting renal function [43]. Additionally, its upregulation in the livers of hyperuricemic mice suggests a role in hyperuricemia development and progression [25]. However, our study found no correlations between serum LPCAT3 levels and renal function markers such as creatinine (Cr) and uric acid (UA). In the context of atherosclerosis, LPCAT3 is essential for regulating phospholipid metabolism, cholesterol homeostasis, cholesterol efflux, and inflammatory responses [47]. Hematopoietic cell-specific Lpcat3 knockout in mice exacerbates atherosclerosis [47], whereas macrophage-specific deletion has no substantial impact on atherogenesis [48]. In human atherosclerotic tissues, reduced LPCAT3 expression correlates with disease severity [49]. Nevertheless, our study found no differences in serum LPCAT3 levels between individuals with and without carotid atherosclerosis, as shown in S2 Table. Despite the lack of significant associations between serum LPCAT3 levels and parameters reflecting hepatic and renal function in our study, it is crucial to consider potential confounding factors, as previous research has emphasized the significance of various variables. In our study comparing the NGT and T2DM groups, significant differences were observed in several aspects, including ALT, Cr, fatty liver status, atherosclerosis status, and inflammatory indicators. These disparities raise the question of whether they could influence the observed differences in serum LPCAT3 levels between the two populations. For example, altered hepatic function, as reflected by changes in ALT levels, may be linked to the metabolic milieu in which LPCAT3 operates, potentially affecting its expression or activity. Similarly, variations in renal function, indicated by Cr levels, could have implications for the systemic regulation of LPCAT3. Moreover, the presence of fatty liver disease and atherosclerosis, which are associated with complex metabolic and inflammatory alterations, might indirectly impact LPCAT3 levels. Inflammatory indicators, often elevated in T2DM and related conditions, could also modulate LPCAT3 expression or function. Given these considerations, these factors warrant in-depth reflection in future research. Larger sample sizes, broader observation indicators, and dynamic multi-system observations may be essential for unraveling the complex relationships between LPCAT3 and these parameters. Such efforts could provide deeper insights into the biological roles and therapeutic potential of LPCAT3 in various pathological conditions.

Conclusion and limitations

This study conducted a preliminary exploration of the alterations in serum LPCAT3 levels among patients with newly diagnosed T2DM and investigated its role in glucose and lipid metabolism. We observed reduced levels of serum LPCAT3 in these patients. However, the association between serum LPCAT3 and T2DM is not independent of adiposity; rather, it is likely confounded or mediated by obesity-related factors. Additionally, while there appears to be a tendency towards a negative correlation between serum LPCAT3 and FBG, BMI, and HDL levels, these relationships become nuanced when adjusting for confounding factors. Specifically, the interplay among glycemic indices, lipid profiles, and anthropometric measures may independently influence or modify the observed associations. Despite the study’s findings, several limitations warrant consideration. Firstly, we must acknowledge the limited explanatory and predictive power of our models. The final stepwise linear regression model explained only 4.9% of the variance in serum LPCAT3 levels (R² = 0.049), suggesting that unmeasured factors are likely the main contributors of LPCAT3 variability. Similarly, the ROC curve analysis yielded an AUC of 0.580, indicating poor discriminatory performance. Furthermore, the instability of predictor relationships due to collinearity, as evidenced by sensitivity to variable inclusion, underscores the weak and unreliable nature of the identified associations. In addition, the substantial standard errors yielded by the PLS analysis provide further evidence of this uncertainty. Secondly, our study design targeted individuals with either NGT or diabetes to delineate distinct phenotypic differences between these two groups. This approach minimized potential confounding factors associated with intermediate metabolic states such as prediabetes, allowing us to confidently attribute observed differences in LPCAT3 levels to the unique metabolic characteristics of NGT and diabetic individuals. However, we acknowledge that excluding prediabetic subjects restricts our ability to comprehensively assess early metabolic changes preceding T2DM diagnosis, thereby hindering a thorough evaluation of LPCAT3’s potential involvement in disease progression, particularly during the pivotal pre-diabetic stage. Thirdly, the cross-sectional nature of our study, which concentrates on serum LPCAT3 levels at a solitary time point, impedes the monitoring of longitudinal fluctuations in LPCAT3 expression over time or within tissues integral to T2DM pathophysiology, such as adipose tissue and the liver. This limitation curtails our capacity to fully elucidate LPCAT3’s role in metabolic regulation. Fourthly, the recruitment of participants from four local medical institutions in the Jiangsu Taizhou region may introduce referral bias. It is plausible that the T2DM cohort includes a disproportionately high number of severe cases, and individuals with heightened health consciousness or research participation enthusiasm might be overrepresented. Additionally, the continuous recruitment process and specific inclusion/exclusion criteria may have led to an overrepresentation of middle-aged individuals. Furthermore, the geographical location of the participating institutions and the socioeconomic environment of the local populace could have influenced the demographic and clinical profiles of the study volunteers. Fifthly, despite our efforts to match the NGT and T2DM groups for gender and age, inherent disparities in characteristics such as BMI, lipid profiles, blood pressure, organ and vascular health, dietary and exercise habits, and potential comorbidities persist. These confounding factors have the potential to skew our study outcomes. To mitigate these limitations, future investigations should adopt a longitudinal design with larger, more heterogeneous cohorts to capture the temporal and tissue-specific dynamics of LPCAT3 levels. Additionally, the utilization of multivariate modeling techniques, incorporating potential confounding variables (e.g., organ function, lifestyle factors) as covariates in regression analyses, will be pivotal in disentangling the independent impacts of glycemic indices, lipids, and anthropometric parameters on LPCAT3 levels. By employing these rigorous methodologies, researchers can ascertain whether LPCAT3 dysregulation precipitates metabolic dysfunction or is a consequence thereof, thus clarifying its role in metabolic health and disease. Such advancements will enhance our comprehension of LPCAT3’s potential as a biomarker or therapeutic target in T2DM and associated metabolic disorders.

Supporting information

Acknowledgments

We extend our sincerest gratitude to Zhou XL, Sun XP, and Ni M for their invaluable assistance in recruiting participants. We are deeply indebted to Zhou NX and her team for their substantial contributions to the data measurement process. During the revision phase of this manuscript, we are particularly grateful to Professor Cheng XB and his team, as well as many anonymous reviewers, for their insightful comments and constructive feedback. Their invaluable suggestions have greatly enhanced the quality and rigor of our work, and we are truly appreciative of their time and expertise.