Introduction

Obstructive sleep apnea (OSA) is characterized by the repetitive collapse of the upper airway, leading to hypoxia, hypercapnia, and sleep arousals1. The prevalence of OSA is notably higher in males, affecting 10%-17% of men compared to 3%-9% of women across various age groups2. If left untreated, OSA significantly elevates the risk of developing cardiovascular, metabolic, and cognitive disorders3,4,5,6,7,8. Additionally, OSA patients are more prone to experiencing daytime sleepiness, depression, accidents, and stroke9. However, early diagnosis and screening are challenging, as 70–90% of OSA patients are asymptomatic in the initial stages10.

Polysomnography (PSG) is currently the gold standard for diagnosing OSA11 . However, its use in screening high-risk individuals is limited by economic and time constraints. Recently, questionnaires have emerged as an alternative method for identifying those at higher risk of OSA, but their effectiveness is compromised by a lack of specificity, limiting their utility as a standalone diagnostic tool12. Similarly, home sleep studies offer a more accessible option, but they are hindered by a relatively high false-negative rate13,14. This underscores the urgent need for a more convenient and cost-effective approach to managing OSA clinically.

CHI3L1 (40 kDa), also known as YKL-40 or human cartilage glycoprotein 39, is a heparin- and chitin-binding glycoprotein15. You et al. reviewed the literature on YKL-40 and its role in diseases, summarizing that YKL-40 is produced by various cell types, including macrophages, neutrophils, fibroblast-like cells, hepatic stellate cells, endothelial cells, and cancer cells16. Numerous studies have demonstrated that YKL-40 levels are significantly elevated in the circulation of patients suffering from a variety of diseases associated with infection, inflammation, and tissue remodeling16,17. Elevated YKL-40 levels have been implicated in vascular endothelial dysfunction and injury through their effects on cell migration and tissue remodeling18. Given the association between OSA and both inflammatory processes and endothelial dysfunction, these findings suggest that CHI3L1 could serve as a valuable biomarker for predicting OSA and possibly hypertension.

Supporting this hypothesis, Mutlu et al. observed that serum CHI3L1 levels were significantly higher in OSA patients compared to controls and were positively correlated with AHI and oxygen desaturation19. Additionally, Peter et al. found elevated levels of CHI3L1 IgA and secretory IgA (sIgA) in the serum of patients with Crohn’s disease (CD), suggesting that CHI3L1 could be a novel biomarker for CD, marking the first instance of CHI3L1 being identified as an autoantigenic target in CD20.

Given the amplification effect of humoral immunity, autoantibodies are generally more abundant and stable than their corresponding antigens21. Despite this, limited research has explored whether well-studied proteins associated with OSA can act as autoantigens, potentially triggering the production of autoantibodies or even replacing these proteins as markers for OSA. Our previous studies have shown that a combination of autoantibodies against C-reactive protein (CRP), tumor necrosis factor-α (TNF-α), IL-8, and IL-6 in peripheral blood could serve as effective biomarkers for OSA22. Building on these findings, the current study aims to investigate the expression levels of autoantibodies against CHI3L1 (CHI3L1-Ab) in OSA patients, analyze their relationship with clinical characteristics, and ultimately develop a logistic regression model that incorporates CHI3L1-Ab, age and body mass index (BMI) to identify high-risk OSA patients.

Methods

Study population and serum collection

Between November 2018 and December 2021, a total of 366 individuals were recruited for this study, including 333 patients diagnosed with obstructive sleep apnea (OSA) and 33 healthy controls. All participants underwent polysomnography (PSG) at the Sleep Center of the First Affiliated Hospital of Zhengzhou University. All study procedures were approved by the Ethics Committee of Zhengzhou University. Written informed consent was obtained from all participants prior to enrollment, after they were fully informed of the study’s objectives, procedures, and potential risks. Exclusion criteria included the presence of peripheral vascular disease, liver disease, kidney disease, inflammatory conditions, acute infections, pregnancy, a history of treatment for OSA, or being under the age of 18.

A total of 416 patients were initially selected from our center. However, individuals with peripheral vascular disease, liver disease, kidney disease, inflammatory conditions, acute infections, pregnancy, a history of OSA therapy, or who were under 18 years old were excluded from the study. Additionally, patients with missing specimens were excluded. Ultimately, 333 patients with OSA were included, along with a control group of 33 individuals over 18 years old with an AHI of less than 5/h, recruited from the general population (Fig. 1).

Fig. 1
figure 1

Flowchart for Patient Screening.

Full size image

Basic demographic and clinical information, including sex, body mass index (BMI), smoking status, drinking status, and histories of hypertension, diabetes, cardiovascular disease, and cerebrovascular disease, was obtained from clinical records for all subjects.

Serum samples were collected by centrifuging blood at 3000 rpm for 5 min at room temperature. The supernatant was then stored at − 80 °C for subsequent analysis.

Polysomnography

Overnight PSG was conducted using equipment from SOMNOmedics GmbH (Randersacker, Germany) at the sleep center in the First Affiliated Hospital of Zhengzhou University. The SOMNOmedics GmbH sleep monitoring device features a comprehensive electrode configuration, including electroencephalography (EEG), electrooculography (EOG), electromyography (EMG), electrocardiography (ECG), respiratory sensors, pulse oximetry, snore sensors, and body position sensors, to provide detailed and accurate assessment of sleep patterns and disorders. The PSG was scored according to the guidelines provided by the American Academy of Sleep Medicine23. Apnea was defined as a reduction in nasal airflow of more than 90% lasting over 10 s, while hypopnea was characterized by a 30% or greater reduction in breathing amplitude for more than 10 s, accompanied by a decrease in oxygen saturation of more than 3%. The apnea–hypopnea index (AHI) was used to measure the average number of apneas and hypopneas per hour during the recording. OSA was defined as an AHI of five or more events per hour, with predominantly obstructive respiratory events. Additionally, the lowest pulse oxygen saturation (LSpO2), oxygen desaturation index (ODI), time spent with oxygen saturation below 90% (T90), and total sleep time (TST) were recorded. OSA severity was classified as mild (AHI 5–15 events/hour), moderate (AHI > 15–30 events/hour), and severe (AHI > 30 events/hour).

Indirect enzyme-linked immunosorbent assay (ELISA)

Purified recombinant CHI3L1 was purchased from CLOUD-CLONE (Wuhan, China). An indirect ELISA was conducted to measure the expression level of antibodies against CHI3L1. The recombinant CHI3L1 protein was diluted in coating buffer to a concentration of 0.125 μg/mL, following the procedure described in a previous study22. All samples were randomly placed in duplicates on the plates. The optical density (OD) values of the autoantibody were measured at 450 nm and 620 nm, with the OD value of the autoantibody calculated as the difference between the OD readings at 450 nm and 620 nm. Blank controls and quality controls were included on each plate to ensure experimental stability and accuracy.

Statistical analysis

All data analyses were performed using R v4.2.3 and GraphPad Prism 9.5. Depending on the data distribution, differences between groups were assessed using either the independent t-test or the Mann–Whitney U test. Group differences were analyzed using the Chi-squared test for large sample sizes and Fisher’s exact test for small sample sizes. The OD value of CHI3L1-Ab was normalized using the z-score normalization method for use in logistic regression. Univariate (stepwise) logistic regression and receiver operating characteristic (ROC) curve analysis were employed to screen candidate variables and evaluate their predictive capability, respectively. Multivariate logistic regression analysis was conducted to identify the optimal model for screening high-risk OSA patients. All statistical tests were two-sided, with statistical significance defined as P < 0.05.

Results

Clinical characteristics

In terms of OSA severity, 333 patients were divided into three subgroups: mild, moderate, and severe OSA. Detailed information for all participants is presented in Table 1. The OSA group was older than the control group (P < 0.001), and there was a significant difference in BMI between OSA patients and the control group. As shown in Fig. 2A, OSA patients exhibited significantly higher CHI3L1-Ab levels compared to healthy controls (P< 0.05), while no significant differences were observed among mild, moderate, and severe OSA groups. Moreover, as shown in Fig. 2B, the AUC for distinguishing OSA from the control group was 0.721 (0.626–0.816). Severe OSA patients showed the highest AUC (95% CI 0.736 [0.637–0.832]) compared to the overall OSA group and other subgroups, highlighting its efficacy in discriminating severe OSA.

Table 1 Baseline characteristics of all participants.
Full size table
Fig. 2
figure 2

The distribution of CHI3L1-Ab and its diagnostic capability for OSA. (A) The level (OD value) of CHI3L1-Ab in control and different OSA subgroup. (B) The ROC curves of CHI3L1-Ab in different OSA subgroup. AUC area under the curve, 95%CI 95% confidence interval.

Full size image

The association between CHI3L1-Ab levels and clinical factors in OSA patients

Chi-squared test and Fisher’s exact test were used to evaluate the association between CHI3L1-Ab levels and categorical clinical characteristics. As presented in Table 2, no statistically significant associations were observed between CHI3L1-Ab levels and clinical variables including age, sex, BMI, smoking status, alcohol consumption, hypertension, diabetes mellitus, coronary artery disease, or cerebrovascular disease (all P > 0.05).

Table 2 Association between CHI3L1-Ab levels and clinical factors in OSA patients.
Full size table

The diagnostic capability of CHI3L1-Ab and clinical factors

Logistic regression was used to develop a diagnostic model for OSA, incorporating CHI3L1-Ab and other clinical factors such as age and BMI. Figure 4 illustrates that the model achieved an AUC (95% CI) of 0.846 (0.778–0.914) in distinguishing OSA from controls, with a sensitivity of 73.0% and a specificity of 84.8%. The AUC increased to 0.878 (95% CI 0.815–0.942) for distinguishing severe OSA from controls, the highest among comparisons with mild and moderate OSA versus controls. As displayed in Table 3, there was a significant association between elevated CHI3L1-Ab levels (per SD increment) and the prevalence of OSA (odds ratio [OR]: 2.49, 95% CI 1.64–3.79, P < 0.001).

Table 3 The logistic regression analysis of OSA predictions.
Full size table

A nomogram was constructed based on the multivariate logistic regression model incorporating age, CHI3L1 autoantibody levels, and BMI as predictors. Using the regplot package in R, we visualized the contribution of each variable to the total risk score, which was then translated into an estimated probability of OSA.

As shown in Fig. 3, each predictor contributes a certain number of points, which are summed to generate a total score. This score corresponds to a predicted probability of OSA on a logistic scale. For example, a representative patient with an age of 45 years, CHI3L1-Ab level of 0.3 OD, and BMI of 25 kg/m2 had a total score of approximately 94.7, corresponding to a predicted probability of 0.84.

Fig. 3
figure 3

Nomogram for predicting the probability of obstructive sleep apnea (OSA).

Full size image

Discussion

Obstructive sleep apnea (OSA) is a prevalent yet vastly underdiagnosed condition, despite the American Academy of Sleep Medicine (AASM) recommending polysomnography (PSG) as the gold standard for diagnosis. Studies have shown that nearly 90% of OSA patients remain undiagnosed and untreated across both developing and developed countries24,25,26. This highlights the critical need for reliable, non-invasive biomarkers to improve early detection, risk stratification, and treatment monitoring. Over the past decade, research efforts have increasingly focused on identifying biomarkers for OSA, and among these, CHI3L1-Ab has shown significant potential for predicting disease susceptibility, aiding diagnosis, evaluating risk, prognosis, and monitoring treatment response27.

In our study, we observed that CHI3L1-Ab levels were significantly elevated in OSA patients compared to healthy controls. These findings suggest that CHI3L1-Ab could be a useful biomarker for identifying individuals at risk for OSA, as well as for determining disease severity. Moreover, when CHI3L1-Ab levels were combined with other demographic and lifestyle factors—such as age and body mass index (BMI)—the biomarker’s predictive capacity was further enhanced, as demonstrated by multivariate logistic regression analysis.

The diagnostic accuracy of CHI3L1-Ab, measured by the area under the curve (AUC), was 0.721 (95% CI 0.626–0.816), indicating moderate discriminatory power between OSA patients and healthy individuals. While this finding is promising, it is consistent with the broader understanding that individual biomarkers may have limited diagnostic utility when used in isolation. However, when CHI3L1-Ab was incorporated into a logistic regression model alongside age and BMI, the model’s AUC increased to 0.846 (Fig. 4), significantly improving its predictive accuracy. This suggests that CHI3L1-Ab, when combined with other clinical factors, may offer a robust approach for predicting OSA and distinguishing affected individuals from healthy controls.

Fig. 4
figure 4

The ROC curve analysis of the model consisting of screening different OSA patients. AUC area under the curve, 95% CI 95% confidence interval

Full size image

Our previous research demonstrated that a logistic regression model incorporating multiple autoantibodies, such as CRP-Ab, IL-6-Ab, IL-8-Ab, and TNF-α-Ab, achieved a superior diagnostic performance with an AUC of 0.876 (95% CI 0.846–0.906)22. This highlights the value of using a panel of biomarkers in combination with demographic factors to enhance diagnostic precision. In this context, our current study builds on this approach by further validating the utility of CHI3L1-Ab in a multivariate model. To enhance clinical applicability, we further visualized the logistic regression model as a nomogram (Fig. 3), allowing for intuitive, point-based estimation of individual OSA risk. This visual tool can assist clinicians in stratifying patients based on probability rather than odds, thereby improving communication and decision-making in clinical settings. The incorporation of CHI3L1-Ab into this framework supports its potential as a practical biomarker for personalized OSA risk assessment. These results underscore the potential of integrating multiple biomarkers and demographic factors into a diagnostic framework for OSA, offering a more comprehensive tool for early detection, especially in populations at high risk for the disease.

Despite the promising results, our study has several limitations. One notable limitation is the imbalance in the sample size, with a relatively small number of healthy controls (n = 33) compared to the number of OSA patients (n = 333). This discrepancy may introduce statistical bias, potentially limiting the generalizability of the findings. Future studies with larger, more balanced cohorts are essential to validate these results and ensure broader applicability.

Furthermore, while we observed a clear association between elevated CHI3L1-Ab levels and OSA, the exact role of CHI3L1-Ab in the pathogenesis of OSA remains unclear. OSA is characterized by intermittent hypoxia, which induces a cascade of inflammatory and oxidative stress responses. Chronic hypoxia, a hallmark of OSA, leads to the release of pro-inflammatory cytokines, such as interleukins (ILs) and tumor necrosis factor-alpha (TNF-α), along with the production of reactive oxygen species (ROS)28,29. These processes contribute to tissue injury and immune dysregulation, which are known drivers of CHI3L1 expression in various cell types.

Although CHI3L1 is a native glycoprotein, its expression is markedly upregulated under conditions of active tissue injury. For instance, Bonneh-Barkay et al.30 demonstrated that CHI3L1 transcription was predominantly observed in reactive astrocytes, with high expression during the acute phase of neurological diseases and subsequent decline in chronic stages. This temporal expression pattern suggests that CHI3L1 is closely associated with active tissue remodeling and inflammation. In the context of OSA, the repetitive hypoxia-reoxygenation cycles characteristic of CIH result in persistent oxidative stress, immune activation, and tissue remodeling—hallmarks of chronic but actively injurious inflammation. Thus, although OSA is a chronic yet actively progressive disease, it maintains a biologically "acute-like" inflammatory microenvironment that may sustain CHI3L1 overexpression.

Moreover, Sforza et al.31 (2016) underscored that CIH in OSA triggers a multifaceted pathological response involving sympathetic overactivity, oxidative stress, and systemic inflammation, all of which can alter tissue architecture and protein expression. These changes may expose previously hidden or modified CHI3L1 epitopes, promoting immunogenicity in predisposed individuals. Although direct evidence of CHI3L1 immunogenicity in OSA is lacking, the convergence of persistent inflammation, oxidative damage, and aberrant protein expression provides a biologically plausible framework for the emergence of CHI3L1-specific autoantibodies in this setting.

To fully understand the implications of these findings, future research must explore the mechanistic pathways linking CHI3L1-Ab to the pathophysiology of OSA. Investigating how hypoxia-induced inflammation, oxidative stress, and immune dysregulation influence CHI3L1 expression and autoantibody production could shed light on new therapeutic targets aimed at mitigating the inflammatory and oxidative damage caused by OSA. Additionally, understanding these mechanisms will help clarify whether elevated CHI3L1-Ab levels are a byproduct of OSA’s pathological processes or if they play an active role in disease progression.

A further limitation of our study is the lack of direct measurement of circulating CHI3L1 protein levels. Evaluating both CHI3L1 and CHI3L1-Ab simultaneously would provide a more comprehensive view of the antigen–antibody relationship and the immunological context of OSA.

Moreover, the study population lacked ethnic diversity, being exclusively Chinese, which inherently limits the generalizability of our findings to broader populations. To address this, future investigations should incorporate samples from multiple medical centers spanning diverse geographic regions and ethnic backgrounds. Such efforts will be crucial in validating whether CHI3L1-Ab functions as a universally applicable biomarker for OSA or if its diagnostic utility varies across different demographic groups. Furthermore, the relatively lower body mass index (BMI) and higher sleep efficiency observed in our cohort may not fully represent populations with greater metabolic burdens or more severe sleep fragmentation, thereby restricting extrapolation to those groups. Additionally, sex-related differences in immune response and OSA prevalence underscore the importance of conducting sex-stratified analyses in future studies to better understand potential variations in CHI3L1-Ab expression and its clinical relevance. Collectively, expanding cohort diversity and addressing these factors will enhance the external validity and clinical applicability of CHI3L1-Ab as a biomarker for OSA.

In summary, our study demonstrated that OSA patients exhibit significantly higher levels of CHI3L1-Ab compared to healthy controls, and that CHI3L1-Ab, when combined with clinical and demographic factors, shows predictive accuracy for distinguishing OSA patients from healthy individuals. These findings suggest that CHI3L1-Ab could be a valuable biomarker for OSA, particularly when integrated into a multivariate diagnostic framework.

While the current findings highlight its diagnostic potential, the broader clinical relevance of CHI3L1-Ab—such as its prognostic value in predicting disease progression or cardiovascular complications—remains to be explored. As our study was cross-sectional and limited to baseline assessments, future prospective investigations are needed to determine whether CHI3L1-Ab levels are associated with long-term outcomes and could inform risk stratification or therapeutic decision-making in OSA management. Furthermore, research into the underlying mechanisms, particularly those related to hypoxia-induced inflammation and immune dysregulation, may yield important insights into disease pathophysiology and novel intervention targets.