The association between chemosensitivity and the 10-year risk of type 2 diabetes in male patients with obstructive sleep apnea Authors: Lixia Wang, Lu Dai, Xiaona Wang, Junwei Guo, Rong Huang, Yi Xiao Published: 2024 ABSTRACT AbstractPurposeObstructive sleep apnea (OSA) is associated with a variety of diseases, including type 2 diabetes (T2D). Chemosensitivity is an important component of the pathophysiological mechanisms of OSA, and it is not only elevated in patients with OSA but also in those with T2D. This study aimed to investigate the association between chemosensitivity and the risk of developing T2D in patients with OSA.MethodsA total of 135 male participants with OSA and without pre-existing T2D were enrolled in this study. Peripheral chemosensitivity was evaluated using the rebreathing test. Data on demographics, polysomnographic parameters, and clinical characteristics were collected. The QDiabetes-2018 risk calculator was employed to calculate the 10-year T2D risk. The association between peripheral chemosensitivity and 10-year T2D risk was examined using multivariate logistic regression.ResultsA total of 64 participants had moderate-to-high 10-year risk of T2D. In the fully adjusted model, participants situated within the second and fifth quantiles of peripheral chemosensitivity levels demonstrated a higher risk of developing T2D, with OR of 4.87 (95% CI, 1.22–19.43) and 5.26 (95% CI, 1.27–21.68) respectively. However, across varying levels of peripheral chemosensitivity, no significant difference in the 10-year T2D risk was observed among different severities of OSA.ConclusionHigher peripheral chemosensitivity was associated with an increased 10-year T2D risk, as calculated using a risk calculator based on clinical variables. For outcomes that reflect a moderate-to-high 10-year risk of T2D, the severity of OSA did not significantly affect the risk, irrespective of whether patients exhibited relatively low or high chemosensitivity.Supplementary InformationThe online version contains supplementary material available at 10.1007/s11325-024-03221-y.Keywords: Obstructive sleep apnea, Type 2 diabetes, Peripheral chemosensitivity, Hypercapnic ventilatory response Introduction Obstructive sleep apnea (OSA) is a common sleep-disordered breathing, affecting nearly 1 billion people worldwide. It is characterized by the repeated complete or partial collapse of the upper airway, resulting in bouts of combined hypercapnia and hypoxia [1]. As understanding of OSA deepens, emerging research has highlighted the role of non-anatomical pathophysiological mechanisms in its occurrence and development among most patients. These mechanisms include increased airway collapsibility, low arousal threshold (LAT), high loop gain (LG), and poor upper airway muscle recruitment responses [2, 3]. Notably, approximately 33.0–36.0% of OSA patients exhibit high LG, indicating unstable ventilatory control [4, 5]. LG describes stability of a system controlled by feedback loops that modulate output, including plant gain, controller gain, and the feedback delay between the controller and plant [6]. In the ventilatory control system, controller gain reflects chemoreflex sensitivity, and plant gain reflects the responsiveness of the respiratory system to blood gases per unit change [2]. The activity of chemoreceptor, also known as chemosensitivity, has been verified to play a significant role in the instability of respiratory motor output during sleep and contributes significantly to the major clinical problem of OSA [7]. Type 2 diabetes (T2D) is a prevalent and serious chronic disease that imposes a significant health-care burden [8]. Previous studies have shown an increased risk of T2D in individuals with OSA, even after adjusting for confounding factors [9–11]. Numerous studies have shown that hyperglycemia, hypoglycemia, or prediabetic conditions can lead to an increase in peripheral chemosensitivity, evidenced by measuring the activity of carotid bodies (CBs, the peripheral chemoreceptors) or by exposing participants to different levels of O2 and CO2 [3, 12–14]. However, other studies have shown that CBs do not directly perceive changes in blood glucose to increase peripheral chemosensitivity [15, 16]. Instead, changes in HbA1c (glycosylated hemoglobin), plasma fasting insulin levels, and homeostasis model assessment of insulin resistance (HOMA-IR) are factors that correlate with peripheral chemosensitivity [12, 15]. Therefore, peripheral chemosensitivity is closely related to glucose metabolism disorders and the development of diabetes. Nonetheless, current research on the potential impact of different chemosensitivity levels on the risk of developing T2D in the context of OSA remains limited. Up to now, several models have been developed to assess the risk of diabetes, such as the Finnish Diabetes Risk Score, the Australian T2D Risk Assessment Tool [17], the Atherosclerosis Risk in Communities [18], and the Diabetes Population Risk Tool [19]. Given the differences in genetic factors and diabetes risk factors between China and Western countries, diabetes risk assessment models tailored for Western populations may not be applicable to Chinese individuals. Therefore, we selected the QDiabetes-2018 risk calculator, which includes Chinese as one of the selectable ethnicities, to evaluate the T2D risk in our study participants [20]. After thorough validation, QDiabetes-2018 risk calculator has emerged as one of the most discriminative models across various countries. It stands out due to its incorporation of a comprehensive set of predictors, leveraging a wealth of information. Developed from a large and diverse sample, this model provides precise estimates of the associations between predictors and outcomes [21, 22]. This study aimed to investigate the association between peripheral chemosensitivity and the 10-year T2D risk in Chinese male OSA patients. We employed the QDiabetes-2018 risk calculator to evaluate the 10-year T2D risk and utilized the rebreathing test to measure peripheral chemosensitivity. Specifically, we hypothesized that higher peripheral chemosensitivity could be associated with an increased 10-year T2D risk. Methods Participants This study utilized data from the sleep center of Peking Union Medical College Hospital (PUMCH). Given the male predominance in clinical practice, our focus was exclusively on male OSA patients aged between 25 and 84 in this study. We excluded patients who: (1) had a rate of the forced expiratory volume in the first second (FEV1) to the forced vital capacity (FVC) < 0.7 in the lung function test (LFT); (2) had pre-existing diabetes; (3) had fasting blood glucose (FBG) levels of less than 2 mmol/L or more than 7 mmol/L on the morning of polysomnography (PSG) test or the following day; (4) had recently received treatment for OSA, including continuous positive airway pressure treatment; (5) had nervous system diseases or uncontrolled cardiovascular disease; (6) had a total sleep time (TST) < 4 h; (7) refused or were unable to perform the rebreathing test, or had incomplete and poor quality PSG data. This study adhered to the principles of the Declaration of Helsinki. Approval was obtained from the ethics committees of PUMCH (JS-3573), and all subjects were fully informed of the study requirements and provided written informed consent prior to enrolment. Overnight PSG and data collection All participants attended overnight PSGs (Embla N7000, Natus Medical Incorporated, Orlando, FL, USA) from 11 p.m. to 6 a.m. at our sleep center, monitored according to standard protocols. Monitoring indices included electroencephalogram, electrooculogram, chin electromyogram, electrocardiogram, body position, chest and abdomen movements, and SpO2. All PSG data were recorded and scored by experienced sleep technologists using the standard criteria recommended by the American Academy of Sleep Medicine [23]. The apnea-hypopnea index (AHI) was calculated as the total number of apnea and hypopnea episodes per hour of electroencephalographic sleep. OSA diagnosis was based on the International Classification of Sleep Disorders-Third Edition, with patients considered to have moderate-to-severe OSA when AHI ≥ 15/h [24]. We scored non-rapid eye movement sleep (N1, N2, N3), rapid eye movement sleep (REM) sleep, sleep efficiency, and TST. Sleep stages (N1, N2, N3, and REM) were expressed as percentages of TST. The oxygen desaturation index (ODI) was calculated by dividing the total number of ≥ 3% SpO2 drops by the total hours of sleep. Mean oxygen saturation by pulse oximetry (mSpO2), lowest oxygen saturation by pulse oximetry (LSpO2), and the percent of time spent with SpO2 below 90% (T90) were obtained. The arousal index was calculated as the total number of arousal episodes per hour of electroencephalographic sleep. The LAT score and the presence of LAT were based on previously reported methods [25]. Demographic data of each participant were collected, including age, body mass index (BMI), neck circumference (NC), and waist circumference (WC). Information on smoking and alcohol consumption was also recorded. All participants completed the Epworth Sleepiness Scale (ESS) questionnaire and Pittsburgh Sleep Quality Index (PSQI). Central obesity was defined as a WC of ≥ 85 cm for men in China [26]. A NC of ≥ 40 cm was found to be a powerful criterion for the stratification and prioritization of patients diagnosed with OSA [27]. Blood examinations, including serum lipid profiles, glycosylated hemoglobin, insulin, and blood glucose levels, were measured using standard procedures. According to the laboratory standards of PUMCH, insulin levels > 17.2 µU/mL and HbA1c levels > 6.3% were defined as elevated. The HOMA-IR was calculated by taking the product of fasting insulin and fasting glucose levels (expressed in millimoles per liter) and dividing the result by 22.5. The QDiabetes-2018 risk calculator, employed to determine the 10-year T2D risk, incorporates various factors such as age, ethnicity, BMI, smoking, family history of diabetes in a first-degree relative, cardiovascular disease, treated hypertension, and regular use of corticosteroids, atypical antipsychotics, and statins. It also considers conditions such as schizophrenia or bipolar affective disorder, and learning disability. For women, the presence of gestational diabetes and polycystic ovary syndrome are also considered. These factors form the basis of Model A. Additional models incorporate fasting blood glucose (Model B) or HbA1c (Model C). Model B was specifically utilized in this study, due to its superior discrimination and sensitivity characteristics [20]. A 10-year T2D risk of ≥ 10% was categorized as moderate-to-high risk [22]. Chemosensitivity and lung function test All participants underwent rebreathing test twice in the morning while in a fasted state, following Duffin’s method [28]. The procedure has been previously detailed [29]. They completed the hyperoxia test first, followed by the hypoxia test after a 30-minute rest period. While seated, participants breathed through a nose peg and mouthpiece connected to a closed one-way circuit with a 6 L plastic rebreathing bag containing a gas mixture of 6% CO2, 4%O2, and balanced nitrogen. To maintain end-expiratory oxygen partial pressure at 50 mmHg or 150 mmHg during the test, 100% O2 was fed into the rebreathing bag. The test lasted 3–5 min or was terminated if end-tidal carbon dioxide partial pressure (PETCO2) exceeded 60 mmHg or minute ventilation exceeded 100 L/min. Data collection was facilitated by a volume transducer and gas sampling port connected to the cardiopulmonary exercise cart (MasterScreen CPX; Jaeger, Hoechberg, Germany), which were connected to the mouthpiece and rebreathing bag. Breath-by-breath measurements were used to monitor expired gas concentrations and minute ventilation. The overall chemosensitivity was evaluated by relating PETCO2 to minute ventilation during the hypoxia test using linear regression. The central chemosensitivity was evaluated by relating PETCO2 to minute ventilation (VE) during the hyperoxia test using linear regression. Upon completion, an.xls file containing time, PETCO2, and VE data was exported. The slope of the VE-PETCO2 curve, calculated using linear regression, represented the corresponding chemosensitivity. The slope under hyperoxic conditions represented central chemosensitivity, while the slope under hypoxic conditions represented overall chemosensitivity. The hypercapnic ventilatory response (HCVR) used in this study, defined as the difference between overall and central chemosensitivity, representing peripheral chemosensitivity, expressed as kPa/L/min. After a 30-minutes rest period, all participants underwent the LFT using the MasterLab Body 6.0 device (Vyasis, Würzburg, Germany). The cardiopulmonary exercise cart was used to assess participants’ pulmonary function, with data on FEV1% predicted and FEV1/FVC ratio were collected. Statistical analysis Continuous variables were represented as mean ± standard deviation or median (P25, P75), depending on whether the data followed a normal or skewed distribution. Categorical variables were represented as number (percentage). The normality of the variable distribution was tested using the Kolmogorov-Smirnov test. Between-group comparisons were conducted using independent samples t-test, Mann–Whitney U-test, chi-squared test, or Fisher’s exact test, depending on the type and distribution of the data. HCVR quantiles were determined by the quantile binning method, which ranked the data and divided it into equal-sized intervals based on these ranks. This ensured that each quantile contained an approximately equal number of observations, reflecting the distribution of the data. As the group number increases from Q1 to Q5, the corresponding HCVR values also increase (Q5 > Q4 > Q3 > Q2 > Q1). The association between HCVR quantiles and moderate-to-high 10-year risk of T2D was assessed using multivariate logistic regression in three adjusted models. Variables included in the QDiabetes-2018 risk calculator were not included for adjustment. Model 1 adjusted for central obesity (WC ≥ 85 cm) and NC ≥ 40 cm; Model 2 adjusted for insulin levels > 17.2 µU/mL and HbA1c levels > 6.3%, in addition to all covariates in Model 1; Model 3 adjusted for LSpO2, mSpO2, ODI, and T90, along with all covariates in Model 2. SPSS software version 26 (IBM Corporation, Armonk, NY, United States) was used for all the statistical analyses. A two-sided P < 0.05 was considered to reflect statistical significance. Results Baseline characteristics A total of 213 male participants underwent overnight PSG and the rebreathing test, with 135 eligible participants included in the analysis (Fig. 1). The included sample had a median age of 42 years, a median BMI of 27.06 kg/m2, a median HCVR of 0.31 kPa/L/min, and a median AHI of 40.60 /h. Furthermore, 64 (47.41%) subjects were categorized as having a moderate-to-high 10-year risk of T2D according to the QDiabetes-2018 risk calculator. The characteristics of the entire dataset (Table 1) showed that those with a higher 10-year T2D risk tended to have lager WC and NC, higher insulin levels, and more severe nocturnal hypoxia, in addition to the factors included in the QDiabetes-2018 risk calculator. However, there were no significant differences in serum lipid profiles and sleep patterns, including REM, N1, N2, and N3. Flow chart of this study. Abbreviations: OSA, obstructive sleep apnea; T2D, type 2 diabetes Table 1Characteristics of OSA patients with different risk levels of T2DVariablesLow T2D risk (N = 71)Moderate-to-high T2D risk (N = 64) P Demographics Age, y39.00 (34.00, 46.00)48.50 (38.00, 52.75)< 0.001 History Cigarettes smoking, n (%)0.016No43 (60.56)23 (35.94)Former8 (11.27)13 (20.31)Current20 (28.17)28 (43.75)Alcohol, n (%)0.171No32 (45.07)21 (32.81)Former3 (4.23)7 (10.94)Current36 (50.70)36 (56.25)HTN, n (%)13 (18.31)26 (40.63)0.004 Symptoms and measurements BMI, kg/m225.95 (24.24, 27.77)28.23 (26.11, 31.86)< 0.001NC, cm40.00 (38.00, 41.00)41.00 (39.00, 43.00)0.002WC, cm96.00 (90.00, 100.00)101.00 (95.25, 109.75)< 0.001NC ≥ 40 cm, n (%)41 (57.75)47 (73.44)0.056WC ≥ 85 cm, n (%)63 (88.73)63 (98.44)0.024EES12.00 (8.50, 15.00)14.00 (9.00, 18.00)0.261PSQI7.00 (5.00, 9.00)8.00 (6.00, 9.00)0.16510-year risk of T2D, %3.90 (1.90, 5.80)36.15 (16.03, 62.40)< 0.001 Laboratory examination TC, mmol/L5.12 ± 1.075.03 ± 1.030.639TG, mmol/L1.65 (1.30, 2.51)1.80 (1.30, 2.65)0.663HDL-c, mmol/L1.08 ± 0.201.05 ± 0.0220.485LDL-c, mmol/L3.22 ± 0.963.23 ± 0.910.970FBG, mmol/L5.10 (4.90, 5.30)6.20 (5.70, 6.60)< 0.001Insulin, µIU/ml9.90 (6.75, 14.10)13.40 (9.00, 20.80)0.002HOMA-IR2.20 (1.55, 3.31)3.48 (2.29, 6.10)< 0.001HbA1c, %5.30 (5.00, 5.45)5.70 (5.40, 6.10)< 0.001Insulin levels > 17.2 µIU/ml, n (%)9 (12.68)24 (37.50)0.001HbA1c > 6.3%, n (%)2 (2.82)9 (14.06)0.004 Polysomnogram Moderate-to-severe OSA, n (%)57 (80.28)49 (76.56)0.599TST, min404.80 (364.50, 443.50)405.45 (352.88, 442.03)0.960AHI, /h40.20 (18.10, 58.90)42.30 (15.88, 66.28)0.245Obstructive apnea, /h30.20 (7.80, 46.60)20.90 (9.15, 58.63)0.721Central apnea, /h0.00 (0.00, 0.40)0.00 (0.00,0.30)0.934Mixed apnea, /h0.30 (0.00, 2.70)0.30 (0.00, 2.43)0.721Apnea, /h30.50 (10.50, 55.10)25.30 (9.35, 64.25)0.519Hypopnea, /h5.30 (2.10, 9.60)5.85 (2.95, 13.05)0.185N1, %7.00 (3.90, 13.55)9.00 (5.40, 14.20)0.219N2, %48.20 (38.90, 60.85)52.20 (40.65, 65.80)0.246N3, %24.50 (13.30, 32.05)17.20 (8.60, 29.75)0.160REM, %16.55 ± 6.0015.37 ± 6.810.287REM-AHI, /h45.30 (19.70, 55.45)48.80 (16.45, 63.75)0.616SE, %92.10 (81.75, 96.95)91.30 (80.90, 96.50)0.548mSpO2, %96.10 (94.70, 97.65)95.00 (92.70, 97.10)0.006LSpO2, %83.00 (77.00, 88.00)81.00 (71.00, 87.50)0.097T90, %0.60 (0.05, 3.10)3.20 (0.10, 13.65)0.020ODI, /h30.80 (11.70, 54.60)39.60 (14.90, 65.00)0.084Arousal index0.90 (0.57, 1.60)0.91 (0.53, 1.41)0.727LAT score1.00 (0.00, 2.00)1.00 (0.00, 2.00)0.883LAT, n (%)25 (35.21)22 (34.38)0.919Abbreviations: T2D, type 2 diabetes; OSA, obstructive sleep apnea; BMI, body mass index; NC, neck circumference; WC, waist circumference; HTN, hypertension; TC, total cholesterol; TG, triglyceride; LDL-c, low-density lipoprotein cholesterol; HDL-c, high-density lipoprotein; FBG, fasting blood glucose; HbA1c, glycosylated hemoglobin; HOMA-IR, homeostasis model assessment of insulin resistance; ESS, Epworth Sleepiness Scale; PSQI, Pittsburgh Sleep Quality Index; TST, total sleep time; AHI, apnea-hypopnea index; N1, non-rapid eye movement stage 1; N2, non-rapid eye movement stage 2; N3, non-rapid eye movement stage 3; REM, rapid eye movement; SE, sleep efficiency; mSpO2, mean values of peripheral blood oxygen saturation; LSpO2, lowest oxygen saturation by pulse oximetry; T90, percent of time spent with SpO2 below 90%; ODI, oxygen desaturation index; LAT, low arousal threshold Characteristics of OSA patients with different risk levels of T2D Abbreviations: T2D, type 2 diabetes; OSA, obstructive sleep apnea; BMI, body mass index; NC, neck circumference; WC, waist circumference; HTN, hypertension; TC, total cholesterol; TG, triglyceride; LDL-c, low-density lipoprotein cholesterol; HDL-c, high-density lipoprotein; FBG, fasting blood glucose; HbA1c, glycosylated hemoglobin; HOMA-IR, homeostasis model assessment of insulin resistance; ESS, Epworth Sleepiness Scale; PSQI, Pittsburgh Sleep Quality Index; TST, total sleep time; AHI, apnea-hypopnea index; N1, non-rapid eye movement stage 1; N2, non-rapid eye movement stage 2; N3, non-rapid eye movement stage 3; REM, rapid eye movement; SE, sleep efficiency; mSpO2, mean values of peripheral blood oxygen saturation; LSpO2, lowest oxygen saturation by pulse oximetry; T90, percent of time spent with SpO2 below 90%; ODI, oxygen desaturation index; LAT, low arousal threshold Association between peripheral chemosensitivity and 10-year T2D risk We evaluated the association between quantiles of HCVR and 10-year risk of T2D using multivariate logistic regression analysis, as shown in Table 2. To achieve optimal adjustment, we gradually incorporated variables with a p < 0.1 from Table 1 into the logistic regression model. Participants situated within the fifth quantile of HCVR demonstrated a heightened proportion of moderate-to-high risk for T2D over a 10-year span, with an unadjusted odds ratio (OR) of 3.40 (95% confidence interval [CI], 1.07–10.78). Participants situated within the second quantile also showed an increased risk, though it was less significant (P = 0.070) with an unadjusted OR of 2.83 (95% CI, 0.92–8.73). After gradually adjusting for NC ≥ 40 cm and WC ≥ 85 cm (Model 1), insulin levels > 17.2 µU/mL and HbA1c levels > 6.3% (Model 2), and variables related to nocturnal hypoxia (Model 3), the association between the second and fifth quantiles of HCVR and the moderate-to-high 10-year risk of T2D remained significant in Model 3 [OR, 4.87 (95% CI, 1.22–19.43); OR, 5.26 (95% CI, 1.27–21.68)]. Due to the lack of significant differences in AHI and the proportion of moderate-to-severe OSA participants among different 10-year T2D risk groups in Table 1, these factors were not included as confounding variables in Table 2. However, based on the clinical significance and previous studies showing that OSA severity was associated with the incidence of T2D [30, 31], we also attempted to include these two variables separately in the analysis. The association between HCVR and the 10-year risk of T2D, as previously described, remained significant (data not shown). Another logistical regression using the third quantile of HCVR as the reference group also showed that participants situated within the second and fifth groups were significantly associated with increased T2D risk after adjusting for all confounders, as shown in Table S1. Table 2Unadjusted and adjusted associations of quantiles of HCVR with moderate-to-high risk of T2D in OSA participantsHCVR quantilesModel 0OR (95%CI)Model 1OR (95%CI)Model 2OR (95%CI)Model 3OR (95%CI)Q1 (< 0.08 KPa/L/min)1111Q2 (0.08∼0.21 KPa/L/min)2.83 (0.92, 8.73)3.36 (1.05, 10.77) *4.58 (1.23, 17.07) *4.87 (1.22, 19.43) *Q3 (0.08∼0.37 KPa/L/min)1.29 (0.42, 4.04)1.18 (0.37, 3.73)1.83 (0.50, 6.67)1.48 (0.39, 5.64)Q4 (0.37∼0.59 KPa/L/min)1.38 (0.44, 4.32)1.59 (0.49, 5.15)2.37 (0.64, 8.87)2.25 (0.55, 9.11)Q5 (> 0.59 KPa/L/min)3.40 (1.07, 10.78) *3.56 (1.08, 11.73) *4.88 (1.28, 18.65) *5.26 (1.27, 21.68) **P < 0.05. Model 0 was unadjusted; Model 1: adjusted for NC ≥ 40 cm and WC ≥ 85 cm; Model 2: adjusted for HbA1c levels > 6.3% and insulin levels > 17.2 µU/mL, plus all covariates in Model 1; Model 3: adjusted for mSpO2, LSpO2, ODI, and T90, plus all covariates in Model 2Abbreviations: NC, neck circumference; WC, waist circumference; HCVR, hypercapnic ventilatory response; OSA, obstructive sleep apnea; T2D, type 2 diabetes; mSpO2, mean values of peripheral blood oxygen saturation; LSpO2, lowest values of peripheral blood oxygen saturation; ODI, oxygen desaturation index; T90, percent of time spent at peripheral blood oxygen saturation below 90% Unadjusted and adjusted associations of quantiles of HCVR with moderate-to-high risk of T2D in OSA participants *P < 0.05. Model 0 was unadjusted; Model 1: adjusted for NC ≥ 40 cm and WC ≥ 85 cm; Model 2: adjusted for HbA1c levels > 6.3% and insulin levels > 17.2 µU/mL, plus all covariates in Model 1; Model 3: adjusted for mSpO2, LSpO2, ODI, and T90, plus all covariates in Model 2 Abbreviations: NC, neck circumference; WC, waist circumference; HCVR, hypercapnic ventilatory response; OSA, obstructive sleep apnea; T2D, type 2 diabetes; mSpO2, mean values of peripheral blood oxygen saturation; LSpO2, lowest values of peripheral blood oxygen saturation; ODI, oxygen desaturation index; T90, percent of time spent at peripheral blood oxygen saturation below 90% Additionally, to understand the impact of different severities of OSA on the 10-year T2D risk under different HCVR levels, we used the receiver operating characteristics curve to determine the cutoff value of HCVR related to the outcome of moderate-to-high 10-year risk of T2D, which was 0.10 kPa/L/min. If participants’ HCVR was greater than 0.10 kPa/L/min, they were classified into the high HCVR group (N = 106); otherwise, they were classified into the low HCVR group (N = 29). We performed logistic regression analysis to assess the association between high HCVR group and low HCVR group with 10-year T2D risk (Table S2). It showed that an increased HCVR level was an independent risk factor for increased T2D risk [OR, 5.28 (95% CI, 1.69–19.61)], even after adjusting for all confounders in the adjusted Model 3 of Table 2. Association between the severity of OSA and HCVR with 10-year T2D risk Regarding the association between the severity of OSA and 10-year T2D risk in low and high HCVR groups, our results showed that, irrespective of HCVR level, there was no significant difference in the 10-year T2D risk among OSA groups of different severities (Low HCVR: 4.15% vs. 6.00%, P = 0.884; High HCVR: 14.80% vs. 10.20%, P = 0.934), as shown in Fig. 2. Comparison of 10-year T2D risk between participants with mild and moderate-to-severe OSA at different HCVR levels. In both low HCVR (N = 29) group and high HCVR group (N = 106), there was no significant difference in the 10-year T2D risk for patients with mild and moderate-to-severe OSA (Low HCVR: 4.15% vs. 6.00%, P = 0.884; High HCVR: 14.80% vs. 10.20%, P = 0.934). Abbreviations: OSA, obstructive sleep apnea; T2D, type 2 diabetes; HCVR, hypercapnic ventilatory response Moreover, we conducted a data analysis comparing the low HCVR group and high HCVR group without dividing the OSA groups by severity. This analysis revealed that the high HCVR group had a significantly higher proportion of participants with moderate-to-high T2D risk and more severe nocturnal hypoxia (Table S3). Discussion Our study demonstrated that higher levels of peripheral chemosensitivity were independently associated with an increased 10-year T2D risk. However, no significant association was observed between the severity of OSA and 10-year T2D risk, and this lack of association persisted across different levels of peripheral chemosensitivity. This indicated that the 10-year T2D risk remained consistent regardless of OSA severity. These findings suggested a complex interaction between peripheral chemosensitivity, OSA severity, and diabetes risk that warrants further investigation. It has been reported that increased peripheral chemosensitivity can initiate a series of adaptive responses to restore normoglycemia [32]. However, enhanced peripheral chemosensitivity may subsequently increase the risk of metabolic disorders, including T2D, once compensatory adjustments are lost [12, 33]. João and coworkers have suggested that peripheral chemosensitivity could serve as an early biomarker for dysmetabolism in humans [12]. As corroborated by the majority of studies [12, 15], an increase in peripheral chemosensitivity was closely linked to the changes of HbA1c, plasma fasting insulin levels, and consequently HOMA-IR, which were key factors in the onset of T2D. Therefore, a significant correlation existed between peripheral chemosensitivity and the incidence of T2D, aligning with the results of our study. Additionally, the positive association between peripheral chemosensitivity and the risk of T2D persisted after adjusting for high blood insulin levels in our study. This intricate relationship underscores the pivotal role of peripheral chemosensitivity in the pathogenesis of T2D and suggests that this relationship is not entirely influenced by insulin levels, further highlighting the importance of this physiological parameter in the predictive models for T2D risk assessment. However, participants situated within the second group were also significantly associated with an increased T2D risk, whereas those in the third and fourth groups showed no significant association (Table 2, Table S1). This suggested a non-linear relationship between chemosensitivity and T2D risk. Moreover, participants in the second and fifth quantiles had poorer basal glucose metabolism than other groups (Table S4). This implied that they may have different lifestyles, such as eating habits and exercise frequency, which may influence the risk of diabetes. For example, diets high in fat and sugar may increase the risk of T2D [34]. Additionally, certain genetic variants that may increase their diabetes risk cannot be excluded [35]. Thus, more prospective research with a larger sample size and more comprehensive variables is needed. The prevailing consensus in the medical community is that the presence of OSA significantly escalates the risk of developing T2D compared to those without OSA. Approximately 15–30% of patients with OSA are also diagnosed with T2D [10, 30]. Intermittent hypoxia and fragmented sleep can lead to increased sympathetic nervous activity, oxidative stress, systemic inflammation, lipolysis, and sympathetic and adrenocortical activation. These changes can result in β-cell dysfunction and insulin resistance, thereby contributing to the onset of T2D [10]. However, the influence of OSA severity on the risk of T2D has been subject to conflicting results in various studies. A few studies have reported a significant relationship between the increasing severity of OSA and the prevalence of diabetes [31, 36, 37], or that once the severity of OSA reaches a certain stage, the risk of T2D no longer increases in a dose-dependent manner [30]. Other studies indicated that, after adjusting for confounding factors, the incidence of T2D at the end of the follow-up period was not significantly associated with the severity of OSA at the time of initial enrollment [38, 39], which was consistent with our study. One possible explanation for this could be that the severity of OSA is typically measured by AHI, which may not fully reflect the impact of OSA on the risk of diabetes [40]. In fact, research has shown that, in addition to AHI, there are other indicators of OSA severity, such as hypoxia burden, arousal intensity, odds ratio product, and cardiopulmonary coupling, which may be more valuable in predicting the risk of diabetes [40]. Additionally, Amandine et al. [41] reported that chronic intermittent hypoxia could improve whole-body glucose tolerance independently of changes in body weight and insulin sensitivity, indicating metabolic adaptation of the body to chronic intermittent hypoxia [42]. Moreover, the contribution of OSA to T2D development appeared to be gender-dependent, with its impact on males being less significant [43]. Coincidentally, all participants in our study were male. The current study has some strengths and limitations. Among patients with OSA, increased peripheral chemosensitivity was positively associated with 10-year T2D risk, irrespective of the severity of OSA. Consequently, in clinical practice, evaluating peripheral chemosensitivity could be a crucial tool for early screening of T2D, regardless of OSA severity. This approach could facilitate the implementation of appropriate preventive measures and interventions, potentially mitigating the onset of diabetes. However, our study has several limitations. Primarily, this is a cross-sectional study conducted in a single center with a small sample size, and only male participants were enrolled. Moreover, there were few participants with mild OSA in this study. Secondly, for the exclusion of individuals with pre-existing T2D, we relied on self-reported information without clinical validation, which may introduce potential inaccuracies. Thirdly, the major outcome in this study was an indirect metric calculated using the QDiabetes-2018 risk calculator, which has not been validated in large-scale studies in the Chinese population or in populations with OSA. Additionally, this calculator’s minimum age limit of 25 could reduce its testing efficacy, posing a limitation given the trend of diabetes affecting younger populations. Conclusion In patients with OSA, elevaed peripheral chemosensitivity is associated with an increased 10-year T2D risk, as calculated using a risk calculator based on clinical variables, independent of OSA severity. Therefore, regardless of the severity of OSA, assessing peripheral chemosensitivity in the OSA population can serve as a valuable tool for early identification of individuals subtly developing metabolic abnormalities. This approach could facilitate the implementation of early and specific regulatory treatments, potentially mitigating the onset of metabolic disorders. This underscores the importance of comprehensive patient assessment and personalized treatment strategies in managing complex health conditions such as OSA and T2D. Electronic supplementary material Below is the link to the electronic supplementary material. Supplementary Material 1 Acknowledgements The authors would like to thank Yang Zhao, Hairong Zhang, and Lijuan Fan for their help in collecting data. Author contributions Lixia Wang: Data curation, Formal analysis, Writing – original draft, Writing – review & editing; Lu Dai: Data curation; Xiaona Wang: Data curation, Resources; Rong Huang: Methodology, Supervision; Yi Xiao: Conceptualization, Supervision, Writing – review & editing. Funding This study was funded by the National Natural Science Foundation of China (No. 82370052). Data availability The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request. Declarations Ethical approval The study was approved by the ethics committees of Peking Union Medical College (JS-3573) and was conducted in accordance with the Declaration of Helsinki. Informed consent Written informed consent was obtained from all participants. Conflict of interest Dr Xiaona Wang reports a patent for a device for measuring ventilatory control based on hypercapnic ventilatory response issued to ZL 2021 2 1127108.9. There were no financial or other relationships might lead to a conflict of interest in the present article. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1 Data Availability Statement The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.