High sensitivity C‐reactive protein and prediabetes progression and regression in middle‐aged and older adults: A prospective cohort study

ABSTRACT

Background

This study aimed to investigate the effect of systemic inflammation, assessed by high sensitivity C‐reactive protein (hs‐CRP) levels, on prediabetes progression and regression in middle‐aged and older adults based on the China Health and Retirement Longitudinal Study (CHARLS).

Methods

Participants with prediabetes from CHARLS were followed up 4 years later with blood samples collected for measuring fasting plasma glucose (FPG) and hemoglobin A1c (HbA1c). The level of hs‐CRP was assessed at baseline and categorized into tertiles (low, middle, and high groups). Prediabetes at baseline and follow‐up was defined primarily according to the American Diabetes Association (ADA) criteria. Logistic regression models were used to estimate the odds ratios (ORs) and confidence intervals (CIs). We also performed stratified analyses according to age, gender, BMI, the presence of hypertension, and the disease history of heart disease and dyslipidemia and sensitivity analyses excluding a subset of participants with incomplete data.

Results

Of the 2,874 prediabetes included at baseline, 834 participants remained as having prediabetes, 146 progressed to diabetes, and 1,894 regressed to normoglycemia based on ADA criteria with a 4 year follow‐up. After multivariate logistics regression analysis, prediabetes with middle (0.67–1.62 mg/L) and high (>1.62 mg/L) hs‐CRP levels had an increased incidence of progressing to diabetes compared with prediabetes with low hs‐CRP levels (<0.67 mg/L; OR = 1.846, 95%CI: 1.129–3.018; and OR = 1.632, 95%CI: 0.985–2.703, respectively), and the incidence of regressing to normoglycemia decreased (OR = 0.793, 95%CI: 0.645–0.975; and OR = 0.769, 95%CI: 0.623–0.978, respectively). Stratified analyses and sensitivity analyses showed consistent results.

Conclusions

Low levels of hs‐CRP are associated with a high incidence of regression from prediabetes to normoglycemia and reduced odds of progression to diabetes.

This study found that low levels of hs‐CRP are associated with a high incidence of regression from prediabetes to normoglycemia and reduced odds of progression to diabetes. Subgroup analyses showed a stronger association between hs‐CRP and type 2 diabetes in women than in men. Thus, prediabetes may need to be closely monitored in middle‐aged and older women with elevated hs‐CRP. In addition, whether a lower cutoff value for hs‐CRP elevation in women should be proposed is a question worthy of future research.

INTRODUCTION

With high prevalence and associated disability and mortality, type 2 diabetes (T2D) has become a major health problem worldwide ¹ . Prediabetes, or intermediate hyperglycemia, represents a high‐risk state for developing type 2 diabetes, and 5–10% of those progress to diabetes annually ² . However, numerous studies have shown that some prediabetic individuals experience an improvement in glucose tolerance to normoglycemia. In addition to medication‐based therapies and mild lifestyle adjustments, a number of clinical factors, including lipid levels and the fatty liver index, were associated with the transition from prediabetes to normal glucose tolerance. Therefore, identifying factors that may be changed is crucial for stopping the advancement of prediabetes or for encouraging its reversal.

Chronic inflammation has been associated with metabolic diseases such as type 2 diabetes by increasing systemic plasma concentrations of cytokines. CRP is primarily produced by the liver and mature adipocytes and is used as an inflammatory index to measure tissue damage, infection, and inflammation in patients ³ . Previous cohort studies based on a Thai population have reported that high baseline CRP levels may increase the risk of type 2 diabetes ⁴ , and similar results were obtained in a cohort of Japanese and Chinese people ⁵ , ⁶ . Moreover, researchers also found that the trajectory of hs‐CRP was associated with incident diabetes in Chinese adults ⁷ . Another prospective study observed that women with high serum CRP concentrations may have an accelerated progression of diabetes ⁸ . A cross‐sectional study showed a positive association between CRP and glucose intolerance in prediabetes ⁹ . The above evidence demonstrated that CRP, an inflammatory cytokine, plays a non‐negligible role in the progression of diabetes ¹⁰ , ¹¹ . However, it remains unclear whether CRP at baseline plays any role in prediabetes regression or progression.

To test whether plasma CRP can be used to identify high‐risk prediabetes and then develop optional preventive strategies, we employed a large‐scale national cohort, China Health and Retirement Longitudinal Study (CHARLS), to investigate the relationship between hs‐CRP and prediabetes progression and regression in middle‐aged and older Chinese population.

METHODS

Study population

CHARLS is a longitudinal survey of people in China who are 45 years of age or older that is nationally representative and includes evaluations of the social, economic, and health conditions of local residents. 17,708 people participated in the study's national baseline survey, which was carried out between June 2011 and March 2012. Every 2 years, physical measurements are taken, and blood samples are taken once every two follow‐up periods. The description and questionnaire of the CHARLS have been described elsewhere ¹² , ¹³ . In this study, data from the 2011–2012 (baseline) and 2015 waves of CHARLS were used, where blood samples were collected. This study was conducted in line with the Declaration of Helsinki, and the protocol of CHARLS was approved by the Ethical Review Committee of Peking University (IRB00001052–11015). All participants provided written informed consent. Of the 6,740 participants followed up at baseline in the 2011 wave, excluded criteria included: (1) missing data on glucose or hemoglobin A1c at baseline (N = 163); (2) confirmed with diabetes (FPG ≥126 mg/dL, and/or HbA1c ≥ 6.5%, and/or have self‐reported history of diabetic disease, and/or use anti‐diabetic medications, and/or random plasma glucose ≥11.1 mmol/L; N = 924); (3) normoglycemia (FPG < 100 mg/dL and HbA1c < 5.7% and random plasma glucose <7.8 mmol/L; N = 2,769); (4) missing FPG or HbA1c measurements in the 2015 wave (N = 10). Finally, 2,874 prediabetes participants as defined by ADA criteria were included for the present analysis (Figure 1).

Figure 1.

Study flow chart, Of the 6740 participants followed up at baseline in the 2011 wave, the excluded criteria were: (1) missing data on glucose or hemoglobin A1c at baseline (N = 163); (2) confirmed with diabetes (FPG ≥126 mg/dL, and/or HbA1c ≥ 6.5%, and/or have self‐reported history of diabetic disease, and/or use anti‐diabetic medications, and/or random plasma glucose ≥11.1 mmol/L; N = 924); (3) normoglycemia (FPG < 100 mg/dL and HbA1c < 5.7% and random plasma glucose < 7.8 mmol/L; N = 2769); (4) missing FPG or HbA1c measurements in the 2015 wave (N = 10). Finally, 2,874 prediabetes participants as defined by ADA criteria were included for the present analysis.

Assessment of hs‐CRP and data collection

Hs‐CRP data were obtained from the blood sample data, and grouped by the quantile and divided into the following three levels: low‐level hs‐CRP (Tertile 1, <0.67 mg/L), medium level (Tertile 2, 0.67–1.62 mg/L), and high level (Tertile 3, >1.62 mg/L). Demographic information (e.g., age, race, and gender), lifestyle (e.g., smoking and drinking), and medical history (e.g., hypertension, dyslipidemia, diabetes, heart disease, tumor, stroke, and other medication use) were recorded. Anthropometric parameters including body weight, height, and waist circumference (WC) were measured. BMI (kg/m²) was calculated as body weight/(height²). According to the Chinese standard, BMI is classified as normal (BMI <24 kg/m²), overweight (BMI 24–27.9 kg/m²), and obese (BMI ≥28 kg/m²) ¹⁴ . A Body Shape Index (ABSI) reflecting the visceral fat level ¹⁵ , was calculated as WC/(BMI^(2/3))/(height^(1/2)). Systolic blood pressure (SBP) and diastolic blood pressure (DBP) were measured three times in the resting state for each participant. Hypertension was defined as self‐report of physician‐diagnosed hypertension, and/or mean systolic blood pressure (SBP) ≥140 mmHg, and/or mean diastolic blood pressure (DBP) ≥90 mmHg, and/or on anti‐hypertensive drugs ¹⁶ . Fasting blood samples in each wave were collected and FPG, HbA1c, total cholesterol (TC), triglycerides (TG), high‐density lipoprotein‐cholesterol (HDL‐c), low‐density lipoprotein‐cholesterol (LDL‐c), and uric acid (UA) were measured. However, a small proportion of blood samples were not fasted, and their glucose was considered as a random plasma glucose (RPG).

Ascertainment of prediabetes, diabetes, and normoglycemia

In this study, the classifications of prediabetes, diabetes, and normoglycemia were based primarily on the ADA criteria ¹⁷ : prediabetes was defined as FPG 100 mg/dL (5.6 mmol/L) to 125 mg/dL (6.9 mmol/L) or HbA1c 5.7–6.4% (39–47 mmol/mol); diabetes was as FPG ≥126 mg/dL (7.0 mmol/L), and/or HbA1c ≥ 6.5% (48 mmol/mol), and/or self‐reported history, and/or the use of anti‐diabetic medications; and normoglycemia was as FPG < 100 mg/dL and HbA1c < 5.7%. Moreover, for participants with RPG (random plasma glucose), they were considered as having diabetes if the RPG was ≥11.1 mmol/L, and as having normoglycemia if the RPG was <7.8 mmol/L.

Statistical analysis

For continuous variables, data were presented as mean ± SD. Categorical data were presented as number (percentage). For comparisons of characteristics across different hs‐CRP groups, one‐way analysis of variance was performed for continuous variables and χ² tests were applied for categorical variables. Participants were stratified into three groups according to follow‐up outcomes: (i) progression to diabetes, (ii) regression to normoglycemia, and (iii) remaining as prediabetes. Multinomial logistic regression analysis was conducted to obtain the odds ratios (ORs) and 95% confidence intervals (CIs) for the association of hs‐CRP in tertiles with progression to diabetes or regression to normoglycemia. Three different models were introduced: Model 1, without adjustment; Model 2, adjusted for age and sex; and Model 3, additionally adjusted for BMI, presence of hypertension and obesity, disease history of dyslipidemia, heart disease, cancer or malignant tumor and stroke, SBP, DBP, triglycerides, and total cholesterol. Stratified analyses according to gender, age, BMI, the presence of hypertension, and the disease history of heart disease and dyslipidemia were conducted mainly using Model 3. Trend tests were performed to assess whether a dose–response association was present. Sensitivity analysis was conducted to verify whether the absence of relevant covariates affected the results. All statistical analyses were conducted using SPSS version 26.0 (IBM Corporation, Armonk, NY), and P < 0.05 was considered statistically significant.

RESULTS

Baseline characteristics

There were 2,874 participants included with prediabetes (mean age 64.09 ± 9.06 years, 44.7% males) based on the ADA criteria. Participants were divided into three groups according to hs‐CRP levels, of which 969 people were at low hs‐CRP levels, 951 people were in middle hs‐CRP levels, and 954 people were at high hs‐CRP levels. Their baseline characteristics are shown in Table 1. Compared with the low hs‐CRP group, there were significant differences in age, disease history, BMI, ABSI, SBP, DBP, TC, TG, HDL‐C, LDL‐C, UA, FPG, HbA1c, hs‐CRP, and obesity between the medium and/or high hs‐CRP groups (all P < 0.05). Moreover, a trend test showed that except for gender, drinking, tumor and stroke, the other indicators were a dose–response association with the CRP increase (P _trend <0.05).

Table 1.

Baseline characteristics of participants stratified by hs‐CRP level

	Total	Low CRP (Tertile 1 < 0.67)	Middle CRP (Tertile 2 0.67–1.62)	High CRP (Tertile 3 > 1.62)	P value	P for trend
Sample size (n)	2,874	969	951	954
Age (years)	64.090 ± 9.064	63.006 ± 9.180	64.083 ± 8.663*	65.197 ± 9.204*	<0.001	<0.001
Male (%)	1,285 (44.7%)	428 (44.2%)	423(44.4%)	434(45.5%)	0.834	0.560
Smoking (%)	1,059 (36.8%)	336 (34.7%)	343(36.1%)	380(39.8%)	0.055	0.020
Drinking (%)	402 (14.0%)	131 (13.5%)	132(13.9%)	139(14.6%)	0.941	0.781
Disease history
Hypertension	766 (26.7%)	186 (19.2%)	250 (26.3%)*	330 (34.6%)*	<0.001	<0.001
Dyslipidemia	288 (10.0%)	70 (7.2%)	98 (10.3%)*	120 (12.6%)*	<0.001	<0.001
Heart problems	362 (12.6%)	101 (10.4%)	123 (12.9%)	138 (14.5%)*	0.023	0.007
Tumor	26 (0.9%)	6 (0.6%)	6 (0.6%)	14 (1.5%)	0.083	0.051
Stroke	54 (1.9%)	14 (1.5%)	19 (2.0%)	21 (2.2%)	0.448	0.22
BMI (kg/m2)	24.134 ± 4.208	23.114 ± 4.075	24.388 ± 3.967*	24.913 ± 4.370*	<0.001	<0.001
ABSI	0.081 ± 0.013	0.080 ± 0.017	0.082 ± 0.010*	0.082 ± 0.012*	<0.001	<0.001
SBP (mmHg)	131.059 ± 21.021	128.746 ± 20.445	130.438 ± 20.343	134.045 ± 21.920*	<0.001	<0.001
DBP (mmHg)	76.216 ± 11.948	75.102 ± 11.838	76.024 ± 11.919	77.547 ± 11.973*	<0.001	<0.001
TC (mg/dL)	196.389 ± 38.210	193.401 ± 37.778	198.253 ± 37.419*	197.565 ± 39.270	0.005	0.016
TG (mg/dL)	135.127 ± 87.020	118.813 ± 76.457	140.575 ± 89.725*	146.267 ± 91.876*	<0.001	<0.001
HDL‐c (mg/dL)	51.083 ± 15.433	54.984 ± 15.757	50.125 ± 14.606*	48.075 ± 15.092*	<0.001	<0.001
LDL‐c (mg/dL)	118.713 ± 35.579	115.313 ± 33.403	120.997 ± 35.769*	119.888 ± 37.270*	<0.001	0.005
UA (mg/dL)	4.459 ± 1.239	4.178 ± 1.150	4.445 ± 1.162*	4.757 ± 1.329*	<0.001	<0.001
FPG (mg/dL)	108.657 ± 6.330	108.172 ± 6.130	108.717 ± 6.285	109.090 ± 6.544*	0.013	0.001
HbA1c (%)	5.189 ± 0.459	5.150 ± 0.476	5.200 ± 0.440*	5.218 ± 0.457*	0.002	0.001
HbA1c (mol/mmol)	33.213 ± 5.013	32.787 ± 5.207	33.335 ± 4.805*	33.525 ± 4.992*	0.002	0.001
Obese (%)	397 (13.9%)	66 (6.9%)	125 (13.3%) *	206 (21.7%) *	<0.001	<0.001
hs‐CRP (mg/L)	2.582 ± 7.436	0.447 ± 0.138	1.055 ± 0.265*	6.272 ± 12.083*	<0.001	<0.001

They were compared using one‐way analysis of variance or χ² test, as appropriate, with Bonferroni's correction for multiple comparisons. ABSI, A Body Shape Index; BMI, body mass index; DBP, diastolic blood pressure; FPG, fasting plasma glucose; HbA1c, hemoglobin A1c; HDL‐c, high‐density lipoprotein‐cholesterol; hs‐CRP, high‐sensitivity C‐reactive protein; LDL‐c, low‐density lipoprotein‐cholesterol; SBP, systolic blood pressure; TC, total cholesterol; TG, triglycerides; UA, uric acid.

^* P < 0.05, compared with low hs‐CRP.

Hs‐CRP and prediabetes progression and regression

The association of hs‐CRP with prediabetes progression or regression is shown in Table 2. Based on the ADA criteria, using low levels of hs‐CRP as a reference, participants with high levels of hs‐CRP had an increased odds of eventually progressing to diabetes (OR = 2.14, 95%CI: 1.35–3.39) and a decreased odds of regression to normoglycemia (OR = 0.68, 95%CI: 0.56–0.82) in the unadjusted model (Model 1). Trend tests also showed a significant dose–response association. Similar outcomes were obtained after adjusting for age and sex (Model 2). In the adjusted multivariable model (Model 3), the association of hs‐CRP with the odds of regression to normoglycemia remained significant (OR = 0.77, 95%CI: 0.62–0.98, P _trend = 0.015; Table 2). However, we only observed that participants with middle (OR = 1.85, 95%CI: 1.13–3.02) but not high levels of hs‐CRP exhibited an increased risk of progressing to diabetes, when compared with participants with low levels of hs‐CRP.

Table 2.

Hs‐CRP and prediabetes regression and progression

	No. of cases/total	Model 1 †	Model 2 ‡	Model 3 §
		OR (95%CI)	OR (95%CI)	OR (95%CI)
Prediabetes progression
Low (Tertile 1 <0.67)	28/969	1	1	1
Middle (Tertile 2 0.67–1.62)	61/951	2.303 (1.459–3.637)	2.196 (1.374–3.509)	1.846 (1.129–3.018)
High (Tertile 3 >1.62)	57/954	2.136 (1.346–3.388)	2.131 (1.332–3.410)	1.632 (0.985–2.703)
P for trend		0.002	0.003	0.083
Prediabetes regression
Low (Tertile 1 <0.67)	689/969	1	1	1
Middle (Tertile 2 0.67–1.62)	608/951	0.720(0.595–0.873)	0.741(0.610–0.900)	0.793(0.645–0.975)
High (Tertile 3 >1.62)	597/954	0.680(0.561–0.823)	0.695(0.572–0.843)	0.769(0.623–0.978)
P for trend		<0.001	<0.001	0.015

^† Unadjusted.

^‡ Adjusted for age and sex.

^§ Adjusted for age, sex, body mass index, presence of hypertension and obesity, disease history of dyslipidemia, heart disease, cancer or malignant tumor, and stroke, systolic blood pressure, diastolic blood pressure, triglycerides, total cholesterol.

Stratified and sensitivity analysis

Sensitivity analysis after excluding participants with incomplete data showed that it did not affect the primary outcome (Table 3). To assess the effects of age, sex, BMI, hypertension, dyslipidemia, and heart diseases on the association of hs‐CRP with prediabetes regression and progression, we calculated a separate P value for interactions and conducted stratified analysis according to age (<60 vs ≥60 years), sex, BMI (<24 vs 24–27.9 vs ≥28 kg/m²), diagnosis of hypertension, dyslipidemia, history of heart diseases at baseline. We only observed interaction between hs‐CRP and sex on prediabetes regression (P for interaction = 0.035). We further performed stratified analyses to examine the association of tertiles of hs‐CRP with prediabetes regression and progression. Similar trends were obtained in that the higher the hs‐CRP levels in participants with prediabetes, the greater their risk of progression to diabetes and the less likely they were to regress to normoglycemia, although several associations were no longer evident or statistically significant (Table 4).

Table 3.

Sensitivity analysis of hs‐CRP and diabetes progression and regression

	No. of cases/total	Model 1 †	Model 2 ‡	Model 3 §
		OR (95%CI)	OR (95%CI)	OR (95%CI)
Excluding participants with incomplete data ¶
Prediabetes progression
Low (Tertile 1 <0.67)	26/903	1	1	1
Middle (Tertile 2 0.67–1.62)	50/870	2.057 (1.268–3.335)	2.044 (1.260–3.316)	1.846 (1.129–3.018)
High (Tertile 3 >1.62)	49/881	1.987 (1.223–3.226)	1.950 (1.198–3.174)	1.632 (0.985–2.703)
P for trend		0.008	0.010	0.083
Prediabetes regression
Low (Tertile 1 <0.67)	653/903	1	1	1
Middle (Tertile 2 0.67–1.62)	564/870	0.706 (0.577–0.863)	0.713 (0.582–0.872)	0.793 (0.645–0.975)
High (Tertile 3 >1.62)	639/881	0.636 (0.521–0.777)	0.658 (0.538–0.805)	0.769 (0.623–0.948)
P for trend		<0.001	<0.001	0.015

^† Unadjusted.

^‡ Adjusted for age and sex.

^§ Adjusted for age, sex, body mass index, presence of hypertension and obesity, disease history of dyslipidemia, heart disease, cancer or malignant tumor, and stroke, systolic blood pressure, diastolic blood pressure, triglycerides, total cholesterol.

^¶ Incomplete data mainly included age, sex, body mass index, presence of hypertension and obesity, disease history of dyslipidemia, heart disease, cancer or malignant tumor, and stroke, systolic blood pressure, diastolic blood pressure, triglycerides, total cholesterol.

Table 4.

Association between the Hs‐CRP levels and prediabetes progression and regression in subgroups

	Low (Tertile 1 <0.67)	Middle (Tertile 2 0.67–1.62)	High (Tertile 3 >1.62)	P value for trend
	OR (95%CI)	OR (95%CI)	OR (95%CI)
Prediabetes progression
Age, years
<60	1 [Reference]	1.818 (0.718–4.599)	1.760 (0.658–4.712)	0.264
≥60	1 [Reference]	1.859 (1.036–3.337)	1.621 (0.896–2.934)	0.159
Sex
Male	1 [Reference]	1.167 (0.572–2.379)	1.455 (0.730–2.902)	0.281
Female	1 [Reference]	2.636 (1.291–5.381)	1.922 (0.909–4.063)	0.155
BMI, kg/m2
<24	1 [Reference]	1.134 (0.502–2.561)	1.356 (0.617–2.979)	0.45
24–27.9	1 [Reference]	3.153 (1.410–7.049)	2.074 (0.874–4.922)	0.203
≥28	1 [Reference]	0.924 (0.243–3.515)	1.159 (0.340–3.945)	0.718
The presence of hypertension
Yes	1 [Reference]	1.573 (0.676–3.660)	1.325 (0.566–3.102)	0.627
No	1 [Reference]	1.979 (1.071–3.655)	1.802 (0.954–3.404)	0.078
The disease history of heart disease
Yes	1 [Reference]	3.298 (0.613–17.727)	5.191 (1.027–26.241)	0.04
No	1 [Reference]	1.793 (1.068–3.010)	1.394 (0.808–2.407)	0.283
The disease history of dyslipidemia
Yes	1 [Reference]	1.708 (0.309–9.443)	1.581 (0.280–8.936)	0.716
No	1 [Reference]	1.867 (1.115–3.126)	1.597 (0.937–2.722)	0.111
Prediabetes regression
Age, years
<60	1 [Reference]	0.820 (0.567–1.184)	0.929 (0.622–1.387)	0.953
≥60	1 [Reference]	0.759 (0.591–0.975)	0.700 (0.546–0.896)	0.005
Sex
Male	1 [Reference]	0.939 (0.693–1.273)	0.963 (0.709–1.308)	0.816
Female	1 [Reference]	0.684 (0.515–0.908)	0.631 (0.473–0.844)	0.002
BMI, kg/m2
<24	1 [Reference]	0.805 (0.601–1.079)	0.703 (0.524–0.944)	0.018
24–27.9	1 [Reference]	0.889 (0.633–1.249)	0.921 (0.645–1.315)	0.678
≥28	1 [Reference]	0.599 (0.302–1.190)	0.631 (0.329–1.209)	0.272
The presence of hypertension
Yes	1 [Reference]	0.864 (0.569–1.314)	0.984 (0.652–1.486)	0.955
No	1 [Reference]	0.778 (0.612–0.989)	0.702 (0.549–0.898)	0.005
The disease history of heart disease
Yes	1 [Reference]	0.811 (0.449–1.464)	1.074 (0.598–1.929)	0.704
No	1 [Reference]	0.793 (0.636–0.990)	0.727 (0.581–0.911)	0.006
The disease history of dyslipidemia
Yes	1 [Reference]	1.248 (0.655–2.378)	1.067 (0.548–2.078)	0.942
No	1 [Reference]	0.756 (0.608–0.940)	0.760 (0.609–0.948)	0.016

All analyses adjusted for age, sex, body mass index, presence of hypertension and obesity, disease history of dyslipidemia, heart disease, cancer or malignant tumor, and stroke, systolic blood pressure, diastolic blood pressure, triglycerides, total cholesterol.

DISCUSSION

In this prospective study based on middle‐aged and older adults in China, 5.08% participants with prediabetes eventually progressed to diabetes, and 65.90% eventually returned to normal glucose after 4 years of follow‐up. After classifying hs‐CRP levels into three grades, we found the relationship between inflammatory factor hs‐CRP and the progression and regression of prediabetes. Our study is the first to show that higher levels of inflammatory factors, represented by CRP, are associated with increased odds of progression to diabetes and lower levels of CRP are associated with increased odds of regressing to normoglycemia over a 4‐year follow‐up in middle‐aged and older Chinese with prediabetes.

Systemic persistent low‐grade inflammation has been associated with the etiopathogenesis of type 2 diabetes and even with significant consequences. The assessment of tissue damage, infection, and inflammation in patients can be done using serum high‐sensitivity CRP (hs‐CRP) levels ³ . A Japanese study found that elevated blood CRP levels in middle‐aged Japanese people were strongly linked to raised risk factors for prediabetes, particularly elevated HbA1c values ¹¹ . It has been shown that the activation of the innate immune system plays an important role in the development of type 2 diabetes ¹⁸ , and the elevated serum CRP concentration may be a reflection of the activation of innate immune system components in type 2 diabetes ¹⁹ . Our study shows that low concentrations of inflammatory markers represented by CRP can effectively promote regression to normoglycemia in middle‐aged and elderly people with prediabetes, and high levels of CRP can also make prediabetic patients progress to diabetes, which is consistent with the above conclusions. These results suggest that the level of hs‐CRP may be a common indicator to verify the progression of prediabetes. Although the precise mechanism underlying the association between CRP and diabetes is unknown, the following are some possible explanations: First, it is known that a key aspect of the pathogenesis and development of type 2 diabetes is insulin resistance (IR) ²⁰ . CRP has been associated with insulin resistance in people with type 2 diabetes, according to several studies ²¹ , ²² , ²³ . Researchers discovered in a study involving rats that CRP not only functions as a proxy for inflammation but also directly controls leptin levels, insulin sensitivity, and glucose homeostasis ²⁴ . Additionally, inflammation will cause a reduction in vascular permeability and a modification in peripheral blood flow mechanics, which will block the transport of insulin in tissues and cells and encourage the development of insulin resistance ²⁵ , ²⁶ . Second, activated inflammation signaling pathways, such as the nuclear factor (NF)‐κB inhibitor kinase pathway, C‐Jun N‐terminal kinase (JNK) pathway, result in abnormal insulin receptor signal transduction, therefore insulin receptor substrate (IRS) tyrosine phosphorylation, which may be a cause of glucose tolerance and insulin resistance ²⁷ . Third, islet β‐cell dysfunction in type 2 diabetes is associated with β‐cell damage and inflammatory status. The induction pathway of β‐cell injury may be related to abnormal adipocyte metabolism. When adipocyte free fatty acids (FFA) are produced too much, FFA flowing into β‐cells increases. Excessive FFA in the β‐cells may impair mitochondrial and Glut2 function, and even lead to β‐cell apoptosis. When plasma free fatty acids (FFA) are elevated, FFA can activate JNK and NFκB proinflammatory pathways. PKC and ROS lead to increased expression of proinflammatory cytokines, such as TNFα, IL‐6, and IL‐1β, which cause insulin resistance in adipose tissues and other insulin‐sensitive tissues ²⁸ . To date, there have been several clinical trials targeting inflammatory factors in type 2 diabetes ²⁹ , ³⁰ , ³¹ . The study found that IL‐1β gene expression was 100 times higher in β‐cells of patients with type 2 diabetes compared with non‐diabetic people ³² . Treatment with the IL‐1 antagonist Anakinra resulted in lower glycated hemoglobin levels, lower proinsulin to insulin ratios, and lower IL‐6 and CRP levels ³³ . Furthermore, the TNF antagonist Etanercept improved glucose tolerance in rats with a lower homeostasis model assessment in a rat experiment ³⁴ .

Evidence is growing about the relevance of the CRP gene to the risk of type 2 diabetes, despite the fact that there is ample evidence that CRP may have a significant etiological role in the pathogenesis of type 2 diabetes. In a study utilizing allele‐specific PCR to examine CRP gene polymorphisms, it was discovered that both heterozygosity and mutation homozygosity were associated with a higher risk of type 2 diabetes ³⁵ . According to genome‐wide scans, a locus on chromosomes 1q21–q24 was associated with features related to diabetes, according to Vionnet et al. ³⁶ . In addition, Wolford et al ³⁷ reported that haplotypes with low‐risk alleles were related with several indices of insulin secretion in Pema Indians. The putative promoter region of the CRP gene contained these single nucleotide polymorphisms (SNPs). These results imply that changes in plasma CRP levels may be a direct or indirect consequence of CRP genetic variations on insulin sensitivity and glucose metabolism ²⁰ . These studies provided evidence that the CRP gene may play a genetic role in glucose metabolism.

In gender‐specific stratified analyses, inflammatory markers represented by hs‐CRP predicted the progression and regression of prediabetes in the female prediabetes group, while the same results were not found in male participants. In the Mexico City Diabetes Study, CRP was found to be a significant predictor of the development of metabolic syndrome in women. The authors suggest that low‐grade inflammation may interfere more with insulin action in women than in men ³⁸ . The sex difference in the association may have to do with the interaction between sex hormones and inflammation ³⁹ . But this proposition should still be tested in future studies, along with assessments of endogenous sex hormones.

Our study provides a new idea for preventing middle‐aged and elderly patients with prediabetes from progressing to diabetes and promoting their regress to normoglycemia, that is, reducing the level of hs‐CRP in patients with prediabetes can promote their regress to normoglycemia. This has important implications for the association of other inflammatory markers with diabetes risk or regression with prediabetes. In a West of Scotland study, pravastatin, an HMG‐CoA reductase inhibitor, reduced the risk of type 2 diabetes by 30%, possibly due to anti‐inflammatory effects beyond its ability to lower cholesterol ⁴⁰ . Therefore, people with prediabetes may be able to control their blood sugar through monitoring inflammation levels. It has certain clinical significance for the treatment and prevention of prediabetes.

There are still many limitations to our study. First, we used the ADA criteria, the most common of the current criteria for the management of diabetes. We did not conduct other methods to define diabetes, such as WHO and IEC criteria, because the results were consistent across the three criteria ⁴¹ . Second, during the follow‐up period, we excluded many participants who did not meet the follow‐up requirements, and the deletion of these people may also affect the study results. Third, although our model 3 controlled for multiple variables, confounding from unmeasured factors (such as history of fatty liver disease, the use of medications during the period from 2011 to 2015, amount of exercise, eating habits, and body fat percentage) could not be excluded. Fourth, a small number of participants had incomplete baseline covariate variables (e.g., BMI, diastolic blood pressure, and systolic blood pressure). However, sensitivity analyses still yielded the same conclusions. Fifth, our study only included middle‐aged and elderly people and may not be applicable to other populations.

The strengths of this study are specifically the following: First, we used a large Chinese population cohort population, which provides a basis for the credible robustness of our results. There is at present a lot of research on inflammatory factors and diabetes, but we are so far the first study to investigate the relationship of inflammation factors assessed by hs‐CRP and the progression and regression of prediabetes. Third, the controllable factor hs‐CRP is commonly used and an easier measured index, so it can be applied to clinical practice.

In conclusion, this study found that low levels of hs‐CRP are associated with a high incidence of regression from prediabetes to normoglycemia and reduced odds of progression to diabetes. Stratified analyses showed a stronger association between hs‐CRP and type 2 diabetes in women than in men. Thus, prediabetes may need to be closely monitored in middle‐aged and older women with elevated hs‐CRP. In addition, whether a lower cutoff value for hs‐CRP elevation in women should be proposed is a question worthy of future research.

DISCLOSURE

All authors declare that they have no conflict of interest.

Approval of the research protocol: The protocol of CHARLS was approved by the Ethical Review Committee of Peking University.

Informed consent: All the enrolled subjects signed an informed consent.

Approval date of Registry and the Registration No. of the study/trial: This study was approved by the Biomedical Ethics Review Committee of Peking University (IRB00001052‐11015).

Animal studies: N/A.

DATA AVAILABILITY STATEMENT

Data are available and can be downloaded from http://charls.pku.edu.cn/, accessed on 11 February 2023.