Prognostic value of C-reactive protein predicting all-cause and cause-specific mortality: a prospective cohort study in Shanghai, China

Abstract

Objectives

To investigate the role of C-reactive protein (CRP) in predicting all-cause and cause-specific mortality in a community-based Chinese cohort.

Design

A community-based prospective cohort study.

Setting

34 communities in Pudong New Area, Shanghai, China.

Participants

A total of 9360 permanent residents from 34 randomly selected communities were enrolled in 2013. Follow-up began at baseline and continued until death or 30 September 2023, whichever occurred first.

Main outcome measures

The primary outcome of this study was death, as recorded in the Vital Registry of Pudong New Area, Shanghai, China. Associations between CRP and risks of all-cause and cause-specific mortality were studied. Cox proportional hazards models were applied to estimate HR and 95% CI. Adjustments were made for age, sex, area, marriage status, education, current smoking, alcohol consumption, physical activity, body mass index, systolic blood pressure, diastolic blood pressure, type 2 diabetes, dyslipidaemia and chronic kidney disease. Competing risk analyses were performed for cause-specific mortality. Restricted cubic splines were used to assess potential non-linear associations. We evaluated improvements in mortality prediction by calculating changes in the C-index, integrated discrimination improvement and net reclassification improvement after adding CRP to conventional risk factor models.

Results

Over a median follow-up of 10.52 (IQR: 10.43–10.56) years, 920 deaths (9.68 per 1000 person-years) were recorded. After adjusting for traditional risk factors, higher baseline CRP levels were significantly associated with increased risk of all-cause mortality, cardiovascular mortality and cancer mortality. Non-linear associations were observed between CRP levels and all-cause and cancer mortality. The addition of CRP level significantly improved the reclassification and discrimination ability beyond the conventional risk factor models for all-cause mortality and cancer mortality.

Conclusions

Elevated CRP levels, indicative of low-grade inflammation, are an independent risk factor for all-cause, cardiovascular and cancer mortality.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• This community-based prospective cohort study with long-term follow-up enhances the reliability of mortality risk assessments.

• The use of competing risk models and restricted cubic splines provides robust evaluation of cause-specific mortality and non-linear relationships.

• As an observational study, residual confounding may persist despite extensive adjustments.

• C-reactive protein was measured only at baseline, preventing assessment of its dynamic changes over time.

Background

It is well known that inflammatory processes are involved in the pathogenesis of several acute and chronic conditions. C-reactive protein (CRP), as a marker of elevated levels of circulating inflammatory, has been proposed as a predictor of many non-communicable chronic diseases including cardiovascular diseases and cancer.^{1 2} As an important mediator of atherosclerosis, CRP has been used as a biomarker for cardiovascular events and cardiovascular mortality in patients with intermediate risk, although the cost-effectiveness of CRP screening is still an area of controversy.^{3 4} Previous studies revealed that elevated circulating levels of CRP were a potential biomarker to assess risks of overall cancer and several site-specific cancers.^{2 5 6} Yet the long-term effect of CRP on cancer mortality remains conflicting.7,9 However, so far, evidence comes mainly from western cohorts, studies regarding the associations between CRP and mortality in Asian populations are limited.

Compared with Western populations, CRP concentrations are significantly lower in the Chinese population.¹⁰ Cardiovascular disease in the Chinese population is characterised by a higher rate of stroke and lower rate of coronary heart disease.¹¹ Cancer in the Chinese population is characterised by lower incidence but higher mortality; a large proportion of the cancer-related deaths were from the digestive tract cancers and have relatively poorer prognoses.¹² Considering that different levels of CRP may contribute differently to cause-specific mortality risk, it is necessary to evaluate the dose-response relationship and the predictive value of CRP on mortality in the Chinese general population. Thus, the current study was performed to investigate the role of CRP in predicting all-cause and cause-specific mortality in a community-based Chinese cohort of representative participants randomly selected from the general population.

Methods

Study population

This study was the sequential research of the ongoing community-based prospective cohort study performed since 2013 in Pudong New Area, Shanghai, China.¹³ Using multistage stratified random cluster sampling, (1) 38 streets were stratified by socioeconomic status, (2) 12 streets were randomly selected, (3) 34 communities were chosen and (4) 11% of households per community were sampled. After excluding patients with type 1 diabetes and pregnant women, 10 657 from 12 382 eligible individuals participated. From the initial 10 657 baseline participants, we excluded (1) 976 participants with missing covariate data, (2) 88 participants with CRP >15 mg/L (to exclude acute inflammation/infection)¹⁴ and (3) 233 participants not registered as Shanghai permanent residents who were lost to follow-up. A total of 9360 participants were followed until 30 September 2023 (median follow-up: 10.52 years) and included in analyses (figure 1). During follow-up, 920 deaths occurred, including 392 cardiovascular disease-related deaths, 282 cancer-related deaths and 246 deaths from other causes.

Figure 1. Diagram of this study. CRP, C-reactive protein.

Exposure measurements

After an overnight fasting for 12 hours, fasting blood samples were collected from all participants. The levels of serum CRP were measured using an immunometric assay (HITACHI 7170A, Japan). For analytical purposes, the levels of CRP were divided into three groups. CRP less than 1 mg/L was regarded as the low level, 1–3 mg/L as the moderate level and greater than 3 mg/L as the high level.^{15 16}

Covariates

During the baseline assessment, standard demographic data and lifestyle information were collected in all participants via a face-to-face interview conducted by trained health workers. Individuals with at least one cigarette a day in the past 6 months were defined as current smoking. Regular alcohol drinker with at least three times per week in the past 6 months was defined as alcohol consumption. Individuals participating in sports activities for at least once per week in the past 5 years were defined as physical activity. Height, body weight and blood pressure were measured by trained health workers. Body mass index (BMI) was calculated as weight (kg)/height (m²). Systolic blood pressure (SBP) and diastolic blood pressure (DBP) were calculated as the mean of the last two out of three measurements taken at 5 min intervals and were analysed as continuous variables. Type 2 diabetes was defined as self-reported diabetes, or fasting plasma glucose ≥7.0 mmol/L, or 2-hour plasma glucose ≥11.1 mmol/L by OGTT (oral glucose tolerance test) test. Dyslipidaemia was defined as self-reported dyslipidaemia, or TG ≥2.26 mmol/L, or total cholesterol (TC) ≥6.20 mmol/L, or low-density lipoprotein (LDL) ≥4.13 mmol/L, or high-density lipoprotein (HDL) <1.03 mmol/L. Renal function was assessed using the estimated glomerular filtration rate (eGFR), calculated with the Modification of Diet in Renal Disease (MDRD) equation (adjusted for Chinese adults). Decreased kidney function was defined as an eGFR <60 mL/min/1.73 m². Albuminuria was categorised as microalbuminuria (urine albumin-to-creatinine ratio (ACR): 30–299 mg/g) or macroalbuminuria (ACR ≥300 mg/g). Chronic kidney disease (CKD) was defined as either decreased eGFR, albuminuria or both. Participants with missing data on covariates were excluded from analysis.

Outcomes

The primary outcome of this study was all-cause death, which was assessed by data linkage with the Vital Registry of Pudong New Area, Shanghai, China. Underlying death causes were coded according to the International Classification of Diseases, Tenth Revision (ICD-10): codes I00–I99 for cardiovascular deaths, and C00–D48 for cancer deaths.

Statistical analysis

Continuous variables were presented as mean (SD) or median (IQR) and were compared using the one-way analysis of variance (ANOVA) test or Kruskal-Wallis test. Categorical variables were presented as number (%) and compared using the Pearson χ² test. We used the Cox proportional hazards model to estimate the association between CRP and all-cause mortality, adjusted for age, sex, area, marriage status, education, current smoking, alcohol consumption, physical activity, BMI, SBP, DBP, type 2 diabetes, dyslipidaemia and chronic kidney disease. Taking competing risks into consideration, we used the Fine-Gray proportional subdistribution hazards model to evaluate the association between CRP and cause-specific mortality. We used restricted cubic splines in Cox models to test whether there was a non-linear association between CRP as a continuous variable and all-cause/ cause-specific mortality, with the median level of CRP (0.25 mg/L) as the reference value. To examine whether CRP increases the predictive value of mortality prediction models, discrimination and reclassification were measured by change in C-index, integrated discrimination improvement (IDI) and net reclassification improvement (NRI). The 95% CIs for discrimination and reclassification measures were computed by bootstrap. SPSS V.22.0 and SAS V.9.4 were applied for statistical analysis. A p value of <0.05 was considered significant.

Patient and public involvement

Patients and/or the public were not involved in this research.

Results

Baseline characteristics

The baseline characteristics are presented in groups of CRP in table 1. The study population of 9360 participants had a mean age of 57.92 years at baseline. In general, participants with high CRP levels at baseline were more likely to be older, be not married, have a lower educational level and be current smokers; have a higher prevalence of type 2 diabetes, dyslipidaemia or CKD; and have higher BMI, waist circumference, SBP, DBP, TG, TC and LDL cholesterol. They were less likely to have higher HDL cholesterol.

Table 1. Baseline characteristics stratified by C-reactive protein levels.

Characteristics	C-reactive protein			Total n=9360	P value
	<1 mg/L n=7118	1–3 mg/L n=1329	>3 mg/L n=913
Age (years)	57.28 (12.92)	59.8 (12.19)	60.19 (13.22)	57.92 (12.90)	<0.001
Male (%)	2670 (37.51%)	514 (38.68%)	351 (38.44%)	3535 (37.77%)	0.656
Married (%)	6255 (87.88%)	1163 (87.51%)	776 (84.99%)	8194 (87.54%)	0.046
Urban (%)	4285 (60.20%)	802 (60.35%)	555 (60.79%)	5642 (60.28%)	0.942
≥9 years of education (%)	5692 (79.97%)	991 (74.57%)	686 (75.14%)	7369 (78.73%)	<0.001
Current smoking (%)	1144 (16.07%)	226 (17.01%)	179 (19.61%)	1549 (16.55%)	0.023
Alcohol consumption (%)	867 (12.18%)	147 (11.06%)	109 (11.94%)	1123 (12.00%)	0.514
Tea consumption (%)	1961 (27.55%)	363 (27.31%)	285 (31.22%)	2609 (27.87%)	0.059
Physical activity (%)	1749 (24.57%)	331 (24.91%)	224 (24.53%)	2304 (24.62%)	0.965
Type 2 diabetes (%)	1187 (16.68%)	362 (27.24%)	274 (30.01%)	1823 (19.48%)	<0.001
Dyslipidaemia (%)	3192 (44.84%)	774 (58.24%)	542 (59.36%)	4508 (48.16%)	<0.001
Chronic kidney disease (%)	1262 (17.73%)	329 (24.76%)	239 (26.18%)	1830 (19.55%)	<0.001
SBP (mm Hg)	139.44 (22.35)	145.78 (22.07)	147.51 (22.86)	141.13 (22.56)	<0.001
DBP (mm Hg)	85.58 (11.79)	88.46 (11.5)	89.72 (12.64)	86.39 (11.92)	<0.001
TG (mmol/L)	1.29 (0.91,1.87)	1.65 (1.16,2.38)	1.56 (1.06,2.32)	1.36 (0.95,1.99)	<0.001
TC (mmol/L)	5.44 (4.73,6.19)	5.65 (4.88,6.38)	5.56 (4.80,6.35)	5.48 (4.76,6.23)	<0.001
HDL (mmol/L)	1.36 (1.15,1.60)	1.26 (1.08,1.46)	1.26 (1.07,1.48)	1.33 (1.13,1.57)	<0.001
LDL (mmol/L)	3.01 (2.37,3.67)	3.22 (2.64,3.95)	3.24 (2.56,3.92)	3.07 (2.42,3.75)	<0.001
BMI (kg/m2)	24.56 (3.36)	26.44 (3.62)	26.61 (4.24)	25.03 (3.59)	<0.001
Waist circumference (cm)	81.45 (9.17)	86.38 (9.06)	87.29 (10.02)	82.72 (9.52)	<0.001

Data are n (%), mean (SD) or median (IQR).

BMI, body mass index; DBP, diastolic blood pressure; HDL, high-density lipoprotein; LDL, low-density lipoprotein; SBP, systolic blood pressure; TC, total cholesterol; TG, triglycerides.

Outcome of the study participants

Over the course of approximately 10.52 years (IQR: 10.43–10.56 years) of follow-up from 2013 to 2023, a total of 920 deaths (9.68 per 1000 person-years) occurred among participants. Cardiovascular disease was the cause of mortality in a majority of the participants (392 (42.61%)). The cumulative all-cause mortality, cardiovascular mortality and cancer mortality were highest in participants with CRP of >3 mg/L, followed by those with CRP of 1–3 mg/L, compared with those with CRP of <1 mg/L (table 2).

Table 2. HRs of all-cause and cause-specific mortalities according to C-reactive protein levels in cohort participants.

C-reactive protein	Person-years	Death cases	Mortality (1/1000 person-years)	Crude HR (95% CI)	P value	Adjusted HR* (95% CI)	P value
All-cause mortality
<1 mg/L	72 821	585	8.03	ref.		ref.
1–3 mg/L	13 292	177	13.32	1.67 (1.42 to 1.98)	<0.001	1.29 (1.09 to 1.53)	0.003
>3 mg/L	8961	158	17.63	2.23 (1.87 to 2.66)	<0.001	1.53 (1.28 to 1.83)	<0.001
Cardiovascular mortality
<1 mg/L	72 821	244	3.35	ref.		ref.
1–3 mg/L	13 292	73	5.49	1.68 (1.29 to 2.18)	<0.001	1.22 (0.94 to 1.58)	0.139
>3 mg/L	8961	75	8.37	2.58 (1.99 to 3.34)	<0.001	1.71 (1.32 to 2.22)	<0.001
Cancer mortality
<1 mg/L	72 821	182	2.50	ref.		ref.
1–3 mg/L	13 292	58	4.36	1.79 (1.33 to 2.41)	<0.001	1.51 (1.12 to 2.03)	0.007
>3 mg/L	8961	42	4.69	1.98 (1.41 to 2.77)	<0.001	1.56 (1.11 to 2.18)	0.010

^*Adjusted for age, sex, area, marriage status, education, current smoking, alcohol consumption, physical activity, body mass index, systolic blood pressure, diastolic blood pressure, type 2 diabetes, dyslipidaemia and chronic kidney disease.

Association between CRP and mortality

Cox regression analysis suggested that after multivariate adjustment, higher baseline CRP levels were independently associated with increased risk of all-cause mortality. As shown in table 2, significantly higher risk of all-cause mortality was noted in participants with CRP of 1–3 mg/L (HR: 1.29, 95% CI 1.09 to 1.53) or >3 mg/L (HR: 1.53, 95% CI 1.28 to 1.83) than in those with CRP of <1 mg/L. As for cause-specific mortality, similar increased risks were observed; after multivariate adjustment, increased risk of cardiovascular mortality was noted in participants with CRP >3 mg/L (HR: 1.71, 95% CI 1.32 to 2.22), and increased risk of cancer mortality was noted in participants with CRP of 1–3 mg/L (HR: 1.51, 95% CI 1.12 to 2.03) or >3 mg/L (HR: 1.56, 95% CI 1.11 to 2.28).

Cox regression analyses with restricted cubic spline further demonstrated that higher baseline CRP levels were significantly associated with increased risk of all-cause mortality, cardiovascular mortality and cancer mortality, and suggested non-linear associations of CRP levels with all-cause mortality and cancer mortality (figure 2). Compared with participants with CRP of 0.25 mg/L, participants with CRP ≥0.30 mg/L were at higher risk of all-cause mortality (p_overall < 0.001, p_non-linear = 0.008), cardiovascular mortality (p_overall < 0.001, p_non-linear = 0.124) and cancer mortality (p_overall = 0.002, p_non-linear = 0.011).

Figure 2. Multivariable adjusted spline curves for association between CRP and risk for all-cause mortality, cardiovascular mortality and cancer mortality, adjusted for age, sex, area, marriage status, education, current smoking, alcohol consumption, physical activity, body mass index, systolic blood pressure, diastolic blood pressure, type 2 diabetes, dyslipidaemia and chronic kidney disease. Knots were set at the 5th, 50th and 95th percentiles. Reference was set at the median level of CRP (0.25 mg/L). The bar graph shows the distribution of CRP among study participants. CRP, C-reactive protein.

The predictive value of CRP for mortality

Table 3 showed that the addition of CRP level significantly improved the reclassification and discrimination ability beyond the conventional risk factor models for all-cause mortality and cancer mortality. The C-index of the conventional risk factor model for all-cause mortality (0.839 (0.828 to 0.850)) changed after addition of the CRP level (0.841 (0.829 to 0.851)) (p<0.05), with IDI of 0.332% (p<0.05) and NRI of 21.932% (p<0.01). The C-index of the conventional risk factor model for cancer mortality (0.775 (0.755 to 0.910)) changed after addition of the CRP level (0.778 (0.758 to 0.911)) (p<0.05), with IDI of 0.937% (p<0.05) and NRI of 22.564% (p<0.01). The addition of CRP level did not significantly improve the C-index of the conventional risk factor model for cardiovascular mortality.

Table 3. Improvement in predicting all-cause mortality and cause-specific mortality by adding CRP levels to conventional risk factors.

	C-index (95% CI)	Change in C-index (95% CI)	IDI, % (95% CI)	NRI, % (95% CI)
All-cause mortality
Conventional risk factors	0.839 (0.828 to 0.850)	ref.	ref.	ref.
Plus CRP	0.841 (0.829 to 0.851)	0.002 (0.001 to 0.003)*	0.332 (0.108 to 0.624)*	21.932 (15.313 to 26.406)*
Cardiovascular mortality
Conventional risk factors	0.900 (0.831 to 0.912)	ref.	ref.	ref.
Plus CRP	0.901 (0.833 to 0.912)	0.001 (−0.001 to 0.004)	0.703 (0.118 to 1.222)*	23.936 (14.564 to 31.195)*
Cancer mortality
Conventional risk factors	0.776 (0.755 to 0.910)	ref.	ref.	ref.
Plus CRP	0.778 (0.758 to 0.911)	0.003 (−0.001 to 0.005)*	0.937 (0.133 to 1.551)*	22.564 (11.029 to 30.547)*

Conventional risk factors included age, sex, area, marriage status, education, current smoking, alcohol consumption, physical activity, body mass index, systolic blood pressure, diastolic blood pressure, type 2 diabetes, dyslipidaemia and chronic kidney disease.

^*p < 0.05.

CRP, C-reactive protein.

Discussion

In the present community-based prospective cohort study in China, the baseline CRP level was significantly and positively associated with an increased risk for all-cause mortality, cardiovascular mortality and cancer mortality. Moreover, the baseline CRP level had a nonlinear dose-response relationship with the risk of all-cause mortality and cancer mortality. The addition of CRP to the traditional risk factors of all-cause mortality and cancer mortality may improve the reclassification and discrimination ability. The study participants were randomly recruited from community residents in Pudong New Area, a district accounting for 20% of the total population in Shanghai. Moreover, Pudong New Area is the only district with both urban and rural populations at various levels of socioeconomic status, which is highly representative for general population. The results of this study offer references for optimising health promotion strategies in specific populations. Future research could further explore its applicability in a broader range of populations.

Despite the large number of studies, the convincing evidence on the long-term impact of CRP on chronic diseases and mortality risk was limited. A previous meta-analysis showed CRP has continuous associations with the long-term risk of vascular mortality and cancer mortality.¹⁷ The Zwolle Outpatient Diabetes project Integrating Available Care (ZODIAC) cohort study with 11 years of follow-up suggested that hs-CRP (high-sensitivity CRP) was independently associated with all-cause mortality and cardiovascular mortality.¹⁸ In people without diabetes, elevated levels of CRP are related to the development of insulin resistance and type 2 diabetes.¹⁹ In Asian populations with relatively low CRP concentrations, the results of CRP as a predictor of mortality were inconsistent. The China Health and Retirement Longitudinal Study suggested that plasma hs-CRP serves as an independent predictor of all-cause mortality among the middle-aged and elderly populations.²⁰ Another study suggested that hs-CRP >2.16 mg/L in Chinese people with hyperglycaemia predicted an increased risk of all-cause mortality independent of traditional risk factors.²¹ A Japanese cohort study suggested that hs-CRP may be an independent predictor of all-cause, cardiovascular and cancer mortalities in apparently healthy men, but not women.²² Establishing CRP’s causal role in mortality risk remains challenging, as elevated inflammatory responses may stem from multiple mortality-associated factors including ageing, smoking, physical inactivity, obesity, hypertension and socioeconomic status.23,25 Zacho et al suggested that the associations between CRP and mortality risk may reflect underlying chronic inflammatory conditions.²⁶ Besides, elevated levels of CRP may reflect a common biochemical pathway of poor health status, which subsequently leads to increased mortality risk in old age.²⁷ In the present study, the prevalence of type 2 diabetes, dyslipidaemia and CKD was significantly higher in participants with higher CRP than in those with lower CRP; these chronic diseases have also been associated with increased mortality risk. This supports the interpretation that elevated CRP level reflects the exposure of multiple pathogenic factors and could be used to identify individuals with elevated mortality risk. The addition of CRP significantly improved the reclassification and discrimination of participants across clinical risk categories currently used to guide preventive treatment decisions. Confirming the clinical utility of CRP requires further intervention studies.

In the current study, CRP did not improve the discriminatory ability of predicting cardiovascular mortality; however, the reclassification ability was significantly improved. This finding is consistent with previous research which indicates that CRP improved little discrimination ability to the existing methods of cardiovascular risk assessment.²⁸ The reason may be that the overall level of CRP in this study is relatively low. In addition, the role of low-grade inflammation in atherosclerotic plaque initiation, progression and rupture could be explained by many other lines of evidence; as a general marker of inflammation, CRP alone is insufficient to predict cardiovascular mortality risk.²⁹ The European guidelines on cardiovascular disease prevention in clinical practice suggested that participants at moderate, but not low, risk of cardiovascular disease (CVD) should measure hs-CRP.³⁰

Several possible mechanisms of the association between CRP and mortality risk have been reported. Previous research suggested that CRP was not just simply a stable biomarker of systemic immune response with long half-life, but is a contributing factor to such disorders and has numerous critical proinflammatory properties.^{31 32} Through directly activating platelets and triggering the classical complement pathway, CRP plays an important role in the pathogenesis of atherothrombosis and venous thromboembolism.¹ In the chronic inflammatory environment, platelet aggregates aberrantly on the surface of endothelial cells, resulting in immunological dysfunction and local tissue death.³³ Subsequently, the decline in immune system capabilities facilitates endothelial dysfunction, abnormal aggregation and thrombosis, and increases the risk of adverse cardiovascular events.³⁴ In patients with cancer, high serum levels of CRP are associated with the tumour mass, reflecting advanced disease and poor prognosis.³⁵ CRP acts as a mediator of carcinogenesis and cancer progression through activation of several pathways, such as oxidation of protein and DNA, point mutations in tumour suppressor genes, DNA methylation and post-translational variations.^{36 37} Moreover, cancer cells can express CRP which in turn regulate the liver CRP production.³⁸ Apart from directly increasing cardiovascular and cancer mortality risk, chronic subclinical inflammation marked by mildly increased CRP level was involved in the pathologies of insulin resistance and metabolic syndrome, both of which are proven risk factors of all-cause mortality.³⁹

The strengths of this study include its prospective design, representative participants randomly selected from the general population, standardised data collection, long follow-up period and substantial number of certified deaths. The results of this study add to the evidence to show that CRP is an independent risk factor for all-cause and cause-specific mortalities and inflammatory pathway could be considered a potential therapeutic target to promote longevity. This study had several limitations. First, the follow-up period was relatively short; a longer follow-up period is needed to obtain more stable results. Second, this study only analysed a single measurement of CRP at baseline; CRP trajectory during the follow-up period was not taken into account. Third, data on medication use, dietary habits and social/psychological factors were not collected at baseline, potentially resulting in residual confounding.

Conclusions

In conclusion, elevated CRP levels, indicative of low-grade inflammation, are an independent risk factor for all-cause, cardiovascular and cancer mortalities. These findings highlight CRP’s potential as a risk stratification marker to identify high-risk groups for preventive strategies. Whether CRP is a causal driver or merely a biomarker of inflammation requires verification from intervention studies.

Data availability statement

Data are available upon reasonable request.