背景和目的

颈动脉内中膜厚度（Carotid intima-media thickness，CIMT）是指颈动脉内膜和中膜层的总厚度，是一种表征动脉粥样硬化程度的标记物。CIMT与多种心血管疾病危险因素相关，然而在不同年龄人群中心血管疾病危险因素与CIMT关联的异质性仍不清楚。此外，与年龄较大的人群相比，CIMT与心血管事件的关联强度可能在较年轻的人群中更强。然而，既往研究的年龄分组不够精细，且未进一步评估CIMT的预测增量值，因此未能明确CIMT对心血管事件预测作用的年龄趋势。针对上述尚未解决的科学问题，本研究将在无心血管疾病史人群中：（1）探索心血管疾病危险因素与CIMT关联的年龄趋势；（2）评估CIMT与心血管疾病死亡关联的年龄趋势；（3）评估CIMT对心血管疾病死亡预测增量值的年龄趋势；（4）构建包含CIMT的心血管疾病死亡风险预测模型。

方法

本研究依托心血管病高危人群早期筛查与综合干预项目（China Health Evaluation And risk Reduction through nationwide Teamwork，ChinaHEART）。研究对象是该项目2014年11月至2020年12月入选并在基线接受颈动脉超声检查的35–75岁无心血管疾病史社区居民。根据年龄将研究人群分成4组（35–44、45–54、55–64、65–75岁）。CIMT平均值为左右颈总动脉处6段测量值的平均值，CIMT最大值为6段测量值的最大值。

采用横断面研究，探索基线心血管疾病危险因素和CIMT增厚关联的年龄趋势。危险因素包括目前吸烟、中心性肥胖、低体力活动、高血压、糖尿病史、血脂异常，研究结局为CIMT增厚（CIMT＞0.9mm）。采用多因素Logistic回归模型，计算心血管疾病危险因素与CIMT增厚的比值比（Odds ratio，OR）及其95%置信区间（Confidence interval，CI）。使用年龄与心血管疾病危险因素的交互项来评估其交互作用。

采用前瞻性队列研究，评估CIMT与心血管疾病死亡的关联、对心血管疾病死亡预测增量值的年龄趋势和构建包括CIMT的风险预测模型。死亡数据来源于中国疾病预防控制中心的全国死亡率监测系统和死亡登记系统。研究随访时间截止到2021年12月31日。使用竞争风险模型计算CIMT与心血管疾病死亡的风险比（Hazard ratio，HR）及其95%CI，并调整协变量。使用年龄与CIMT的交互项评估其交互作用。计算Harrell一致性指数变化量（The differences of Harrell’s C-index，ΔC index）、连续性净重分类改善度（Net reclassification improvement index，NRI）、综合判别改善度（Integrated discrimination improvement，IDI），评估CIMT在传统危险因素的基础上对心血管疾病死亡的预测增量值。进一步构建包括CIMT的对5年心血管疾病死亡风险预测模型。建模时将数据集1：1随机分为训练集和测试集。模型筛选变量时，CIMT和主要危险因素直接进入模型，待选变量需与包括CIMT和主要危险因素的模型比较，NRI和IDI均大于0才会被纳入模型。计算Harrell一致性指数、Brier分数，进行Greenwood-Nam-D'Agostino检验和描绘校准曲线，评估模型的预测能力。使用列线图对预测模型进行可视化呈现。

结果

本研究共纳入369,478名参与者，平均年龄57.81±9.44岁，女性占比58.63%，CIMT平均值增厚的比例为17.46%，最大值增厚的比例为34.84%。横断面研究结果显示，在全部人群中，各项心血管疾病危险因素均与CIMT平均值增厚存在关联，且这种关联强度呈现明显的年龄趋势，即在较年轻人群中，危险因素与CIMT平均值增厚的关联更强。年龄与目前吸烟、中心性肥胖、低体力活动、高血压、糖尿病史之间存在显著的交互作用（P＜0.001），年龄与血脂异常的交互作用P值为0.095。其中，高血压与CIMT平均值增厚关联最强，高血压的OR值从35–44岁的2.028（95%CI 1.776–2.317）逐渐下降到65–75岁的1.285（95%CI 1.229–1.344）。以CIMT最大值增厚为结局，呈现同样的年龄趋势。

前瞻性队列研究中，中位随访4.7（25分位–75分位：3.5–5.7）年，共有4,723 名（1.28%）参与者因心血管疾病死亡。在全部人群中，CIMT平均值与心血管疾病死亡存在关联，且这种关联同样存在年龄趋势。CIMT平均值每升高1个标准差，心血管疾病死亡的HR值从35–44岁的1.268（95%CI 1.171–1.373）下降至65–75岁的1.141（95%CI 1.100–1.184），年龄与CIMT平均值交互项P值＜0.001。同时，CIMT平均值对心血管疾病死亡的预测增量值存在年龄趋势。CIMT平均值的NRI从35–44岁的22.60%（95%CI 15.56%–29.64%，P＜0.001）下降至65–75岁的7.00%（95%CI -6.82%–20.83%，P＝0.32）。ΔC index、IDI同样呈现年龄趋势，并且显示CIMT平均值仅在＜55岁人群中有预测增量作用。CIMT最大值的结果呈现相同的年龄趋势。在＜55岁人群中构建包括CIMT的预测模型，预测变量还包括性别、目前吸烟、收缩压、低密度脂蛋白胆固醇、糖尿病史、服用降压药物。在训练集中，Harrell一致性指数为0.77，Brier分数为0.0047；在测试集中，Harrell一致性指数为0.76，Brier分数为0.0051。在两个数据集中，Greenwood-Nam-D'Agostino检验P值均＞0.05，校准曲线显示校准度均良好。

结论

在35–75岁的无心血管疾病史人群中，心血管疾病危险因素与CIMT的关联存在年龄趋势，即年龄越小，关联越强。CIMT与心血管疾病死亡的关联存在同样的年龄趋势，即相比于年龄较大的人群，较年轻的人群CIMT与心血管疾病死亡关联更强。CIMT在55岁以下人群中对心血管疾病死亡具有显著的预测增量价值，而在55岁及以上人群中未见作用。在55岁以下人群中，结合CIMT和传统危险因素的预测模型可准确评估心血管疾病死亡风险。本研究的发现表明，在中青年人群中CIMT可评估动脉粥样硬化程度并可作为其心血管风险增强因素，优化个性化风险评估。

Background and objective

Carotid intima-media thickness (CIMT) refers to the total thickness of the intima and media layers of the carotid artery, serving as a marker of atherosclerosis. CIMT is associated with various cardiovascular disease risk factors, but the heterogeneity of these associations across different age groups remains unclear. Additionally, the associations between CIMT and cardiovascular events may be stronger in the younger population than in old ones. However, previous studies used broad age categorizations and did not further evaluate the added predictive value of CIMT, leaving the age-related trends in the predictive role of CIMT for cardiovascular events unresolved. To address these gaps, this study aims to in a population without prior cardiovascular disease: (1) explore the age-related trend in the associations between cardiovascular risk factors and CIMT; (2) investigate the age-related trend in the associations between CIMT and cardiovascular death; (3) investigate the age-related trend in the added predictive value of CIMT for cardiovascular death; (4) develop a cardiovascular death risk prediction model incorporating CIMT.

Methods

The study population consisted of community residents aged 35–75 years without prior cardiovascular diseases who were enrolled in the China Health Evaluation And Risk Reduction through Nationwide Teamwork (ChinaHEART) project between November 2014 and December 2020 and underwent carotid ultrasonography at baseline. Participants were categorized into four age groups: 35–44, 45–54, 55–64, and 65–75 years. The mean of CIMT was defined as the average values of six measurements at left and right common carotid arteries, and the maximum of CIMT as the maximal values of six measurements.

A cross-sectional study was conducted to explore the age-related trend in the associations between cardiovascular risk factors and abnormal CIMT. The risk factors included current smoking, central obesity, low physical activity, hypertension, diabetes, and dyslipidemia. The outcome was abnormal CIMT (CIMT>0.9 mm). Multivariate logistic regression models were used to calculate the odds ratios (ORs) and 95% confidence intervals (CIs) for the associations between cardiovascular risk factors and abnormal CIMT. Interaction terms between age and cardiovascular risk factors were tested.

A prospective cohort study was conducted to evaluate the age-related trend in the associations between CIMT and cardiovascular death, the added predictive value of CIMT for cardiovascular death, and to develop a prediction model incorporating CIMT. The mortality data were obtained from the National Mortality Surveillance System and Vital Registration of Chinese Center for Disease Control and Prevention. Follow-up continued until December 31, 2021. Competing risk models were used to calculate hazard ratios (HRs) and 95%CIs for the associations between CIMT and cardiovascular death, adjusting for covariates. Interaction terms between age and CIMT were used to assess interactions. The differences in Harrell’s C-index (ΔC index), net reclassification improvement (NRI), and integrated discrimination improvement (IDI) were calculated to evaluate the added predictive value of CIMT based on the traditional risk factors. A 5-year cardiovascular death risk prediction model incorporating CIMT was further developed, with the dataset randomly divided in a 1:1 ratio into training and test sets. CIMT and the major risk factors were retained in models directly, while candidate variables would be included if both the NRI and IDI showed improvement compared to the model containing CIMT and the major risk factors. Model performance was assessed using Harrell’s C-index, Brier score, Greenwood-Nam-D’Agostino test, and calibration curves. A nomogram was created to visualize the prediction model.

Results

Among the 369,478 participants included, the mean age was 57.81±9.44 years, and 58.63% were women. The prevalence of abnormal CIMT_mean was 17.46%, while that of abnormal CIMT_max was 34.84%. In the cross-sectional study, all cardiovascular risk factors were associated with abnormal CIMT_mean in overall population, with stronger associations observed in younger age groups. Significant interactions (P<0.001) were observed between age and current smoking, central obesity, low physical activity, hypertension, and diabetes. The P value for interaction between age and dyslipidemia was 0.095. Among these risk factors, hypertension showed the strongest association with abnormal CIMT_mean, with ORs of 2.028 (95%CI 1.776–2.317) in the 35–44 age group, gradually decreasing to 1.285 (95%CI 1.229–1.344) in the 65–75 age group. The results for the outcome of abnormal CIMT_max, showing a similar age-related trend.

In the prospective cohort, with a median follow-up of 4.7 (25th percentile to 75th percentile: 3.5–5.7) years, during which 4,723 individuals (1.28%) died from cardiovascular disease. CIMT_mean was associated with cardiovascular death in overall population, also showing an age-related trend. The HRs for CIMT_mean per standard deviation decreased with age, from 1.268 (95%CI 1.171–1.373) in the 35–44 age group to 1.141 (95%CI 1.100–1.184) in the 65–75 age group. The P value for interaction between age and CIMT_mean was <0.001. Meanwhile, the added predictive ability of CIMT_mean for cardiovascular death showed an age-related trend, with NRI decreasing from 22.60% (95%CI 15.56%–29.64%, P<0.001) in the 35–44 age group to 7.00% (95%CI -6.82%–20.83%, P=0.32) in the 65–75 age group. The ΔC index and IDI also showed an age-related trend, indicating that CIMT_mean provided the added predictive value only in individuals younger than 55 years. Similar age-related trends were found for CIMT_max. A prediction model incorporating CIMT was developed for individuals younger than 55 years, including sex, current smoking, systolic blood pressure, low-density lipoprotein cholesterol, diabetes, and antihypertensive medication. In the training set, Harrell’s C-index was 0.77 and the Brier score was 0.0047; in the test set, Harrell’s C-index was 0.76 and the Brier score was 0.0051. In both datasets, the Greenwood-Nam-D’Agostino test P values were >0.05, and the calibration curves indicated good calibration.

Conclusion

In a population without prior cardiovascular disease aged 35–75 years, stronger associations between cardiovascular risk factors and CIMT were observed in younger individuals. The association between CIMT and cardiovascular death also demonstrated an age-related trend, with a stronger association observed in the younger population. CIMT had the added predictive ability for cardiovascular death in individuals younger than 55 years, but not in those aged 55 and above. In individuals under 55 years old, a predictive model combining CIMT and traditional risk factors can accurately assess the risk of cardiovascular death. These findings suggest that CIMT can be used to evaluate the degree of atherosclerosis and can serve as a risk-enhancing factor to optimize personalized risk assessment in young and middle-aged populations.