Individual Prediction of Cardiovascular Events in Blood or Marrow Transplant Survivors - a BMT Survivor Study Report

BACKGROUND: Blood or marrow transplant (BMT) recipients are at increased risk of accelerated atherosclerosis due to prior exposure to chemotherapy and radiation, and pre-existing/new onset comorbidities. This increases the risk of long-term cardiovascular disease (CVD) including coronary heart disease and stroke. There is a need to identify BMT recipients at highest risk of CVD to develop targeted intervention strategies. We performed a comprehensive evaluation of the risk of late-occurring CVD in BMT survivors and developed a prediction model using the resources offered by BMT Survivor Study (BMTSS). METHODS: BMTSS includes patients who received BMT at one of the three participating US sites between 1974 and 2014, survived ≥2 years after BMT and completed the BMTSS survey. The survey asked participants to report specific chronic health conditions diagnosed by their healthcare provider (including coronary heart disease and stroke as well as cardiovascular risk factors [CVRFs: hypertension, diabetes and dyslipidemia]), along with age at diagnosis. They also provided information on socio-demographics (sex, race/ethnicity, education, income) and health behaviors such as smoking and alcohol intake. Medical records were abstracted for information regarding primary cancer diagnosis, therapeutic exposures (pre-BMT chemotherapy/radiation, transplant conditioning regimens), stem cell source (autologous, allogeneic), graft type (bone marrow, cord blood or peripheral blood stem cells), and history of chronic graft vs. host disease (among allogeneic BMT recipients). All participants provided informed consent. Sixty percent of the cohort was used for discovery and the remainder for replication of the risk model. Cox regression models estimated the model's discrimination based on the concordance (C) statistic. Variables selected in our CVD risk model included: age at BMT, sex, race/ethnicity, chest/ neck/ cranial radiation, health behaviors and CVRFs present at BMT. RESULTS: The study included 3,848 BMT survivors; 53.4% had received an allogeneic BMT; 55.9% were males; 75.2% were non-Hispanic whites. Median age at study participation was 56.3 years (interquartile range: 41.3-65.5y); median follow up from BMT was 9.1 years (interquartile range: 5.9-15.3y). CVD was diagnosed after BMT in 238 participants (117 allogeneic; 121 autologous). Conditional on surviving ≥2 years after BMT, the cumulative incidence of CVD was 5.8% at 10 years (allogeneic BMT: 5.0%; autologous BMT: 6.8%) and 9.8% at 20 years (allogeneic BMT: 7.8%; autologous BMT: 13.2%). Increasing age at BMT (hazard ratio [HR]=1.03/y, 95% confidence interval [CI]: 1.02-1.04, p<0.0001), male sex (HR=1.50, 95%CI: 1.13-1.94, p=0.005), history of smoking (HR=1.54, 95%CI:1.08-2.20, p=0.02), hypertension at BMT (HR=1.68, 95%CI: 1.27-2.22, p=0.0003), dyslipidemia at BMT (HR=1.66, 95%CI: 1.24-2.22, p=0.0006), chest radiation (HR=1.86, 95%CI: 1.06-3.24, p=0.029) and neck/cranial radiation (HR=1.62, 95%CI: 0.99-2.67, p=0.057) were associated with increased CVD risk in multivariable analysis. Inclusion of age at BMT, sex, history of smoking, diabetes, hypertension or dyslipidemia at BMT, and pre-BMT chest/neck/cranial radiation yielded C statistics of 0.69 (95%CI: 0.63-0.75) in the training and 0.67 in the testing set (95%CI: 0.61-0.73) (Table 1). Risk scores were derived by adding parameter estimates of each variable, and a cut point was determined to maximize the C statistics. Survivors were then assigned to statistically distinct risk groups corresponding to cumulative incidence of CVD of 13.4% in high-risk group (risk score >1) 4.3% in the low-risk group (score ≤1) at 10 years post-BMT (Figure 1). CONCLUSION: Traditional CVRFs at the time of BMT as well as history of chest, neck and cranial radiation can be combined to identify BMT survivors at high risk for CVD. The risk prediction model provides a framework for future targeted screening and interventions.

背景：由于先前接触过化疗和放疗以及预先存在/新发的合并症，血液或骨髓移植（BMT）受者发生加速性动脉粥样硬化的风险增加。这增加了包括冠心病和中风在内的长期心血管疾病（CVD）的风险。有必要识别出心血管疾病风险最高的骨髓移植受者，以制定有针对性的干预策略。我们对骨髓移植幸存者发生晚期心血管疾病的风险进行了综合评估，并利用骨髓移植幸存者研究（BMTSS）提供的资源开发了一个预测模型。 方法：BMTSS包括1974年至2014年间在美国三个参与研究的地点之一接受骨髓移植、在移植后存活≥2年并完成BMTSS调查的患者。该调查要求参与者报告其医疗服务提供者诊断出的特定慢性健康状况（包括冠心病和中风以及心血管危险因素[CVRFs：高血压、糖尿病和血脂异常]）以及诊断时的年龄。他们还提供了社会人口统计学信息（性别、种族/民族、教育程度、收入）以及吸烟和饮酒等健康行为信息。查阅病历以获取有关原发癌症诊断、治疗暴露（移植前化疗/放疗、移植预处理方案）、干细胞来源（自体、异体）、移植物类型（骨髓、脐带血或外周血干细胞）以及慢性移植物抗宿主病病史（在异体骨髓移植受者中）的信息。所有参与者均提供了知情同意。队列中60%用于发现风险模型，其余用于复制该模型。Cox回归模型根据一致性（C）统计量评估模型的判别能力。我们的心血管疾病风险模型中所选变量包括：骨髓移植时的年龄、性别、种族/民族、胸部/颈部/头颅放疗、健康行为以及骨髓移植时存在的心血管危险因素。 结果：该研究包括3848名骨髓移植幸存者；53.4%接受了异体骨髓移植；55.9%为男性；75.2%为非西班牙裔白人。参与研究时的中位年龄为56.3岁（四分位间距：41.3 - 65.5岁）；骨髓移植后的中位随访时间为9.1年（四分位间距：5.9 - 15.3年）。238名参与者在骨髓移植后被诊断出患有心血管疾病（117名异体移植；121名自体移植）。在骨髓移植后存活≥2年的条件下，10年时心血管疾病的累积发病率为5.8%（异体骨髓移植：5.0%；自体骨髓移植：6.8%），20年时为9.8%（异体骨髓移植：7.8%；自体骨髓移植：13.2%）。骨髓移植时年龄增加（风险比[HR]=1.03/年，95%置信区间[CI]：1.02 - 1.04，p<0.0001）、男性（HR = 1.50，95%CI：1.13 - 1.94，p = 0.005）、吸烟史（HR = 1.54，95%CI：1.08 - 2.20，p = 0.02）、骨髓移植时高血压（HR = 1.68，95%CI：1.27 - 2.22，p = 0.0003）、骨髓移植时血脂异常（HR = 1.66，95%CI：1.24 - 2.22，p = 0.0006）、胸部放疗（HR = 1.86，95%CI：1.06 - 3.24，p = 0.029）以及颈部/头颅放疗（HR = 1.62，95%CI：0.99 - 2.67，p = 0.057）在多变量分析中与心血管疾病风险增加相关。纳入骨髓移植时的年龄、性别、吸烟史、糖尿病、骨髓移植时的高血压或血脂异常以及移植前胸部/颈部/头颅放疗，在训练集中得到的C统计量为0.69（95%CI：0.63 - 0.75），在测试集中为0.67（95%CI：0.61 - 0.73）（表1）。通过将每个变量的参数估计值相加得出风险评分，并确定一个截断点以使C统计量最大化。然后将幸存者分配到具有统计学差异的风险组，对应于骨髓移植后10年心血管疾病累积发病率在高风险组为13.4%（风险评分>1），在低风险组为4.3%（评分≤1）（图1）。 结论：骨髓移植时的传统心血管危险因素以及胸部、颈部和头颅放疗史可结合起来识别心血管疾病高风险的骨髓移植幸存者。风险预测模型为未来有针对性的筛查和干预提供了一个框架。