Risk prediction methods

我们继续发展、完善和评估国家癌症研究所乳腺癌风险评估工具（BCRAT）。使用来自亚裔美国人乳腺癌研究的数据，我们获得了亚裔美国女性的种族特异性相对风险和归因风险，并将这些与SEER中年龄和种族特异性乳腺癌发病率相结合，得出绝对风险。该模型在来自妇女健康倡议的独立数据中进行了有效性评估。这项工作已经发表并被纳入BCRAT。与一位博士前研究员Elisabetta Petracci一起，我们开发了一个乳腺癌风险模型，其中包括可修改的风险因素，如饮酒、缺乏锻炼和身体质量指数（BMI）。我们估计了通过减少这些风险因素可以获得的绝对风险的降低。我们与一位夏季研究员Mateo Banegas（他是西雅图华盛顿大学的博士生）一起，在妇女健康倡议中量化了拉丁裔妇女的BCRAT表现。我们发现BCRAT在一定程度上低估了风险，但重新校准到最近的SEER率改善了BCRAT的预测。我们分析了华盛顿德系犹太人研究（Washington Ashkenazi Study）的数据，以确定来自已知携带BRCA1或BRCA2基因突变的高风险家族的女性，即使被发现没有携带突变，患乳腺癌的风险是否高于平均水平。由于乳腺癌风险的大多数家族相关性不是由于BRCA1或BRCA2突变，并且由于大多数高风险家族是由于几个成员受到影响而确定的，因此有理由相信，这样的女性仍然比一般人群具有更高的风险，即使风险不像突变携带者那么高。我们的数据和文献回顾表明，这种残留的家族风险可以影响临床管理。有兴趣确定是否增加信息从单核苷酸多态性（snp）可以增加鉴别准确性和有用性筛选风险模型。我们研究了可预见的全基因组关联研究是否会发现足够多的具有足够强关联的snp，从而进一步改善风险预测。我们评估了一系列癌症的几个标准，并得出结论，未来SNP发现的贡献可能是有限的。一项相关评论指出，其他遗传变异，如拷贝数变异或罕见的强变异，可能会提供一些额外的歧视性力量。乳腺癌、子宫内膜癌和卵巢癌具有相同的激素病因学和流行病学危险因素。虽然有几个模型可以预测乳腺癌的绝对风险，但很少有模型可以预测普通人群患卵巢癌的风险，没有模型可以预测子宫内膜癌的风险。使用来自前列腺、肺、结直肠和卵巢（PLCO）癌症筛查试验和AARP-NIH饮食与健康研究的50岁以上白人、非西班牙裔女性的数据，以及来自监测、流行病学和最终结果（SEER）登记处的发病率和死亡率rr，我们建立了模型来估计女性在特定时间间隔内患乳腺癌、子宫内膜癌或卵巢癌的绝对风险。这些模型使用护士健康队列研究的独立数据进行验证。所有模型的危险因素包括胎次和激素替代疗法的使用。此外，乳腺癌模型还包括初产年龄、绝经年龄、乳腺癌或卵巢癌家族史、妇科手术史和良性乳房疾病/乳房活组织检查史、饮酒和体重指数（BMI）；子宫内膜模型包括绝经年龄、BMI、吸烟和口服避孕药（OC）使用情况；卵巢模型包括使用卵巢癌和乳腺癌或卵巢癌家族史。所有模型均经过良好校准(预期(E)与观察(O)癌症的比值为：乳腺癌的E/O=1.03, 95%置信区间[CI] = 0.99 ~ 1.07；子宫内膜癌的E/O=1.00, 95%CI 0.91 ~ 1.11；卵巢癌的E/O=0.91, 95%CI 0.81 ~ 1.01。我们估计了在Kaiser Permanente北加州（KPNC）接受HPV检测和子宫颈抹片检查的33万名妇女患宫颈癌的风险。我们正在进一步估计HPV和巴氏试验结果的新组合在KPNC中的绝对风险，以及多次筛查。我们正在构建一个基于HPV检测的模型，以帮助指导有宫颈癌风险的妇女的诊断检测和治疗。我们正在开发一个统计模型，将筛选方案从自然历史中分离出来。这将使我们能够估计任何筛查方案的风险。我们根据血清炎症标志物的组合估计肺癌的绝对风险。我们正在估计PLCO中肺癌风险基础层的绝对死亡率风险，以评估吸烟者的某些风险基础亚群是否可能从胸部x线肺癌筛查中获得死亡率益处。基于IL28B rs12979860-CC基因型和4个临床预测因子，我们建立了一个模型来预测哪些慢性丙型肝炎患者对干扰素/利巴韦林治疗有持续的病毒学反应。我们提出并发表了两个标准来评估预测疾病发病率风险的模型对筛查和预防的有用性，或评估疾病诊断后管理的预后模型的有用性。第一个标准，遵循PCF的病例比例(q)，是指将发病的个体的比例，包括在最高风险人群中个体的比例q中。第二个标准是需要随访的比例，即PNF(p)，即人们需要随访的处于最高风险的一般人群的比例，以便在注定成为病例的人群中有一定比例的人得到随访。在某些病例中，当死因信息缺失时，我们开发了预测SEER中死于意外癌症的绝对风险的归算方法。与Arpita Ghosh（客座研究员）一起，我们正在开发方法和软件，用于在仅在队列的复杂样本中观察到某些暴露时估计和验证风险预测模型。我们开发了一种基于影响函数的方法来计算绝对风险估计和绝对风险函数的方差。我们将这种方法应用于评估风险因素分布变化对个人和人群绝对风险影响的标准。作为一个例子，我们使用了乳腺癌的绝对风险预测模型除了标准乳腺癌风险因素外，还包括可修改的风险因素。对绝对风险和准则的基于影响函数的方差估计与自举方差估计进行比较。

We continued to develop, refine and evaluate the National Cancer Institutes Breast Cancer Risk Assessment Tool (BCRAT). Using data from the Asian American Breast Cancer Study, we obtained ethnicity-specific relative risks and attributable risks for Asian American women, and we coupled these with age- and ethnicity-specific breast cancer incidence rates from SEER to produce absolute risks. The model was assessed for validity in independent data from the Womens Health Initiative. This work has been published and incorporated into BCRAT. With a predoctoral Fellow, Elisabetta Petracci, we developed a breast cancer risk model that included modifiable risk factors, such as alcohol consumption, lack of exercise and body mass index (BMI). We estimated the reductions in absolute risk that could be obtained by reducing these risk factors. With a summer Fellow , Mateo Banegas, who is a pre-doctoral student at the University of Washington in Seattle, we quantified the performance of the BCRAT for Latina women in the Women's Health Initiative. We found that BCRAT underestimated risk somewhat, but that recalibration to more recent SEER rates improved the predictions of BCRAT. We analyzed data from the Washington Ashkenazi Study to address whether a woman from a high risk family known to carry mutations in BRCA1 or BRCA2 genes had above average risk of breast cancer even if she was found not to carry a mutation. Because most of the familial correlation in breast cancer risk is not due to BRCA1 or BRCA2 mutations, and because most high risk families are ascertained because several members are affected, there is reason to believe that such a woman remains at higher risk than the general population, even though the risk is not as high as for a mutation carrier. Our data and review of the literature indicate that such residual familial risk can affect clinical management. There is interest in determining whether adding information from single nucleotide polymorphisms (SNPs) can increase the discriminatory accuracy and usefulness for screening of risk models. We examined whether foreseeable genome-wide association studies will discover enough SNPs with sufficiently strong associations to make substantial further improvements in risk prediction. We evaluated several criteria for a range of cancers and concluded that the contributions from future SNP discoveries are likely to be modest. A related commentary indicated that other genetic variations, such as copy number variants or rare strong variants, might provide some additional discriminatory power. Breast, endometrial and ovarian cancers share a hormonal etiology and epidemiologic risk factors. While several models predict absolute risk of breast cancer, few models predict risk of ovarian cancer in the general population and none for endometrial cancer. Using data on white, non-Hispanic women over age 50 years from the Prostate, Lung, Colorectal, and Ovary (PLCO) Cancer Screening Trial and the AARP-NIH Diet and Health Study and d RRs with incidence and mortality rates from the Surveillance, Epidemiology and End Results (SEER) registries, we developed models to estimate a womans absolute risk of developing breast, endometrial or ovarian cancer over specific intervals. The models were validated using independent data from the Nurses Health Cohort Study. Risk factors included in all models were parity and hormone replacement therapy use. In addition the breast cancer model included age at first life birth, age at menopause, family history of breast or ovarian cancer, history of gynecologic surgeries and benign breast disease/breast biopsies, alcohol consumption and body mass index (BMI); the endometrial model included age at menopause, BMI, smoking and oral contraceptive (OC) use; the ovarian model included OC use and family history of breast or ovarian cancer. All models were well calibrated (ratio of expected (E) to observed (O) cancers were: E/O=1.03, 95% confidence interval [CI] = 0.99 to 1.07 for breast cancer; E/O=1.00, 95%CI 0.91-1.11 for endometrial cancer; E/O=0.91, 95%CI 0.81-1.01 for ovarian cancer. We estimated cervical cancer risk from 330,000 women undergoing HPV testing and Pap smears in Kaiser Permanente Northern California (KPNC). We are further estimating absolute risks in KPNC for new combinations of HPV and Pap test results, and over multiple screening visits. We are constructing a model based on HPV testing to help guide diagnostic testing and treatment of women at risk of cervical cancer. We are developing a statistical model that will separate the screening protocol from natural history. This will allow us to estimate risk for any screening protocol. We estimated absolute risks of lung cancer based on combinations of serum inflammatory markers. We are estimating absolute mortality risks within lung cancer risk-based strata in the PLCO to assess if some risk-based subsets of smokers might have a mortality benefit from chest x-ray lung cancer screening. We developed a model to predict which patients with chronic hepatitis C would have a sustained virologic response to interferon/ribavirin treatment, based on the IL28B rs12979860-CC genotype and four clinical predictors. We proposed and published two criteria to assess the usefulness of models that predict risk of disease incidence for screening and prevention, or the usefulness of prognostic models for management following disease diagnosis. The first criterion, the proportion of cases followed PCF(q), is the proportion of individuals who will develop disease who are included in the proportion q of individuals in the population at highest risk. The second criterion is the proportion needed to follow-up, PNF(p), namely the proportion of the general population at highest risk that one needs to follow in order that a proportion p of those destined to become cases will be followed. We developed imputation methods for projecting absolute risk of dying from an incident cancer in SEER when cause of death information is missing in some cases. With Arpita Ghosh (Visiting Fellow), we are developing methods and software for estimating and validating risk prediction models when some exposures are observed on only a complex sample of a cohort. We developed an influence function based approach to compute the variances of estimates of absolute risk and functions of absolute risk. We applied this approach to criteria that assess the impact of changes in the risk factor distribution on absolute risk for an individual and at the population level. As an illustration we used an absolute risk prediction model for breast cancer that includes modifiable risk factors in addition to standard breast cancer risk factors. Influence function based variance estimates for absolute risk and the criteria were compared to bootstrap variance estimates.