Long-term prediction of prostate cancer death from kallikreins measured in blood

描述（由申请人提供）：我们将确定在44 - 50岁男性中测量的激肽释放酶形式（前列腺特异性抗原[PSA]和激肽释放酶2 [hK 2]）是否可预测前列腺癌的后续转移或死亡。以前没有研究评估在健康志愿者群体中分析的生物标志物是否预测多年后（通常为15-35年）的转移或癌症特异性死亡。在瑞典的低发病率PSA检测在招募（1974-1986年）和长期随访是我们独特的“自然实验”的先决条件。我们不会问“我们能检测出前列腺癌吗？“但是“我们能预测对生活质量有重要影响或预期寿命缩短的前列腺癌吗？“我们希望预测前列腺癌转移或死亡的风险，以进行风险分层筛查和化学预防策略，这必须在这些努力开始之前进行;即在44- 50岁。我们将首先确定在年龄=50时收集的血液中高度优化的PSA和hK 2测量值是否预测前列腺癌的转移或死亡的后期发展;继续确定第二个样本（6年后收集）中的测量值是否增强这些预测。我们将建立统计模型来预测一个人的转移或死亡的概率，前列腺癌使用单一的，或重复测量在44 - 55岁。如果这些模型在R21阶段有足够的准确性，我们将继续在R33阶段，以确定其对40，000名男性的独立队列的准确性。避免了许多常见的问题。其中包括过度诊断：我们知道PSA以高灵敏度（但低特异性）在早期、可治愈阶段检测前列腺癌，然而，许多PSA检测的癌症在患者死于其他原因之前不会导致患者发病或死亡。研究还受到选择偏倚的影响：有家族史或其他风险因素的男性更有可能参加筛查;以及前置时间偏倚，其中明显的生存优势仅与早期诊断有关。关于过度诊断的关键问题，我们的研究终点是死亡或转移性疾病。由于我们的研究队列中PSA筛查的发生率较低，因此几乎不存在任何选择偏倚。由于本研究的回顾性性质，受试者从未受到任何PSA数据的影响，并且我们不会受到提前期偏倚的影响，因为我们不会比较筛选与未筛选队列之间的生存期。使用20世纪70年代中期至80年代在瑞典采集的血液样本，我们能够证明，44 - 50岁的单一血液测试可以预测25年后的前列腺癌。由于更多的男性被诊断患有前列腺癌，而不是死于这种疾病，目前的建议是扩大我们的工作，以确定我们是否可以预测前列腺癌死亡。如果是这样的话，一个单一的血液测试可以用来确定男性患前列腺癌的风险最高，然后可以特别努力为这些人进行密集的筛查和化学预防。

DESCRIPTION (provided by applicant): We will determine whether kallikrein forms (prostate-specific antigen [PSA] and kallikrein 2 [hK2]) measured in men aged 44 - 50 predict subsequent metastases or death from prostate cancer. No previous study have evaluated whether a biomarker analyzed in a healthy volunteer population predicts metastases or cancer-specific death many years (typically 15-35 years) later. Low incidence of PSA testing in Sweden during enrollment (1974-1986) and long-term follow-up are prerequisites for our unique "natural experiment". We do not ask "can we detect prostate cancer?" but "can we predict prostate cancers with important impact on quality of life or reduced life expectancy?" We wish to predict risk of metastases or death from prostate cancer to risk-stratify screening and chemoprevention strategies, which must be made before these efforts start; i.e. in age 44- 50. We will first determine whether highly optimized measures of PSA and hK2 in blood collected at age =50 predict later development of metastases or death from prostate cancer; go on to determine whether measures in a 2nd sample (collected 6 years later) enhances these predictions. We will build statistical models to predict an individual man's probability of metastases or death from prostate cancer using single, or repeated measures at ages 44 - 55. If these models have sufficient accuracy in the R21 phase, we will go on in the R33 phase to determine their accuracy on an independent cohort of 40,000 men. Many common problems are avoided. These include over-diagnosis: we know that PSA detects prostate cancer at early, curable stages with high sensitivity (but poor specificity), however, many PSA-detected cancers would never cause morbidity or mortality to the patient before he dies of other causes. Studies are also affected by selection bias: men with family history or other risk factors are more likely to attend screening; and lead-time bias, where apparent survival advantage only relates to early time of diagnosis. Regarding the key issue of over-diagnosis, our study endpoint is death, or metastatic disease. There will be little if any selection bias due to low incidence of PSA screening in our study cohorts. Due to the retrospective nature of the study, the participants are never influenced by any PSA-data, and we will not be subject to lead- time bias as we will not compare survival between screened vs. unscreened cohorts. Using blood samples taken in Sweden in the mid-1970s - 1980s, we were able to show that a single blood test at age 44 - 50 can predict prostate cancer up to 25 years later. As many more men are diagnosed with prostate cancer than who die from the disease, the current proposal is to extend our work to determine if we can predict prostate cancer death. If so, a single blood test could be used to identify men at highest risk from prostate cancer and particular efforts could then be made for intensive screening and chemoprevention for these individuals.