Welch H Gilbert, Dey Tanujit
Center for Surgery & Public Health, Department of Surgery, Brigham and Women's Hospital, Boston, Massachusetts.
JAMA Intern Med. 2023 Nov 1;183(11):1255-1258. doi: 10.1001/jamainternmed.2023.3781.
Cancer screening is often promoted as a means to save lives. The question of whether cancer screening truly saves lives is becoming increasingly relevant given the growing enthusiasm for multicancer detection blood tests (ie, liquid biopsies). It is possible in randomized clinical trials for screening to reduce deaths due to the targeted cancer without reducing deaths due to all causes. To explore the feasibility of powering studies for cancer-specific vs all-cause mortality, a series of sample size calculations was performed for selected cancers (breast, colorectal, liver, pancreas, and prostate) and for all cancers combined.
Randomized clinical trials of screening for an individual cancer typically require 100 000 or more participants to test its effect on cancer-specific mortality. Testing all-cause mortality requires trials of more than a million participants. However, the sample size requirements change markedly when considering a randomized clinical trial of screening for all cancers, as is envisioned when using multicancer detection blood tests. In this setting, the question of whether cancer screening reduces all-cause mortality can be reasonably addressed in a trial of fewer than 100 000 participants.
It is not feasible to test all-cause mortality when screening for an individual cancer. However, it is feasible to test all-cause mortality for multicancer screening because cancer deaths are such a large component of deaths in general. Observational data on the effects of cancer screening are misleading. Multicancer screening would entail tremendous costs and potentially substantial harms. For these reasons, a randomized clinical trial is mandatory not only to learn if multicancer screening saves lives, but also to learn how frequently it causes harm.
癌症筛查常被宣传为一种挽救生命的手段。鉴于对多癌检测血液检测（即液体活检）的热情日益高涨，癌症筛查是否真的能挽救生命这一问题变得越来越重要。在随机临床试验中，筛查有可能减少因目标癌症导致的死亡，而不减少全因死亡。为了探索针对癌症特异性死亡率与全因死亡率进行研究的可行性，我们对选定的癌症（乳腺癌、结直肠癌、肝癌、胰腺癌和前列腺癌）以及所有癌症合并情况进行了一系列样本量计算。
针对单一癌症的筛查随机临床试验通常需要100,000名或更多参与者来测试其对癌症特异性死亡率的影响。测试全因死亡率则需要超过100万名参与者的试验。然而，当考虑进行所有癌症筛查的随机临床试验时，样本量要求会显著变化，就像使用多癌检测血液检测时所设想的那样。在这种情况下，在少于100,000名参与者的试验中就可以合理地解决癌症筛查是否降低全因死亡率的问题。
筛查单一癌症时测试全因死亡率是不可行的。然而，测试多癌筛查的全因死亡率是可行的，因为癌症死亡在总体死亡中占很大比例。关于癌症筛查效果的观察性数据具有误导性。多癌筛查将带来巨大成本，并可能造成重大危害。出于这些原因，不仅要通过随机临床试验来了解多癌筛查是否能挽救生命，还要了解其造成伤害的频率。
