Wu et al. reply
Wu et al. reply
复制标题
DOI：
10.1038/nature23303
复制
发表时间：
2017-08
期刊：
影响因子：
64.8
通讯作者：
Song Wu;Wei Zhu;Y. Hannun
中科院分区：
文献类型：
--
作者：
Song Wu;Wei Zhu;Y. Hannun
我们最近报道了癌症的发病率和相应正常组织中干细胞分裂数量之间的强烈相关性1。我们将这种相关性解释为意味着在DNA复制过程中产生的随机基因组改变(称为复制性或内在)，以及由环境(外部)和遗传因素引起的突变，在肿瘤发生1-3中具有重要作用；然而，我们没有量化内在因素和外部因素对任何癌症类型1的贡献。在他们的研究中，Wu等人。4估计了内在因素对许多癌症类型的贡献上限，得出的结论是内在因素在癌症病例中所占比例“不到10-30%”。我们认为，这些作者的几个假设导致他们低估了内在因素的作用，我们进一步表明，他们的方法之一导致了这样的结论：在外在因素没有作用的情况下，外在因素占风险的85%。这条评论的回复是由Wu，S.，朱，W.&Hannun，YA Nature548，http://dx.多伊。Org/10.1038/Nature23303(2017年)。1、Wu et al.选择了七种癌症类型来获得“内在风险下限”(LBIR)线，代表他们认为具有零外部风险的癌症(参考文献中的图3a)。4)。而不是使用参考文献中图3A底部的癌症类型。4为了定义LBIR线，可以类似地使用参考文献中图3a顶部的癌症类型。4.定义“上界外在风险”(Uber)线(方法)。在最接近这条线的七种癌症中，74%是可以预防的(http://www.癌症研究中心。因此，org/cancer-info/cancerstats/causes/preventable)和90%将是这条线上癌症的外在风险的保守估计。使用这条优步路线来确定其他24种癌症的风险，平均而言，可以计算出总风险的80%是由内在因素造成的(图1)。相比之下，使用LBIR行，平均只有7%的总风险被计算为由固有风险造成(参考文献中的扩展数据表1)。4)。如果假设优步线路的外部风险值与流行病学证据不一致(例如，外部风险为25%或≥为99%)，则可归因于内在因素的风险要么极高，要么极低(图1)。因此，基于边界的方法可能会产生大相径庭的结论，这仅仅取决于所选择的是上边界还是下边界，以及边界所代表的外部风险的比例。考虑到需要进行许多不同和复杂的生物实验来确定干细胞分裂的数量，对干细胞分裂终生数量的估计是有争议的。在我们最初的论文中，我们通过评估噪声对所有癌症类型的影响进行了稳健性分析1。由于每个器官中干细胞分裂的终生数量对于LBIR线的定义至关重要，并且对于癌症类型到LBIR线的距离来说，干细胞估计中的噪声可能强烈影响对外部风险的估计。Wu等人。认识到了这个问题，并进行了稳健性分析。然而，我们认为，在定义LBIR线时，没有充分考虑噪声的影响；这条线构成了比较所有其他癌症类型的基线。我们进行了模拟，以评估噪声对用于获得LBIR线的数据和Wu等人得出的结论的影响(方法)。图2a显示了一个典型的模拟，其中外部风险被假设为10%。引入的噪波将代表真实值的绿点转换为红点。所有…的立场
We recently reported a strong correlation between the incidence of cancers and the number of stem-cell divisions in the corresponding normal tissues 1. We interpreted this correlation to mean that random genomic alterations (termed replicative or intrinsic) arising during DNA replication, as well as mutations that arise owing to environmental (extrinsic) and inherited factors, have important roles in tumorigenesis 1–3; however, we did not quantify the contribution of intrinsic versus extrinsic factors to any cancer type 1. In their study, Wu et al. 4 estimated an upper bound for the contribution of intrinsic factors to many cancer types, concluding that intrinsic factors account for “less than 10–30%” of cancer cases. We believe that several of the assumptions made by these authors led them to underestimate the role of intrinsic factors, and we further show that one of their methods leads to the conclusion that extrinsic factors account for> 85% of the risk in situations in which extrinsic factors have no role. There is a Reply to this Comment by Wu, S., Zhu, W. & Hannun, YA Nature 548, http://dx. doi. org/10.1038/nature23303 (2017).On the basis of the data in ref. 1, Wu et al. chose seven cancer types to obtain a ‘lower bound intrinsic risk’(LBIR) line, representing cancers that they assumed to carry a zero extrinsic risk (figure 3a in ref. 4). Instead of using the cancer types at the bottom of figure 3a in ref. 4 to define an LBIR line, one could analogously use cancer types at the top of figure 3a in ref. 4 to define an ‘upper bound extrinsic risk’(UBER) line (Methods). 74% of the seven cancer types nearest this line are preventable (http://www. cancerresearchuk. org/cancer-info/cancerstats/causes/preventable) and 90% would therefore be a conservative estimate for the extrinsic risk in the cancers on this line. Using this UBER line to determine the risk of the 24 other cancer types, on average, 80% of the total risk can be calculated to result from intrinsic factors (Fig. 1). In contrast, using the LBIR line, on average, only 7% of the total risk is calculated to result from intrinsic risks (extended data table 1 in ref. 4). If one assumes values of extrinsic risk for the UBER line that are inconsistent with epidemiologic evidence (for example, extrinsic risk< 25% or≥ 99%), the risks attributable to intrinsic factors are either extremely high or extremely low (Fig. 1). Boundary-based approaches can therefore yield widely variant conclusions simply depending on whether an upper or lower boundary is chosen and the fraction of extrinsic risk the boundary is assumed to represent. The estimates for the lifetime number of stem-cell divisions are noisy given the many different and complex biologic experiments required for their determination. In our original paper, we performed a robustness analysis by assessing the effect of noise on all cancer types analysed 1. Because the lifetime number of stem-cell divisions in each organ is critical for the definition of the LBIR line, and for the distance of a cancer type from the LBIR line, noise in the stem cell estimates could strongly affect estimates of extrinsic risk. Wu et al. recognized this problem and performed a robustness analysis. However, we believe that the effects of noise were not taken sufficiently into account for defining the LBIR line; this line forms the baseline to which all other cancer types are compared. We performed simulations to evaluate the effect of noise on the data used to obtain the LBIR line and on the conclusions reached by Wu et al.(Methods). A typical simulation in which the extrinsic risk was assumed to be 10% is shown in Fig. 2a. The introduced noise transforms the green dots, representing the true values, into the red dots. The positions of all the …