FlyTitle: Oncology

The most comprehensive genetic map of cancers ever made shows how hard they will be to crack


经济学人双语版-肿瘤测绘 The topography of tumours

PERHAPS MORE than any other, cancer is seen as a disease of genes gone wrong. So, as genetic-sequencing technology has become cheaper and faster, cancer scientists are using it to check which changes to genes cause tumours to spread.


The latest insights from one group, the international Pan-Cancer Analysis of Whole Genomes (PCAWG), are revealed last month in Nature. In an analysis of the full genomes of 2,658 samples of 38 types of tumour taken from the bladder to the brain, the researchers give a blow-by-blow account of how a series of genetic mutations can turn normal cells into runaway clones. It provides the most comprehensive analysis yet of where to find this damaging disruption to DNA and, by unpicking the genetics of what makes cancer tick, just how hard it will be to tame.


For each of the cancer samples, the team produced a read-out of the tumour genome—the 3bn or so individual DNA letters—and compared it with the genome sequences of healthy cells taken from the same patients. In this way they could look for the genetic signatures of the cancer cells, where specific mutations had warped the genetic information.


Most mutations in the genome are harmless. But driver mutations, where genetic changes cause a cell to multiply more easily and faster than other cells, can trigger tumour growth. Many driver mutations have been found over the past decade and a handful have been translated into new medicines. In a fifth of breast cancers (pictured), for example, a driver mutation in the gene HER2 makes cells produce more of a protein on their surface that encourages them to grow and divide out of control. A series of drugs, including Herceptin, target this protein, and lead to significantly improved survival rates. The same HER2 mutation also appears in some lung cancers, raising hopes that similar therapies could work against that disease.

基因组中大多数的突变都是无害的。但发生“驱动突变”时,基因变化会导致突变细胞比其他细胞更容易并更快速地繁殖,可能触发肿瘤生长。过去十年中, 人类已经找到了不少驱动突变,并基于其中的几个研制出了新药。例如,在五分之一的乳腺癌中(见图),HER2基因发生的驱动突变使细胞表面产生更多的蛋白质,导致细胞的生长和分裂失控。包括赫赛汀(Herceptin)在内的一系列靶向药物就是针对这种蛋白质,显著提高了存活率。部分肺癌中也出现了同样的HER2突变,因此类似的疗法也有望用于治疗肺癌。

The problem is that most cancers have multiple driver mutations. Indeed, the PCAWG work found that on average each cancer genome carried four or five. And with some clever genetic archaeology they also found that some driver mutations can occur years before symptoms appear.


To discover this, researchers used a new concept called “molecular time” to reconstruct the cellular evolution of tumour cells. By comparing the DNA of cells within tumours, the researchers could place mutations in chronological order based on how many cells they appeared in. Earlier mutations occur more frequently. For example, driver mutations in a gene called TP53 were found to have originated at least 15 years before diagnosis in types of ovarian cancer, and at least five years before in types of colorectal and pancreatic cancer. Driver mutations in a gene called CDKN2A were found to have occurred in some lung cancers more than five years before diagnosis. In theory, that provides a window in which to find people at risk of developing these diseases, and perhaps prevent the cancer ever appearing.


The new study closes down talk that significant numbers of unknown driver mutations could lurk in the relatively unexplored regions of the human genome. One such driver mutation in non-coding DNA was found in 2013—a mutation in the TERT gene across many different cancer types. To check for more like this, the consortium sequenced and analysed all the DNA letters of these non-coding regions (which account for 98% of human DNA) for the first time. They found that non-TERT driver mutations occurred at a rate of less than one per 100 tumours in these regions.


Peter Campbell of the Wellcome Sanger Institute in Cambridge, Britain, and a member of the PCAWG consortium, says an important contribution of the study is that by sequencing so many tumours it has raised the number of patients in whom a genetic contribution to their cancer can be identified from less than 70% to 95%. The goal, he says, is for genome sequencing of tumours to become routine. Efforts to introduce this are under way in some countries, including Britain, the Netherlands and South Korea, he adds.

研究团队成员、英国剑桥的威康桑格研究所(Wellcome Sanger Institute)的彼得·坎贝尔(Peter Campbell)指出,这项研究的一大贡献在于,通过对这么多肿瘤测序,把可以识别出其所患癌症的遗传因素的病患比率从不到70%升至95%。他表示,研究的目标是将肿瘤基因组测序常态化。他补充道,包括英国、荷兰和韩国在内的一些国家正在努力推行这一做法。

Results, results, results


Insights are all very well, but what about cold, hard clinical progress? Turning genome sequences into meaningful predictors of cancer will require comparisons between samples from tens of thousands of patients, say the researchers, along with data on their treatments and survival rates. Processing this would be beyond the reach of any single organisation. Instead, a follow-up project is planned that includes national funding agencies, charities and corporate partners from more than a dozen countries around the world. It aims to link full sequences of 200,000 cancer patients to their clinical data by 2025.■