By Monica Heger
This article was originally published July 12.
Analyzing changes in copy number from tumor sequence data may allow researchers to detail mutation progression and pinpoint early mutations that could aid in diagnosis and help guide targeted therapies, according to recently published research.
In a proof-of-principle study published in Cancer Discovery in June, a team from the University of California, San Francisco, and Lawrence Berkeley National Laboratory tested the method on cutaneous squamous cell carcinoma samples. They sequenced the exomes of eight tumors and matched normal samples, then analyzed the tumor samples for mutation copy number and chromosomal duplications to figure out which simple mutations occurred before and which occurred after the duplication event.
"This is one of the first methods that enables researchers to use sequence data to understand which abnormalities happened earlier in the cancer and which occurred later," said Raymond Cho, health sciences assistant clinical professor at UCSF and a co-senior author of the study with Paul Spellman from Lawrence Berkeley National Laboratory. Knowing which mutations occur early is "critical to early diagnosis and personalizing treatment strategies," he added.
The team chose to analyze cutaneous squamous cell carcinoma because skin cancers typically acquire thousands of mutations, which gave the team many markers to analyze and also enabled them to do exome sequencing, rather than whole-genome sequencing.
"The higher the mutational density, the easier it is to use this archaeological approach and do the statistics to tell if something is early or late," Cho said.
The team theorized that examining regional chromosomal duplications would allow them to identify which mutations occurred before and after the duplication by determining the copy number of the specific mutation — mutations whose copy number is doubled occurred before the event, while those that are haploid occurred after. Additionally, the duplication event itself can be time-stamped by looking at the ratio of heterozygous to homozygous mutations in regions of copy-neutral loss of heterozygosity.
The researchers first looked at mutations in areas of copy-neutral loss of heterozygosity, where a regional chromosomal duplication replaces the matching portion of the paired chromosome. Half of the cases had a copy-neutral loss of heterozygosity in chromosome 17, all of which also had TP53 mutations.
In all four cases, the TP53 mutations were in "high allelic abundance" when compared to the other somatic mutations, "indicating that TP53 mutations occurred and were duplicated before other mutations arose," the authors wrote.
Additionally, of the 63 somatic mutations found in that region, 59 occurred after the loss of the second wild-type allele of TP53. The loss of both copies of TP53 also preceded the majority of somatic mutations found throughout the rest of the exome.
This indicates that "it's not after you lose one copy, but after you lose the second copy that you can get tens of thousands of mutations throughout a cancer," said Cho.
Three other samples that did not have copy-neutral LOH on chromosome 17 nevertheless also had mutations to TP53.
Aside from the duplication on chromosome 17, there were a number of other chromosomal aberrations. In those samples, there were seven other duplications. However, the researchers were able to determine that they occurred after the duplication event on chromosome 17 and after the mutations to TP53.
"Loss of the second TP53 allele appears to precede not only a vast expansion of simple mutations but also the development of chromosomal aberrations," they wrote.
While the TP53 mutations seemed to be the major drivers, in total, the team found 486 nonsynonomous mutations for which they could determine copy number and that fell in a region of copy number loss of heterozygosity. Those included mutations to CDKN2A, TW1, HRAS, NOTCH1, NOTCH2, and PKHD1.
To validate the method, the team examined sequence data from 10 ovarian cancer samples that were sequenced recently as part of the Cancer Genome Atlas project (CSN 6/29/2011). In three samples with a clear copy-neutral loss of heterozygosity on chromosome 17, there was evidence "for complete loss of TP53 as the earliest event."
While the researchers believe that determining early mutations will be important for diagnosing and potentially guiding treatment of cancer, the key finding of this study — that mutations to TP53 are among the earliest — does not bode particularly well for treatment.
[ pagebreak ]
"In this instance, it highlights a problem area more than a solution," said Cho. "People have been trying to [target TP53] for a long time and it's been complicated."
Cancers like melanoma, which are characterized by a high mutation rate, should be "treated a little differently than cancers that don't have that characteristic," said Cho. Targeting therapies to block specific mutations may not be effective in these cases, since this type of cancer is "capable of generating tons of mutations, and new mutations would be able to get past that block."
The method could be useful, however, in early assessment of whether a pre-cancerous lesion is at risk for progressing to cancer. "Many things that are getting diagnosed as cancer early might not progress to cancer," said Cho.
For instance, there is "evidence that a single mutation is present in a lot of abnormal skin cells, even just badly sun-damaged skin, but it seems like it's the second loss of TP53 that makes them incredibly mutation prone," he said.
Cho's group is not the only group to try and determine which mutations came early and cancer and which came later.
Peter Campbell, a group leader for the Wellcome Trust Sanger Institute's cancer genome project, is working on a similar method.
He told Clinical Sequencing News in an e-mail that reconstructing the order that mutations arise in cancer can help "identify the early driver mutations."
His team is developing developing bioinformatics algorithms to piece together the order of mutations in cancer. His algorithms seek to reconstruct the order of genomic rearrangements and then time these rearrangements by the distribution of mutations.
The approach is similar to the Cancer Discovery method, but also allows researchers to temporally order "mutation signatures." For instance, he said, his team has used the algorithm to analyze a previously sequenced melanoma genome, which had the C-to-T mutations associated with ultraviolet light, and identified the emergence of a later signature of G-to-T mutations.
Additionally, other groups have studied tumor samples at various time points throughout the course of a patient's disease to try and figure out the order of mutations.
For example, a group from the University of Oxford is sequencing tumor samples of patients with B-cell chronic lymphocytic leukemia throughout treatment to identify the mutations that are present early on, as well as those that respond to therapy and that arise after treatment (CSN 5/17/2011).
Such approaches, however, are only feasible on a small scale, since they require the constant sampling of patients, while the UCSF method could be applied to any cancer sequencing data, such as that being generated by the TCGA or other large-scale cancer sequencing projects.
Going forward, Cho said that the team will do exactly that — apply the method to TCGA data to analyze other types of cancer, such as colon, lung, and melanoma.
Have topics you'd like to see covered by Clinical Sequencing News? Contact the editor at mheger [at] genomeweb [.] com.