Monitoring of cancer patients is an important part of their treatment and this is traditionally done with tests like Computed Tomography imaging (CT) or biomarker analysis. NGS of cancer amplicons, exomes and/or genomes is being discussed as a realistic addition, and possibly an alternative, to these traditional methods. A major hurdle to their use is the fact that analysis of cancers can be hugely complicated by tumour heterogeneity and that serial biopsy of patients to obtain material for sequencing is not a realistic option for many patients. Although the analysis of circulating tumour cells has received a lot of interest, collecting the cells is no easy task (there have been some great papers that show what might be possible from CTCâ€™s, the MALBAC
method published in Science is a current favourite of mine). Nitzan Rosenfelds group
at the Cambridge Institute has been pioneering the use of circulating tumour DNA (ctDNA) sequencing as a more amenable method to understand cancer genomics.
Almost exactly one year ago I had the pleasure of being an author on an exciting piece of work published by Nitzan Rosenfeldâ€™s group at CRI in Science Translational Medicine (â€œthe liquid biopsyâ€, in the news here and here). In a recent NEJM paper the Rosenfeld group collaborated on analysis of 52 metastatic breast cancer patients using targeted ctDNA sequencing. In that study they compared the sequencing to CT-scanning and analysis of CA15-3 (a circulating cancer antigen) and analysis of circulating tumour cells. The ctDNA sequencing was able to identify progressive disease much earlier than CT imaging.
Today Muhammed Murtaza from the Rosenfeld group is lead author on a paper in Nature
that describes his collaborative work on tumour exome sequencing from cell-free circulating tumour DNA (ctDNA) as a non-invasive method to track tumour evolution during treatment. The ctExome is here!
In this proof-of-concept study Murtaza et al demonstrate that ctExome-seq provides a robust non-invasive method to track genome-wide changes in somatic mutations and CNAâ€™s. They followed six patients (2 Breast Cancer, 3 Ovarian Cancer and 1 NSCLC) with advanced cancer over 1-2 years (433 days median clinical follow-up), sequencing exomes from ctDNA collected during treatment monitoring. ctExomes were sequenced when tumour mutant allele fraction was high (as determined by digital-PCR and TAM-seq testing). Comparison of CNV profiles showed a good match between the plasma exomes and tumour whole genomes. The correlation of mutant allele frequencies was high where mutations were found in both plasma and tumour samples (see figure 3 from the paper below).
|Figure 3 from Murtaza et al Nature 2013. a) Plasma exome-CNV, b) tumour WGS-CNV
|How much DNA do you need to sequence an exome:
As little as 2.3ng of ctDNA extracted from about 2ml of blood plasma was used to generate around 100x exome coverage. The exome sequencing was made possible by using a relatively new library prep method from Rubicon
(Iâ€™ve written a follow-up post on Thruplex which you can find here
Murtaza looked to see what changes were visible in the mutational profile during clonal evolution of the tumours, from DNA collected at initial presentation (up to 9 years earlier), metastatic biopsy and ctExomes. In the paper they presented several examples:
- Case 1 BrCA: After paclitaxel treatment an increase in the levels of a PIK3CA mutation, these mutations are known to be involved in paclitaxel resistance.
- Case 2 BrCA ER+, Her2+: After tamoxifen + trastuzumab treatment an initial increase in the levels of a MED1 mutation, this is an ER co-activator and these mutations are known to be involved in tamoxifen resistance. After secondary therapy with lapatinib + capecitabine a mutation in GAS6 was linked to activation of the AXL tyrosine-kinase receptor which is known to be involved with lapatanib resistance.
- Case 4 OvCA: After cisplatin treatment an increase in the levels of a RB1 mutation, which was also seen in 95% of reads from the metastatic biopsy. RB1 loss is known to be involved in chemotherapy resistance.
- Case 6 NSCLC: After gefitinib treatment an EGFR mutation was detected with digital-PCR, this mutation is known to inhibit binding of gefitinib to EGFR and is the main driver of gefitinib resistance.
How might this affect cancer treatment:
The fact that tumours appear to be inherently â€œleakyâ€ and shed DNA from all locations, primary and metastatic, should allow ctDNA analysis to generate a comprehensive picture of tumour burden. The ability to find mutations known to be responsible for resistance to therapy coupled with the ability to detect relapse earlier than CT-imaging will hopefully lead to a more rapid evolution of treatment. As more and more patients are sequenced in this manner we will get an even greater understanding of the evolution of tumours under selective pressure from cancer therapies. This should identify patients most likely to benefit from experimental cancer medicines; they will be identified as not responding to first-line therapies earlier and consequently have more opportunity to move onto novel therapies that may help. However we need to remain realistic, as it is clear that tumours are likely to evolve resistance as quickly as we develop new therapies. This is an arms race, similar to plant:pathogen or bug:antibiotic. Ultimately we are going to need good combination therapies with multiple drugs targeting multiple pathways to stop cancer in its tracks.
How might we use NGS to “triage” cancer-genome sequencing: In both the NEJM paper and the Nature paper a message comes through about using different next-gen sequencing methods to triage the sequencing of patients cancer genomes.
Amplicon-seq : Exome-seq : Whole Genome Seq
Many tumours will yield trackable mutations for follow-up studies using fast and low-cost targeted PCR amplicon sequencing. Those patients whose tumours do not show anything in a targeted screen are very likely to have something useful show up in exome sequencing; which can still be completed at reasonable cost. And if that still does not yield a usable NGS-based biomarker then whole genome sequencing can be performed. This offers an opportunity to target the depth of sequencing to the patients that need it most and make the use of NGS in the clinic as cost- and time-efficient as possible. Perhaps clinical whole-genome sequencing might be considered as a last resort?