The first long-term study of how lung cancer evolves is revealing that therapies targeting multiple proteins in tumour cells could help to outpace the disease.
TRACERx (Tracking Cancer Evolution through Therapy) is the first large-scale, longitudinal study of the evolution of non-small cell lung cancer (NSCLC) in individual patients. Led by chief investigator Charles Swanton, the project, set up in 2014, aims to follow 840 people being treated at 14 UK NHS hospitals, from diagnosis through to cure or relapse and, tragically, death. Some tantalising results have begun to emerge which help to explain why early diagnosis is so crucial for effective treatment, as clinicians have noted for decades.
Most of researchers’ understanding of cancer evolution has come from observing cellular changes in biopsies of tumours or alterations in DNA from tumour cells. Such samples are typically taken for diagnosis, and provide only snapshots of a complex process. By contrast, TRACERx aims to monitor this process as it unfolds. The study compares molecular changes in the responses of tumour cells and also immune cells, across multiple tumour regions and as the tumour progresses, and by doing so can reconstruct the tumour’s evolutionary history.
Scientists in the TRACERx project hope that clinical practice might soon benefit from the emerging insights, which can be combined with those from previous studies on cancer evolution. As well as explaining a person’s immune response against their cancer, the study could guide treatments for targeting the many genetic mutations that accumulate in intractable late-stage cancers. This is vitally important, as more than 80% of people who are diagnosed with stage-four lung cancer survive for less than five years.
Analysis of tumours from the first 100 people enrolled in the study has revealed many genomic changes. These include chromosome deletions and duplications, and in nearly 75% of tumours, the doubling of whole genomes (Jamal-Hanjani et al, 2017). Genome doubling is a potent trigger of chromosomal instability, which drives parallel evolution in the tumours, and leads to intra-tumour heterogeneity in the form of both mutational and copy-number diversity among subclones (Dewhurst et al, 2014). This ability to catalyse chaos means the presence of chromosomal instability is the strongest predictor of poor outcome.
We see our role as taking biological discoveries from the Crick into the clinic at our partner hospitals, and this has been really helped by the Translation team and the Crick’s pro-active attitude. All the clinical trials we have developed can
be traced back to the laboratory programme, and were initiated from the Crick.”
Point mutations in the tumours could be broadly classified into three types of mutational signatures: from tobacco exposure; from an unknown cause associated with ageing; and from mutagenesis due to APOBEC cytidine deaminases, whose normal role is to deactivate invading viruses as part of the immune response. Mutations induced by smoking tend to dominate the ‘trunk’ of the tumour’s evolutionary tree. These founder, or truncal, mutations are present in every tumour cell, and for the most common type of NSCLC — adenocarcinomas that form in mucus-secreting glands — the number of smoking-related mutations in the trunk correlates with the number of cigarettes that the person has smoked. As the cancer advances, APOBECs cause haphazard branch mutations to accumulate in some cells (Jamal-Hanjani et al, 2017).
Whole-genome doubling often occurs early on in patients with lung cancer who have a history of smoking (Jamal-Hanjani et al, 2017), but the evolutionary driver for such a dramatic event was unknown. Lopez and collaborators (Lopez et al, 2020) provided evidence that it is a defence against the accumulation of deleterious mutations in tumours exhibiting chromosomal instability, where essential genes can become haploid due to loss of heterozygosity (LOH). In a pan-cancer analysis which included the TracerX 100 cohort, tumours with haploid LOH had profound evidence of negative selection against further mutation in essential genes—which would be a lethal event for the tumour—but this disappeared after genome doubling. Further, tumours with a low frequency of haploid LOH also had a low frequency of genome doubling, and those with high haploid LOH had a high frequency of genome doubling.
How do tumours adapt to the immune environment and avoid destruction? McGranahan et al (2016) demonstrated that chromosomal instability is one possible explanation for a tumour’s ability to evade the immune system. Chromosomal instability is correlated with an increase in the number of subclonal, rather than clonal neoantigens. Cancers with a high burden of subclonal neoantigens responded poorly to immunotherapy based on checkpoint blockade, in contrast to those whose neoantigens were mostly clonal. Clonality may be important for a number of reasons: subclonal neoantigens will not be present at all sites of disease, and will not be presented by all tumour cells, making effective immune control less likely. Further, a minimum cell fraction may be necessary to elicit T-cell-mediated immune rejection. There is also evidence that clonal diversity results in an immune suppressive microenvironment (Wolf et al, 2017).
Analysis of the TRACERx dataset has demonstrated that in addition to the scattergun deployment of subclonal neoantigens, early stage NSCLCs can evolve other cell intrinsic mechanisms for immune evasion, sculpting themselves in response to the particular immune challenges they face. In so-called immune “hot” tumours, characterised by a brisk lymphoid infiltrate, tumours are concealed from the immune system by defects in antigen presentation: HLA loss occurs in 40% of early stage lung cancers, usually as a subclonal event, and is permissive for branched evolution associated with expansion of mutations predicted to bind the lost HLA allele (McGranahan et al, 2017). Deleterious mutations in other parts of the antigen presentation machinery are also found in “hot” tumours (Rosenthal et al, 2019). In contrast, “cold” tumours, with little or no lymphoid infiltrate, tend to select against clones harbouring neoantigens, through DNA copy number loss or transcript repression; strikingly, there is marked selection against neoantigen mutations in genes essential for NSCLC viability (Rosenthal et al, 2019). Consistent with the importance of immune evasion for tumour progression, patient outcome is worse for tumours with evidence of an immune evasion event.
How might the discoveries from TRACERx and previous studies on cancer evolution guide clinical practice? The overriding message is that it will take multiple approaches to outpace cancers that have sophisticated evolutionary mechanisms, such as lung cancer. Each approach will have to focus on different stages of the disease to improve outcomes. Currently, half of people with early-stage NSCLC relapse within five years of surgery with curative intent, due to the presence of undetected minimal residual disease. The TRACERx study has developed a sensitive method to identify and treat patients at risk of relapse. Truncal DNA mutations common to all lung cancer cells in a patient’s tumour are identified, and following surgery, these mutations can be detected in the blood as circulating free DNA (ctDNA) if there is minimal residual disease (Abbosh et al, 2017).
Together with US company Archer Diagnostics, Swanton and colleagues have developed a proprietary biomarker assay to seek out ctDNA harbouring these truncal mutations. Joint IP is owned by the Crick/UCL and Natera. The new assay is now in clinical trials following FDA breakthrough approval: AstraZeneca has invested more than £200m in two phase III trials, and Roche recently showed that the assay can be used to specifically identify which patients will have a survival benefit with the use of a checkpoint inhibitor after bladder resection for bladder cancer.
Such trials could test the hypothesis that post-operative chemotherapy is beneficial only to patients who have tumour DNA in their blood after surgery, and therefore have minimal residual disease. For every 100 people who are given chemotherapy after surgery, between five and 15 extra patients will be alive (depending on tumour stage) after five years, compared with those who do not receive chemotherapy. Currently, doctors cannot predict who those individuals might be.
The TRACERx study also offers insight into how immunotherapies for lung cancer could be refined. Currently, malignant cells are treated by cancer therapies that target just one oncogenic protein on their surface. Most people’s tumours become resistant to such therapies after 18 months, through acquired resistance mutations, either in the targeted protein itself or in a downstream pathway. Knowing that the immune system adapts rapidly in tandem with the tumour, clinicians could now target not just one but many such truncal mutations. This might decrease the chances of resistance developing, and could have the advantage of not harming normal tissue. It might also provide a way to keep abreast of complex, fast-evolving cancers.
To target such truncal mutations, Achilles Therapeutics, a Crick/UCL spinout company founded by Swanton and three UCL researchers (Sergio Quezada, Karl Peggs and Mark Lowdell) have developed the first precision tumour-infiltrating lymphocyte therapy. The therapy uses the bioinformatics pipeline developed by Swanton and colleagues at the Crick to identify clonal neoantigens in the TracerX study, and two first-in-human phase I/ II trials, in advanced melanoma and NSCLC, have recently launched. Once a patient’s clonal neoantigens are identified, T cells recognising them are purified from patients, cultured and expanded in the laboratory and transferred back into the patient to amplify his or her immune response against the truncal mutations.
TRACERx and other studies have revealed the complex interactions between malignant tumour cells and normal cells in the surrounding environment, which can even come to depend on each other for survival and growth (Abdul-Jabbar et al 2020). Understanding of these dynamics is still rudimentary, but to assess them over space and time would require frequent sampling of a patient’s tumour, which is impossible. Therefore, the TRACERx project includes a national autopsy study called PEACE that helps to resolve this problem and sample more deeply.
Scientists on the PEACE study are analysing the tumour material of 20 patients from TRACERx who gave consent for their tissues to be used in research after their death. Work in progress is beginning to reveal how cancer cells spread from the primary tumour to distant sites in the body. TRACERx is also illuminating how cells that are shed from the tumour at surgery reflect the future site of spread months later (Chemi et al, 2019).
Together with its sister project, TRACERx Renal (see Tracking the evolution of clear cell renal cell carcinoma through multi-regional sequencing), this comprehensive mapping of the evolutionary process driving cancer is unprecedented in its insights into the disease. Future approaches include the imaging of “normal” cells in the microenvironment of a growing tumour, and single-cell nucleic acid sequencing, preferably in situ to avoid disruption of the tumour architecture. With their close links to the clinic, the TracerX projects should improve the survival and quality of life of cancer patients suffering from currently intractable tumours.