A prospective study of patients with non–small cell lung cancer undergoing treatment reveals that chromosomal instability and tumor heterogeneity drive disease recurrence. Researchers involved in the TRACERx trial documented that genetic diversity by sequencing multiple regions of the primary tumor and then created bespoke ctDNA tests that allowed for early detection of relapse and metastasis.

Lung tumors rife with chromosome instability are more likely to return after surgery, according to the first published analyses from TRACERx—the first trial ever to prospectively track genetic heterogeneity and tumor evolution within individual patients with cancer. The findings, described in two new papers, also build the case for using bespoke blood tests to detect disease recurrence earlier than standard radiographic methods, and to retrace the individual tumor subclones responsible for relapse and metastasis.

“What this means is we have to go back to the lab and think differently about how we approach cancer from a therapeutic perspective,” says Alberto Bardelli, MD, a cancer geneticist at the Institute for Cancer Research and Treatment and the University of Torino in Italy, who was not involved in the study. “If we can find ways to restrict chromosome instability, then we might be able to limit the ability of tumors to evolve, which could impact patient outcomes.”

Led by Charles Swanton, MD, PhD, a cancer geneticist at the Francis Crick Institute and University College London (UCL) in the UK, and launched in 2014, TRACERx aims to recruit 842 patients with non–small cell lung cancer eligible for surgery and, if appropriate, adjuvant chemotherapy and/or radiation. So far, Swanton and his colleagues have enrolled 520 patients; the new papers analyze data from the first 100 participants.

For each patient, the team resected the primary tumor, completed whole-exome sequencing of two to eight tumor regions, and constructed phylogenetic trees to trace each cancer's clonal and subclonal evolution. In this way, they showed that intratumor heterogeneity was widespread and largely driven by genome doubling and subsequent chromosomal instability (N Engl J Med 2017 Apr 26 [Epub ahead of print]).

Certain driver mutations, including those in EGFR, MET, and BRAF, almost always occurred early in the evolution of the cancer—which could explain why targeting these alterations typically yields robust and uniform clinical responses—whereas other genetic changes were more commonly found on individual branches of the tumor evolution tree. Still others routinely underwent parallel but independent gene amplifications.

From a prognostic viewpoint, not all of these mutational events were equally informative. Only the extent of structural diversity in the genome, as measured by copy-number heterogeneity, proved to be significantly linked to relapse risk: Patients with greater-than-average copy-number heterogeneity were almost five times more likely to experience disease recurrence or death than those with below-average levels. The proportion of subclonal point mutations had no such prognostic value.

However, those subclonal single-nucleotide variants did enable Swanton and his team to create blood tests customized to each patient's tumor. As evidenced by the presence of circulating tumor DNA (ctDNA) in the blood, these liquid biopsy assays not only found signs of relapse up to a year before routine follow-up (which involved clinical assessment and chest scans), but also pinpointed the exact tumor subclone or subclones responsible for resisting chemotherapy and seeding recurrence (Nature 2017 Apr 26 [Epub ahead of print]).

According to Mariam Jamal-Hanjani, MD, PhD, a study author from UCL, this information could enable oncologists to intervene earlier with treatments directed at specific gene alterations in the recalcitrant tumors. “Taking multiple tumor biopsies from patients in whom cancer has returned or progressed is often not feasible or even safe,” she says, “but perhaps taking regular blood samples to detect when the cancer recurs and to guide treatment based on what we find in the ctDNA is a real alternative.”

To illustrate her point, Jamal-Hanjani refers to a 70-year-old patient who had detectable ctDNA within a month of surgery—and about 5 months earlier than relapse was spotted clinically. A liquid biopsy revealed which subclones from her original tumor had developed resistance to adjuvant chemotherapy. The researchers went back to the original exome sequences and saw that the one responsible for her relapse contained an ERBB2 amplification, indicating that the patient might benefit from trastuzumab (Herceptin; Genentech) or another anti-HER2 therapy.

Christopher Abbosh, MD, another UCL study author, notes that even when no actionable mutation is found, the phylogenetic knowledge could be valuable. “If you know which subclone is responsible for seeding the relapse, then you could target the parts of that subclone that are immunogenic with a vaccine,” he says. Last year, Swanton and others at UCL launched a biotech company called Achilles Therapeutics to develop and commercialize neoantigen-based vaccines using data that emerges from the trial.

Abbosh also suggests that “ctDNA can be used as a marker of adjuvant chemotherapy efficacy, which could enable the evaluation of novel therapies for NSCLC in the adjuvant setting.” He notes that the 70-year-old woman, like two other patients, had a steady increase in the number and frequency of tumor-specific mutations in her blood despite surgery and adjuvant chemotherapy; all three went on to experience relapse within a year. In contrast, another patient who also had detectable ctDNA soon after surgery saw his ctDNA levels drop to zero after completing the adjuvant treatment; more than 2 years later, he remains cancer-free.

Geoffrey Oxnard, MD, a thoracic oncologist at Dana-Farber Cancer Institute in Boston, MA, notes that prior studies had documented tumor heterogeneity and ctDNA shed, but this prospective analysis “grounds those biological phenomena with clinical meaning,” he says. “This reorients us into understanding how we can use chromosome heterogeneity and instability as maybe a more fundamental indicator of prognosis.”

But Oxnard worries about how feasible it will be for most clinicians to apply the methods in routine practice. “The next thing,” he says, “is turning this fundamental biological finding into a practical diagnostic that can inform clinical decision-making.”–Elie Dolgin