Scientists have reconstructed how acute myeloid leukaemia (AML) develops over the decades before diagnosis, showing that in two-thirds of cases the first cancer-driving mutation had been acquired before the age of 20 – more than 50 years before diagnosis.
The discovery could help predict whether an individual will develop the disease years – or even decades – before symptoms arise, raising the possibility of preventing the disease before it develops.
In research published today in Nature, a team at the University of Cambridge and UCL used blood samples collected annually, some dating back 15 years before diagnosis, as a 'fossil record' of the disease's development, tracing it step by step. Using ultra-sensitive DNA sequencing and mathematical modelling, they estimated when each mutation arose and how quickly the affected cells multiplied and competed with other blood cells in the years prior to AML diagnosis.
AML is an aggressive blood cancer affecting around 2,700 people in the UK each year. Only around one in six survive five years after diagnosis. Many healthy older people carry blood cells with mutations also found in leukaemia, a condition known as clonal haematopoiesis, but only a small proportion develop AML. What has not been clear is why, because the process has never been tracked year by year in the run-up to diagnosis.
The researchers used samples from the UK Collaborative Trial of Ovarian Cancer Screening (UKCTOCS), in which around 50,000 women gave a blood sample every year for up to 11 years. The team studied 47 women who later developed AML, at an average age of 72, alongside matched samples from 46 women who stayed blood cancer-free.
This first mutation, acquired before age 20 in two-thirds of the women who developed AML, was followed by decades of slow growth – cells carrying it took years to double in number. Each additional mutation increased their growth rate around two- to three-fold, with cells carrying four cancer-driving mutations doubling in a matter of weeks.
The growth of these groups of cells, known as clones, proved remarkably predictable. Once a clone carrying a cancer-driving mutation was established, it expanded at a steady rate, slowing only when it ran out of room or was overtaken by a faster-growing clone. In practice, this means a clone's growth rate, measured from a few blood samples today, can be used to estimate how large that clone may become years and even decades into the future.
In at least a third of the AML cases, a clone had come to dominate the blood years before diagnosis, a phenomenon known as a 'somatic sweep'. No such sweeps were seen in the women who remained blood cancer-free. The team confirmed the link with AML risk using UK Biobank data, finding that people with these sweeps were at greatly increased risk of developing the disease. The findings highlight the importance of recognising this potential warning sign, which standard genetic analysis often overlooks by mistaking these mutations for inherited variants.
Professor Jamie Blundell from the Early Cancer Institute at the University of Cambridge, who led the study, said: "Serial blood samples are a fossil record of the blood's evolution. They let us go back in time and reconstruct when each mutation arose and how fast it grew. What surprised us most is how predictable the growth of these clones is. Once a clone is established, its growth rate today tells you where it will be years, even decades, from now."
He added: "Because the growth is exponential, even modestly slowing a high-risk clone over the decade or more before diagnosis could substantially reduce the chance it becomes leukaemia. This opens up strategies we can exploit to prevent these clones ever becoming leukaemia."
Dr Caroline Watson, also from the Early Cancer Institute, and first author of the study, said: "As a haematologist, I've seen how suddenly AML can change someone's life. This study shows that the changes leading to AML can build over decades, with the affected cells growing in surprisingly predictable ways."
Dr Watson is also an Honorary Consultant Haematologist and sees patients at the Cambridge Clonal Haematopoiesis Clinic at Addenbrooke's Hospital, the first clinic focusing on leukaemia prevention in Europe. She added: "At our clinic, we already assess and monitor people who carry blood-cell mutations that increase their risk of leukaemia. Today's findings give us real hope of improving how we identify which of them are at greatest risk and point to a window in which prevention could one day be possible."
Professor Usha Menon, from the UCL Institute of Clinical Trials and Methodology, who led UKCTOCS, said: "While UKCTOCS was designed to assess ovarian cancer screening, a key outcome was the establishment of a unique bioresource of serial serum samples. This resource has provided unprecedented opportunities to study how cancers develop over time and to discover and validate biomarkers for earlier detection. We owe an enormous debt of gratitude to the participants whose willingness to contribute their samples and data over many years made this possible. Their generosity is the foundation of this achievement."
The women were diagnosed with AML between the ages of 53 and 83, and the study did not include samples taken at time of diagnosis. In around one in five cases, expanding populations of mutated cells were not detectable until one to two years before AML diagnosis, suggesting that some cases may offer a shorter warning period.
The findings point towards opportunities for early detection and prevention. The team's modelling suggests that even a small reduction in a clone's growth rate, compounded over the ten or more years between the second mutation and diagnosis, could substantially reduce the risk of progression to leukaemia. This raises the possibility of preventative treatments that slow these cells' growth before leukaemia develops. Testing whether this can be achieved safely and effectively in people identified as high risk is the team's next step.
The research was largely funded by UK Research & Innovation and Cancer Research UK.
