Acute myeloid leukemia: mutations half a century before diagnosis

Acute myeloid leukemia can appear to strike suddenly. The disease often progresses rapidly, leaving little time between the first symptoms and diagnosis. Yet the cellular history behind it may stretch back decades. According to a new study published in Nature, the first mutation contributing to leukemia development was estimated to have arisen before age 20 in two-thirds of the women who eventually developed the disease. On average, that initial event occurred more than 50 years before diagnosis.

The researchers uncovered this history using blood samples originally collected for an entirely different purpose.

A blood archive with an unexpected second life

For years, a major clinical trial in the United Kingdom collected annual blood samples from tens of thousands of women to investigate ovarian cancer screening.

The samples were stored, creating an unusually detailed biological record of the participants' lives.

Some of those women were later diagnosed with acute myeloid leukemia (AML). This gave researchers at the University of Cambridge and University College London an opportunity that cancer scientists rarely have: access to blood samples collected repeatedly from individuals long before they developed the disease.

The team examined 249 samples from 47 women who subsequently developed AML, comparing them with samples from 46 age-matched women who did not develop a blood cancer.

Some participants had samples available from as many as ten annual time points. The earliest samples had been collected up to 15 years before diagnosis. The median age at AML diagnosis was 72.

A conventional genetic analysis might reveal which mutations were present in a particular blood sample. With repeated samples, however, the researchers could go further and examine how the proportion of cells carrying each mutation changed over time.

They used highly sensitive duplex DNA sequencing, a technique designed to detect rare genetic variants while minimizing sequencing errors. They then combined these measurements with mathematical models of evolutionary competition.

This allowed them to estimate when important mutations first appeared and how quickly the affected cell populations expanded.

In effect, the stored samples provided a partial record of the blood's evolutionary history.

A mutation is not the same as leukemia

Blood production depends on stem cells in the bone marrow. When one of these cells acquires a mutation that improves its ability to survive or reproduce, its descendants may gradually become more numerous.

A population descended from one original cell is called a clone. When that population expands, the process is known as clonal expansion.

Such expansions become increasingly common with age. Many otherwise healthy adults carry blood-cell clones with mutations in genes that are also involved in leukemia. This phenomenon is called clonal hematopoiesis.

In the new study, mutant clones were detected in 72% of the women who remained free of blood cancer and in 89% of those who eventually developed AML.

The presence of a mutation alone was therefore a poor way to distinguish between people who would develop leukemia and those who would not.

The crucial difference lay in how those clones evolved.

Early mutations generally gave cells only a modest growth advantage. Some clones took several years to double in size and could remain a relatively small component of the blood for decades.

By fitting evolutionary models to the repeated measurements, the researchers estimated that the earliest driver mutations often arose surprisingly early in life.

In two-thirds of future AML cases, the first such mutation was estimated to have occurred by age 20.

This does not mean the researchers directly detected leukemia-associated mutations in blood samples taken during childhood. Their estimates were reconstructed from samples collected much later in life. Nor does it mean that these individuals had leukemia as teenagers.

What appears to have started early was the expansion of a cell population carrying a mutation that could contribute to a much longer sequence of events.

As additional mutations accumulated in the same lineage, the pattern changed dramatically.

Each successive driver mutation was associated, on average, with a roughly two- to threefold increase in the clone's growth rate. Some clones carrying four driver mutations doubled in weeks or months rather than years.

One particularly revealing case involved a clone estimated to have originated when a woman was approximately 39 years old.

Initially, that clone doubled about every 2.4 years. Nearly three decades later, its descendants acquired additional driver mutations, and the estimated doubling time shortened to approximately eight weeks.

The woman was diagnosed with AML at age 70.

This illustrates how a disease with an extraordinarily long cellular prehistory can eventually emerge as an aggressive clinical illness.

When one clone takes over blood production

Another important finding involved the competition between different mutant clones.

In some future AML patients, one clone expanded so extensively that it effectively displaced other blood-cell populations.

Researchers call this phenomenon a somatic sweep, a form of selective takeover in which most blood production ultimately descends from a single ancestral stem cell.

Evidence of such takeovers was present in at least one-third of the women who later developed AML. None were detected among the control participants.

The researchers also examined genetic data from the UK Biobank. Their analysis found a strong association between these large somatic expansions and the subsequent risk of myeloid leukemia.

This suggests that the structure of the blood-cell population may provide information that the simple presence or absence of a mutation cannot.

There is also a technical complication.

When a genetic variant is present in a very large proportion of blood cells, conventional sequencing pipelines may mistake it for an inherited variant. Algorithms designed to exclude inherited mutations can therefore remove exactly the kind of signal that might indicate an unusually large mutant clone.

A potentially valuable warning sign could be hidden by routine data filtering.

Still, even a dominant mutant clone does not guarantee that leukemia will develop. Its biological significance depends on the mutations involved, their combinations, the clone's growth dynamics, and other factors.

Could leukemia be prevented before it develops?

Perhaps the most promising aspect of the study is how much of this complex cellular evolution could be explained by relatively simple mathematical rules.

Once a clone was established, its subsequent expansion was often remarkably predictable from its estimated growth advantage and competition with other clones.

The researchers could reproduce observed changes in clone size across multiple years without requiring an elaborate model for every individual case.

This raises an intriguing possibility: future clinical assessments might consider not only which mutations a person carries, but also how rapidly the affected cell populations are growing.

A rapidly expanding clone with several driver mutations may carry very different implications from a small, slowly growing clone that remains stable for years.

The authors also explored a potential preventive strategy.

Their modeling suggests that even a modest reduction in the growth rate of a high-risk clone, maintained over many years, could substantially alter its future expansion and potentially reduce the probability of malignant progression.

That is a theoretical possibility, not an established preventive treatment. The study did not demonstrate that slowing such clones in patients prevents AML.

Several limitations also affect how broadly the findings can be applied.

The main longitudinal analysis involved 47 women, and the earliest mutation dates were inferred mathematically rather than directly observed. The sequencing approach did not capture every possible genetic alteration, and the cohort did not include men.

Importantly, in approximately 21% of cases, the suspected pre-leukemic clones were detectable only in the final one or two years before diagnosis, or were not detected earlier with the sequencing method used.

This suggests that not all AML cases necessarily follow the same long, readily detectable evolutionary path.

Nevertheless, the study provides a much more detailed picture of how AML can emerge.

The first mutant blood-cell clones may appear in early adulthood or even childhood, expand slowly for decades, and compete with other clones before acquiring additional genetic changes that dramatically accelerate their growth.

The blood samples originally collected for ovarian cancer research made it possible to reconstruct that sequence retrospectively.

For some patients, the earliest detectable roots of leukemia were estimated to have existed more than half a century before the disease appeared in their medical records.