Advertisement

The Long, Slow Process of Carcinogenesis

  • 20 Jan 2022
  • By Derek Lowe
  • 3 min read
  • Comments

When someone is diagnosed with cancer, it's a natural response for them to wonder what they did wrong, and how they could have avoided it. And while there are some behaviors and environmental effects that have been definitively tied to increased risk, for the huge majority of cancers there is no way to point at any one thing. Sadly (and understandably) that doesn't stop people from going back over their life histories wondering what it was they did (or neglected to do) that led them to the diagnosis they have now.

Here's a new paper that might help put this question into a better perspective. It's a look at myeloproliferative neoplasms (MPNs) - blood cell cancers such as AML or CML where a single type of blood cell is produced in huge numbers. These occur when something goes wrong with the corresponding stem cells in the blood marrow, and this tissue specificity along with the ability to do early diagnosis with blood tests makes these disorders valuable places to study cancer mechanisms in general. It's clear that there are mutations in the stem cells ("driver mutations" that lead to a cancer phenotype), and for many years it appeared that these might occur late in life and not that long before diagnosis. The studies of increased leukemia risk in survivors of the Hiroshima and Nagasake atomic bombings originally supported this view, but long-term follow-up (see that link) shows a complex situation with regard to radiation exposure, age at the time of the bombings, time elapsed since 1945, and the type of leukemia that developed. And it's long been known that people who do not show signs of actual leukemia can harbor one or more of these driver mutations. Some of these people do indeed go on to develop MPNs, which suggests that there might be a longer "multi-hit" process that could go on for many years.

This work supports that idea. The team studies twelve MPN patients, whose tissue samples provided over a thousand different clones of malignant blood cells. Sequencing these turned up over 580,000 mutations (!), and the paper puts these into a phylogenetic framework to reconstruct the sequence of what the key mutations were and when they might have taken place. Using rates of mutation as a clock, some of them appear to go back even to before birth - the key JAK2V617F mutation, long associated with these malignancies, is estimated to have shown up anywhere from the 33rd week of gestation up to the age of 11. The DNMT3 mutation, similarly, seems to have appeared from the 8th week of gestation (!) out to about the age of 8. Additional driver mutations layer on top of these early events over the years to come - the mean latency between the JAK2 mutation and diagnosis of cancer, for example, was about thirty years.

There's still a lot of variability inside that number. The rates of clonal expansion for these various mutated cells vary widely by the mutation type, and among a given mutation they vary among patients themselves. This suggests that there are additional factors are at work, either genetic ones or environmental ones in the bone marrow, that influence the effect of having one of these drivers. But in all cases, it seems clear that it takes many years for MPNs to develop - the diagnoses that are made in the clinic are capturing the end result of what is often a decades-long process of accumulated mutations and clonal expansion. 

This suggests that targeting therapies towards these mutated cells earlier in life, before the patients involved even have cancer at all, could be a really useful strategy. And it also suggests that this framework doesn't apply only to blood cancers, either (it's just easier to prove there). As the authors note, all sorts of other apparently normal tissues reveal mutations when you sequence them, and you'd expect this sort of long, slow pileup of mutations in various clonal expansion lines to be operating there, too. 

Postscript: you may have already noted the difference between these results and the situation with childhood leukemias. Some of these (especially childhood ALL) are much more treatable, as is well known, and that's because they tend to be driven by a whole different set of mutations, manifesting disease much more quickly, many of which lead to cell populations that are more responsive to chemotherapy, radiation, and other treatments. Some of these mutations can go back to before birth as well, but the whole process is clearly much faster, and the cancerous cells themselves thus have different properties and vulnerabilities.