Chemotherapy leaves genomic scars in relapsed childhood cancers

A tumour that survives treatment can retain a record of that treatment in its DNA. In many relapsed and metastatic childhood cancers, researchers found characteristic combinations of mutations that could be linked to previous therapies. The clearest marks came from platinum-based drugs.

Chemotherapy is intended to damage cancer cells so severely that they die or stop dividing. Some cells, however, survive. Their genomes may then contain not only the mutations that were already driving the cancer, but also new damage acquired during treatment.

This recurring pattern is called a mutational signature. It is not a single mutation. It is a recognizable distribution of DNA substitutions, insertions and deletions produced by a particular type of damage and the way the cell attempts to repair it. Ageing, faulty DNA repair, radiation and some drugs can each leave a different molecular handwriting.

Treatment as a source of mutations

The study analysed 611 whole-genome-sequenced tumours from 544 patients enrolled through precision-oncology programmes in Canada, Australia and the United States. The cohort covered roughly 100 childhood tumour types. Fifty-seven per cent of samples were collected after therapy had begun, and 44% represented progression, relapse, metastasis or a second malignancy rather than the original primary tumour.

The researchers compared primary tumours sampled before or after treatment with advanced tumours, again separated by prior therapy. Medical records were reviewed to reconstruct drug dates, doses and routes of administration, producing more than 3,200 detailed treatment records. This allowed the genomic findings to be related not only to whether treatment had occurred, but also to its timing and intensity.

Post-treatment tumours carried about twice the total burden of somatic mutations seen in treatment-naive tumours. Among 69 signatures involving point mutations and small insertions or deletions, 15 appeared only in treated tumours. In this young cohort, chemotherapy and radiotherapy were the only detectable external sources of mutagenesis.

That does not mean treatment created every mutation present at relapse. Tumours continue to evolve through their own internal processes, while therapy selects cells that are able to survive. The study shows that a measurable layer of drug-specific DNA damage is added to that evolution.

Why platinum drugs stood out

Cisplatin, carboplatin and related drugs bind to DNA and form chemical crosslinks between nearby bases. These lesions obstruct DNA replication and gene transcription. If a cell survives and repairs the damage imperfectly, it can be left with a characteristic pattern of mutations.

Of the 611 tumours, 173 had been exposed to platinum therapy. Seventy-one, or 41%, carried previously recognized platinum signatures, and two newly described signatures increased the number of identifiable cases. Within this cohort, platinum drugs accounted for about 60% of all mutations attributed to therapy.

With whole-genome sequencing, the earliest clear platinum signature appeared 91 days after treatment began, once a tumour contained at least 1,454 single-base substitutions. Deeper targeted sequencing detected faint signals earlier. By 12 months, 35% of platinum-treated tumours carried a detectable platinum signature; by 18 months, the proportion had reached 48%.

The signature was not universal. Even among tumours sampled after sufficient exposure and containing enough mutations for detection, 27% had no recognizable platinum pattern. This may reflect biological differences between tumours, technical sensitivity, or resistant clones that were still too small to dominate the sample.

A treatment scar is not yet a clinical test

In a separate paediatric outcome cohort of only 15 patients, platinum signatures were associated with progressive disease and later recurrence. A similar association was found in 166 adults with metastatic cancers who were treated with platinum drugs again. Signature-positive tumours also showed higher activity of genes involved in detoxifying platinum compounds and pumping them out of the cell.

These observations support a link between treatment-related mutational scars and drug resistance, but they do not prove that the newly acquired mutations caused relapse. The paediatric outcome dataset was very small, and validation in adults cannot substitute for a dedicated childhood-cancer study.

The main cohort was deliberately enriched for difficult-to-treat and relapsed cancers, included around 100 diagnoses and involved many combination regimens. Half of treated children received at least four different agents, making it difficult to separate the effect of one drug from the others. Most of the analysis also compared different tumours sampled at different times rather than following the same tumour serially in every patient.

The findings do not justify reducing chemotherapy doses now, and they do not establish a ready-to-use blood test for emerging resistance. They show something more fundamental: a tumour that survives therapy is genetically different from the tumour that entered treatment. Those changes can be measured, and they may eventually help clinicians follow cancer evolution more closely.