NEW YORK (GenomeWeb) – The mutations that are present in medulloblastomas at relapse differ from those present at diagnosis, according to an international team of researchers. The team, led by the Hospital for Sick Children's Michael Taylor, added that this underscores the need to revisit any tailored therapies being used.
Genomic studies of recurrent medulloblastoma — which is nearly always fatal — have been lacking due to the morbidity and discomfort associated with surgery at recurrence. Because of that, the researchers noted, most targeted therapies tested at recurrence assume that the recurrent tumor is similar to the initial tumor, an notion their new study appearing in Nature questions.
By sequencing a mouse model of recurrent medulloblastoma, Taylor and his colleagues found that genetic mutations present in the initial tumors generally don't overlap with those present in the recurrent tumor. When they extended their study to a cohort of human medulloblastoma cases, the researchers further observed genetic divergence between the major clone present at diagnosis versus that present at recurrence.
"Targeted therapy is unlikely to be effective in the absence of the target, therefore our results offer a simple, proximal, and remediable explanation for the failure of prior clinical trials of targeted therapy," Taylor and his colleagues wrote in their paper.
The major clone at recurrence, they added, likely arose through selection on a minor clone that was present in low numbers at diagnosis.
He and his colleagues studied a mouse model of medulloblastoma in which the mice have Sleeping Beauty transposons inserted in the Math1 compartment of their developing cerebellums, which leads to the development of disease. They then mimicked the treatment patients receive by performing subtotal tumor removal on the mice followed by irradiation. The mice were then monitored for recurrence.
Some 60 percent of the mice relapsed.
By sequencing initial and recurrent tumors from the mice that relapsed, the researchers identified 23 genic Common Insertion Sites (gCISs) in 11 primary tumors and 40 gCISs in the recurrent tumors. The gCISs present in the diagnostic samples are "extremely different" from those in the recurrence samples, they noted.
Only the known medulloblastoma tumor suppressor CBP was found in all compartments, and only Trp53, Arid1b, and Tcf4 were in both recurrent compartments.
To determine whether what they'd found in mice also holds for human patients, Taylor and his colleagues performed whole-genome sequencing on a small cohort of treatment-naïve primary human medulloblastoma and recurrent medulloblastoma tumor pairs, both with and without matched germline DNA.
After analyzing a suite of genetic changes — including somatic SNVs, copy number aberrations, and large-scale structural rearrangements — the researchers again noted a "striking" genetic divergence between therapy-naïve and recurrent medulloblastoma tumors.
Only a small portion of genetic events was shared by therapy-naïve and matched recurrent medulloblastoma tumors, they added.
The post-therapy samples, they noted, exhibited a significant increase in the portion of clonal mutations. In addition, two-thirds of the patients harbored a mutation in at least one gene for which a cancer drug was been identified, the researchers found.
"Cumulatively, our data demonstrate that putative drug targets discovered in therapy-naïve tumors are not, as previously assumed, representative of the targets present at recurrence," Taylor and his colleagues wrote. "A number of targetable events are present in the recurrent tumor as subclonal mutations, indicating that a combination therapy approach may be necessary to minimize evolution of resistance."
Using the EXPANDS algorithm and a complementary statistical analysis called PyClone, the researchers modeled the clonal divergence of the primary and recurrent tumors and examined the extent of clonal selection that occurred. From this, they uncovered three general patterns: first, subclonal lineages in the primary tumor expanded to become clonal at relapse; second, clonal lineages in the primary tumor became subclonal at relapse; and, third, lineages in the primary tumor retained the same cellular prevalence at relapse.
Further, each tumor harbored a significant incidence of mutations that were restricted to the dominant clone. Based on their mouse model findings, the researchers suggested that some of these mutations might be present as subclones at diagnosis. Indeed, they reported that for 16 out of 20 patients, they found evidence for a significant expansion of clones that were initially present at less than 5 percent, indicating that clonal selection is common after therapy in medulloblastoma.
"Our data demonstrate that the model of tumor biology as static is not valid for medulloblastoma, which instead demonstrates striking evolution over time," Taylor and his colleagues wrote. "Clinical trials of targeted therapy based on targets no longer present in the dominant clone at the time of recurrence would seem doomed to failure."
He and his colleagues thus suggested that future clinical trials conduct an additional biopsy to determine whether the target is still present in the dominant recurrent clone.