WGS
Whole-genome sequencing of tumor and matched normal tissue or blood.
Infrastructure › GOMAP
GOMAP, the GISTT-omics Atlas Program, collects the data that the open questions about GIST require. The hypotheses come first; they decide which samples are collected and how many.
Proposal for discussion · numbers are targets, not samples collected
The Cancer Genome Atlas (TCGA), run by the US National Cancer Institute and the National Human Genome Research Institute from 2006 to 2018, characterized more than 20,000 primary cancers and matched normal samples across 33 cancer types. Its genomic, epigenomic, transcriptomic and proteomic data, more than 2.5 petabytes, are still among the most used resources in cancer research (NCI).
GIST was not part of the program. There is an unmet need for a platform with well-curated GIST omics data, linked to treatment and outcome.
Whole-genome sequencing of tumor and matched normal tissue or blood.
Whole-transcriptome sequencing.
Genome-wide DNA methylation.
Where fresh-frozen tissue is available.
A whole-slide image for every case; treatment, response, survival and, ideally, the site and time of first metastatic spread.
| Hypothesis | What it needs | Where in GOMAP |
|---|---|---|
| GIST is a remarkably genome-driven cancer. | Whole-genome sequencing of tumor and matched normal for every case, so that drivers and secondary events are seen in full, not only through a gene panel. | All cohorts |
| Oncogenic progression before treatment, not selection under treatment, determines clinical behavior: metastatic spread and response to treatment, including the response of metastases. | Pre-treatment tissue linked to long, well-documented outcome. | Genotype table: A vs B, C vs D |
| Primaries may differ from their metastases. | Primary and metastasis from the same patient. | Column B, longitudinal cohort |
| Primaries are heterogeneous, and where a biopsy is taken matters for predicting metastatic spread. | Several regions of the same primary, each sequenced and imaged. | Heterogeneity cohort |
| 6–12 months of imatinib do not change the genome, but they change the transcriptome and possibly the epigenome. | Paired samples from the same tumor before and after imatinib. | Longitudinal cohort |
| Liver and peritoneal metastases may respond differently, and might need to be treated differently. | Liver and peritoneal lesions from the same patients. | Liver vs peritoneum cohort |
| The primary mutation matters for prognosis, for prediction, and for treatment aimed at cure. That includes the different exon 11 types (557–558 deletions, other deletions, point mutations, duplications and insertions), exon 9, 13 and 17, and the PDGFRA genotypes. | Enough cases per genotype to compare within it, not only across it. | Genotype table |
| The same is needed for all other genotypes: SDH-deficient, NF1, NTRK and the other rare drivers. | Every available case of the rare genotypes. | Genotype table, last rows |
| Genotype | ALocalized, no relapse | BPrimary that spread | CMetastatic, long benefit | DMetastatic, short survival | Total |
|---|---|---|---|---|---|
| KIT exon 11, deletion 557–558 | 40 | 40 | 30 | 30 | 140 |
| KIT exon 11, other deletions | 30 | 30 | 20 | 20 | 100 |
| KIT exon 11, point mutations | 30 | 30 | 20 | 20 | 100 |
| KIT exon 11, duplications and insertions | 20 | 10 | 10 | 5 | 45 |
| KIT exon 9 | 20 | 25 | 20 | 20 | 85 |
| KIT exon 13 or 17 (primary) | 10 | 10 | 5 | 5 | 30 |
| PDGFRA exon 18 D842V | 30 | 15 | 10 | 10 | 65 |
| PDGFRA, other (exon 12, 14, 18 non-D842V) | 15 | 10 | 5 | 5 | 35 |
| SDH-deficient | 15 | 20 | 15 | 10 | 60 |
| NF1 | 10 | 5 | 5 | 5 | 25 |
| NTRK, BRAF and other rare drivers | all available cases, about 20 | ~20 | |||
| Total | 220 | 195 | 140 | 130 | ~705 |
Every case: WGS with matched normal, WTS and methylation of pre-treatment tissue; proteomics where fresh-frozen tissue exists; one whole-slide image.
Why these numbers. The central comparison is A against B: primaries that never spread against primaries that did. Pooled across KIT exon 11, that is about 120 against 110 cases, enough to detect a secondary alteration found in 15% of one group and 35% of the other, with at least 80% power at a two-sided significance level of 0.05. Within each exon 11 type, 30 to 40 cases per column detect differences of about 10% against 40%.
C against D, pooled across the KIT genotypes (about 105 against 100), has the same power for the question of what predicts long benefit. For the rare genotypes the numbers are descriptive: enough to see whether they follow the KIT pattern, not to test it on their own.
| Cohort | Design | Proposed size | Omics |
|---|---|---|---|
| Longitudinal: baseline to late line | 30 patients with a biopsy before and after 6–12 months of imatinib (for example at resection after preoperative treatment), and 30 patients with biopsies at diagnosis and at each progression (at least three time points). | 60 patients | WGS at baseline and at progression; WTS and methylation at every time point. |
| Liver vs peritoneum | Patients with a liver and a peritoneal metastasis sampled at the same time point. | 30 patients | WGS, WTS and methylation of both lesions. |
| Paraffin vs fresh-frozen | The same tumor region preserved both ways; agreement of point mutations, copy number and structural variants decides how archival paraffin tissue can be used. | 30 tumors | WGS of both, WTS of both where RNA quality allows. |
| Heterogeneity of primaries | 10 primary tumors, each cut into 10 regions; neighboring sections are scanned, so that every region has its own digital image. | 10 tumors, 100 regions | WGS, WTS and methylation per region; one whole-slide image per region. |
A whole-slide image is hosted for every case. The Persistence Atlas of the Road to the Cure labs and reGISTTry are planned contributors.
To contribute cases or data, please contact info@gistt-race.org.
“The most effective way to do it is to do it.”attributed to Amelia Earhart, aviation pioneer