GISTT RACE · Research Alliance for CurE

Trial

A synopsis of the GISTT-RACE platform trial, and the tasks the working groups need to settle before the first patient can enroll.

Synopsis

Working title
GISTT-RACE platform trial
Aim
One academic, international trial platform that tests the most promising treatments for cure, chosen by tumor biology rather than commercial interest.
Design
  • Prospective master protocol with adaptive design
  • Seamless transition into phase III
  • Growing historical controls; aligned with industry where helpful
  • Minimal burden for patients to be included
Populations
  • Localized GIST: move to neoadjuvant treatment in resectable disease, with window-of-opportunity cohorts
  • Metastatic GIST, first line: focus on potentially resectable disease
  • Prospective molecular and clinical cohorts
Endpoints
Pathological response, ctDNA and functional imaging as novel endpoints; functional endpoints from resection specimens; the burden of treatment as patients experience it, with the planned annoyance scale.
Samples and data
Imaging, pathology and molecular data from every patient; on-treatment biopsies to be considered; specimens collected for research.
Built-in validation
Prognostic scores, including AI-based ones, validated on the platform.
Target start
2028 (2024 roadmap)

Trial templates

One master protocol, three populations — and one trial that is already running at a single center, as the template for a biomaterial-intensive cohort. The three designs are draft concepts, not protocols: the arms, the schedule and the endpoints are placeholders until the working groups settle them. What matters here is the shape, and where a trial meets the platforms. Open a tile to see the design.

Localized GIST

Draft concept

Resectable disease. Treatment moves in front of surgery, so that the resection specimen becomes the readout.

Show the designHide
Readouts along the way
Pathological responseImmunogenic cell deathMetabolismPersistence
Entry
Diagnosis and biopsyPathology, mutation, imaging, plasma
Assignment
ExperimentalWindow of opportunity
ImatinibStandard
No treatmentDirect to surgery
Local therapy
ResectionSpecimen to pathology, biobank and omics
After surgery
Risk-adapted adjuvantIntensity from the risk model
Observation
Follow-up
Imaging and ctDNAInto the registry

Endpoints: Primary readout at resection, with recurrence-free survival as the long-term endpoint.

Metastatic GIST, first line

Draft concept

Focus on potentially resectable metastatic disease, with local therapy and maintenance inside the trial.

Show the designHide
Readouts along the way
Pathological responseImmunogenic cell deathMetabolismPersistence
Entry
Diagnosis and biopsyPathology, mutation, imaging, plasma
Assignment
ExperimentalWindow of opportunity, then combination
ImatinibStandard
Local therapy
Surgery, ablation or radiotherapyWhere the disease becomes resectable
Maintenance
Experimental maintenanceAimed at persistent disease
Imatinib aloneStandard
Follow-up
Imaging and ctDNAInto the registry; progression assigns the next cohort

Endpoints: Pathological response and target inhibition in resected tissue; ctDNA and functional imaging as early endpoints.

Resistance

Draft concept

Progression on standard therapy. The resistance mechanism, not the line of therapy, assigns the cohort. Not yet discussed in the consortium.

Show the designHide
Readouts along the way
Pathological responseImmunogenic cell deathMetabolismPersistence
Entry
ProgressionRe-biopsy and ctDNA define the resistance mechanism
Assignment
Matched experimental cohortBy secondary mutation or bypass pathway
Standard next lineBy label and genotype
Local therapy
Resection or ablationOf the resistant lesion where feasible
Not feasibleSystemic treatment continues
On treatment
Continue until progressionOn-treatment biopsy and serial ctDNA
Follow-up
Next assignmentBack into the registry and the next cohort

Endpoints: Clearance of the resistant clone on ctDNA, target inhibition in the resected lesion, and time to the next treatment.

Last line, biomaterial-intensive: PATHFINDERS

Running · single center, not open to other sites

Running at one center and not open to other sites. It serves as the template for a biomaterial-intensive cohort: how much tissue, plasma and imaging one patient can contribute in late-line disease, and what that delivers to the platforms before the master protocol exists.

Show the designHide
Readouts along the way
Pathological responseImmunogenic cell deathMetabolismPersistence
Entry
Progression in late lineEvery lesion tracked on serial CT
Profiling
Resistance profilectDNA, PET and long-read sequencing of tissue
Imaging aloneWhat routine care does today
Local therapy
Local treatment of resistant lesionsWhere only a few lesions drive progression
Decision
Treatment chosen on the profileRechallenge, off-label through the tumor board, or a trial
Follow-up
Profile into the registryThe profiles define the cohorts of the platform trial

Endpoints: Feasibility of profiling in routine care, and the resistance profiles themselves. As a template it sets what a biomaterial-intensive cohort asks of a site: serial plasma, a re-biopsy at progression, PET during treatment, and tissue for long-read sequencing. See the PATHFINDERS synopsis.

How the trials and the platforms connect

The same five phases run through all the templates. At each one, the trial hands something to a platform and takes something back.

PlatformEntryAssignmentLocal therapyAfter / on treatmentFollow-up
reGISTTryRegistration, baseline dataAssignment recordedProcedure and specimen logTreatment and toxicityOutcome, and controls for later trials
Model platformBiopsy to modelsFresh tissue to models and PDXResistant lesions to modelsNovel variants into the pipeline
GOMAPWGS, WTS, methylome at baselinePaired sample under treatmentOn-treatment biopsySample at progression
Imaging and AIBaseline CT, PET or MRIRisk model assigns intensityResponse before surgeryFunctional imagingValidation of prognostic scores
Trial platformEligibility by master protocolArms added, adapted, closedQuality standards for local therapyAdaptive rulesSeamless transition into phase III

The growing control backbone

A patient who receives standard treatment inside one template, or who is followed in the registry under the same rules, is a control for every later comparison. If every cohort contributes to the same pool, that pool keeps growing, and fewer patients have to be randomized to the control arm later on.

One core data set for everyone

Every patient in every template, and every registry patient who meets the same criteria, provides the same core: diagnosis, mutation, imaging at fixed intervals, treatment and outcome. Without an identical core, nothing below works.

Concurrent randomization stays the primary comparison

Each template keeps a randomized control arm. The pool supplements that arm, it does not replace it.

The pool supplements the control arm

Patients from earlier cohorts and from the registry who match on the prespecified factors are added to the concurrent controls, with a cap on how much weight they can carry.

Fewer patients on the control arm over time

As the pool grows, randomization can shift from 1:1 toward 2:1 or 3:1 in favor of the experimental arm, so that more patients receive the new treatment for the same statistical power.

How it accumulates
Localized: no-treatment and imatinib arms
Metastatic: imatinib arm
Resistance: standard next line
Registry and PATHFINDERS: patients on standard treatment, same core data
One control backboneSame core data, same imaging intervals, same outcome definitions
First cohorts: 1:1→ 2:1→ 3:1 as the pool grows

What has to be settled first

  • Standards of care change over time, so a control from 2028 is not a control from 2033. Borrowing has to be limited, time-stratified and prespecified.
  • Patients who enter a trial differ from patients who do not. Matching on the recorded factors cannot remove what was never recorded.
  • Regulators accept external controls in some settings and not in others. The design has to be agreed with them before the first patient, not after.
  • How much borrowing, by which method (for example dynamic borrowing or matching), and the cap on it: to be decided by the working groups.

What is needed for patients to enroll

Each central question has a working group. Its task for the trial defines one part of what a patient needs, or receives, to take part.

#Working group (central question)Task for the trialWhat it settles for enrollment
1How do we make a fast and correct diagnosis of GIST and the specific subtype?Define minimal standards for diagnosisWho is eligible: a confirmed diagnosis of GIST and its subtype
2What clinically useful information can and should be taken from a tumor, a biopsy, plasma or other sources to best prognosticate a patient or predict treatment?Define standards for biobanking, decentralized and centralized testing including digital data storageWhat each patient provides at entry: tissue, plasma and data
3What are therapeutic vulnerabilities beyond the truncal event (KIT/PDGFRA)?Come up with window of opportunity format to study those vulnerabilitiesWho can enter a short window-of-opportunity cohort before surgery
4How do we most efficiently and selectively inhibit the KIT and PDGFRA oncogenic signal?Correlative project to study the target inhibition from surgical specimenHow surgical specimens are handled to measure target inhibition
5Who should receive a medical treatment following or preceding surgery of a localized GIST?Define a risk prognostication tool with the highest possible accuracy in order to reduce both over and undertreatment.Which patients with localized GIST qualify for treatment before or after surgery
6Which treatment should patients with localized GIST who are at risk for metastases receive to cure GIST?Define current state of the art for every GIST genotypeThe standard treatment each genotype receives
7When should patients with GIST receive staging or follow-up studies during or following treatment for localized and metastatic disease?Define imaging (and maybe plasma-based) studies during treatmentThe imaging (and plasma) schedule during treatment
8How do we best characterize resistance in treatment refractory GIST?Define diagnostic approaches and testsWhich tests at progression assign a patient to a cohort
9How can we predict and outrun resistance before it appears?Design a protocol that treats resistance before it arisesEntry rules for cohorts that treat resistance before it arises
10What is the role of surgery to cure?Define limits and prerequisites for surgery in metastatic diseaseWho qualifies for surgery in metastatic disease
11Why do some tumor cells survive treatment without being resistant, and how do we eliminate them?Set up a translational program and define the type and number of samples needed to answer the goalThe type and number of samples each patient contributes
12Which models do we need to develop and which should we prioritize on our road to the cure?Add correlative studies that may help to develop those modelsOptional samples for model development
13Can and should radiotherapy be part of the armamentarium of treatments to CURE GIST?If confirmed that protocol needs a section that defines the role and a standard in a metastatic settingWho qualifies for radiotherapy in metastatic disease
14How do we know, for each patient, how much treatment is needed?Choose and validate the risk model used for cohort assignment (proposed)Which risk model assigns patients to a treatment intensity
15How do we run trials fast enough and appreciate the individual genotype of GIST?Write the master protocol and the rules for adding and closing arms (proposed)The master protocol: how arms are added, adapted and closed
16What is the best treatment for GIST without KIT or PDGFRA mutations?Define cohorts and standards for GIST without KIT or PDGFRA mutations (proposed)Cohorts for GIST without KIT or PDGFRA mutations

GIST is a global challenge

GIST occurs everywhere, but most trials run in a few high-income countries. If every patient on earth could be included, the trial has to be designed around the differences between health systems, not despite them. Design ideas for discussion:

A minimal core, optional modules

The core protocol asks only for what every site can deliver: diagnosis, CT imaging, routine labs and follow-up. ctDNA, PET or MRI, research biopsies and biobanking are modules for sites that can offer them.

Central diagnostics, free of charge

Tissue and plasma are shipped to reference laboratories, so no patient is excluded because mutation testing is not available locally.

Drug supply by the trial

Including standard therapy where it is not reimbursed, so that taking part never depends on insurance or income.

Standard of care as a variable, not an exclusion

Local standards are recorded, randomization is stratified by region and health system, and comparisons use concurrent controls within each stratum.

Endpoints that travel

Primary endpoints that can be measured with CT and routine labs anywhere; advanced endpoints as secondary or modular.

Decentralized where possible

Consent in local languages, shared virtual tumor boards, local imaging with central reading, and support for travel. Affiliated sites connect local practices and hospitals from the day of diagnosis.

Data that stays where it must

Federated analysis that respects each country’s data protection law, instead of moving all data to one place.

Fair partnership

Local co-investigators and authorship, training, and access to effective treatment after the trial.

Open questions

  • How do we define control arms where access to imatinib and later-line drugs differs between countries?
  • How do we assure the quality of surgery across centers with very different volumes?
  • One sponsor for all, or national sponsors under one master protocol?
  • How are per-patient costs funded fairly between high- and low-resource sites?

The GISTT-RACE website develops along the way.

To join a working group, please contact info@gistt-race.org.