Fore Biotherapeutics · Exploratory Scientific Intelligence · Requirements

Questions for Jessica

Follow-ups drawn from Jessica Jang’s reply of 28 August 2026. Each of her six points leaves something underdetermined that only she can settle, and settling it is what lets us scope the work.

Step 1 · Requirements Source: Jessica Jang, 28 Aug 2026 For: Jessica · cc Ping, Michael

Her six pointsWhat each one leaves open

Point 1

Biomarkers — how these are being used in clinical trials as a proxy to monitor tumor growth.Jessica Jang

Which biomarkers do you mean — ctDNA and VAF, protein markers, methylation signatures, something else?
Are you interested in them replacing an imaging timepoint, supplementing it, or acting as an early signal between scans? These are three different evidentiary bars, and the third is much easier to meet than the first.
Is this about how the field uses them, or how Fore should be using them in our own trials?
Which trials have you seen do this convincingly? One worked example tells us what good looks like.

Point 2

Technological advances to measure mutations from plasma of low-shedding tumors, such as thyroid or CNS.Jessica Jang

You say plasma specifically. Should CSF be in scope for CNS, or do you want this kept to blood?
Which technologies are you already tracking — ultra-deep sequencing, fragmentomics, methylation-based detection?
Is this “watch where the field is going” or “we need this to work for a Fore trial”? The urgency and the depth of the answer differ substantially.
Why thyroid? Is there a Fore interest there we should know about, or is it illustrating the low-shedding problem?

Point 3

For all of the questions surrounding resistance, combinations, and other interacting cellular pathways, we’ll have to manage this for each individual tumor type.Jessica Jang

Which tumour types are actually in scope, and in what priority order? This sets the size of everything downstream — every question below gets asked once per type.
Does “manage for each tumour type” mean genuinely separate analyses, or one analysis tagged by type?
Are there types where you already know the resistance picture, so we don’t spend effort re-deriving it?

Point 4

Changes in treatment or diagnostic landscape for each tumor type (may be different regionally as well).Jessica Jang

Which regions matter — US and EU, or do you need Japan and China as well?
Treatment landscape and diagnostic landscape are separate problems. Is one more urgent than the other?
What counts as a change worth flagging — an approval, a guideline update, a reimbursement decision, a new assay becoming available? Without a threshold, continuous monitoring produces a feed nobody reads.

Point 5

Comparison of direct mechanism of action between Plixorafenib and other competitor BRAFi or RAFi, biochemical structure, resistance mechanisms, etc.Jessica Jang

Which competitors specifically? Are any already well enough understood internally to deprioritise?
Do you want this once, as a reference comparison, or kept current as competitors publish?
By “biochemical structure,” do you mean binding-mode and structural detail, or a working summary at the level of monomer versus dimer?

Point 6

We may expand competitive landscape to include MEKi or RASi if we are targeting a similar population (e.g. use of MEKi for BRAF fusions in pediatric LGG).Jessica Jang

What is your test for “similar population” — shared indication, shared alteration, shared line of therapy? Without a rule the competitive set has no boundary and will drift.
Beyond MEKi in paediatric LGG, where else do you expect the overlap to matter?
Does this extend to combination regimens, or only single agents?

Across all sixQuestions that apply to every point

For each of these, what do you do today to get an answer, and how long does it take? This tells us what we would actually be replacing, and where the time is really going.
Which sources do you personally go to? Which would you trust enough to act on, versus only use as a lead worth chasing?
For each point, the two or three publications you are reasoning from — as DOIs or PMIDs rather than titles. Titles do not resolve cleanly. This is also the fastest way to capture what Fore already knows, so the work starts from your reasoning rather than beside it. Please flag anything under licence or embargo.
What decision does each question feed? Knowing what changes if the answer is X rather than Y is what makes an investigation worth running.
Is any of this tied to a date — SNO, a board meeting, a protocol amendment?
How would you want findings delivered: an alert when something changes, a weekly digest, or on demand when you ask?
What would make an answer trustworthy to you — citations, a stated confidence level, showing where the evidence disagrees with itself?
What have you already investigated, so we don’t hand back work you have done?
Do we hold subscriptions we should be using rather than relying on open sources?

What your answers already settled — please correct anything misread

Scope is four tumour types, analysed genuinely separately rather than tagged: CNS, CRC, melanoma, papillary thyroid. Pathways differ, so the analyses stay apart.
The competitive set is your seven — vemurafenib, encorafenib, dabrafenib; claturafenib, mosperafenib, NXP200; tovorafenib — kept current as they publish or present, and extending to combinations, not only single agents. MEK and RAS inhibitors still to be named.
Monitoring triggers on approvals and guideline updates across US, EU, Japan, China, Australia, Canada and Korea, with treatment landscape ahead of diagnostic.
Every finding will carry a citation and a source tier, drawn from the line you set: primary literature, conference material, patents and disclosures as actionable; news and second-hand reporting for awareness only, never as scientific grounding.
ctDNA and VAF are a field watch, not a dependency of any current trial — with real-time supplementing of imaging as the direction of travel, not a present requirement.
Paywalled full text goes through RightFind rather than the work settling for abstracts.

Twenty-five of the twenty-nine questions above are answered. That is enough to start shaping the work, which is why the six lines here are worth a moment’s check — if any of them misstate you, that is far cheaper to fix now than three weeks into a build.

Round two · 31 AugustWhat your answers open up

You answered seventeen of these on 31 August. Every answer closed one question and opened others — these are those. Same rule as before: rough beats polished.

Your answer · Point 3 CNS, CRC, Melanoma, papillary thyroid carcinoma

Is that list already in priority order as written, or would you rank it differently? Four types run sequentially is a very different commitment from four in parallel.
CNS is the widest of the four. Which histologies — pleomorphic xanthoastrocytoma, ganglioglioma, paediatric low-grade glioma, high-grade glioma, or anything CNS carrying a BRAF alteration?
Which alteration classes travel with each type — V600E only, or fusions and class II/III as well? This decides the competitive set as much as Point 6 did. BRAF-fusion paediatric LGG and V600E CRC share almost no competitors.
Melanoma: cutaneous only, or acral, mucosal and uveal too? And is it in scope because Fore may go there, or as the reference case where BRAFi biology is best characterised?
Papillary thyroid: does anaplastic thyroid come with it, or is that a separate watch?

Your answer · Point 3 Genuinely separate analyses, as pathways seem to be different

When a finding plausibly transfers — an MEK-reactivation mechanism found in CRC that could apply in CNS — should it be pushed across, or do you want the four kept strictly walled off?
If a mechanism turns out to be shared across all four, is that itself a finding worth flagging?

Your answer · Point 3 We have a general idea of the pathways for each tumor type, but I’d like the analysis to be re-run in case there’s something we’ve missed.

What is that general idea, per type, even roughly? Re-running against nothing means we hand back what you already have. Knowing your current picture is what lets us find the gap in it.
Is the thing you expect to have missed more likely a mechanism, a paper, or a competitor’s data you have not seen? Those are three different searches.

Your answer · Point 5 Early generation BRAFi: vemurafenib, encorafenib, dabrafenib. Paradox breakers: claturafenib, mosperafenib, NXP200. Pan-RAFi: tovorafenib.

No MEK or RAS inhibitors are named, though your Point 6 opens that door. Which specifically — trametinib, selumetinib, mirdametinib, binimetinib? Any RAS(ON) inhibitors?
Do discontinued and shelved programmes stay in the watch set? A competitor stopping is a signal, and the discontinuation record is one of the more informative datasets in this space.
Preclinical and patent-stage assets, or only agents already in the clinic?
Anything you want deliberately excluded because Fore tracks it well enough elsewhere?

Your answer · Point 1 Right now we are testing ctDNA for mutations as a retrospective analysis. It would be great if we could implement this in real-time to supplement traditional imaging as an earlier signal.

What would you actually do differently if ctDNA rose between scans — an unscheduled MRI, a discussion, nothing yet? This is the decision the biomarker work has to serve. Without it we cannot tell a useful finding from a merely interesting one.
Which assay and vendor is the retrospective FORTE-201 analysis running on, and what turnaround would “real-time” have to mean to be usable?
What size of VAF change would you treat as real rather than noise? If that threshold is not established, is establishing it part of what you want?
Which trials have you seen do this convincingly? (Still the one open question under Point 1 — one worked example tells us what good looks like.)

Your answer · Point 2 In an ideal world we could collect CSF for the CNS patients … but CSF samples so far have been difficult to collect for us.

Is the collection problem protocol design, site willingness, or patient burden? Only one of those is something an evidence agent can help with.
Should we watch CSF-ctDNA methods anyway, on the assumption collection becomes feasible later?

Your answer · Point 2 We have enrolled some patients with thyroid cancer in Forte 201, but may open a larger study focused on thyroid in the future.

Is that decision live, and on what timeline? If agent output could inform a real go/no-go, that is the sharpest test case available and we would build toward it first.

Your answer · Point 5 Binding pockets, structural details that might affect how they work (e.g. plixorafenib binding to L505 residue in BRAF to push alphaC helix out to prevent paradoxical activation), monomer vs dimer binding

Do you want structural work — PDB structures, docking, modelling — or a rigorous literature summary of what others have published and shown? These are very different builds. The first needs structural biology in the loop; the second does not.
Where a competitor has never disclosed a co-crystal, is inference from SAR and patent claims useful to you, or would you rather we mark it unknown?

Your answer · delivery I think a weekly digest would be the best cadence … we can also follow up on specific topics when needed.

Who receives it — you alone, or Ping and Michael as well?
Where should it land: email, the FRED hub, Teams?
How long is short enough that it actually gets read?
What would justify breaking the weekly cadence and alerting you the same day?
Some of this touches embargoed material. Does the digest need confidentiality tiering, or is the audience small enough that it does not?

Your answer · Point 4 Ideally all the regions, including US, EU, Japan, China, Australia, Canada, Korea · Treatment landscape is more urgent · Approvals and guideline updates.

Seven regions means seven regulators. Do you need PMDA, NMPA, MFDS, TGA and Health Canada monitored in their own local-language sources, or is English-language coverage of them sufficient? The first is a materially larger build than the second.
Which guideline bodies count — NCCN, ESMO, ASCO, the WHO CNS tumour classification, national bodies? NCCN alone revises constantly. Without naming the bodies, “guideline updates” is either a trickle or a flood.
Do you want every version bump, or only changes that move something BRAF-relevant?
Does the diagnostic landscape drop out entirely for now, or stay on at lower priority?

Your answer · sources and trust Pubmed, conference websites, literature searches, patents, some public investor decks are all trusted sources. Second hand sources such as news reports are good for knowing the regulatory approval landscape or updates on diagnostic testing assays or competitive landscape, but are not sources of scientific rigor.

We would build that distinction straight into the output: every item tagged either grounding or awareness. Is a two-tier split right, or do patents and investor decks deserve a middle tier of their own? A patent claim and a peer-reviewed result are both primary, but they are not equally reliable about what a molecule actually does.
When something surfaces in news only, should it appear at all, or stay suppressed until a primary source confirms it?
Four sources from the original note you did not mention — ClinicalTrials.gov, WHO ICTRP and EU CTIS, cBioPortal, and the NCI Genomic Data Commons. In or out?

Your answer · today’s workflow, and trust Usually I’ll search through pubmed first … followed by searches for abstracts/presentations at conferences. · Citations are a must, especially when the data can be conflicting.

Roughly how long does that loop take you for one real question — an afternoon, a week? This is the baseline any of this would be measured against, and it is the one number we still do not have.
When sources genuinely conflict, do you want the disagreement surfaced as the finding, or a single best-supported answer with the conflict noted underneath?
RightFind resolves paywalls, but someone has to order the document. Should the system queue requests for you to approve, or simply flag “full text needed” and leave the ordering to you?

Your answer · timing Not at this moment. This is for future decisions on clinical trials, and will be an ongoing process.

Nothing forces the pace, which is the main risk to work like this — it drifts into a feed nobody reads. The nearest real decision you mentioned is the possible larger thyroid study. Is that the right thing to aim the first output at? One real decision sharpens the work more than an indefinite watch ever will.
Failing that, what would tell you six weeks from now that this was worth doing?

From the original briefThree specialisations your points never touched

The 28 August note proposed nine agent specialisations. Your six points cover Biomarkers, Resistance, Tumor Intelligence, Competitive Intelligence and Pathway Biology. Three were never mentioned, and whether they are in scope roughly halves or doubles the build.

Safety. Nothing in your six points touches it, though paradox-driven toxicity is the thing plixorafenib is designed to avoid. In scope or not?
Real-world evidence. In scope? And does Fore hold access to Flatiron, TriNetX or similar? Without a licence this agent has very thin material to work with.
Clinical trials as its own watch. Competitive trials are implied by Point 6 — do you also want registry monitoring in its own right: new starts, enrolment changes, quiet terminations?

From the original briefPatents, embargo, and the one ask still open

Patents. In scope? Reading competitor patents raises questions worth routing past legal before we start rather than after.
Conference abstracts often surface months ahead of a paper. Is pre-publication material fair game, and does anything Fore holds under embargo change that answer?
One ask from the original note is still completely open: what do you wish you knew? Everything else on this page is us guessing at the shape of the problem from the outside. That one is you telling us directly, and it is the question most likely to change what gets built.

Round two — if this needs narrowing again, these four

What decision each question feeds. Still the largest gap. It is what separates a finding worth sending from one that is merely interesting.
Alteration class per tumour type. V600E versus fusion versus class II/III — sets the competitive set for all four types.
Safety and RWE: in or out. Decides whether this is a five-agent build or a nine-agent one.
The papers behind each point. Still open from round one. The starting corpus.

If the list needs narrowing, these four shape everything else

Which tumour types, and in what order. Point 3 — sets the scale of the whole effort.
What defines the competitive set. Point 6 — without it the boundary drifts.
What you do today, and how long it takes. Establishes the baseline we would be improving on.
The papers behind each point. The starting corpus. Everything else can follow.