TLSA / foralumab-na-spms — Intranasal foralumab for non-active secondary progressive multiple sclerosis

Program analysis · bpiq_drug_id 13497 · prepared 2026-08 · USD · framework v5.6.0 · Coverage: CLEARED

Company context, financials, ownership, options and the full pipeline: ../company.md.

How to read this. Written for a reader who has not studied pharmacology or finance. Terms specific to this program are in the Glossary below; general terms are defined once in framework/04-glossary.md. Tags: [VERIFIED — source] / [UNVERIFIED] / [WEB ESTIMATE — source, date].

Glossary

Executive summary

• What it is (one sentence): A fully human anti-CD3 antibody, sprayed into the nose rather than injected, intended to induce regulatory T cells that travel to the brain and calm the chronically activated microglia believed to drive disability in non-active secondary progressive multiple sclerosis.
• The event and when (as disclosed): Topline results from INFORM-MS (NCT06292923), a 54-patient randomised, double-blind, placebo-controlled Phase 2a. Disclosed as “late Q3/early Q4 2026” by the company on 2026-06-25, and as “Late Q3 2026” in the BPIQ catalyst_date_text field. No day has been disclosed. This analysis judges the window as 2026-09-15 at the earliest, 2026-10-21 as the single best guess, 2026-11-30 at the latest, at PERIOD precision and MEDIUM confidence — see the Readout section, which is what every timing decision below reads. The likeliest date is the opening day of the joint ACTRIMS-ECTRIMS meeting in Toronto, because the company has said in writing that it will present these data there. Nothing about the window moved this refresh: no release since 2026-06-25 changes the guidance, and the registry record is unchanged.
• The main reason it could work: This is the first controlled test of a mechanism that has produced a consistent, reproducible imaging signal in every uncontrolled setting it has been tried in — ten na-SPMS patients now published in a peer-reviewed journal, fourteen more in an expanded-access program, and three of three in a separate multiple system atrophy study. The disease has no approved therapy in the United States, and the one competitor that reached the FDA with a positive Phase 3 was rejected in December 2025 on liver safety.
• The main risk: The registered primary efficacy measure is a PET imaging biomarker in roughly eighteen patients per arm, not a disability outcome. Every efficacy observation to date is open-label and comes from one academic group whose two most senior members hold documented financial relationships with the sponsor — and, new this refresh, whose senior author also holds a federal grant on this drug’s own mechanism (A.5b). The ClinicalTrials.gov record still contradicts itself on whether all patients or only a sub-study site undergo PET, so the evaluable sample for the primary comparison may be materially smaller than fifty-four.
• What it means for the stock: Dominant materiality — see ../company.md C.2. The shares sit at 7.8% of their 52-week range and, for the first time in this corpus’s coverage of the name, have moved: $0.97 → $1.00 on the chief executive buying about $986,000 of stock in three consecutive open-market sessions. That is positioning, not data. The company carries a going-concern qualification, so a financing is near-certain either way. Attribution could not be determined for this program: the Alzheimer’s row on the same ticker carries a period-end placeholder that overlaps this window, and two placeholders touching is evidence of not knowing rather than a finding. The two sides of this event are far apart, the expected value sits above spot on asymmetry alone, and the modal outcome is still a fall. No directional edge is claimed.

0. Program-tier coverage — CLEARED

Company-tier coverage is in ../company.md C.0, which this refresh extended with the three new v5.6.0 mandatory rows (EDGAR submissions, EDGAR XBRL share count, FINRA short interest).

Two further program-tier sources were gathered in the same pass and carry no NOT CALLED state, per 02-connectors.md § Readout sources and § KOL sources: the congress match against data/congresses.json (ACTRIMS-ECTRIMS 2026, admissible here because the company has said in writing it intends to present) and the modelled readout arithmetic. CT.gov search_investigators and the PubMed conflict searches were re-run for A.5b.

The regulatory tier does not apply to this program, and is not silently skipped. 02-connectors.md makes that tier — openFDA drugsfda, Federal Register documents, optionally FAERS — mandatory only where the catalyst is a regulatory event such as a PDUFA date or an advisory committee meeting. This program’s catalyst is a trial readout: the topline release from INFORM-MS. No application is pending for this drug, so openFDA would return the sponsor’s absence rather than a submission history, and no advisory committee is scheduled, so the Federal Register would return nothing about it. Those rows are therefore not called and carry no state at all, by design rather than by omission.

A BLOCKED tool is not a NOT CALLED tool. No mandatory row reads NOT CALLED at either tier, so this program is CLEARED and the prediction below locks.

A. Scientific & clinical assessment

A.1 Mechanism of action and scientific rationale

Foralumab is a monoclonal antibody — a laboratory-made protein designed to stick to one specific target — and its target is CD3, part of the receptor machinery on the surface of T cells [VERIFIED — ChEMBL CHEMBL1743020, molecule type Antibody, USAN stem "-mab (-umab)", 2026-08-10]. Antibodies against CD3 have been used in medicine for decades, but almost always by injection into the bloodstream, where they broadly suppress or deplete T cells and cause a short, unpleasant burst of immune activation.

Tiziana’s idea is different, and it is the whole thesis. The drug is sprayed into the nose. The nose contains lymphoid tissue specialised for deciding what the body should tolerate — the same machinery that stops you mounting an immune attack against inhaled pollen or food proteins. Presenting an anti-CD3 antibody there is intended to induce regulatory T cells, the immune system’s off-switch, rather than to deplete T cells [VERIFIED — Chitnis et al., "Nasal administration of anti-CD3 monoclonal antibody modulates effector CD8+ T cell function and induces a regulatory response in T cells in human subjects", Frontiers in Immunology 2022, DOI 10.3389/fimmu.2022.956907, retrieved via PubMed 2026-08-10]. Those regulatory T cells are then meant to traffic to the brain and damp down the chronically activated microglia believed to drive disability accumulation in progressive multiple sclerosis.

How well is the target validated? In three parts, and they are not equally strong.

The molecular target is validated. An anti-CD3 antibody, teplizumab (Tzield), is approved by the FDA for delaying the onset of type 1 diabetes, which establishes that engaging CD3 with an antibody changes human autoimmune disease [VERIFIED — teplizumab is an approved anti-CD3 monoclonal antibody; class mechanism reviewed in Kuhn and Weiner, "Therapeutic anti-CD3 monoclonal antibodies: from bench to bedside", Immunotherapy 2016, DOI 10.2217/imt-2016-0049]. The limit of that verification matters: it validates CD3 as a druggable immune target, in a different disease, by a different route. It says nothing about whether nasal dosing produces a useful effect in the brain.

The endpoint’s disease relevance is validated, and by a source with no connection to this company. Independent work from the Turku PET Centre shows that TSPO binding in normal-appearing white matter and perilesional tissue is significantly higher in MS patients who go on to worsen than in those who stay stable, and predicts one-year EDSS worsening with an area under the curve of 0.76 to 0.78 [VERIFIED — Radford-Smith, Airas et al., BMJ Neurology Open 2025, DOI 10.1136/bmjno-2025-001026, n=87 with an independent 37-patient validation cohort, retrieved in full via PubMed 2026-08-07]. The limit of that verification also matters, and it is the important one: it establishes that the signal this trial measures tracks the disease, not that a twelve-week drug course can move it, and not that moving it changes what happens to a patient. See A.5b.

Independent human genetic validation of CD3 in multiple sclerosis could not be checked this session either. Open Targets, the connector that answers that question, was BLOCKED with Rate limit exceeded for client: global on three attempts, for the fourth consecutive sweep. That claim is therefore [UNVERIFIED].

The route — nasal mucosal tolerance producing a measurable central nervous system effect — is the novel and unproven half. It has produced consistent uncontrolled observations and has never been tested against placebo in this disease.

The exact scientific step this readout must prove. That twelve weeks of nasal foralumab produces a fall in [18F]PBR06 PET signal — the imaging marker of microglial activation — that is statistically larger than what the placebo group shows over the same twelve weeks. Not that patients felt better, not that disability stabilised, and not that the drug was well tolerated. Those are either secondary endpoints or, in the case of tolerability, near-certain to pass and therefore uninformative.

The honest scientific risk. Four things, in order of how much they matter.

First, this is a biomarker, and biomarkers separate from placebo less reliably than their proponents expect once blinding is applied. TSPO PET quantification carries substantial measurement variability between subjects, and the trial excludes only the lowest-affinity binders of the tracer — patients with a Thr/Thr polymorphism in the TSPO gene — leaving mixed-affinity heterozygotes in the analysis [VERIFIED — ClinicalTrials.gov NCT06292923 exclusion criterion 14, 2026-08-10]. The independent Turku group states plainly that TSPO-PET’s “broader clinical application faces feasibility challenges due to high technical demands and costs” [VERIFIED — BMJ Neurology Open 2025, DOI 10.1136/bmjno-2025-001026].

Second, the sample is small. Fifty-four patients randomised one-to-one-to-one is about eighteen per arm [VERIFIED — ClinicalTrials.gov NCT06292923, 2026-08-10]. Worse, the registry record contradicts itself about who is scanned: the primary outcome timeframe says imaging happens “after Cycle 4 at the PET sub-study site”, while the same outcome’s description says “Subjects at all sites will undergo PET imaging”. If PET is in fact a sub-study at one of seven sites, the evaluable sample for the primary efficacy comparison could be a fraction of fifty-four, and the trial would be underpowered by construction. This is flagged rather than resolved, because the record does not resolve it, and it is unchanged across four sweeps.

Third, every efficacy observation so far is open-label and comes from essentially one research group, and that group’s relationship with both the sponsor and the mechanism is now documented from two independent directions. The senior author on essentially every foralumab clinical paper is the chair of Tiziana’s scientific advisory board, who has received consulting fees and stock options from the company; the principal investigator on both sibling na-SPMS studies sits on the same board and consults for the company [VERIFIED — conflict-of-interest statement, Frontiers in Immunology 2022, DOI 10.3389/fimmu.2022.956907, read in full 2026-08-07]. New this refresh: the same senior author holds an active National Institute on Aging grant titled “Mechanisms underlying the neuroprotective effect of nasal administration of anti-CD3,” $805,654 in FY2026 [VERIFIED — NIH RePORTER, project 5R01AG084596-03, read 2026-08-10]. That is a federal, peer-reviewed grant rather than sponsor money, and it cuts two ways: it is independent scientific endorsement of the mechanism, and it is a second reason the same laboratory’s own scientific standing rises and falls with this drug. Both readings are recorded; neither is collapsed into the other. That combination is the situation in which regression to the mean is largest and in which an open-label prior deserves the steepest discount. See A.5b for the full disclosure.

Fourth, the tracer used here is not the tracer the independent literature is built on. Almost all published MS TSPO work uses [11C]PK11195; INFORM-MS uses [18F]PBR06, a second-generation tracer [VERIFIED — ClinicalTrials.gov NCT06292923; BMJ Neurology Open 2025 states "the majority of TSPO-PET imaging studies in the context of MS have used the [11C]PK11195 radioligand"]. Second-generation tracers are expected to have a better signal-to-noise ratio, which cuts in this trial’s favour, but it also means the external prior about effect sizes and variability does not transfer cleanly.

A.2 Clinical development plan, timeline, feasibility, resourcing

The arithmetic of the topline window is tight, and that is the timing finding. The last patient received their first dose on 2026-06-25 [VERIFIED — company PR 2026-06-25]. The treatment period is four three-week cycles, so twelve weeks [VERIFIED — ClinicalTrials.gov NCT06292923]. That puts the last patient’s final dose around 2026-09-17, with the post-treatment PET scan after that. Getting from there to a topline release by 2026-09-30 — the end of “late Q3” — leaves under two weeks for the final scan, image reconstruction, quantification, database lock, unblinding and analysis. It is not impossible, but it is aggressive, and the company itself widened its guidance from “Late Q3 2026” to “late Q3/early Q4 2026” on the very day it announced last-patient dosing. Expect the early-Q4 half of that window rather than the late-Q3 half, which is why the Readout section below puts the likeliest date at the congress in late October rather than at the end of September. Five weeks now separate this analysis from the earliest edge of that window.

Trials

Feasibility matrix

Resourcing sufficiency. For this readout, yes — the trial is fully dosed and the remaining cost is analysis. For anything after it, no. The company carries a going-concern qualification and a $2.7M working-capital deficit, and cannot fund a registrational Phase 3 from its balance sheet; see ../company.md C.3, which this refresh extended with the shelf capacity read directly from the F-3 ($250M base, $100M at-the-market with Jefferies). Part of the reason reported burn is low relative to seven registered programs is that some trials are run and funded by academic sponsors — the Alzheimer’s Phase 2 is sponsored by Brigham and Women’s Hospital rather than by Tiziana [VERIFIED — ClinicalTrials.gov NCT06489548, 2026-08-10], a US Department of Defense grant supports the spinal-cord-injury work [VERIFIED — company PR 2025-09-15 via BPIQ press feed], and the mechanism itself carries an NIA R01 in the senior author’s own laboratory [VERIFIED — NIH RePORTER 5R01AG084596-03, 2026-08-10]. That structure is genuinely capital-efficient, and it also means the company controls less of its own timeline than the program count suggests.

A.3 Target label, target product profile, strategic questions, designations

A.3a Target label. Adults with non-active secondary progressive multiple sclerosis — no relapses and no new MRI lesions for at least two years, with documented continued disability accumulation despite prior disease-modifying therapy, EDSS 2.5–6.5.

A.3b Target product profile

A.3c Strategic Go/No-Go questions. The set below is the Go-to-Phase-III set, because that is the decision this readout feeds: whether to commit to a registrational trial with a clinical disability endpoint. Several answers are honestly “not answerable until the readout”, and that concentration of unanswered questions is itself the finding.

Pre-Phase-III (Go-to-Phase-III / registration):

A.3d Regulatory designations.

• Fast Track designation, granted by the FDA on 2024-07-24 for intranasal foralumab in non-active SPMS. Fast Track means more frequent meetings with the FDA and eligibility for rolling review of a future application. It does not lower the evidence bar and it is not an opinion on efficacy [VERIFIED — company PR 2024-07-24 via BPIQ press feed and BPIQ historical catalysts].
• Orphan Drug Designation — filed May 2024, still not confirmed granted. If granted, and if the product were ultimately approved for the designated indication, it would carry up to seven years of US market exclusivity. The filing is documented; a grant is not, and a repeat search on 2026-08-10 — twenty-seven months after a filing the FDA reviews within ninety days — still found no announcement of one. A grant would ordinarily be announced by a company that announced the filing, so the silence is weak evidence against [WEB ESTIMATE — Tiziana press release "Tiziana Life Sciences Files for Orphan Drug Designation for Intranasal Foralumab", May 2024, via BioSpace and the company's own site, re-checked 2026-08-10].

A.4 Target product profile valuation matrix

Calibration examples from the framework, not claims about this asset: in oncology each incremental month of overall survival is worth roughly $150–300M in peak-sales potential; oral formulations command roughly 15–25% price premiums over injectables.

A.5 Evidence quality and endpoints

Endpoints

A.5b Key opinion leaders.

Panel as of. 2026-08-10 — the date the investigator, conflict and funding searches below were run. The funding column is new: 02-connectors.md § KOL sources added NIH RePORTER as an optional source at v5.6.0, and this is its first use in this corpus.

Investigators

The pivotal trial itself still yields no investigator row. CT.gov get_trial_details on NCT06292923 returns no overall officials and contacts: null on all seven sites, and search_investigators on the condition returns nobody attached to that NCT — re-checked 2026-08-10, unchanged. The three people below are therefore named from the two sibling na-SPMS foralumab studies run by the same sponsor and from the program’s own publication record, and each row says which. Two of them hold disclosed relationships with Tiziana beyond any trial payment; the third has none on record, which is a different fact from having none.

Independent voices

Judgement

Dissent

Empty. No named voice disagrees with the MIXED reading above, and inventing one would be worse than recording none. Dissent is required to be non-empty only for SUPPORTIVE_CONTESTED.

B. Commercial assessment

B.0 Current treatment algorithm

A patient who reaches non-active secondary progressive MS has usually been through relapsing disease first and has been treated with one or more disease-modifying therapies — interferons, glatiramer acetate, an oral such as fingolimod or dimethyl fumarate, or a B-cell antibody such as ocrelizumab or ofatumumab. Those drugs work by suppressing inflammatory attacks. When the patient stops having attacks and stops forming new MRI lesions but keeps getting worse anyway, those drugs have nothing left to act on.

At that point, in the United States, there is no approved disease-directed option. Siponimod is approved for active SPMS, meaning patients still having relapses or new lesions — precisely the population this trial excludes. Ocrelizumab is approved for primary progressive MS, a different course. What remains is symptomatic care: physical therapy, walking aids, and treatment for spasticity, fatigue and bladder dysfunction.

Where foralumab would fit: first-line in a line of therapy that currently does not exist. It would not displace anything; it would be added into an empty space. That is the strongest single element of the commercial case, and it is why a technically modest Phase 2a matters more here than the same trial would in a crowded indication.

The European picture changed in June 2026. Tolebrutinib was approved in the EU as Cenrifki for SPMS without relapses, so the European white space is now partly occupied [WEB ESTIMATE — Sanofi PR 2026-06-23]. The US white space is not, because the FDA issued a complete response letter for the same indication in December 2025 and, as of 2026-08-10, has still not approved the drug [WEB ESTIMATE — Sanofi PR 2025-12-24; MS News Today and Pharmacy Times coverage, re-checked 2026-08-10].

B.1 Competitive landscape and positioning

Where this asset wins, in plain sentences. It wins on route and on tolerability. A nasal spray taken at home three days a week, with 37.4 patient-years of exposure and no serious treatment-related adverse events, is a genuinely different product from a daily oral BTK inhibitor requiring liver-enzyme monitoring. The FDA’s rejection of tolebrutinib in December 2025 was a safety rejection — a “substantial and unusually high” risk of severe, sometimes fatal liver injury, with no subgroup showing a clearly favourable benefit-risk — rather than a rejection of the indication or of the disability endpoint. Eight months later that drug is still not approved in the United States and no resubmission decision date has been announced, which means the US white space has now been empty for longer than the previous version of this analysis could observe. That distinction matters both ways. It means the US white space is durable for a safety reason a clean drug is well placed to exploit. It also means there is no competitor precedent bearing on what efficacy package the FDA would accept in this indication.

The single fact the thesis rests on. That a placebo-controlled reduction in PET microglial activation is real and translates into disability benefit. Every other element — the empty market, the clean safety, the convenient route, the insider buying — is worth nothing if that link does not hold, and this trial tests only the first half of it. The independent literature validates that the signal tracks the disease and is silent on whether a drug can move it (A.5b).

Calibration example from the framework, not a claim about this asset: oncology first-movers in novel mechanisms have shown roughly 3.2× higher peak-sales potential but about 1.8× higher development risk.

B.2 Addressable market

The launch market would be the United States first, because that is where the trial ran, where Fast Track applies and where the competitor was rejected. Europe second, and now against an approved competitor — and with no EU trial of this drug registered at all, confirmed this refresh by a CTIS search returning zero records [VERIFIED — CTIS search, containAll "foralumab", 2026-08-10], which means any European path starts from nothing.

B.3 Value and feasibility

B.3a Expected peak sales. Built bottom-up as eligible patients × annual net price × peak penetration. Every figure below is conditional on a chain of events that has not happened: a positive controlled readout, then a registrational Phase 3 with a clinical disability endpoint, then approval, then pricing, then uptake. It is not risk-adjusted, and it excludes the Alzheimer’s, multiple system atrophy and amyotrophic lateral sclerosis programs entirely.

Price anchors: US MS oral disease-modifying therapy net prices averaged about $82,000 per year in 2021, against list prices of about $104,000 [WEB ESTIMATE — Neurology Clinical Practice, "Changes in List and Net Prices for Multiple Sclerosis Disease-Modifying Therapy, 2013 to 2021", DOI 10.1212/CPJ.0000000000200597]. Ocrelizumab’s list price is about $68,000 per year [WEB ESTIMATE — WebMD, undated]. ICER identified $16,500–$34,900 per year as the cost-effective range for MS antibodies [WEB ESTIMATE — ICER MS evidence report].

These figures are carried forward unchanged, and this refresh records a reason to treat the patient column as the weakest of the three inputs rather than the firmest: the prevalence meta-analysis this build rests on reports a pooled figure of 22.42 per 100,000, below the 27–45 range the patient counts above were built from. Correcting for that alone would cut every scenario by roughly a fifth. The figures are not restated here, because doing so on one re-read of a secondary summary would substitute one unverified number for another; the discrepancy is recorded as the finding.

B.3b Feasibility

B.4 Product-development risk

Framing questions.

1. Does the development plan support the target product profile claims? Only partly. The profile claims a therapy that slows disability in non-active SPMS. The trial reading out measures brain imaging over twelve weeks, with disability as an underpowered secondary. A win here supports the mechanism claim, not the benefit claim, and no plan for the trial that would support the benefit claim has been disclosed or funded.
2. Will the identified risks affect the target product profile? Yes, one of them decisively. If the PET endpoint does not separate from placebo, there is no profile left to defend, because nothing else in the program is controlled evidence.
3. If a risk cannot be mitigated, is the asset still differentiated from competitors? On route and tolerability, yes — a home nasal spray with a clean safety record remains differentiated whatever this trial shows, and the competitor’s still-unresolved liver-safety rejection makes that differentiation more commercially relevant than it was. But differentiation without efficacy is not a product.

Readout

Sources

The modelled estimate. The registry’s primary_completion_date is 2026-09. Rule 34’s default, absent a benchmark entry, is primary completion plus two to four months to database lock and analysis. Taking the end of September as the anchor gives 2026-11-30 to 2027-01-31. A second, tighter arithmetic runs off the dosing schedule rather than the registry: the last patient’s first dose was 2026-06-25 [VERIFIED — company PR 2026-06-25], the treatment period is four three-week cycles [VERIFIED — CT.gov NCT06292923], so the last dose falls around 2026-09-17 and the post-treatment PET after that; adding four to eight weeks for reconstruction, quantification, database lock, unblinding and analysis gives 2026-10-15 to 2026-11-12. Both are stated because they disagree, and the disagreement is the point: the registry-based default is more conservative than the dosing-based one by about six weeks, and the window below spans them rather than choosing.

Window

Basis. Unchanged from the 2026-08-07 analysis, because every source behind it is unchanged. Earliest is the first day of the company’s own “late Q3”, which is the earliest date any source supports and is the conservative edge a run-up exit must be timed against — even though the dosing arithmetic makes a release that early close to impossible. Likeliest is 2026-10-21, the opening day of the joint ACTRIMS-ECTRIMS meeting the company has said in writing it will present at: it sits inside “early Q4”, inside the dosing-based modelled range, and is the one date on the calendar the company has committed to in public. Latest is 2026-11-30, the near end of the registry-based modelled range, which also absorbs a third slip of the kind this program has already made twice. Precision is PERIOD because no source names a month for the readout itself — the tightest company wording straddles two quarters, and the congress commitment fixes a venue rather than a release date. Confidence is MEDIUM rather than LOW because four independent sources bracket the same three-month span, and rather than HIGH because two documented slips and a self-contradicting BPIQ row are on the record.

Disagreement. UNRESOLVED. The BPIQ structured field and the CT.gov primary completion date both point at September; the company’s own most recent wording, its congress commitment and both modelled arithmetics point at October or later. The gap is roughly six weeks and it is reported, not averaged and not resolved by picking one (rule 24). The direction of the disagreement is consistent — every source that is newer than the BPIQ structured field points later than it does — which is why the window’s likeliest sits in October rather than at the midpoint.

Date slippage. Two slips across four dated statements, parsed from the BPIQ note field on drug id 13497 and corroborated by the company’s own releases, oldest first. Four statements produce three transitions and two of those three were an actual slip; the 2026-05-14 entry restated the previous guidance unchanged. No new statement has been added since 2026-06-25, so the count is unchanged at two.

Attribution

Status.

Computed by lib/clustering.mjs’s attributionFor over this ticker’s pipeline (../company.md C.2) for bpiq_drug_id 13497, with CATALYST_CLUSTER_MIN_MONTHS = 6, re-run 2026-08-10. Transcribed, not estimated.

Conflicts

The Alzheimer’s row is the only other has_catalyst: true row on this ticker. Its catalyst_date of 2026-12-31 is BPIQ’s period-end placeholder for the bare text “2026”, so the clustering pass widens it to the whole of 2026-07-01 to 2026-12-31 — which swallows this program’s entire readout window, giving a gap of zero days. Neither side rests on a disclosed date: this program’s own window carries PERIOD precision and the Alzheimer’s row carries catalyst_date_is_exact: false, so the conflict is not confirmed and the status is INDETERMINATE rather than CONTAMINATED. Two placeholders touching is evidence of not knowing where either event lands, not evidence that they collide. One thing did change underneath it this refresh without changing the computed status: the Alzheimer’s trial’s own registry primary completion date of 2026-06 has now passed with no announcement, so the event that row stands for is arguably nearer than its placeholder says. That is not a disclosed date either, so it cannot confirm the conflict — it is a reason to expect this status to be revisited rather than a reason to change it now. The five has_catalyst: false rows — MSA, ALS, Crohn’s, COVID-19 and milciclib — do not enter the computation at all.

Note. Not required for INDETERMINATE, and deliberately not written.

Market and timing for this event

Company-level figures — price, market capitalisation, the 52-week range, ownership and the options chain — are in ../company.md and are cited rather than repeated here.

• Plain takeaway. The market is still not positioned for this event in either direction, but for the first time there is something to report on the positioning side. The shares moved $0.97 → $1.00 across the two sessions before this analysis, on roughly 2.7× the trailing average dollar volume (../company.md C.1, C.4), and the disclosed cause is the chief executive buying 988,423 shares for about $986,000 in three consecutive open-market sessions, coded P on Forms 4 (../company.md C.5). Everything else still says nobody is crowded in: 7.8% of the 52-week range, no strike within 2.5× of spot on either options vendor (../company.md C.6), zero hedge-fund holders, and short interest that fell 20.4% at the 2026-07-15 settlement to 1,245,738 shares, 4.62 days to cover.
• Months to this catalyst. About 1.2 months to readout.window.earliest (2026-09-15), 2.4 months to likeliest (2026-10-21) and 3.7 months to latest (2026-11-30), from 2026-08-10. Every one of those is measured from the Readout window above, never from catalyst_date (rule 23). readout.precision is PERIOD, so treat all three as the edges of a judgement rather than as dates anyone has disclosed: no source names a month for the readout itself.
• Expected move around this event. Not computable. ../company.md C.6 records the chain as unusable on two independent vendors this refresh — no strike within 2.5× of spot, and implied-volatility sentinel values on both BPIQ (0.01488 floor) and Yahoo (0.500005 on calls, 0.00001 on puts). A bracket rather than a point estimate: this name has printed single-day moves between −9.3% and +27.4% on uncontrolled news (../company.md C.7), and a first controlled readout should produce a larger move than any row in that table. No percentage is quoted.
• Nearest comparable past reaction. There is no true analogue, and that is the finding. Every row in ../company.md C.7 is an uncontrolled result, a site opening, a regulatory filing or a preclinical study. The company has never reported a placebo-controlled result. The closest in kind is 2026-05-19 — one-year expanded-access clinical data plus a sell-side target raise on the same day — which moved the stock +0.7%. That row is good evidence that uncontrolled clinical data no longer move this stock, and it is not evidence about how a controlled one would.
• Materiality. Dominant, as recorded in ../company.md C.2. This is the only program whose timing is pinned by more than a placeholder, the only placebo-controlled study the company has ever run to completion, and the company carries a going-concern qualification. The stock-direction call below is consistent with that: a dominant program means the reaction should approximate a full re-rating in either direction, which is why the two scenario ranges are far apart and why no direction is claimed between them.
• Date slippage. Two slips in four months, unchanged this refresh — see readout.slips in the Readout section above for the dated sequence. The cumulative drift is roughly four to six months, and the second slip was announced on the same day as a positive execution milestone.

Spot. $1.00, read 2026-08-10, cited from ../company.md C.1 and C.4.

Scenario prices

Expected value. Applying the 35% probability below to the midpoint of each range: 0.35 × $2.40 + 0.65 × $0.425 = $1.12, which is +11.6% against the $1.00 spot. This is arithmetic, not advice, and it is not a price target. It is positive only because the two ranges are asymmetric — about +140% against about −58% — and not because the outcome is judged more likely good than bad. The modal outcome in this record is a miss. The percentage is four points lower than the 2026-08-07 lock for one reason and one only: the spot it is measured against rose 3%, while the scenario ranges did not move.

Run-up

Entry and exit

Predicted move and peak

Priority score drivers

Priority score. Transcribed from lib/runup.mjs’s priorityScore, never hand-computed.

Priority score 10 · formula_version 1.0.0

Ten out of a hundred is a low rank, and it is low for one reason: the date. Six of the seven drivers read well — a genuinely unmet indication, a dominant program, a stock at the bottom of its range with nobody positioned but its own chief executive — and the seventh, date_confidence at 20, multiplies all of that down. That is the formula working as designed rather than a verdict on the science. A run-up you cannot time is a run-up you should not size. The score is unchanged from the 2026-08-07 lock, and the two one-point driver moves (squeeze 48 → 47, priced-in 93 → 92) are too small to shift it.

Settlement. Left null at lock time. exit is also null, and this refresh can record why without the hand-written workaround the last one needed: resolveExit("T-5", window, bars) now returns null correctly when the price cache ends before the window opens, where on 2026-08-07 it returned a wrong date five bars back from the end of the series. The library was fixed after that run and names it in its own comment. Settled only on an explicit user request, from a confirmed primary source, against the committed price cache.

Verdict

What I would do. Watch.

Why. The mechanism is genuinely novel and the market it targets is genuinely empty in the United States, which is a rare combination, and the FDA’s rejection of tolebrutinib — still unresolved eight months later — keeps that emptiness durable for a reason a drug with 37.4 clean patient-years is well placed to exploit. But the trial reading out measures brain imaging in roughly eighteen patients per arm over twelve weeks, the registry record does not agree with itself about how many of those patients are even scanned, and every efficacy observation to date is open-label work from a group whose two most senior members sit on the sponsor’s scientific advisory board, one of them as its chair with stock options, and whose senior author also holds a federal grant on this drug’s own mechanism. Underneath that sits a balance sheet whose own auditors say it cannot meet its liabilities for twelve months without more money, with $100M of at-the-market capacity now quantified from the shelf rather than assumed. A real thesis, a weak test of it, a conflicted evidence base and a forced financing is a combination to watch closely and to size very small if at all. The one genuinely new fact on the positioning side — the chief executive buying about $986,000 of stock in three consecutive sessions immediately before this lock — is worth reading, and it is not evidence about the data.

What would change this. A topline release that reports a statistically significant, dose-ordered separation from placebo on PET microglial activation and a directionally consistent EDSS or MFIS movement, together with a disclosed plan and a partner or financing for a registrational trial with a clinical disability endpoint. Any one of those without the others is not enough. On the other side, a release that claims the trial “met its primary endpoint” while describing only safety and nasal tolerability would be a negative disguised as a positive, and should be read as one.

What to watch.

• Now to the readout: any 6-K or press release announcing a financing, an at-the-market drawdown or a partnership. Under the going-concern qualification this can come at any time, the shelf has $100M of at-the-market capacity available, and it would say a great deal about management’s own confidence. It is also the single most likely event to end the run-up call above before its exit rule fires. Watch the filing feed rather than the press feed: an ATM drawdown need not be announced.
• From mid-August 2026: publication of the ACTRIMS-ECTRIMS 2026 programme and abstract titles. The main abstract deadline has passed and a late-breaking round was advertised, but as of 2026-08-10 no programme is public. Whether a title names the placebo-controlled PET comparison is the first checkable signal about what will actually be presented.
• From 2026-09-15: whether topline arrives inside the “late Q3” half of the window. It probably will not — the dosing arithmetic in A.2 leaves under two weeks — and a third slip would make three in six months.
• 2026-10-21 to 2026-10-23: the joint ACTRIMS-ECTRIMS meeting in Toronto, where the data are planned for presentation. This is the readout.window.likeliest date.
• At the readout: whether the release reports the placebo-controlled PET comparison with a p-value, or reports within-arm change from baseline only. The second would be an evasion.
• At the readout: how many patients contributed evaluable PET scans. If that number is far below fifty-four, the sub-study ambiguity in the registry record was real.
• Any time: a ClinicalTrials.gov update to NCT06292923 resolving the “all sites” versus “PET sub-study site” contradiction, naming an overall official, or posting results.
• Any time: a readout from NCT06890923, the open-label multicentre sibling whose primary completion date of 2026-04-30 passed more than three months ago with nothing announced — and from the academic Alzheimer’s study NCT06489548, whose own 2026-06 primary completion has now passed too. Two silent registry milestones on one molecule is a pattern, not a coincidence.
• Any time: further Form 4 filings. The chief executive’s buying is now the only positioning signal on this ticker with a date and a price on it, and whether it continues into the window is checkable weekly on EDGAR.

Locked prediction

• Outcome-direction (will the readout succeed on its primary endpoint?): miss — probability 35%, band 20–52% [UNVERIFIED — modelled]. The probability is that the settlement definition below is met, which requires a statistically significant placebo-controlled separation on the PET primary — not that the company describes the trial as successful. It is held exactly where the 2026-08-07 lock put it, and that is deliberate. Nothing this refresh found bears on whether the drug separates from placebo: the registry record is unchanged, no new clinical data exist, the peer-reviewed publication Europe PMC surfaced is the same open-label ten-patient study already counted, and the NIH grant found on the senior author is a fact about funding rather than about efficacy. Where a refresh finds nothing on the outcome side, the honest move is to leave the number alone rather than manufacture movement from connector novelty.
• Stock-direction (which way do the shares move?): no-edge — confidence none — window 2026-09-15 to 2026-11-30, basis: readout.window.earliest to readout.window.latest from the Readout section, never catalyst_date. Materiality is dominant (../company.md C.2), so the move should approximate a full re-rating in either direction. Attribution could not be determined for this program: the Alzheimer’s row on the same ticker (bpiq_drug_id 17171) carries a period-end placeholder whose widened half-year swallows this entire window, and two placeholders touching is evidence of not knowing rather than a measured collision (rule 36).
• Scenario prices: positive $1.80–$3.00 · miss $0.30–$0.55
• Expected value: $1.12, +11.6% against spot $1.00 (arithmetic, not advice)
• Run-up: entry $1.00 on 2026-08-10, exit rule T-5 trading days before readout.window.earliest — predicted move −10% to +22%, predicted peak 0% to +32% around 2026-09 — priority score 10, formula_version 1.0.0
• Settles on the INFORM-MS (NCT06292923) topline release reporting a statistically significant reduction versus placebo, at the trial’s pre-specified alpha, in change from baseline in [18F]PBR06 PET microglial activation at Week 12, for at least one of the 50 µg and 100 µg doses. A release meeting only the safety or Total Nasal Symptom Score co-primaries, or hitting only a secondary clinical endpoint, scores as a miss.
• Locked: yes · Settled: no

Why no-edge sits beside an expected value 11.6% above spot: the two answer different questions, and here they genuinely diverge. The expected value is positive because the payoff is asymmetric — roughly +140% against roughly −58% — which is a consequence of the shares still sitting at 7.8% of their 52-week range. The direction question is answered by where the probability mass sits, and 65% of it is on a fall. Calling this “up” would report a lottery ticket with positive expected value as though it were a forecast. The honest statement is that this is a bimodal event with a positive expected value and a modal outcome of a large loss, that attribution could not be determined for it, and that no direction is claimed.

This record supersedes TLSA-13497-2026-08-07, which was locked under framework v5.5.0 at a $0.97 spot. The earlier record is kept exactly as it was locked.

Program data-quality flags

• Open Targets search_entities BLOCKED on three attempts, verbatim: Rate limit exceeded for client: global. Fourth consecutive sweep. Independent genetic validation of CD3 in multiple sclerosis is [UNVERIFIED] in this analysis, and target validation rests instead on the approval of a different anti-CD3 antibody in a different disease.
• The PubMed connector still does not return the peer-reviewed na-SPMS paper, and Europe PMC does. The Neurology Neuroimmunology & Neuroinflammation article (DOI 10.1212/NXI.0000000000200543) is absent from PubMed’s 15 “foralumab” hits on a fourth consecutive sweep, while Europe PMC returns it as PMC12815464 with its full 26-author list and a 2026-01-16 first-publication date. The optional Europe PMC row added at v5.6.0 closed a gap three previous sweeps could only flag. Its full text was still not read this session.
• Europe PMC returns 88 hits where PubMed returns 15, on the identical query. The extra hits are mostly reviews and preprints, including two 2026 reviews of PET imaging in MS that bear on this program’s endpoint and that PubMed’s foralumab query does not reach. The two indexes are not interchangeable and this analysis now uses both.
• CT.gov discloses no investigator for the pivotal trial, re-checked 2026-08-10. get_trial_details on NCT06292923 returns no overall officials and contacts: null on all seven sites, and search_investigators finds nobody attached to that NCT. Every name in A.5b’s Investigators table therefore comes from a sibling trial or from the publication record, and each row says which. This is an absence in the registry, not an absence of investigators.
• lib/runup.mjs’s resolveExit bug reported by the 2026-08-07 analysis has been fixed. It now returns null when the last cached bar predates readout.window.earliest, where it previously returned a date five bars back from the end of the series; the library names the 2026-08-07 refreshes in its own comment as the case that prompted the fix. runup.exit is null here because the function says so, not because it was overridden by hand.
• The ClinicalTrials.gov record for NCT06292923 contradicts itself on PET coverage. The primary outcome timeframe says imaging occurs “after Cycle 4 at the PET sub-study site”; the same outcome’s description says “Subjects at all sites will undergo PET imaging with the radiotracer [18F]PBR06”. The evaluable sample for the primary efficacy comparison is therefore unknown, and could be far below 54. Reported, not resolved, and unchanged across four sweeps.
• The enrolment figure disagrees between sources, and the disagreement has persisted. ClinicalTrials.gov records enrolment of 54; the company’s own release of 2026-06-25 says “up to 48 patients”, and the 2026-05-21 release said 48 enrolled. The registry figure of 54 is used here because it is the protocol figure, and the discrepancy is flagged rather than averaged.
• The trial’s primary endpoint is described differently by the registry and by the company. ClinicalTrials.gov lists three primary outcome measures — adverse event count, Total Nasal Symptom Score, and the PET microglial change. The company’s release of 2026-06-25 says “The primary endpoint of the trial is the change in microglial activation as measured by positron emission tomography (PET) scans”. This gap is why the settlement definition above keys explicitly on the PET comparison: a company could truthfully say the trial “met its primary endpoint” on the safety co-primary alone.
• The catalyst date disagrees across four sources and the disagreement is UNRESOLVED. See the Readout section’s Disagreement row for which source points which way. It is reported, not resolved and not averaged (rule 24).
• BPIQ catalyst_date (2026-09-30) is a period-end placeholder, not a disclosed day (rule 23), and is used for no timing decision in this document. It survives only as the bpiq row of readout.sources[].
• The BPIQ drug name for this program is dirty and must never be used as a key. The same molecule appears across rows as "Intranasal Foralumab (anti-CD3 mAb)", "Intranasal Foralumab " (trailing space), "Foralumab" and "Foralumab (anti-CD3 mAb) ". This analysis keys on bpiq_drug_id 13497 throughout.
• The SPMS prevalence input behind B.3a reads lower in its own source than in this analysis. The BMC Neurology 2022 meta-analysis is quoted here as 27–45 per 100,000; a repeat search on 2026-08-10 returns its pooled estimate as 22.42 per 100,000 (99% CI 18.30–26.95). The peak-sales scenarios are not restated on one re-read of a secondary summary, and the discrepancy is recorded as the finding. The non-active share of SPMS remains unverified on a fourth sweep.
• A conflict disclosure is only as good as the journal’s own practice, and the funding search does not cover everyone named. Tarun Singhal’s and Maria Anagnostouli’s conflict entries rest on searches that returned nothing, not on statements saying there is nothing; NIH RePORTER was queried on Weiner and Chitnis but not on Singhal, so his federal-funding status is unchecked rather than clear. conflicts_checked records where each search was made; it does not certify what it did not find (02-connectors.md § KOL sources).
• The Front Immunol 2022 conflict statement is nearly four years old. The disclosures it carries for Weiner and Chitnis were current at 2022-11-23, and it also discloses a third consultant, Clare M. Baecher-Allan, who is not recorded in A.5b because she is neither an investigator on any of these trials nor a commentator on the endpoint. A board seat or a consulting relationship can end, and a newer one can begin, so those as_of dates are recorded rather than treated as current.
• No third party has published a probability on this specific event that this sweep found. The available third-party views are a price target ($9, Lucid Capital Markets, Buy, 2026-05-19, still the only published target and therefore also the consensus) and a ratings action (Wall Street Zen, downgrade to Sell, 2026-04-18), which point in opposite directions and neither of which is expressed as a probability.
• The committed price cache is three trading days stale relative to the spot this prediction locks at. data/prices/TLSA.json ends at 2026-08-05 ($0.99) while the lock reads $1.00 from BPIQ on 2026-08-10. Every price-history-derived figure here (the 52-week range, the position within it, average dollar volume, the priced_in driver) is computed from the committed cache per the framework’s own discipline; the spot is not. Refreshing the cache is outside this analysis’s lane.