28
Milestones
16
Met at status date
15
The program can move
3
Nobody can move
Contents
1. What a Milestone Is
A milestone is a state, not an event, and not a piece of work. “Database locked” is a milestone because it is either true or it is not. “Phase 3 progressing well” is not a milestone; it is an opinion with a date attached. The distinction sounds pedantic until a program is three weeks from a gate and needs to know whether something happened.
| Rule | Why it holds |
|---|---|
| A milestone is binary | Ninety per cent complete is zero per cent complete. Partial credit is what turns a schedule into a narrative, and it is the mechanism by which a program reports green for eighteen months and then slips six. |
| It has one named owner | Not a function — a person. A milestone owned by “Clinical Operations” is owned by nobody at the moment it is missed. |
| Its evidence standard is fixed when the milestone is DEFINED | Not when it is claimed. The same discipline the Committee applies to gate conditions, for the same reason: the party who wants the milestone closed should never also be the party who decides what closing it requires. |
| It is dated from the schedule, not typed in | Every date in the register below is read from the integrated schedule. The two documents cannot disagree, because there is only one set of dates. |
| A deliverable is a thing; a milestone is a moment | The Statistical Analysis Plan is a deliverable. “SAP approved” is a milestone. Programs that conflate the two end up tracking documents that exist in draft forever. |
The test for a well-formed milestone: could two people who disagree about the program's health
still agree on whether this one has been met?
If the answer is no, the milestone is a status report wearing a milestone's clothes, and it will be argued about at precisely the moment the program can least afford the argument.
If the answer is no, the milestone is a status report wearing a milestone's clothes, and it will be argued about at precisely the moment the program can least afford the argument.
2. The Register
28 dated commitments across the life of the program. 16 met at the status date, 12 forecast.
| Ref | Date | Milestone | Owner | Type | Evidence standard | At 15 Oct 2026 |
|---|---|---|---|---|---|---|
| M-01 | 31 Dec 2022 | Gate 1 — IND-Enabling Readiness | C. Tyrrell | Internal | Minuted Committee decision with the scored criteria attached | Met |
| M-02 | 30 Jun 2023 | IND submitted | Dr. P. Raghunathan | Regulatory | Agency acknowledgment of receipt with an IND number assigned | Met |
| M-03 | 30 Jul 2023 | IND in effect | Dr. P. Raghunathan | Statutory | Day 30 elapsed with no clinical hold communicated — silence IS the evidence | Met |
| M-04 | 30 Sep 2023 | Phase 1 first participant dosed | Dr. R. Molyneux | Internal | First dose recorded in the EDC against a signed consent | Met |
| M-05 | 13 Apr 2024 | Phase 1 topline & dose selection | Dr. S. Aldridge | Internal | Topline tables reviewed and the dose recommendation signed by the CMO | Met |
| M-06 | 22 Apr 2024 | End-of-Phase-1 meeting held | Dr. P. Raghunathan | Regulatory | FDA official minutes received — not the sponsor's notes | Met |
| M-07 | 31 May 2024 | Gate 3 — Phase 2 Readiness | C. Tyrrell | Internal | Minuted Committee decision; Stage 3 tranche released | Met |
| M-08 | 31 Aug 2024 | Phase 2 first participant randomized | Dr. R. Molyneux | Internal | IRT randomization record against a signed, dated consent form | Met |
| M-09 | 31 Dec 2024 | Phase 2 last participant randomized | Dr. R. Molyneux | Contractual | IRT record; triggers a CRO milestone payment | Met |
| M-10 | 24 Oct 2025 | Phase 2 database lock | Dr. F. Achterberg | Internal | Lock certificate countersigned by data management and the study statistician | Met |
| M-11 | 09 Dec 2025 | Phase 2 dose selection confirmed | Dr. S. Aldridge | Internal | Dose rationale memo accepted by the Committee as Gate 4 evidence | Met |
| M-12 | 09 Jan 2026 | EOP2 briefing package submitted | Dr. P. Raghunathan | Regulatory | Agency acknowledgment; the 30-day clock before the meeting starts here | Met |
| M-13 | 19 May 2026 | End-of-Phase-2 meeting held | Dr. P. Raghunathan | Regulatory | FDA official minutes; agreed iPSP due within 60 days | Met |
| M-14 | 30 Jun 2026 | Gate 4 — Phase 3 Initiation | C. Tyrrell | Internal | Minuted decision, conditions recorded with verification methods fixed at issue | Met |
| M-15 | 30 Jun 2026 | Phase 3 first site activated | Dr. R. Molyneux | Internal | Green-light letter issued — contract executed, IRB approved, site trained | Met |
| M-16 | 30 Aug 2026 | Registration batches released | Dr. K. Oyelaran | Internal | Three batch records reviewed and dispositioned by Quality Assurance | Met |
| M-17 | 30 Sep 2027 | Phase 3 last participant randomized | Dr. R. Molyneux | Contractual | IRT record; largest single CRO milestone payment in the program | — |
| M-18 | 30 Dec 2027 | Process validation complete | Dr. K. Oyelaran | Internal | PPQ report approved; variability within the Stage 1 prediction | — |
| M-19 | 30 Jul 2028 | Statistical analysis plan approved | Dr. F. Achterberg | Internal | Signed SAP, dated BEFORE the lock certificate | — |
| M-20 | 05 Aug 2028 | Pivotal database lock | Dr. F. Achterberg | Internal | Lock certificate; unblinding authorization issued only after it | — |
| M-21 | 10 Aug 2028 | Pre-NDA meeting held | Dr. P. Raghunathan | Regulatory | FDA official minutes; content and format agreed | — |
| M-22 | 30 Sep 2028 | Gate 5 — NDA Submission | C. Tyrrell | Internal | Minuted decision; Stage 5 tranche released | — |
| M-23 | 12 Oct 2028 | NDA submitted | Dr. P. Raghunathan | Regulatory | Submission receipt from the electronic gateway | — |
| M-24 | 11 Dec 2028 | Filing acceptance | Dr. P. Raghunathan | Statutory | Day-74 letter confirming filing and the review clock | — |
| M-25 | 12 Jun 2029 | Pre-approval inspection closed | Dr. I. Solberg | Regulatory | Inspection closeout; any Form 483 observation responded to within 15 working days | — |
| M-26 | 11 Oct 2029 | Action date | FDA | Statutory | The agency's action — nothing the sponsor produces is evidence of this | — |
| M-27 | 11 Oct 2029 | Gate 6 — Approval & Launch | C. Tyrrell | Internal | Minuted decision taken on approval as an INPUT, not a forecast | — |
| M-28 | 01 Nov 2029 | First commercial shipment | L. Whitcombe | Contractual | Shipment confirmation against a released, serialized lot | — |
3. Four Kinds of Milestone, and Who Can Actually Move One
The most useful column in the register is not the date. It is the type, because the type tells you what happens if the program wants to move it.
| Type | Count | Share | What it means for the program |
|---|---|---|---|
| Internal | 15 | 54% | The program can move it. It still costs credibility. |
| Contractual | 3 | 11% | Moving it costs money — it triggers a payment or a change order. |
| Regulatory | 7 | 25% | The agency controls the date or the acceptance. The sponsor controls only readiness. |
| Statutory | 3 | 11% | Arithmetic performed by somebody else. Not movable by anyone in this program. |
Only 15 of 28 milestones are the program's to
move.
The rest are held by a contract, by an agency, or by statute. That ratio is worth sitting with, because the instinct under schedule pressure is to look for the date that can give — and 13 of these cannot, at any price the program is able to pay.
The statutory three are the sharpest case. Filing acceptance is sixty days from submission and the action date is ten months from filing acceptance. Nobody in this program can move either. They are arithmetic performed by somebody else, and the only sponsor behavior that touches them is the kind that makes them worse.
The rest are held by a contract, by an agency, or by statute. That ratio is worth sitting with, because the instinct under schedule pressure is to look for the date that can give — and 13 of these cannot, at any price the program is able to pay.
The statutory three are the sharpest case. Filing acceptance is sixty days from submission and the action date is ten months from filing acceptance. Nobody in this program can move either. They are arithmetic performed by somebody else, and the only sponsor behavior that touches them is the kind that makes them worse.
The contractual three deserve their own note, because they carry a conflict that is easy to miss. Both Phase 2 and Phase 3 last-participant-randomized trigger milestone payments to Meridian. That is a normal and sensible contracting structure — it aligns the vendor with the thing the program actually needs. It also means the party that benefits from the milestone being declared is the party in the best position to declare it. Which is why the evidence standard for both is the IRT randomization record rather than a CRO status report: the system of record is neutral, and the sponsor reads it directly.
4. The Evidence Standard
Every row in the register carries one, and they are deliberately specific. Three patterns in them are worth naming.
| Pattern | Example | The reasoning |
|---|---|---|
| The evidence is external to the program | “FDA official minutes received — not the sponsor's notes” (M-06, M-13, M-21) | The sponsor's record of a meeting is a recollection. The agency's minutes are the record, they arrive within thirty days, and where the two differ the sponsor's is not the one that counts. |
| The evidence is an absence | “Day 30 elapsed with no clinical hold communicated” (M-03) | An IND is an exemption, not a permission. Nobody sends a letter saying it worked. Silence is the evidence, which is a genuinely uncomfortable thing to record as a milestone and is exactly why it should be recorded as one. |
| The evidence is a countersignature | “Lock certificate countersigned by data management and the study statistician” (M-10, M-20) | Two functions with different incentives both have to be willing to sign. A single signature on a database lock is a control that has been designed to be passable. |
The evidence standard is fixed at definition and is not renegotiable at claim time.
This is the same rule the Committee applies to gate conditions, and it exists for the same failure. When a milestone is a week late, there is always a reasonable case for a slightly weaker proof: the data is in hand even if the certificate is not signed, the meeting happened even if the minutes have not arrived, the batches are made even if Quality has not dispositioned them. Each case is individually reasonable.
A standard that can be renegotiated by the person who needs it relaxed is not a standard. The program's answer is that the milestone is simply not met yet, which is a smaller and more honest statement than it feels like at the time.
This is the same rule the Committee applies to gate conditions, and it exists for the same failure. When a milestone is a week late, there is always a reasonable case for a slightly weaker proof: the data is in hand even if the certificate is not signed, the meeting happened even if the minutes have not arrived, the batches are made even if Quality has not dispositioned them. Each case is individually reasonable.
A standard that can be renegotiated by the person who needs it relaxed is not a standard. The program's answer is that the milestone is simply not met yet, which is a smaller and more honest statement than it feels like at the time.
5. Position at the Status Date
16 of 28 met, every one of them on or close to its planned date, and the program is nonetheless carrying a known schedule exposure. Both things are true, and holding them together is the point of this section.
| Position | |
|---|---|
| Milestones met | 16 of 28 |
| Next milestone due | Phase 3 last participant randomized — 30 Sep 2027, Dr. R. Molyneux |
| Milestone most at risk | M-17 Phase 3 last participant randomized — 76 participants behind curve, and treatment duration is fixed at 68 weeks after it |
| Milestones the program could move if it had to | 15 |
A clean milestone record is not the same as a healthy program, and this is the artifact where
that gap is most visible.
Every milestone met so far was met because it was defined as a state with a proof. None of them measures whether the program is on course for the ones ahead. The enrolment shortfall does not show up anywhere in the register until M-17 is missed — at which point it is roughly two years too late to respond.
Milestones are lagging indicators by construction. That is not a flaw; it is what makes them trustworthy. But it means a program that manages only its milestone register is steering by the wake, and it is why the register is read alongside the leading indicators in the enrolment dashboard rather than instead of them.
Every milestone met so far was met because it was defined as a state with a proof. None of them measures whether the program is on course for the ones ahead. The enrolment shortfall does not show up anywhere in the register until M-17 is missed — at which point it is roughly two years too late to respond.
Milestones are lagging indicators by construction. That is not a flaw; it is what makes them trustworthy. But it means a program that manages only its milestone register is steering by the wake, and it is why the register is read alongside the leading indicators in the enrolment dashboard rather than instead of them.
6. The One Ordering That Is Not Negotiable
Most of the register could be resequenced under pressure without damaging anything but confidence. Two rows could not.
| Ref | Milestone | Date |
|---|---|---|
| M-19 | Statistical analysis plan approved | 30 Jul 2028 |
| M-20 | Pivotal database lock | 05 Aug 2028 |
The SAP is signed before the database is locked, and the database is locked before anyone is unblinded. The gap between those two milestones is 6 days, which is tight, and under schedule pressure it is exactly the kind of gap somebody proposes to close by running the two in parallel.
Reverse those two and the trial result becomes unfalsifiable.
An analysis plan written after the data is visible is not a plan; it is a description of a choice already made, and no reviewer can tell the difference between a pre-specified analysis and a well-chosen one after the fact. The entire evidentiary value of a pivotal trial rests on the sequence being provable: blind data review assigns the analysis populations while still blinded, the SAP is signed, the database is locked, and only then is anyone unblinded.
This ordering is asserted in the fact base. If a future edit dates the SAP after the lock, nothing in this suite builds — which is a strange place to enforce a scientific principle, and the right place, because it is the only one that cannot be talked out of it.
An analysis plan written after the data is visible is not a plan; it is a description of a choice already made, and no reviewer can tell the difference between a pre-specified analysis and a well-chosen one after the fact. The entire evidentiary value of a pivotal trial rests on the sequence being provable: blind data review assigns the analysis populations while still blinded, the SAP is signed, the database is locked, and only then is anyone unblinded.
This ordering is asserted in the fact base. If a future edit dates the SAP after the lock, nothing in this suite builds — which is a strange place to enforce a scientific principle, and the right place, because it is the only one that cannot be talked out of it.
Everything else in this register is a commitment. This one is a constraint, and the difference is worth carrying: a commitment can be renegotiated with the people it was made to, and a constraint cannot be renegotiated with anybody.