5
eCTD modules
46%
Content in hand
24
Months to submission
5
Refuse-to-file grounds
Contents
1. An NDA Is Not a Document
A New Drug Application is a structured electronic dossier of five modules and many thousands of individual files, submitted through a gateway in a defined format. It is not written; it is assembled, from content produced over the preceding seven years by every function in the program.
| Position at 15 October 2026 | |
|---|---|
| Target submission | 12 October 2028 — 24 months away |
| Weighted content in hand | 46% |
| Most advanced module | Module 4 — nonclinical, 95%, finished since Gate 2 |
| Least advanced module | Module 2 — the summaries, 10%, and correctly so |
| The module carrying the most risk | Module 3 — CMC, 62%, gated by validation and stability |
46% is not a progress figure and should not be read as one.
Most of what is missing cannot exist yet. The pivotal clinical study reports require a database lock that is 21 months away; the summaries in Module 2 summarize content that has not been written. A readiness percentage against content that is structurally impossible to have is a number that measures the calendar, not the program.
Which is why every module below carries a gating dependency rather than a target date. The useful question is not “how complete is Module 5” — it is “what has to happen before Module 5 can be completed, and is that thing on track.”
Most of what is missing cannot exist yet. The pivotal clinical study reports require a database lock that is 21 months away; the summaries in Module 2 summarize content that has not been written. A readiness percentage against content that is structurally impossible to have is a number that measures the calendar, not the program.
Which is why every module below carries a gating dependency rather than a target date. The useful question is not “how complete is Module 5” — it is “what has to happen before Module 5 can be completed, and is that thing on track.”
2. The Five Modules
| Title | Owner | Content in hand | Gated by | Note | |
|---|---|---|---|---|---|
| Module 1 | Administrative and Prescribing Information | Dr. P. Raghunathan | ████░░░░░░ 35% | Labeling negotiation, which follows the data | Forms, patent and exclusivity statements, proposed labeling, REMS assessment. ⚠ Mostly assembly, and the one section here that is NOT assembly is the proposed label. |
| Module 2 | Common Technical Document Summaries | Dr. P. Raghunathan | █░░░░░░░░░ 10% | Every other module — it summarizes them | ⚠⚠ Written LAST and read FIRST. The Clinical Overview (2.5) is roughly 30 pages and is the part of the dossier most likely to be read in full by a reviewer. |
| Module 3 | Quality — CMC | Dr. K. Oyelaran | ██████░░░░ 62% | Process validation and 24-month stability | ⚠ The module that gates the filing independently of anything clinical. Furthest advanced and carrying the most named risk. |
| Module 4 | Nonclinical Study Reports | Dr. M. Sørensen | ██████████ 95% | Nothing — the studies are complete | Effectively finished since Gate 2. Reports exist, are signed, and need only to be formatted into the dossier. |
| Module 5 | Clinical Study Reports | Dr. S. Aldridge | ███░░░░░░░ 28% | Database lock, then 12 weeks to final CSRs | Phase 1 and Phase 2 reports are final. The two pivotal reports and the integrated summaries do not exist and cannot before lock. |
Module 2 is written last and read first, and that inversion is the single most useful thing to
understand about a submission.
The Common Technical Document summaries — the Quality Overall Summary, the Nonclinical Overview, and above all the Clinical Overview at 2.5 — are what a reviewer reads to form a view. They are roughly thirty pages against a dossier of many thousands, and they are written after everything they summarize exists.
So the part of the submission with the most influence per page is produced under the most schedule pressure, by the smallest number of people, at the very end. That is a resourcing observation as much as a regulatory one, and it is why the regulatory function's headcount does not fall away after the CSRs are done.
The Common Technical Document summaries — the Quality Overall Summary, the Nonclinical Overview, and above all the Clinical Overview at 2.5 — are what a reviewer reads to form a view. They are roughly thirty pages against a dossier of many thousands, and they are written after everything they summarize exists.
So the part of the submission with the most influence per page is produced under the most schedule pressure, by the smallest number of people, at the very end. That is a resourcing observation as much as a regulatory one, and it is why the regulatory function's headcount does not fall away after the CSRs are done.
The assert that keeps this honest: Module 2 may never be more ready than the least ready module it summarizes. If it were, somebody would be writing a summary of content that does not exist — which is not merely premature, it is the mechanism by which a summary comes to say something the underlying data does not support.
3. The Sequence That Cannot Be Compressed
| Step | What it produces | When |
|---|---|---|
| Database lock | Pivotal data frozen | Jul 2028 |
| Unblinding | Only after lock and a signed SAP | Jul 2028 |
| Topline results | First view of whether the claims are supportable | Jul 2028 |
| Clinical study reports | 12 weeks from lock | Sep 2028 |
| Integrated summaries | New analyses, produced in parallel with the CSRs | Sep 2028 |
| Module 2 summaries | Written last, from everything above | Sep 2028 |
| Gate 5 | Committee authorizes submission | Sep 2028 |
| Submission | eCTD to the electronic gateway | Oct 2028 |
Roughly eleven weeks separate the last clinical data point from the submission, and almost
none of it is compressible.
Lock, unblind, topline, clinical study reports, integrated summaries, Module 2 summaries, gate, submit. Each step consumes an input the previous one produces. The two pivotal reports are written in parallel by separate teams precisely because they cannot be written faster, and the integrated summaries run alongside them rather than after them for the same reason.
The only genuine slack in the whole sequence is work done before the lock — Modules 3 and 4, the administrative content in Module 1, the shell tables and analysis programs. Everything prepared in advance is a week not needed at the end, and everything left until after lock is a week on the critical path.
Lock, unblind, topline, clinical study reports, integrated summaries, Module 2 summaries, gate, submit. Each step consumes an input the previous one produces. The two pivotal reports are written in parallel by separate teams precisely because they cannot be written faster, and the integrated summaries run alongside them rather than after them for the same reason.
The only genuine slack in the whole sequence is work done before the lock — Modules 3 and 4, the administrative content in Module 1, the shell tables and analysis programs. Everything prepared in advance is a week not needed at the end, and everything left until after lock is a week on the critical path.
This is why Module 4 sits at 95% and is not a source of comfort. Nonclinical finished years ago; its readiness reflects work that was never at risk. The modules that determine the submission date are 3 and 5, and both are gated by things the program cannot accelerate — stability accruing at one month per month, and 68 weeks of treatment on the last participant randomized.
4. Refuse-to-File
The failure mode this assessment exists to prevent. A refuse-to-file is not a rejection of the drug — it is the agency declining to start the review at all, on the grounds that the dossier is not capable of being reviewed.
| Ground for refuse-to-file | What it means | Position |
|---|---|---|
| Incomplete on its face | Required content missing from a module | The most common and the most preventable. A completeness review against the guidance checklist catches it. |
| No agreed pediatric study plan | PREA requirement unmet | ⚠ Agreed iPSP is in place from the End-of-Phase-2 meeting. This one is closed. |
| Studies not adequate and well-controlled | Design or conduct inadequate on its face | Judged from the protocol and the CSR, not from the result. Two adequate pivotals is the conventional standard and this program has them. |
| Data not submitted in the required format | SDTM/ADaM non-conformance | ⚠ Mechanical, and entirely avoidable. Conformance is checked before submission with the same validator the agency uses. |
| Unsigned or unauthorized | Administrative failure | Rare, humiliating, and the reason the submission checklist has a signature block. |
A refuse-to-file does not restart the review clock. It removes it.
There is no clock until the submission is accepted, so a refuse-to-file costs the full time to remediate and resubmit, plus a fresh sixty-day filing review, plus the ten-month review that had not started. On this program's arithmetic that is comfortably more than a year, against a patent that expires on a fixed date.
And every ground on that list is preventable. None of them is about whether the drug works. They are completeness, format, and administrative failures — which is exactly why a completeness review against the published guidance checklist is run before submission rather than being trusted to the people who assembled the dossier.
There is no clock until the submission is accepted, so a refuse-to-file costs the full time to remediate and resubmit, plus a fresh sixty-day filing review, plus the ten-month review that had not started. On this program's arithmetic that is comfortably more than a year, against a patent that expires on a fixed date.
And every ground on that list is preventable. None of them is about whether the drug works. They are completeness, format, and administrative failures — which is exactly why a completeness review against the published guidance checklist is run before submission rather than being trusted to the people who assembled the dossier.
Two of the five grounds are already closed. The agreed pediatric study plan has been in place since the End-of-Phase-2 meeting, and the program has two adequate and well-controlled pivotal trials, which is the conventional evidentiary standard. The remaining three are all mechanical and all resolved by checking rather than by hoping.
5. Data Format Conformance
The one refuse-to-file ground that is purely technical, and the easiest to leave until too late.
| Requirement | What it involves | When it is checked |
|---|---|---|
| SDTM conformance | Study data tabulation model, to the required version | ⚠ Continuously during the study, not at the end. A conformance problem discovered after lock is expensive to fix and cannot be fixed by re-collecting. |
| ADaM conformance | Analysis datasets traceable back to SDTM | As the analysis programs are written, against the SAP. |
| define.xml | The machine-readable metadata that makes the datasets interpretable | ⚠ Not documentation about the datasets — it is how they are read. A dataset without it is a large amount of data nobody can use. |
| Validator run | The same conformance tool the agency uses, run before submission | Every issue it reports is either fixed or explained in a reviewer's guide. An unexplained conformance error invites the question of what else was not checked. |
Conformance is checked with the agency's own validator, and the findings are either fixed or
explained — never silently shipped.
Some conformance findings are legitimate: a study design feature the standard does not anticipate, a variable the model has no home for. Those go in the reviewer's guide with a rationale. What cannot happen is an unexplained finding, because a reviewer running the same validator sees the same list and now has to decide whether it reflects a data problem or an attention problem.
Some conformance findings are legitimate: a study design feature the standard does not anticipate, a variable the model has no home for. Those go in the reviewer's guide with a rationale. What cannot happen is an unexplained finding, because a reviewer running the same validator sees the same list and now has to decide whether it reflects a data problem or an attention problem.
6. What This Assessment Is For
| This assessment | gate5-readiness |
|---|---|
| Tracks the dossier — can the submission be assembled | Tracks the gate criteria — may the Committee authorize it |
| Owned by Regulatory Affairs | Owned by the Program Director |
| Answers: what is missing, and what is it waiting on | Answers: is the program permitted to proceed |
| A module at 28% is expected and fine | ⚠ A must-meet criterion unsatisfied produces RECYCLE |
| Updated continuously as content is produced | Updated monthly, tabled at the gate |
The two documents will disagree, and the disagreement is informative.
It is entirely possible for the gate criteria to be satisfied while the dossier is not assembled — that is the normal state right up until the final weeks. It is also possible, and far more dangerous, for the dossier to look nearly complete while a gate criterion is unsatisfied, because assembly progress is visible and cheerful while a missing registration batch is neither.
Reading only the module percentages would produce a confident program with an unfileable submission. The gating column exists so that the two are always read together.
It is entirely possible for the gate criteria to be satisfied while the dossier is not assembled — that is the normal state right up until the final weeks. It is also possible, and far more dangerous, for the dossier to look nearly complete while a gate criterion is unsatisfied, because assembly progress is visible and cheerful while a missing registration batch is neither.
Reading only the module percentages would produce a confident program with an unfileable submission. The gating column exists so that the two are always read together.
Both are tabled at Gate 5. The Committee is not asked to assess the dossier — that is Regulatory Affairs' professional judgment — but it is entitled to see whether the assembly is tracking against a submission date the program has committed to, and to hear directly from the owner of Module 3 whether the two things gating it are on course.