Most PDF problems in a submission are mechanical: a font that was not embedded, a bookmark that resets the zoom, a link that opens the right document on the wrong page. None is hard to fix. All are expensive when they are found during final publishing.
Key takeaways
- Agencies check PDF properties as part of technical validation; failures can stop a sequence.
- The checks fall into six groups — file, text, navigation, links, content and files.
- Run them where documents are created and approved, not during final publishing.
- Fix at the source and re-render, or the defect returns with the next version.
Why PDFs fail
A submission’s PDFs are produced by many people with many tools: medical writers in Word, CMC teams exporting from lab systems, partners sending files of their own. Each tool has its own defaults, and few of them match what a regulatory agency expects. The result is a steady stream of small defects that are trivial individually and costly in aggregate, because they surface when the sequence is validated — usually days before a deadline.
The checks, group by group
File and version. Use a PDF version every target region accepts, with no password, security restrictions or encryption, and Fast Web View enabled so large files open at the first page. Text and fonts. Every page should be text-searchable, every font embedded, and scanned pages run through OCR.
Navigation. Bookmarks should mirror the document’s headings, the file should open showing bookmarks and the first page, and bookmarks should not reset the reader’s zoom. Links. Hyperlinks should use relative paths inside the submission, every target should exist, and each link should land on the page or table it cites — the check most often skipped, because a link to the right file on the wrong page passes a simple existence test.
Content and metadata. No embedded scripts, multimedia or attachments unless a region allows them; no author names or internal metadata left in the file; and a file type the agency accepts in that section. Files and pages. Lowercase names without spaces or special characters, names and paths within the specification’s length limits, and landscape pages rotated for reading.
A link to the right file on the wrong page passes most checks and still sends the reviewer the wrong way.
Move the checks upstream
The cheapest place to catch a PDF defect is where the document is created and approved, not where the sequence is assembled. Checking on upload, fixing mechanical problems in a new version, gating approval on the result and verifying every link by rule moves the work to the people who can fix it at the source — and leaves final publishing for review rather than repair.
Fix at the source. Patching a PDF inside the sequence fixes this sequence only. Correct the source document and re-render, or the same defect will be back in the next version.
How DnXT approaches it
DnXT Document QC checks documents against your rules when they are uploaded, fixes mechanical problems such as Fast Web View and bookmark zoom in a new version, can block approval while a high finding is open, and verifies every hyperlink against its destination text. DnXT Publisher then validates the assembled sequence against the target region’s own criteria.
Catch it before it reaches the sequence
Bring a set of documents and see DnXT Document QC and Auto Link Verification run on them.