Digest — 2026-09-06, The Clinical
Four things crossed the escalation bar. The first one is a miss by last week's run, and it is recorded as such. The last one has a deadline eight days out.
1. Amulet 360 has a CE Mark. It has had one since 27 August, and last week's run said it did not.
Abbott announced CE Mark for the Amulet 360 Left Atrial Appendage Occluder on 2026-08-27, from Abbott's own newsroom. The 2026-08-30 clinical run ran a targeted news sweep for exactly this and reported "no new Amulet 360 announcement, IFU, size matrix, or CE-mark statement was found." The release had been live for three days.
Your charter's rule for this item is anything at all, until it launches. A keyword news sweep has now failed that rule once. Standing method change recorded in the entry: read Abbott's own newsroom index directly on every run until the device launches.
What actually changed.
The sentence this corpus has carried since 2026-08-09 — Abbott's own "not available for sale in any geography" — is now wrong for Europe. The accurate sentence, and both halves have to travel together:
CE marked in Europe as of 27 August 2026. Investigational in the United States, IDE only. US launch named by Abbott's CEO for within twelve months of July 2026.
Two things in the release are worth your attention beyond the headline.
The DRT discrepancy has partially resolved, and it resolves through timing. Cardiac Interventions Today's VERITAS write-up states device-related thrombus as "one event observed at 45 days; zero through 7 days." If that is right, Abbott's "one DRT event" is a 45-day figure and the paper's 2.4% covers a fuller adjudication. That is reported from a secondary source — neither Abbott nor the paper says it. Your guidance does not change: say 2.4%. But you now have a reconciliation to offer instead of only a discrepancy to concede.
The loose closure definition is house wording, not a slip. The CE Mark release again pairs 93.9% with "no leaks >3 mm." The JACC paper defines that same 93.9% as 0 mm leak. Abbott has now published the looser definition alongside the paper's number twice, seven months apart. Assume it will be in every piece of Amulet 360 material you are handed.
One failed check, stated: Abbott's VERITAS clinical-evidence page now returns 404, so the "not available for sale in any geography" quotation could not be re-verified and may have been revised.
→ substrate/clinical--amulet-360
2. The Merlin software version question is closed for the US — it is v28.9.1
You can stop hedging on this one. The 2026-08-30 digest told you to say "v28.9.1 or v28.9.2 rev 1 or higher" and that the exact build was unconfirmed. That was right at the time. It is no longer necessary in the US.
Abbott's own US-facing Cardiovascular Product Advisories index — read for the first time this run; every previous read in this corpus was of the international index — gives the resolution as:
"Updated Merlin™ PCS 3650 programmer software (Model 3330 v28.9.1 Rev 1 or higher)"
That agrees with the HRS-carried notification. Two US-facing sources now say v28.9.1. The v28.9.2 figure is from the ANSM-hosted French copy of the EU notice.
Each number appears in its own geography's document, which makes a regional build the strongest available reading. It is a reading, not a sourced claim — Abbott has not reconciled them.
Practical version, for a device clinic: in the US, confirm v28.9.1 Rev 1 or higher. If a clinic shows you a European notice specifying v28.9.2, the answer is not confusion — "the EU notice carries a different build number; anything at or above what your notice specifies is the point."
One thing that stops this from becoming a rule about sources. On the November 2024 AVEIR DR event, Abbott's US index says the fix was v28.2.1 rev 2 where HRS says v28.3.2 rev 1 — the disagreement runs the other way. So the lesson is narrower and more useful: Merlin remedy version numbers differ between channels often enough that you read the version off the programmer and compare it to the document that clinic actually received.
Also newly recorded, and it further defuses the "still not terminated" question: Z-0753-2025, the November 2024 recall, is itself still "Open, Classified" nearly two years on.
→ substrate/clinical--merlin-pcs-3650-recall
3. PVI-SHAM-AF — the first sham-controlled ablation trial, and its primary endpoint was negative
This is the single most quotable EP result of the year and it will be put to you. ESC Congress 2026, reported 30 August, simultaneously published in The Lancet.
262 symptomatic AF patients, randomised 2:1 to catheter ablation or a sham procedure — conscious sedation for an equivalent duration plus vascular access, no catheter placement — across nine sites in Germany and Poland, double-blind.
| Ablation | Sham | |
|---|---|---|
| AFEQT score, baseline → 6 months | 61.3 → 81.1 | 59.2 → 74.9 |
| Between-group difference | p = 0.36 | |
| Free of AF at 6 months | 73% | 52% |
The trial did not show ablation does not work. It showed the six-month symptom benefit is not attributable to the ablation. The rhythm effect is intact and large. What failed to separate was a patient-reported questionnaire, because the sham arm also improved by about 16 points.
Three things to have ready, all from the trial's own numbers.
- 73% vs 52%. Lead with it. The rhythm effect is real.
- Six months, and AFEQT. A short window and a symptom instrument. CABANA, CASTLE-AF and EAST-AFNET 4 asked different questions. Do not let a QoL null be restated as an outcomes null — and do not overstate it in the other direction yourself.
- 937 of 1,199 invited patients declined. Say it before someone else does. The patients willing to risk a sham are not the ones with the heaviest symptom burden. It is a real limitation and volunteering it is more credible than being handed it.
Do not dismiss it. Double-blind, sham-controlled, in The Lancet, funded by a hospital group and a university rather than by a device company. It is the best evidence of its type that exists.
And it is not a PFA argument. It tested pulmonary vein isolation as a strategy, not an energy source. The modality mix was not reported. If anyone tries to make it about catheters, it is not.
Numbers are reported — thelancet.com returned 403 to this run's fetch. The citation is recorded for a future upgrade.
→ substrate/clinical--pvi-sham-af
4. The CY2027 PFS comment window closes in eight days — 14 September 2026
Verified against the Federal Register document itself this run, not a summary. CMS-1848-P, FR doc 2026-14327. DATES section, verbatim:
"To be assured consideration, comments must be received at one of the addresses provided below, by September 14, 2026."
Comment period confirmed OPEN. No correction or extension notice on the document page. 41,624 comments received to date.
This is the rule proposing that RPM/RTM be billable only by practice-employed staff. As written it does not reach the 93293–93299 CIED family — the corpus established that on 2026-08-09 by full-text search of the rule, and nothing has changed it. Your exposure through the Vector contract work is real but not immediate. The live risk is extension in a later cycle, not enactment in this one.
Whether to comment is your decision and no agent's. After 14 September it is nobody's. The SaMS window (CMS-1850-P) closed on 31 August; that one is gone.
→ substrate/clinical--cms-cy2027-remote-monitoring
Also written this run, below the bar
SyncAV PMT — Abbott's own CRT optimisation algorithm, ESC Hot Line 10, presented by Niraj Varma on 31 August. ESC describes overall efficacy as "robust and comparable between arms," with benefit in a PR ≥185 ms subgroup (mortality, HF hospitalisation, persistent AF, negative echo response). The registered primary endpoint, from the UK HRA record, is change in LVESV at 12 months — and no source retrieved gives the sample size or any primary-endpoint statistic.
The hallway version will be "SyncAV reduced mortality." That is not what was reported. The defensible and more useful sentence is the qualified one, because PR interval is a measurable pre-procedural variable and a subgroup like that is a programming indication rather than a weakness. Whether PR ≥185 ms was pre-specified is not established and this corpus does not guess.
Escalation-relevant enough to write up now, listed here rather than above because it is thin on retrievable substance. → substrate/clinical--syncav-dynamic-av-optimization
LAACS-2 — surgical LAA closure during open-heart surgery, ESC Hot Line 12. Primary endpoint missed: stroke or TIA 4.28% vs 5.04%, HR 0.85 (0.53–1.37), p=0.517, over a median 4.0 years in 1,500 patients. The half with CHA₂DS₂-VASc above the median had a 56% reduction (p=0.018).
The number that matters when this is quoted at you: only 4.1% of LAACS-2 patients had atrial fibrillation. It was prophylactic closure in a general cardiac surgery population, not stroke prevention in AF. It bears on surgical practice; it does not bear on Amulet's indication. And the subgroup runs toward the mechanism — closure helped where appendage-derived stroke was most plausible. Do not say "LAACS-2 supports Amulet." Say it fails to undermine it and its subgroup is directionally consistent. → substrate/clinical--laacs-2-surgical-laa-closure
Instruments — one problem diagnosed, two admitted
ClinicalTrials.gov is diagnosed rather than just broken. The corpus has carried "unreachable" since 2026-08-30. Tested two ways this run:
clinicaltrials.gov/api/v2/studies/NCT…→ ROBOTS_DISALLOWED, refused at the target.clinicaltrials.gov/study/NCT…→ HTTP 200, but the record is client-rendered and the fetch returns only header, metadata and glossary.
The host is reachable, the API path is policy-blocked, the record path returns an empty shell. So NCT-ID retrieval is not coming back through this route. Every trial-registration fact in this corpus that was established by direct NCT lookup cannot currently be re-verified — including the VERITAS record. Registry facts now have to come from publications, sponsor pages or regulator records. This closes the fleet question the Technologist handed back untested on 2026-09-04.
PMC is still blocked — sixth consecutive run. The Assert-IQ Europace 2026 supplement (PMC13300832), which the 2026-08-30 digest named as the first thing to read this run, returned a reCAPTCHA at PMC and HTTP 429 at Europe PMC's REST service. Not read. Still the first thing to read next run — it is where Abbott's unsupported 91% pause-reduction and 98.7% figures would have to come from, if they come from anywhere.
Two checks this run did not perform, stated so no one assumes they were done: openFDA was not re-queried at all (Amulet PMA, Merlin recall status, and the Abbott recall sweep are unchecked since 2026-08-30), and the FDA Safety Communications page was not re-read.