Uthando Nethemba

The TTR silencers — siRNA vutrisiran succeeded, ASO eplontersen missed, but the deciding variable was the trial population, not the modality

—

by

itembalabantu
in

Evidence-first notes on bioscience and deep tech, at the edge of the lab and the market. Information only — not investment advice, not medical advice. All knockdown, hazard-ratio and subgroup figures are attributed to the sponsoring trial, journal or company; several eplontersen numbers are a company press release rather than peer-reviewed data (noted inline), with full data pending. These are cross-trial observations, not head-to-head comparisons.

The 30-second version

  • What. Two liver-directed TTR silencers knock down transthyretin to nearly the same depth — eplontersen (an antisense oligonucleotide, ASO) −84%, vutrisiran (an siRNA) −88%, patisiran >80% — yet in hard-outcome cardiomyopathy trials they split: vutrisiran succeeded in HELIOS-B (overall HR 0.72, p=0.01) while eplontersen missed its primary endpoint in CARDIO-TTRansform (AstraZeneca/Ionis press release, 2026-07-08). The surface headline reads “siRNA wins, ASO loses.”
  • So what. That headline is not supported by the evidence. Knockdown is equivalent, so neither modality nor knockdown depth is the deciding variable. The monotherapy (no-stabilizer) subgroup hazard ratios are near-identical — vutrisiran 0.67 vs eplontersen 0.71 — and both lose their effect in the stabilizer-combination stratum (HELIOS-B combination HR 0.79, non-significant; CARDIO-TTRansform “no effect observed in the stabilizer arm”). The deciding variable is the background-therapy mix — the threshold of proving an add-on benefit on top of a tetramer stabilizer that has become standard of care — not the molecule.
  • Now what. The hidden variable is enrollment-era standard-of-care drift: HELIOS-B enrolled a ~60%-monotherapy population and succeeded; CARDIO-TTRansform enrolled a more recent population with ~81% stabilizer exposure (57% at baseline plus 24% starting on-study) and missed. The eplontersen monotherapy HR 0.71 is a nominal subgroup sitting under a failed primary endpoint (multiplicity-uncorrected → hypothesis-generating), disclosed PR-only (Businesswire, 2026-07-08); decisive full data are due at ESC 2026-08. The honest verdict is proceed-with-caveats: “ASO-inferior / siRNA-superior” is not established.

The five-minute read

Same silencer target, same knockdown depth, split outcomes

This series’ central puzzle: two drugs that both silence hepatic TTR production deeply (into the 80s of percent) reached opposite conclusions in cardiomyopathy hard-outcome trials. The siRNA vutrisiran succeeded in HELIOS-B (overall HR 0.72, p=0.01); the ASO eplontersen missed its primary endpoint in CARDIO-TTRansform (AstraZeneca/Ionis press release, 2026-07-08 to -09). Read the two trials against their own subgroups, however, and a different account emerges: the variable that split success from miss was not the modality (siRNA vs ASO) and not knockdown depth, but the threshold of proving an add-on benefit on top of a stabilizer that had become standard of care. The decisive alignment is that the two drugs’ monotherapy subgroups are almost identical — vutrisiran HR 0.67 vs eplontersen HR 0.71.

One caveat belongs up front, at headline level. The eplontersen numbers are press-release-only. The full effect size and confidence intervals are pending ESC 2026-08. The monotherapy HR 0.71 quoted below is a nominal subgroup under a failed primary endpoint — hypothesis-generating, not confirmatory. And the 0.67-vs-0.71 proximity is a cross-trial observation (different endpoints, timepoints and sample sizes), not a head-to-head test.

Drug (modality) Overall result Monotherapy (no-stabilizer) subgroup Stabilizer-combination stratum
vutrisiran (siRNA) — HELIOS-B HR 0.72 (0.56–0.93), p=0.01; all-cause death 0.65 (0.46–0.90) — success HR 0.67 (0.49–0.93), P=0.02 (~60%, 395 patients) HR 0.79 (0.51–1.21) — non-significant (diluted)
eplontersen (ASO) — CARDIO-TTRansform Primary endpoint missed (PR-only; full data ESC 2026-08) HR 0.71 nominal (CI undisclosed; under a failed primary → hypothesis-generating) “No treatment effect observed” (PR)
Knockdown depth (neuropathy population) eplontersen −84% (18 mo) ≈ vutrisiran −88% (18 mo) ≈ patisiran >80% (max 96%) — equivalent; cardiomyopathy head-to-head is unverified
The monotherapy hazard ratios are near-identical (0.67 ≈ 0.71) and both drugs lose their effect once a stabilizer is added. What differs between the two trials is the population’s stabilizer mix — HELIOS-B ~60% monotherapy vs CARDIO-TTRansform ~81% stabilizer exposure — not the molecule or the knockdown depth. Eplontersen figures are PR-only (Businesswire 2026-07-08); full data pending ESC 2026-08. Cross-trial, not head-to-head.

Deep dive

1. Background — vutrisiran’s HELIOS-B (success)

Trial: N=655, vutrisiran 25 mg subcutaneous once every 12 weeks (quarterly); primary endpoint = composite of all-cause death plus recurrent cardiovascular events; follow-up to 42 months (NEJM 2025;392:33–44 / JACC 2025).

  • Overall: HR 0.72 (95% CI 0.56–0.93), p=0.01; all-cause death at 42 months HR 0.65 (0.46–0.90).
  • Monotherapy subgroup (no baseline tafamidis, ~60% / 395 patients): HR 0.67 (0.49–0.93), P=0.02 (NEJM reports p=0.016). CV events alone: RR 0.68 (0.53–0.86), P<0.01.
  • Tafamidis-combination subgroup (~40%): HR 0.79 (0.51–1.21) — same direction but non-significant (CI crosses 1.0).

The point worth emphasizing first: even within HELIOS-B, the stabilizer-combination stratum was already diluted (0.79, non-significant). The reason HELIOS-B succeeded overall is that ~60% of its sample was monotherapy and that stratum was robust (0.67), pulling the pooled result up. In other words, “success versus miss” already looks like a function of the background-therapy mix from inside this single trial.

2. What this analysis newly establishes — eplontersen’s CARDIO-TTRansform (missed, PR-only)

Trial: N=1,432, eplontersen 45 mg subcutaneous once every 4 weeks (monthly), 130 sites across 20 countries, 1:1 randomization; primary endpoint = composite of CV death plus recurrent CV events at Week 140 (≈2.7 years). Stratification: NYHA (I/II vs III), genotype (wild-type vs variant), 6-minute walk (≤350 vs >350 m), and stabilizer use (yes vs no) (AstraZeneca/Ionis press release 2026-07-08 to -09; baseline characteristics in EJHF).

  • Primary endpoint: not met. Adding eplontersen on top of modern standard of care (most patients on a stabilizer) “did not provide a significant benefit” (PR wording).
  • Background therapy (the key numbers): 57% of each arm was on a stabilizer at baseline, with a further 24% starting a stabilizer on-study → effectively ~81% stabilizer exposure. Pure monotherapy was a minority (~19%).
  • Baseline-stabilizer arm: no treatment effect observed (PR).
  • Monotherapy (no-stabilizer) subgroup: HR 0.71 (nominal) — but a subgroup under a failed primary endpoint, so multiplicity-uncorrected and hypothesis-generating. The CI is not disclosed in the PR (unverified).

The PR-only boundary matters: the monotherapy 0.71 is a point estimate only. The CI, p-value and event counts cannot be checked until the full data at ESC 2026-08. The statement “it works in monotherapy” is not confirmable at this time.

3. Strengths and limits — testing the causal hypothesis: background-therapy threshold vs modality vs knockdown

Three hypotheses, narrowed on the evidence.

(a) Did the modality itself (siRNA vs ASO) decide it? — weak support. The mechanisms differ: eplontersen is an ASO that recruits RNase H to cleave TTR mRNA, while vutrisiran is an siRNA using Argonaute 2 (Ago2); both are GalNAc-conjugated for liver targeting. But knockdown depth is effectively equivalent — eplontersen serum TTR −84% (18 months, NEURO-TTRansform, neuropathy), vutrisiran −88% (18 months, HELIOS-A, neuropathy), patisiran >80% (up to 96%). A review states the clinical differences come “from trial design, population and dosing schedule, not molecular mechanism” (Medicine mini-review, 2024). This points toward rejecting the modality/depth hypothesis. (Caveat: the head-to-head knockdown in the cardiomyopathy population remains unverified — the figures above are from neuropathy populations.)

(b) The add-on-on-stabilizer threshold (most likely) — strong support. The decisive alignment: the two drugs’ monotherapy HRs are near-identical — vutrisiran 0.67 vs eplontersen 0.71. Conversely, both lose their effect in the stabilizer-combination stratum — HELIOS-B combination 0.79 (non-significant), CARDIO combination “no effect observed.” The two trials show the same-direction biology; only the sample’s stabilizer mix differs:

  • HELIOS-B: ~60% monotherapy → pooled success.
  • CARDIO-TTRansform: ~81% stabilizer exposure → the diluted stratum dominates the sample → pooled miss.

The hidden variable is enrollment-era standard-of-care drift. HELIOS-B enrolled when tafamidis penetration was low, so monotherapy predominated; CARDIO-TTRansform enrolled a more recent population in which stabilizers had hardened into standard of care. The same drug was therefore tested against a higher bar — “add-on on top of a stabilizer.” This reads not as “ASO does not work” but as a trial-epidemiology problem: it has become harder to demonstrate incremental benefit when knockdown is layered onto a heart whose tetramer is already stabilized.

(c) Timepoint, endpoint definition, power — secondary variables. CARDIO’s primary timepoint of Week 140 (≈2.7 years) vs HELIOS-B’s up to 42 months — amyloid deposition reverses slowly, so a shorter observation window can close before event curves separate. The endpoint denominators also differ: HELIOS-B all-cause death + CV vs CARDIO CV death + CV. In a population whose CV death has already been lowered by stabilizers, a CV-death-based composite has sparse events, which can erode power — a secondary factor that interacts with the background-therapy threshold.

Conclusion: the primary variable that split the outcomes was not modality and not knockdown depth but the background-therapy mix (the add-on threshold), amplified by endpoint definition and observation window. Declaring “ASO inferior” is not supported by the evidence.

4. Neighbouring domains — patisiran (APOLLO-B): surrogate vs hard outcome

Patisiran (a first-generation siRNA, IV every 3 weeks) tested APOLLO-B (N=360, 21 countries) with a primary endpoint of 12-month change in 6-minute walk (6MWT) — a surrogate/functional endpoint. Result: a median between-group difference of 14.69 m (95% CI 0.69–28.69), P=0.02, plus KCCQ quality-of-life P=0.0397, both met. Yet the FDA issued a Complete Response Letter (non-approval) in October 2023 — the reason was not safety or quality but “insufficient evidence of clinical meaningfulness and short cardiomyopathy follow-up.” Two lessons follow: (1) meeting 6MWT ≠ hard-outcome approval — regulators require separate evidence that a surrogate is clinically meaningful; (2) so the subsequent silencer trials (HELIOS-B, CARDIO) raised the bar to death/CV hard composites, and in doing so surfaced the add-on threshold problem. Patisiran’s surrogate success followed by non-approval effectively foreshadowed the rising threshold of the hard-outcome era.

5. Commercialization and competitive context

  • Maturity: vutrisiran (Alnylam, ALNY) carries an approved hard-outcome result in ATTR-CM; eplontersen (AstraZeneca AZN / Ionis IONS) missed its cardiomyopathy primary endpoint, with full data pending ESC 2026-08. These are neutral, trial-attributed facts, not efficacy rankings.
  • The disclosure asymmetry: HELIOS-B is peer-reviewed (NEJM 2025; JACC 2025); the eplontersen CARDIO-TTRansform result is press-release-only (AstraZeneca/Ionis/Businesswire, 2026-07-08 to -09), with the decisive effect size and confidence intervals still to come at ESC 2026-08. Any commercial read should wait for the full data.
  • Read-through for the modality debate: because knockdown is equivalent and the monotherapy HRs are near-identical, a plausible design implication is that future silencer trials may need to enroll stabilizer-naive/monotherapy populations deliberately to demonstrate a hard-outcome benefit — a trial-design question, not a molecule-ranking one.
  • Company statements are limited to neutral, trial-attributed description; competitive or efficacy-ranking statements are not buy/sell signals. The eplontersen effect size and CIs are [unverified] pending ESC 2026-08.

6. The skeptic’s bottom line

  • PR-only boundary: the eplontersen miss and the monotherapy HR 0.71 are press-release-only with undisclosed CIs; decisive data are due ESC 2026-08.
  • Subgroups, not confirmations: the monotherapy signals (in both trials) are subgroups — and CARDIO’s sits under a failed primary endpoint, so it is multiplicity-uncorrected and hypothesis-generating.
  • Knockdown comparison is off-population: the 84%/88% figures are from neuropathy populations; the cardiomyopathy head-to-head is unverified.
  • Cross-trial, not head-to-head: the 0.67-vs-0.71 proximity spans different endpoints, timepoints and sample sizes — do not treat it as a quantitative identity.
  • Neutral-framing note: the ALNY success vs AZN/IONS miss must be stated as factual and neutral. “siRNA superior / ASO inferior” as a security implication is blocked.

7. What to watch (falsifiable)

  • P1 — ESC 2026-08 full data: if the eplontersen monotherapy HR reproduces with an upper CI bound <1.0 while the stabilizer stratum is flat → the add-on-threshold hypothesis is strengthened. If the benefit also disappears in monotherapy → the reading shifts toward a drug/modality-specific failure. (Verification point: 2026-08.)
  • P2 — deliberate enrollment: if a future silencer trial intentionally enrolls a stabilizer-naive/monotherapy population and hits a hard outcome → the threshold hypothesis is validated; failure in such a population would falsify it.
  • P3 — cardiomyopathy head-to-head knockdown: aligning serum-TTR reduction in the cardiomyopathy population should show that any siRNA-vs-ASO difference lies in dosing/durability, not depth. A significant depth gap would revive hypothesis (a).
  • Also watch: how, absent a head-to-head, cross-trial hazard ratios can be compared honestly — whether synthetic control arms or adaptive designs are the answer (carried to Part 4’s methodology axis).

References

Disclosure

This post is for information only and is not investment advice, and not medical advice. Treatment decisions should always be made with your own clinician.

COI note: this post describes listed companies (Alnylam ALNY, AstraZeneca AZN, Ionis IONS) and their sponsored trials in a descriptive, neutral context. Every knockdown, hazard-ratio and subgroup figure is attributed to the sponsoring trial, journal (NEJM, JACC) or company. The eplontersen CARDIO-TTRansform result — the primary-endpoint miss and the monotherapy HR 0.71 — is a company press release (AstraZeneca/Ionis/Businesswire, 2026-07-08 to -09), not peer-reviewed data; the full effect size and confidence intervals are pending ESC 2026-08 and are labeled PR-only and [unverified]. The monotherapy subgroup sits under a failed primary endpoint and is multiplicity-uncorrected (hypothesis-generating). The knockdown figures (−84% / −88% / >80%) are from neuropathy populations; the cardiomyopathy head-to-head is unverified. All quantitative claims are attributed to the vendor, author, trial or press release. Cross-trial observations (including the 0.67-vs-0.71 proximity) are not head-to-head comparisons, and “siRNA superior / ASO inferior” is not established and must not be framed as a security implication. Competitive and outcome statements are factual, neutral descriptions and are not buy/sell implications for any security. The author holds no position in, and has no financial interest in, the companies named.

Comments

Leave a comment