A methodological audit: Where the glasses may be rose-colored https://authors.elsevier.com/a/1nYxN4r9Rkz1wZ
The open-label problem
KOANEWA is a Phase Ib, open-label, non-randomized, single-dose study (ClinicalTrials.gov identifier: NCT07401121). The CENTAUR trial was open-label in its earliest iterations, and its 25% ALSFRS-R signal dissolved entirely under the blinded conditions of the PHOENIX Phase III trial (https://www.amylyx.com/news/amylyx-pharmaceuticals-announces-topline-results-from-global-phase-3-phoenix-trial-of-amx0035-in-als).(p2),(p3) Because the subjective ALSFRS-R is highly susceptible to expectation bias from both patients and investigators, unblinded designs evaluating high-visibility novel gene therapies face severe validation challenges. Without a sham or vehicle control arm – which, while ethically challenging to justify in a Phase Ib safety study, is not impossible to approximate through delayed-treatment or waitlist designs – any functional signal from KOANEWA cannot be attributed to CTx1000 with statistical confidence.
The biomarker gap
The field has converged on neurofilament light chain (NfL) as the leading surrogate end point in ALS, and NfL reduction formed the biomarker backbone of tofersen’s accelerated approval.(p5) KOANEWA’s secondary end points include NfL in cerebrospinal fluid and plasma (ClinicalTrials.gov identifier: NCT07401121). The fundamental limitation is that NfL is a downstream marker of axonal damage: a readout of neuronal death, not of TDP-43–14-3-3 complex formation, pathological aggregate clearance, or target engagement at the molecular level. Crucially, because there is currently no validated living-patient biomarker for TDP-43 aggregate clearance itself, NfL shifts remain ambiguous: a reduction suggests slower neurodegeneration but does not confirm specific 14-3-3/degron target engagement, while an unchanged NfL cannot distinguish between a lack of molecular efficacy, an unengaged target, or poor regional biodistribution.
A more mechanistically proximal alternative is emerging. Because nuclear TDP-43 loss of function produces measurable cryptic-exon-containing RNA species (in STMN2, UNC13A, and other transcripts), these cryptic transcripts and their potential protein products are increasingly discussed as candidate biofluid biomarkers of TDP-43 functional status, distinct from NfL’s non-specific readout of neuronal injury.(p11),(p12) Developing a validated assay along these lines – whether a cerebrospinal fluid assay for cryptic RNA species, the TDP-43–14-3-3 complex itself, or a positron emission tomography (PET) ligand for TDP-43 pathology – might be as crucial to the field’s long-term progress as CTx1000 itself. Celosia’s clinical development plan should explicitly address this gap, rather than relying on NfL as a proxy for mechanism.
The exclusion criteria paradox
KOANEWA requires disease onset within two years, excludes C9ORF72 repeat expansion carriers (the most common genetic form of ALS-FTD, accounting for approximately 40% of familial and 5–10% of sporadic ALS cases(p17)), and imposes an anti-AAV9 titer cutoff of ≤1:50 at enrollment (ClinicalTrials.gov identifier: NCT07401121). These criteria create a highly selected, non-generalizable patient population. Early-onset, non-C9ORF72, seronegative TDP-43 proteinopathy patients might respond differently from the broader sporadic ALS population in whom any eventual therapy must demonstrate real-world value. It is also worth noting that many other ALS trials already restrict enrollment to genetically defined or otherwise selected subpopulations, so KOANEWA’s recruitment feasibility is unlikely to be limited by these filters, even though the resulting evidence base will be narrow. Figure 2 illustrates the KOANEWA eligibility criteria and their non-generalizability for the ALS patient population.
Posted in