Abdelgaied et al 2026 — Negative metformin pilot RCT in RRMS

One-paragraph summary

Negative randomized pilot trial of metformin as adjuvant therapy to interferon β-1a in 30 Egyptian patients with relapsing-remitting multiple sclerosis (RRMS). Trial design: experimental arm received IFN-β1a + 2 g/day metformin orally; control arm received IFN-β1a alone. Six-month follow-up. Primary endpoints: serum neurofilament light chain (sNFL), nuclear factor κB (NF-κB), T2 lesions on MRI, and Expanded Disability Status Scale (EDSS). Result: no statistically significant differences between arms in any endpoint — sNFL change −32.8 vs −32.8 (p=0.99); NF-κB change −64.9 vs −61.6 (p=0.8); EDSS change 0 vs 0 (p=1.0); MRI lesion progression nonsignificantly different (p=0.23). Tolerability was good: most common side effects were diarrhea and abdominal pain; no incidence of lactic acidosis. Authors conclude that metformin add-on did not improve outcomes in this trial and call for larger studies. For the project, this is the first negative human-trial result on a v0.3 cure-path arm and serves as an important counterweight to the Y 2026 metformin-clinical-trial-landscape enthusiasm and the Chaichana 2026 624K-population PCC-prevention signal. Same epistemic role as Ang 2014 played for H2 in v0.2 — a careful negative-result anchor that says "stratification, dose, and timing matter." Crucially, the Chaichana 2026 paper documents PCC-prevention efficacy (early metformin, prophylactic, large cohort), while this paper tests already-established RRMS (late metformin, therapeutic, small cohort) — the simplest interpretation is that metformin works prophylactically / in early-disease windows but has limited efficacy in established neurodegenerative disease.

Claims as triples

• metformin — absent_in → multiple_sclerosis [evidence: 6-month pilot RCT in established RRMS, no significant sNFL/NFkB/EDSS/MRI improvement (n=30); confidence: emerging (negative result, small cohort)]
• metformin — modulates → multiple_sclerosis [evidence: this paper provides negative data point; existing positive evidence (Kazakou, Qin, AlAseeri reviews) remains; confidence: contested]

Methods note

Single-center randomized pilot trial. Population: 30 RRMS patients in Egypt. Arms: IFN-β1a + 2g/day oral metformin (experimental) vs IFN-β1a alone (control). Allocation method and blinding not specified in the abstract — likely open-label given pilot framing. Outcomes: serum NfL (Simoa or equivalent), NF-κB activity (likely activity assay or pathway-marker ELISA), MRI T2 lesion count, EDSS clinical score. 6-month follow-up. Statistical: median (IQR) for continuous endpoints, comparative tests as appropriate. Trial registered: ClinicalTrials.gov ID NCT06812585.

Limitations

• Small sample size (n=30); underpowered for detecting modest effects. The authors explicitly call for larger studies.
• Already-established RRMS population. Patients had IFN-β1a as background therapy; metformin tested as add-on. Not a treatment-naive design. May not detect a metformin effect that requires earlier-disease intervention.
• Six-month follow-up is short for MS outcomes. Many MS efficacy trials run 12-24 months. Whether longer follow-up would have shown metformin effect is open.
• Single center, single ethnicity (Egyptian). Generalizability limited.
• No mechanism-biomarker stratification. Patients were not stratified by EBV serology, HERV expression, AMPK-pathway markers, or other potential responder predictors. The negative result may mask real efficacy in a stratified subset.
• Open-label likely, no placebo arm — limits causal interpretation even with the negative result.

Open questions raised

• Would metformin show efficacy in early-disease MS (clinically isolated syndrome / first relapse) rather than established RRMS? Chaichana 2026's PCC-prevention finding suggests early-window efficacy.
• Would metformin show efficacy in MS patients stratified by AMPK-pathway markers (mitochondrial dysfunction biomarkers, GDF15 baseline)? Stratified retrial design becomes the natural next step.
• Does metformin's positive PCC-prevention signal (Chaichana 2026, 624K cohort, HR 0.36) reflect a different mechanism than what's being tested in established RRMS — i.e., metformin works prophylactically in pre-disease/peri-onset windows but not in established neurodegenerative disease? This is the most consequential interpretation question.
• For the project's FM-application framing: does this negative result generalize to FM, where the cure-path arm timing is by definition therapeutic (treating established FM) rather than prophylactic? Unclear — but reinforces the case for stratifying any FM metformin trial by HERV+ / EBV+ / autoantibody+ subset rather than treating all-comers FM.

Triangulation notes

• Important counterweight to the metformin enthusiasm. v0.3 §12.9 is explicitly framed as a watch-list arm; this paper validates that framing rather than refuting it. The arm should remain a candidate for stratified pilot testing in HERV+/EBV+ FM rather than all-comers FM.
• Same epistemic role as Ang 2014 played for H2 in v0.2: a careful negative-result anchor that says stratification matters. Both trials enrolled unstratified populations and produced null results; both leave the door open for stratified retrials.
• Compatible with the Chaichana 2026 PCC-prevention finding under the "early window" interpretation. The most parsimonious unified read of the metformin literature: metformin prevents post-viral chronic conditions when given prophylactically (Chaichana 2026, 624K cohort) but has limited efficacy in established neurodegenerative disease (this paper). For FM, this argues for prophylactic / early-onset trials rather than treating chronic established FM with metformin.
• Tempers v0.3 §12.9 framing — the metformin arm should be promoted to "active candidate with stratified-trial design" rather than to "established cure-path candidate." The v0.3.1 white paper revision should reflect both the positive preclinical/population evidence and this negative human-trial result.

Bridges

• Tempers B17 — metformin / AMPK-activator class ↔ FM. Bridge remains active but the human-evidence base is now mixed (positive preclinical, positive population-scale prophylactic, negative human pilot in established disease).
• Reinforces the project's stratification discipline. All-comers metformin trials in established neurodegenerative disease may produce null results that don't refute the underlying mechanism but waste resources.

Status as a negative-result anchor

This paper joins Ang 2014 (ketotifen RCT in unstratified FM, NEGATIVE) as the project's second formally-ingested negative-result anchor. Both share the structural feature: unstratified enrollment + negative result + tolerability data + authorial call for stratified retrial. Both reinforce the project's central strategic recommendation that the biomarker-mapping cohort study should precede any new interventional trial in FM.