Finding. In Ndufs4 knockout Leigh mice, two rescue strategies part company the moment you stop them. Pexidartinib and rapamycin keep helping after the drugs are gone. Take the animals off chronic 11% oxygen and disease starts fast and then runs faster. Before onset the brainstem does not look inflamed. Wipe out leukocytes and the molecular disease signature goes with them. Macrophages and monocytes look like the effectors. Hypoxia, in this reading, sits upstream of immune activation. That is a mechanistic split and a translation warning.
Why this paper matters
Leigh syndrome is the common pediatric face of genetic mitochondrial disease, and the Ndufs4(-/-) mouse is where most modern “this might work” ideas are born. Two of those ideas are immune targeting (rapamycin, high-dose pexidartinib) and chronic mild hypoxia. Olkhova, Kayser, Dimitriou and Johnson ask the question that clinics will ask: what happens when you stop?
If hypoxia is a switch that holds inflammation off, cycling a child off low oxygen could synchronize a crash. If CSF1R blockade or rapamycin remodels the immune set-point, a finite course might be thinkable. Those are different products. The paper also hands future users a method: hypoxia-cessation to line up inflammatory onset.
What they actually measured
Brainstem immune profiles before and after onset, and under pexidartinib. Evidence that macrophages/monocytes drive pathology, aligned with recent genetics. Pre-onset animals without inflammatory signs. Leukocyte elimination that clears the molecular disease signature. Distinct post-developmental treatment windows, then cessation: pexidartinib and rapamycin persist; hypoxia cessation detonates and accelerates disease.
They do not claim to have solved how hypoxia works, only where it sits relative to immune activation. That epistemic humility is part of the value.
How to read the score
Low eighties. Pediatric GMD, the field’s workhorse Complex I model, a translationally load-bearing on/off design. Confidence is medium for the upstream-of-immune inference and high for the cessation phenomenology in this mouse. Score 83. Heuristic 72 undersold it.
What to do with it
If you work on Leigh, Ndufs4, hypoxia therapy, or neuroimmune mitochondrial disease, this is required reading. Pull the brainstem immune panels and the cessation survival curves. Do not write a protocol that cycles children off hypoxia based on this brief. Do not equate pexidartinib “full suppression” in a mouse with a human CSF1R plan. The directional implication is that immune targeting and hypoxia are not the same lever, and hypoxia’s benefit may vanish when oxygen returns.