Lu et al. state 'Compared with glaucomatous eyes that received either PBS or AAVs with no OSK induction (?OSK), the OSK-treated glaucomatous eyes (+OSK) presented with a restored axon density equivalent to that in the non-glaucomatous eyes' and 'the optomotor response assay indicated that half of the visual acuity lost from increased intraocular pressure was restored.' Careful inspection of their data as presented do not support these statements.
Mice had bead-induced glaucoma for four weeks, were then intravitreally injected with a viral vector overexpressing three factors (OSK) and followed four more weeks. Retinal ganglion cell (RGC) bodies in retina and axons in the optic nerve were counted and two functional tests were performed. Regarding RGC density, the loss of RGC somas at four weeks (Fig. 9b) is shown for only five eyes with values differing by < 5% among them. This remarkable lack of variation is inconsistent with either normal variation as seen in their control RGC densities (and our mouse studies) or with the typical variability in RGC cell loss among mouse glaucoma eyes. More importantly, Figure 9d shows that despite injection of OSK vector, there was significant loss of RGC bodies at eight weeks that was not significantly different from the control vector or saline injection controls. In fact, there was no 'rescue or restoration' of RGC somas.
Assessment of axons was given as 'density' and not total axon number, the latter being the definitive method for axon counting. Various treatments (such as +OSK) may induce changes in axon density, but not in overall axon count (or vice versa), due to alterations in nerve astrocytes (Schaub et al. IOVS 2017) which affects optic nerve cross-sectional area. In addition, axon counts cannot be carried out in the same eyes at four weeks (prior to vector injection) and at eight weeks, so the axonal loss in different eyes is being compared, in as few as six per group, far smaller than needed to assure that variability has been accounted. In conducting bead glaucoma studies in > 1,000 mice, we find samples < 20 eyes per group are insufficient reliably to detect treatment effects.
Individual IOP exposure may be responsible for observed differences and are not presented. While the authors show one IOP graph in non-vector, bead-treated eyes (Fig. 3b), they do not account for IOP as a covariate in RGC axon and soma data. Perhaps animals with OSK injections, studied at eight weeks, had lower IOP exposure than the four week group, so any 'recovery' could simply be failure to achieve as much damage. Indeed, the investigators removed 'mice that had [ …] clouding or an oedematous cornea.' IOP increase in mice causes axial elongation, corneal steepening and edema, which are greater with higher induced IOP. The investigators may have systematically removed higher IOP eyes, minimizing damage at eight weeks. How many eyes were not included and was this more common in +OSK vector groups?
Therefore, these data require replication with the addition of numerous critical controls
They state: 'the optomotor response assay indicated that half of the visual acuity lost from increased intraocular pressure was restored (Fig. 3c,d).' While precise descriptive statistics are not provided, from Figure 3, control mice scored ~37 cycles/degree (cpd), while mice after four weeks IOP elevation scored ~25 cpd. The same +OSK group at eight weeks was only ~27 cpd, a 16% increase in the 12 cpd lost at four weeks; not 'half restored' (as claimed) and still 27% below baseline control values. In optomotor data, one should compare the acuity loss at four weeks to its value at eight weeks for each eye, not as group data. Again, individual IOP was not included in appropriate statistical models for acuity and ERG analysis. Furthermore, it would be informative to show correlative RGC soma and axon loss in the functional study animals. Presumably, mice could see from both eyes during the optomotor testing. Since only one eye had experimental glaucoma, its true effect on function is problematic. If one eye were closed (e.g., lid suture) we could determine that vision effects were actually due to change in the glaucoma eye.
Therefore, these data require replication with the addition of numerous critical controls.