Neural stem cells in models of spinal cord injury.

Replication of published studies is an important and respected aspect of the conduct of science. Most would argue that the interpretation of "negative" outcomes is still more challenging than the interpretation of "positive" findings, however, due to uncertainty in knowing precisely why a hypothesized outcome was not observed: in particular, are "negative" findings in replication studies a result of invalidity of the original experimental hypothesis, or due to a methodological failure, insensitivity of the applied instruments of analysis, or other factors? These points must be carefully considered. Steward and colleagues report findings of a study in which multipotent neural progenitor cells were grafted to sites of T3 complete transection. Unlike our study, cells failed to fill the lesion site, leaving collagenous rifts between rostral and caudal graft components. This "anatomical" failure precluded formation of neural relays across the lesion site, and was predictably associated with a failure to detect functional improvement. In summarizing outcomes of the study, Steward and colleagues did not clearly link the failure to achieve graft continuity in the lesion cavity with functional outcomes, despite the central role of this observation in cogently interpreting results of the replication study. In addition, the authors stated that they failed to replicate our report of "extensive" host axonal regeneration into grafts, but we did not report "extensive" host anatomical regeneration; moreover, underexposed images may have contributed to Steward's underestimation of host axonal penetration. The authors also stated that our original study excluded some animals from functional analysis, and this is incorrect. While replication studies are important and necessary, this particular report contained several errors, and the failure to form a continuous neural progenitor cell bridge across the lesion site limited the ability to conclude whether continuous grafts can restore function. In subsequent experiments we too have observed rift formation in animals grafted at long delays (>2weeks) after SCI, and we confirm that animals with rifts do not exhibit functional improvement; we are developing methods to remove or prevent rift formation. The replication study confirmed the cardinal finding of our original report: that early-stage neural precursors extend very large numbers of axons over remarkably long distances through the lesioned adult spinal cord. Published by Elsevier Inc.