Granulocyte colony-stimulating factor

[Drs. Borlongan and Hess respond:]

Although we welcome the clinical trial by Woei-Cherng Shyu and associates¹ and the European clinical trials cited in our commentary,² we reiterate that critical assessment of these trials is warranted. As we pointed out,² given the small number of patients in the study by Shyu and associates, and the variable onset of treatment, conclusions about the efficacy of this treatment cannot be drawn. Nonetheless, this is the first demonstration of feasibility and safety of G-CSF in human stroke patients. In view of the limited therapeutic window for thrombolytic therapy with tissue plasminogen activator, the feasibility and safety of administering G-CSF between 1 and 7 days after stroke is a significant clinical advance.

Julian Harriss argues that at the outset, the patients assigned to receive the treatment had on average “more than 10% better stroke scale scores” than the patients who received the placebo. Shyu and associates¹ noted that there were no statistically significant differences at baseline between the 2 groups of patients, who were randomly assigned to the treatment conditions. We confirmed, through our own statistical analysis, that there were no significant differences between the groups at baseline for the 4 stroke scales used. Closer examination of the baseline scores (see Table 2 in the article by Shyu and associates¹) reveals that for 3 of the 4 stroke scales (National Institutes of Health Stroke Scale, ESS, ESS Motor Subscale), the patients with the worst scores were in the G-CSF group; for the Barthel Index, 3 of the patients with the second-worst score were in that group.