Neurology -- Correspondence for Silver et al., 67 (5) 748-755

Correspondence published:

Effects of rivastigmine on cognitive function in patients with traumatic brain injury: James M. Noble, M.D., W. Allen Hauser, M.D. (3 December 2006)

Reply from the Authors: Jonathan M. Silver, Philip D. Harvey (3 December 2006)

Effects of rivastigmine on cognitive function in patients with traumatic brain injury 3 December 2006

James M. Noble, M.D.,
Gertrude H. Sergievsky Center, Department of Neurology, Columbia University Medical Center, New York
710 W 168th St, New York NY 10032,
W. Allen Hauser, M.D.
Send Correspondence to journal:
Re: Effects of rivastigmine on cognitive function in patients with traumatic brain injury

jnoble{at}neuro.columbia.edu James M. Noble, M.D., et al.

Efforts to identify treatments for traumatic brain injury (TBI) should be commended since the condition often removes the young from social and functional independence. However, in attempting to identify novel treatments, inappropriate conclusions should not be drawn when the study rigor does not warrant them. We are concerned by the potentially misleading abstract conclusions regarding rivastigmine and TBI reported by Silver et al. [1]
As described in the abstract and manuscript results sections, the primary cognitive outcome measures showed no difference compared to placebo at any time. The final statement "rivastigmine shows promising results in the subgroup of patients with traumatic brain injury with moderate to severe memory deficits" seems inconsistent with the results presented in the paper.
The abstract conclusion reflects what the authors described as "additional analyses…planned to identify subgroups more likely to benefit from treatment." Referring to the corresponding clinical trial identifier # NCT00171795 (and the open-label 26 week extension, CTI #NCT00219245)[2], there were no planned secondary analyses based on pre-treatment severity of illness. That these patients with moderate to severe memory deficits had greater improvement hinges upon post-hoc analysis of 81 patients with relatively worse baseline scores on two of twelve reported tests. Several concerns arise from these analyses.
First, the characterization of people meeting "moderate to severe impairment" would not be clinically possible; for clinical usefulness, these criteria are crucial. The results of the other ten tests in this group, including several others testing aspects of memory, are not discussed. Given no difference overall, improvement on selected tests in severely impaired patients begs the question of impact on mildly impaired patients. Was there a subgroup that actually did worse while on treatment? How many different cuts were made before a "severely affected" subgroup became defined? Wide variances, small sample sizes, and use of a single direction standard-error bar chart for data presentation, make p-value interpretations of the reported tests dubious. A table listing results of each test performance relative to a defined pre-study cognitive function (mild vs. moderate-severe) would be more informative to the reader.
Given that TBI phenotype and recovery periods widely vary, perhaps future study design should include a lead-in period of serial neuropsychological tests to establish pre-treatment cognitive trends, pre- determined post-treatment analyses, and standard neuropsychological test cutoffs and clinical criteria for degree of cognitive impairment.
References
1. Silver JM, Koumaras B, Chen M, et al. Neurology. 2006;67:748-755.
2. http://www.clinicaltrials.gov, using search terms "Rivastigmine" or "Traumatic Brain Injury"
Disclosure: The authors report no conflicts of interest.

Reply from the Authors 3 December 2006

Jonathan M. Silver,
New York University School of Medicine
40 East 83rd Street, Suite 1E, New York, NY 10028,
Philip D. Harvey
Send Correspondence to journal:
Re: Reply from the Authors

jonsilver{at}aol.com Jonathan M. Silver, et al.

We thank Drs. Noble and Hauser for their comments on our article and the opportunity to clarify the subgroup analysis. We agree that the statistically significant results are from a post-hoc analysis; therefore, all of the caveats associated with post-hoc analyses apply. Currently, there are no treatments effective for TBI so we think that the use of "promising" to describe treatment with rivastigmine was suitable. While categorization of the severity of the initial traumatic brain injury has been standardized, the characterization of the severity of impairments is difficult.
With respect to specific questions about study design, the patients were assessed twice during the run-in phase and so it is unlikely that our results are a random retest artifact. The primary endpoint of the study was the percentage of responders on memory tests and so the post-hoc analysis reported the results of assessments of memory. Although not statistically significant, all of the neuropsychological (NP) tests administered (12 in total), except for CANTAB-RVIP A’, showed greater improvement in the severely impaired patients for rivastigmine versus placebo.
In some tests, the severely impaired patients receiving rivastigmine showed improvement at week 12 compared with the patients receiving placebo, who showed deterioration on some indices (eg, CANTAB reaction time, CANTAB-RVIP mean latency, HVLT retention, and WAIS digit span). Thus, other NP tests showed a trend for similar results of greater improvement with rivastigmine for the more impaired patients, but these results were not reported in the paper.
Only two subgroups were analyzed: patients with greater than 25% impairment and patients with less than 25% impairment from baseline. For the less impaired patients (N=62), the percent responders with at least a 4-point improvement on HVLT total trials 1–3 was greater for those receiving rivastigmine versus those receiving placebo; however, this post-hoc result was not statistically significant.
For the other NP tests, patients receiving placebo showed more improvement than those receiving rivastigmine. For Trails B, the improvement for patients with less than 25% impairment was statistically better for treatment with placebo versus rivastigmine.
We agree with most of the points made by Drs. Noble and Hauser. However, rivastigmine offers at least a hint that it might be beneficial in more severely impaired patients with TBI. Further clinical trials are needed to further elaborate on these results.
Disclosure: Disclosure: The article to which this correspondence refers supported by Novartis Pharmaceuticals Corporation. J. Silver has received honoraria from Novartis for consulting services. P. Harvey has received honoraria from Novartis. Disclaimer: The views expressed in the article to which this Correspondence refers are those of the author and do not reflect the official policy of the Department of the Army, Department of Defense, or US Government.