In The News
In The Pipeline
Researchers Look to Neuregulins for Stroke Clues
Findings indicate Glial Growth Factor 2 may improve treatment days after stroke.
By Zac Haughn, Senior Associate Editor
Neuregulins could play a role in repairing damage caused by stroke, researchers at Acorda Therapeutics have found. Their recent findings indicate that treatment with neuregulins could potentially repair damage and restore function in patients up to a week after an ischemic event. Practical Neurology talked with Anthony O. Caggiano, MD, PhD, Vice President, Research and Development at Acorda Therapeutics to find out more.
What exactly has your company studied/found in regards
to neuregulins in stroke patients?
This program is applying Glial Growth Factor 2 (GGF2), which belongs to a large family of proteins known as neuregulins. In a range of pre-clinical research programs, neuregulins have been shown to bind to ErbB receptors with the potential to stimulate repair in both the nervous and cardiac systems.
In published studies in preclinical models, GGF2 has been shown to protect the brain from the consequences of stroke, restore function and stimulate remyelination in models of MS. The benefits in stroke have been demonstrated in both transient and permanent stroke models. In addition, neuregulins have shown the ability to restore cardiac function and improve survival in preclinical models of heart failure caused by myocardial infarction, rapid pacing, and viral and chemically induced cardiomyopathies.
Acorda has initiated a Phase I single-dose clinical trial of GGF2 in patients with heart failure. The company is also continuing preclinical studies of potential neurology indications for GGF2 and other neuregulin growth factors.
Why are you optimistic about neuregulins in
the treatment of stroke?
The only approved therapy to treat stroke is tPA which enzymatically disrupts the clot causing the stroke. Treatment with tPA is limited to a few hours after the event. Our research on the use of GGF2 is especially compelling because we are targeting promotion of recovery even when treatment does not start for up to seven days after a stroke. The potential benefits of treatment with GGF2 several days after a stroke have been demonstrated in preclinical stroke models. In addition, GGF2 is in clinical trials for another indication and so this program has the benefit of extensive previous non-clinical research related to dosing, efficacy and safety and a certain amount of human safety and tolerability data.
What are some of the challenges you face in development?
A major challenge in clinical research related to acute stroke is the need to deliver a neuroprotectant to patients immediately after an ischemic event, presenting considerable challenges to enrollment and limited utility in the real health care setting. Research related to non-acute intervention including rehabilitation has been limited and so in many ways represents “new territory” in clinical research. However, this focus does potentially position us to be better able to design and successfully execute a clinical program in stroke that will target patients up to several days after an event.
What is next for the study of GGF2?
The progress and results in pre-clinical research thus far have been encouraging. Acorda is working to advance the research program for GGF2 to develop a better sense of dosing range and to replicate previous results in larger animal models. The Phase I trial in heart failure patients was initiated in December 2010.
This is a safety and tolerability study in heart failure; findings from this study will provide us with important insights related to safe dosing levels for treatment in stroke.
In The News
MS Drug Meets Primary Endpoint
Actelion announced that its experimental drug, a selective S1P1 receptor agonist named ponesimod, has met its primary endpoint in a Phase IIb dose-finding study in patients with relapsingremitting multiple sclerosis and will enter late stage testing.
The study assessed efficacy, safety, and tolerability of three ponesimod doses (10mg, 20mg or 40mg) versus placebo, administered orally once daily for 24 weeks, ultimately reducing the number of new active inflammatory lesions in the brain. The company noted that with 464 patients enrolled, this was the largest dose-finding study conducted in this autoimmune disorder.
Ponesimod significantly reduced the cumulative number of new active lesions on monthly MRI brain scans performed from weeks 12 to 24, with the most effective dose at p<0.0001.
Laquinimod Progress Stumbles
Results from the Phase III BRAVO study showed that laquinimod was not more effective than a placebo and thus failed to meet its primary endpoint.
Teva said the randomization process for BRAVO was adequately performed, but they found MRI scans showed the patients in the laquinimod and Avonex groups had more brain lesions, an indication they had more severe MS. When this imbalance was adjusted, the company says, laquinimod demonstrated a significant reduction in the annualized relapse rate (21.3 percent, p=0.026), in the risk of disability progression as measured by Expanded Disability Status Scale (EDSS) (33.5 percent, p=0.044) and in brain volume loss (27.5 percent, p<0.0001).
Teva and its partner Active Biotech will still seek approval of laquinimod in the US and Europe.
Positive Data for Sumavel DosePro
Data from a new study showed that “patients currently treated with triptans and less than very satisfied with their acute migraine therapy experienced a statistically significant and clinically relevant increase in satisfaction with therapy and enhanced confidence in treatment after use of Sumavel DosePro for up to four migraine attacks.”
Published online in the August edition of Headache, researchers followed 212 patients in their open-label, multicenter study. Sumavel DosePro was self-administered for four migraine attacks (over a 60-day period) involving moderate or severe baseline pain by adult migraineurs who currently were using triptans (any form, any dosage).
They found that Patient Perception of Migraine Questionnaire, revised (PPMQ-R) satisfaction—the primary endpoint—increased significantly from baseline to the end of treatment (mean ± SD 65.7 ± 19.8 vs 73.7 ± 29.1, P = .0007), an improvement that met the criterion for clinical significance. From baseline to the end of treatment, PPMQ-R scores also improved significantly for Efficacy (62.2 ± 17.6 vs 76.2 ± 23.7, P < .0001), Functionality (59.0 ± 22.3 vs 73.8 ± 25.3, P < .0001), and Tolerability (83.9 ± 13.1 vs 86.4 ± 15.0, P = .02), but declined for Ease of Use (82.6 ± 15.3 vs 67.8 ± 27.6, P < .0001).
Novel Analytical Approach May Influence
Trials and Patient Care
A novel analytical approach appears effective for the analysis of clinical trials data and could eventually lead to new methods of patient monitoring and therapeutic selection, research suggests. Hierarchical Bayesian cognitive processing models (HBCP) provide precise measurements of cognitive change and can do so in very small samples. A dramatic example of the power of HBCP models was reported by the Hoag Neurosciences Institute (HNI) at the Annual Meeting of the American Academy of Neurology (AAN) in April.
A previous, 18-month, FDA Phase III drug trial of 1,649 Alzheimer’s Disease (AD) patients found no difference in the rate of cognitive decline between patients treated with either Flurizan™ (Myriad) or placebo. In contrast, the Flurizan-treated AD patients’ overall dementia severity—measured using the sum of the item scores of the Clinical Dementia Rating Scale—declined more than those treated with placebo. Researchers at HNI analyzed a sub-sample of 14 of these AD patients who came from HNI. The previous analytical methods used in the FDA trial continued to show no Flurizan vs. placebo treatment difference in this small sample. However, when HNI researchers applied HBCP models to the cognitive data from the trial (the ADAS-Cog), they found that Flurizan, compared to placebo, caused a greater decline in memory, and that it worsened further after the trial ended. The researchers predicted that drugs like Flurizan, which belong to the class of gamma secretase modulators, may also have a harmful effect on memory. Their prediction was confirmed a few months later when the Phase III trial of another gamma secretase inhibitor, Segamacestat (Lilly), was halted because of greater cognitive decline compared to placebo.
HBCP models can be used for new clinical investigations or to re-assess existing trial data, according to William R. Shankle, MS, MD, FACP, program director, Memory and Cognitive Disorders at HNI. Dr. Shankle points out that one potential application of HBCP models is to evaluate the results of small samples of patients treated with a new candidate drug to determine if it merits the high costs required to conduct FDA Phase II and III trials. In the case of Flurizan and Segamacestat, hundreds of millions of dollars could have been saved if this approach had been taken.
The success of the HBCP model used by the researchers was due to its ability to model the underlying cognitive processing required for memory performance. Dr. Shankle notes that models can be developed for various other cognitive processes, such as object recognition, language, and executive function, in order to assess or predict the influence of potential therapies in these specific domains. Other applications of HBCP models, Shankle says, include early detection and differential diagnosis of cognitive impairment.
Shankle and his colleagues at HNI and the University of California, Irvine have also shown how HBCP models can be used to relate how a cognitive process affects functional abilities. They used a delayed recognition memory task in which subjects are read a list of words and asked to choose which of them they had previously been shown during several learning trials. The HBCP model in this case used two underlying processes involved in delayed recognition memory performance to predict the performance of over 500 patients with AD or a related condition who were assessed more than 1500 times. They found that these underlying processes accurately predicted the severity of functional impairment for these patients. To date, there are few studies showing how a cognitive ability affects functional capacity, said Dr. Shankle.
As studies like Dr. Shankle’s confirm the power of HBCP models, they may be more widely used in clinical investigations to determine the benefits and risks of novel therapies. It remains to be seen if and how the FDA will utilize the results of HBCP models. Dr. Shankle predicts that eventually, HBCP models will be adopted by the FDA, at least partly to protect large numbers of patients from being treated with potentially harmful medications, as was seen with Flurizan and Segamacestat. “The results are impressive. In the Flurizan analysis, HBCP modeling results were 100 times more sensitive than current metrics required by the FDA,” he states. Until then, companies could use models internally to support decisions to pursue, design, or abandon large clinical trials.
Whether applied to a new sample of patients or used to assess data from previous research, “the models can greatly improve the power of the analysis,” Dr. Shankle says. Another benefit of HBCP models is that they “handle any kind of missing data, which are common in clinical trials for many reasons,” he adds. This maximizes the use of the available data and does not require the ad hoc approaches that are currently used in clinical trials, such as the “last observation carried forward” approach.
“The underpinnings of this work dates back to the early 20th century in theoretical quantum physics,” Dr. Shankle explains. Now the science is being applied to “put together a model of how a particular disease works.” Models and the data derived from their application can help to improve the scientific community’s understanding of diseases overall, Dr. Shankle observes. “Currently, the medical community is aware only of risk factors associated with AD, but there are no risk factors proven to cause AD,” he says. Models can be built to identify specific causes of specific diseases. In fact, data may eventually be “passed into software that can monitor an individual’s risk, track their cognition, and suggest interventions that can effectively reduce the chance of developing cognitive impairment and AD.” This is an intriguing possibility in light of recent calls by the AAN to improve the detection of Alzheimer’s disease by up to 20 years, well before cognitive decline begins and when available and future therapies may offer the most benefit. Dr. Shankle believes modeling could help predict which individuals are at highest risk for AD and may help identify those in the subjective cognitive impairment and mild cognitive impairment stages of dementing diseases.
Use of modeling is not limited to the study of dementia or cognitive deficits. “There is no limit to the number of models one can apply. Even if a model doesn’t fit well, you’ve learned something,” Dr. Shankles explains. “There are methods that allow you to test which of a set of models is the best one and whether a better model may exist.”