Article In Brief
Investigators are pushing for a phase 3 trial to test the efficacy of infusing glial cell line-derived neurotrophic factor in patients with Parkinson's disease, despite negative findings from a phase 2 trial. They believe the results have promise, though independent experts do not agree.
Researchers have spent almost two decades studying the infusion of glial cell line-derived neurotrophic growth factor (GDNF) into the brains of patients with Parkinson's disease (PD) to see whether the growth factor would protect dopamine neurons and their support cells from dying. Despite some promising results in animal models of the disease, study after clinical study has failed to meet the primary outcome measures in humans, no matter how the drug was delivered.
Now, two new studies by researchers in the United Kingdom were just published with similar negative results from a phase 2 double-blind study, but the research team is pushing to move into a phase 3 study with hopes that the benefits they observed in patients are more than just a robust placebo response.
“Does it work?” asked Steven Gill, MD, a neurosurgeon the Translational Health Sciences at the Bristol Medical School in the United Kingdom and senior author of the studies published in Brain on March 1 and the Journal of Parkinson's Disease on February 26.
Dr. Gill said that the standard rating scales do not adequately reflect disease severity and are a poor method of evaluating efficacy. He said that many of the patients on active drug were carrying out activities that they hadn't been able to do in years. Moreover, the brain scans showed that the drug reached the target. “PET scans showed 100 percent increase in 18F DOPA in the motor putamen demonstrating restoration of dopamine neurons,” he said.
“There is no other explanation for the patients' improvements,” said Dr. Gill. “It took us a long time to figure this out. We found the right materials, flow rate and delivery method, and we had no infections during the study.”
But independent experts, some of whom have conducted their own research involving GDNF, said media hype around the latest findings have led to some misunderstanding of the findings and what they mean.
The History Leading to Current Trials
Dr. Gill and his colleagues published findings from their first GDNF study—a phase 1 safety study comprising five patients in an open-label study—in 2003. The neurotrophic growth factor was infused continuously through micro-catheters into each putamen, which is deprived of dopamine in Parkinson's disease. They reported that this led to a 39 percent improvement in the off-medication motor sub-score of the Unified Parkinson's Disease Rating Scale (UPDRS) and a 61 percent improvement in the activities of daily living sub-score.
The research team followed the patients for another two years and reported that the improvements continued and that there were no serious clinical side effects or significant changes in cognition. Functionally, the patients were reporting improvement in their activities of daily living. One patient who died of an unrelated heart attack was shown to have restoration of dopaminergic neurons at the site of infusion in the putamen as well as in the substantia nigra consistent with the clinical benefit.
Excited by the open-label findings, Amgen, makers of GDNF, launched a double-blind placebo-controlled trial in 2003 using a different catheter system. The multicenter study enrolled 34 patients. A year later, the company announced they were stopping the study. The patients on the active drug did not reach the predetermined endpoint to show significant benefit and the scientists had detected neutralizing antibody in two of the study participants, and this was worrisome. The company had other reasons to stop the trial: Preclinical findings from non-human primates found a loss in neurons in the cerebellum following high dose exposure and subsequent GDNF withdrawal. Patients and the Parkinson's disease community were upset, and lawsuits were filed to try to force the company to provide the drug through a compassionate use application. The courts ruled in favor of Amgen.
Failure of the Amgen study was attributed to inadequate delivery because the larger catheter used resulted in much of the GDNF refluxing back along the catheter/tissue interface and not adequately penetrating the brain tissue, said Dr. Gill. In the non-human primate study, the catheter leak back resulted in a significant dose of GDNF entering the cerebrospinal fluid and collecting around the skull base causing toxic damage to the cerebellum, he added.
Dr. Gill was still convinced that the growth factor would work given the right delivery method and the optimal dose. The problem with the method used was that the continuous slow infusion would result in a build-up of GDNF at the catheter tip causing local toxicity and poor distribution covering no more than 10 percent of the putamen, he said. The solution that he came up with was to drive the drug throughout the putamen using controlled pressure via micro-catheters and then stop the infusion. This would need to be repeated every four weeks to maintain a biological effect.
Findings in Brain
To achieve this, Dr. Gill designed a drug delivery system comprising four micro-catheters that were implanted with a surgical robot into the putamen and connected to a small bone-anchored port that emerges through the skin behind the ear. The drug was delivered by connecting external infusion pumps to individual catheters via an administration set that is connected to the port when required.
In 2013, with charitable funding from Parkinson's UK, the Cure Parkinson's Trust, and Funding Neuro, the team in Bristol commenced a randomized double-blind trial and recruited 41 patients between the ages of 35 and 75. The study participants had motor symptoms for at least five years and their symptoms during the off- and on-state UPDRS ratings were in the moderate disease severity range. Half the patients were enrolled in active drug or placebo for nine months and then all patients received active drug for another nine months. The once-a-month delivery method was enough for the half-life to clear from the brain. Blood samples were taken throughout the study to look for GDNF plasma concentrations and any antibodies that might develop in response to the treatment. The findings from this study were published in Brain.
Forty-one patients were randomized to receive bilateral infusions of GDNF (120 mg per putamen) or placebo for the first 40 weeks. Then everyone received the drug for the next 40 weeks. Every eight weeks before the next infusion, patients were asked to complete three-day diary recordings and they had assessments of motor function in their off state, which was followed by a levodopa challenge. The researchers had MRIs of the patients done to ensure that the doses were administered bilaterally. The researchers took baseline recordings on 18F-DOPA PET and again at the end of the study.
The primary outcome measure was the percentage change from baseline to week 40 in the off-state motor UPDRS score, with the study powered to demonstrate a 20 percent difference between the groups. The change in the mean off-state score in the UPDRS motor score was 17.3 percent in the GDNF group and 11.8 in the placebo group, which was not statistically significant.
The PET findings at baseline were what they expected to find in the pathology of PD. At 40 weeks, those in the placebo arm showed no changes. Those who were receiving GDNF showed significant changes in both sides of the posterior putamen, central putamen, and anterior putamen, said Dr. Gill. “Unfortunately, these changes did not correlate with the motor UPDRS scores, which were not significantly different.”
The secondary endpoints—including an absolute change from baseline in off-state UPDRS motor scores, absolute and percentage change from baseline in UPDRS score in the on state, UPDRS activities of daily living and changes in the disease diary ratings over time, timed motor tests, and changes on non-motor symptoms—were also no different between patients receiving the drug or those on placebo.
Dr. Gill said that they went back to conduct a post-hoc analysis and found that nine (43 percent) of the patients in the active medication arm had at least a 10-point improvement in their off motor score, which he said was clinically meaningful. There were no patients in the placebo arm that achieved this level of benefit, he said, adding that the infusions were well-tolerated and there were no drug-related serious adverse side effects that differed between the placebo and active drug group.
No one in the GDNF group discontinued the medication during the study, although one patient had a mildly symptomatic putamenal stroke with the initial test dose and another a small asymptomatic hemorrhage in both putamina during the initial test infusion. Both were withdrawn from the study before randomization.
The scientists also reported three device-related side effects, including one infection around the port site that required hospitalization and oral antibiotics, and two hypertrophic skin reactions that required surgical thinning. During the early part of the study, several patients reported some loosening of the port fixation, and this led Dr. Gill to modify the design with a stabilizing element to retro fit it in a few individuals. The team realized that more education was needed on how to keep the port site clean.
“Overall, the success of the implantable drug delivery system opens the possibility for its use in delivering other medications safely into the brain,” said Dr. Gill.
The question, of course, is why the delivery of this particular growth factor did not have a significant clinical effect in these patients despite effective delivery and evidence of increased dopamine uptake on the PET scans. Do they need to design this intervention for patients in an earlier stage of disease? Was it an adequate dose? Was nine months long enough? They could show a bimodal distribution of absolute responders and non-responders on the UPDRS measures, but there were no clues about what made some patients improve and not others. The scientists concluded that “further GDNF investigations will need to address potential reasons why our clinical primary endpoint was not reached despite apparently optimizing putamenal therapeutic delivery.”
“It failed in all of its endpoints except fluorodopa PET, and fluorodopa PET has no clinical association with functional outcome. There are numerous patients who have had increased fluorodopa PET in other trials without clinical benefit.”
—DR. JEFFREY H. KORDOWER
“Even if the very modest positive signals emphasized by the investigators generated sufficient enthusiasm to fund another expensive trial, I don't think that a more positive result would encourage widespread use of this treatment given the cost and technically demanding, invasive nature of its application.”
—DR. ANTHONY E. LANG
Adding an Open-Label Extension Study
The open-label extension study, reported in the Journal of Parkinson's Disease, was designed to give everyone in the study the active drug from week 40 to week 80. The researchers did not know the results of the double-blind study when they began this second phase. Patients who accepted enrollment into the open-label study did not know whether they had received the active drug in the first study. Trained raters were also blinded to the patient's condition, prior treatment assignment, and outcome measures.
Every patient in the extension trial received ten doses of GDNF at four-week intervals. (Patients in both studies were maintained on their initial Parkinson's medications.) The clinical measurements were the same in this study at baseline and every eight weeks.
All 41 patients opted to continue in the extension phase of the study. They all completed the trial, as well. Almost 98 percent of the group kept to the infusion schedule. Twelve percent of the infusions were interpreted or terminated early due to mechanical pressure increases in the infusion lines.
Again, there were some differences in those who had 80 weeks of active medicine versus those with 40 weeks, including the observation that patients in the active/active GDNF arm had fewer increases in the levodopa medications they were taking compared with those on placebo/GDNF. It was impossible to know whether this was due to a placebo response because everyone knew they were getting the active drug.
The investigations also said that investigator bias in the extension phase could also have led to the changes noted between 40 and 80 weeks.
“There was a 30 percent improvement in UPDRS motor scores, which is clinically meaningful,” said Dr. Gill. “There was progressive improvement month by month. The PET scans showed improvement and patients were getting better. If you look at the data, it's difficult to attribute this to a placebo effect alone. Hard as I have looked, I've not found any study where patients continue to improve month-by-month over 18 months on a placebo.”
The investigators said that the unexpectedly large 11.8 percent improvement in the motor scores of the placebo arm could in part be explained by the mechanical disturbance of the putamen from the infusions. Micro trauma is known to stimulate the release of neurotrophic factors such as GDNF and brain-derived neurotrophic factor. These could lead to increased synaptogenesis and axonal sprouting in dopaminergic neurons. If so, said Dr. Gill “one should consider including a best medical therapy arm as a control group in future studies, which has been done in studies of deep brain stimulation.”
He said that his team is pushing for a phase 3 study that will include a dose escalation. The dose delivered in the study was constrained by the toxicity data that they had at the commencement of the trial, however recent toxicology studies have indicated that they can now safely increase the monthly dose by four-fold, Dr. Gill explained.
“People misunderstand a failed study. Science is about the balance of probabilities,” he added. “When you fail to meet a pre-determined endpoint, it may mean that you have asked the wrong question, failed to deliver sufficient drug to the correct location at the optimal frequency, failed to observe the patients for long enough, or even produced a biological effect in your control group. This is an extremely difficult problem to solve and translate into a therapy. Nevertheless GDNF is the only molecule with the potential to restore dopaminergic, serotoninergic, noradrenergic, and cholinergic neurons, all of which are affected in Parkinson's. Having seen remarkable improvements in all aspects of the disease in individual patients, I will not give up until I get it right.
“Many of the patients say that they know they are better,” he added. “Now that the study is finished they are desperately trying to get access to the drug.” He said that patients and private foundations are helping to raise money for the phase 3 study. Dr. Gill said that they are also talking about implementing an alternative evaluation of the patients in the study. Dr. Gill invented the surgical system and is on the advisory board for Renishaw, the company that makes the device. Other scientists are using the technology to test the delivery of other growth factors into the brain.
Independent experts were less sanguine about the findings, however. “It is another GDNF trial that has failed,” said Jeffrey H. Kordower, PhD, the Alla V. and Solomon Jesmer professor of neurological sciences and director of the section of neurobiology at Rush Medical College. “It failed in all of its endpoints except fluorodopa PET, and fluorodopa PET has no clinical association with functional outcome. There are numerous patients who have had increased fluorodopa PET in other trials without clinical benefit. For scientists to tell their patients that ‘I know it works’ is totally misleading and unsupported by the science. Also, these types of unsubstantiated statements drive expectations and the placebo effect, which was substantial in this trial.”
C. Warren Olanow, MD, chairman emeritus of the department of neurology at Mount Sinai School of Medicine, agreed. “To say it works in the lab is a great leap of faith to say it will work in humans. Every single GDNF trial has failed. If double-blind studies are negative, it speaks louder than the proclamation of a neurosurgeon saying that he knows it works.”
On February 28, BBC aired the first in a two-part series called The Parkinson's Drug Trial: A Miracle Cure? BBC spent six years documenting the latest UK study, speaking to patients and the scientists, including Dr. Gill and neurologist Alan Whone, MD, the lead author of the studies.
“Unfortunately, the failure of this study has been lost in the hype by the investigators saying that they know the drug works,” added Anthony E. Lang, MD, FAAN, the Jack Clark chair for Parkinson's disease research and director of the Edmond J. Safra program in Parkinson's disease at the University of Toronto. Dr. Lang has carried out GDNF trials, which also had negative results.
“Scientists must have equipoise when they plan a clinical trial and talk about their findings. I have been dealing with this a long time,” he said of GDNF's negative findings. “The methodology that Drs. Gill, Whone, and their colleagues used to test the drug was excellent. The design of the device was very novel and seemed to appropriately address the criticisms they had about the previous trials. They seemed to do all the right things, and if the treatment was truly efficacious it should have worked, but it didn't. Now they are saying it was the dose. Maybe it didn't work because it doesn't work.”
“I personally believe that this treatment should now be allowed to die a natural death,” Dr. Lang said. “Even if the very modest positive signals emphasized by the investigators generated sufficient enthusiasm to fund another expensive trial, I don't think that a more positive result would encourage widespread use of this treatment given the cost and technically demanding, invasive nature of its application.”
“There have been many attempts to deliver GDNF in humans to see if the laboratory findings in animals can be replicated in humans,” said John G. Nutt, MD, FAAN, professor of neurology and physiology/pharmacology at Oregon Health and Science University. “The clinical results of this blinded, 40-week comparison of placebo and GDNF study followed by the open (unblinded) 40 weeks found no difference between placebo and GDNF. The discussion is about whether this is an adequate test of GDNF as treatment for PD with the conclusion that because of only one dose tested it is not.”
Dr. Gill is the medical director of Renishaw PLC and is the inventor of the drug delivery system from which he may have a future royalty share. He serves on the scientific advisory board of MedGenesis Therapeutix Inc, for which he is reimbursed with share options. Drs. Nutt, Olanow, and Kordower had no competing interests.
Dr. Lang has received fees for consulting with Lundbeck, Sun Pharma, Kallyope, Retrophin, Paladin, Seelos, Theravance, Roche, and Corticobasal Degeneration Solutions. He serves on advisory boards for Jazz Pharma, PhotoPharmics, Sunovion, and has received honoraria from Sun Pharma, AbbVie, and Sonovion.