In March 2015, Neuralstem (CUR) released top-line data from the company’s Phase 2a study (NCT01730716) testing NSI-566 in 15 patients with ALS. The Phase 2a study was conducted at Emory University in Atlanta, Massachusetts General Hospital in Boston, and the University of Michigan in Ann Arbor. The trial began in September 2013 and treated the last patient in early August 2014 (patient # 15). Below is a quick summary of the protocol.
The Phase 2 study was designed to treat up to 15 patients in five different dosing cohorts. The first 12 patients received injections in the cervical region of the spinal cord only, where the company believes NSI-566 could help preserve breathing function, in escalating doses ranging from five injections of 200,000 cells per injection, to 20 injections of 400,000 cells each (8 million cells total). The final three patients (Cohort-5) in the trial received both cervical and lumbar injections, for a total of 40 injections of 400,000 cells each, or a total of 16 million cells transplanted. All the patients were ambulatory and without bulbar ALS. The primary assessment was change in ALSFRS-r rating scale compared to a baseline run-in period. Lung function, as measure by seated vital capacity (SVC) was also measured. There was no placebo-control in the trial so we are comparing the data versus the pooled resource open-access ALS clinical trials database (PROACT). We also show Brainstorm Cell Therapeutics (BCLI) recently reported top-line data from their Phase 2a study.
We view the results of the Phase 2a study as somewhat disappointing. Although there are clearly signs that several patients responded very well to NSI-566 and the implantation surgery, we are concerned with the trajectory of several other patients that look like they did far worse following the procedure than the historic PROACT database would suggest for placebo patients. Let’s start with a quick look comparing the “pooled” data from Neuralstem’s Phase 2a study to Brainstorm’s Phase 2a study:
Pooled data from Neuralstem’s Phase 2a trial showed that – on average – patients declined at a rate of -1.72 points on the ALSFRS-r scale per month. We are taking this data directly from Neuralstem’s press release and most recent investor presentation at the Barclay’s Global Healthcare Conference. On initial analysis, this looks far worse than the average of all patients (pooled data) from Brainstorm’s Phase 2a study and the PROACT database of 1,339 placebo patients.
However, we note both Neuralstem and Brainstorm reported a “responder” vs. a “non-responder” analysis with their top-line results. Despite what looks to be an advantage for Brainstorm’s NurOwn™ product when compared to Neuralstem’s NSI-566 on pooled data, the responder vs. non-responder numbers looks pretty similar. Since Brainstorm has not presented individual patient data, we cannot say at this time what individual patient curves looks like for the NurOwn™ Phase 2a trial. Neuralstem has made available incomplete graphs of the 15 patients enrolled in the Phase 2a trial – incomplete in a sense that not all patient data is reported at the nine month mark. Regardless, we present those graphs below:
We can see from the graphs above that some patients (#303, #314, #311, #304) looks like they responded very well to NSI-566. Based on “eyeballing” the numbers from the chart, these four patients look to have rather impressive disease stabilization. However, “eyeballing” data one cannot ignore the fact that several patients in the non-responder graph did rather poorly, with declines in the 2 to 4 points per month range after only six months (#308, #302, #301, #310, #305). These sorts of declines in ALSFRS-r are far worse than one would suspect by looking at the PROACT database of just placebo-controlled patients. For example, Brainstorm reported the PROACT database of patients (n=1339) had an average decline at six months (180 days) of -0.82 points per month. Our analysis of the PROACT database for only patients testing on ALSFRS-r at 10.4 months shows an average (mean) decline of -1.19 ± 1.25 points per month. The median (50th percentile) was -0.92 points per month. Neuralstem clearly had a number of patients (eight in total by eyeballing the data off the charts) that declined at a rate faster than -1.19 points per month on ALSFRS-r. The 25th percentile of the data is a decline of -1.55 points per month. A full standard deviation (± 1.25 points per month) to the downside tells us that 84% of placebo patients will decline at a rate of less than 2.44 points per month, and Neuralstem notes in its investor presentation that non-responders declined at a rate of 3.04 points per month (company actually noted -0.1 points per day). That is a major concern.
At this time it is premature to say that Neuralstem’s NSI-566 or the surgery does harm to patients. The data is too premature and we do not know enough about all the individual patient histories to make that claim. For example, we know from the Phase 1b trial with NSI-566 that bulbar ALS patients (onset in the corticobulbar area of the brainstem) decline rapidly regardless of any treatment. Neuralstem did not specifically exclude bulbar ALS patients from this Phase 2a trial. In the table above we can see that bulbar patients decline at a rate of 1.24 points per month. It is possible that this was a primary reason for some of the non-responders.
Neuralstem’s press release announcing the top-line data also notes that one patient did experience a surgical serious adverse event. We suspect (company will not confirm until the full data set is released) that this was either patient #305 or #310 – two of the worse non-responders. The press release is also packed with comments from Neuralstem management and study investigators calling the cells and the surgery safe and well-tolerated, and the data encouraging. Finally, we note that after each patient an independent data safety monitoring board (DSMB) reviewed the procedure and allowed the trial to continue to completion. If NSI-566 was truly doing harm we think the DSMB would have halted the trial prior to the release of top-line data.
What we do know about ALS is that it is a highly variable disease. Notwithstanding Stephen Hawking, who has been alive with ALS for an astonishing 52 years, the three-year survival rate for ALS is around 50% (source:Alsa.org). Top-line data from Neuralstem’s Phase 2a study looked at patients at nine months. Brainstorm’s stopped at six months! The graph below is taken from Leuven University Hospital and shows the outcome of 30 randomly selected ALS patients in Belgium. We draw several conclusions from this graph: 1) ALS patient outcome is highly variable, 2) bulbar patients decline rapidly once their ALSFRS score drops below 30, 3) nine months is not an acceptable endpoint to determine efficacy for any ALS treatment, 4) some patients experience months (or even years) or disease stabilization while others decline rapidly for no apparent reason.
Data from the PROACT and the Leuven University Hospital tells us that running clinical trials in ALS is a daunting task. Perhaps that’s why 99% fail. Both Neuralstem and Brainstorm will be running placebo-controlled Phase 2 trials in 2015. Neuralstem has already stated that this Phase 2 trial will seek to enroll 50 patients. We suspect that the three centers that participated in the Phase 2a study, Emory University, Michigan, and MGH will likely participate in the upcoming Phase 2 study. Analysis of the Phase 2a data will continue, with a full data presentation likely coming in April or May 2015.
The goal for Neuralstem going forward is to refine the responder patient population and come up with entry requirements and exclusion criteria that significantly tilt the odds in the company’s favor for the Phase 2 trial. For example, Neuralstem believes that predictors of responder vs. non-responder outcomes are evident in the Phase 1 and Phase 2a data. High initial muscle strength as determined by grip-strength test looks like a predictor of success. Vital capacity may be another predictor or who will respond to NSI-566. Eliminating bulbar ALS patients from the upcoming Phase 2 is another likely protocol change from the Phase 2a.
In conclusion, we are not concerned with the responder vs. non-responder path forward. Lots of drugs have responders and non-responders. Non-responder is a fancy way of saying “the drug didn’t work” for these patients, but that is fine. Again, notwithstanding Stephen Hawking, ALS is a fatal disease. Ninety percent of the patients in the Leuven University Hospital analysis were dead at four years. A cell therapy, like NSI-566, that extends lives in 50% of the patients is a blockbuster. There are cancer drugs, Nexavar® for example, that sell in the billions by extending survive only a few months. Investors should view treatment options for ALS similar to terminally-ill cancer patients. NSI-566 and NurOwn™ are not designed to be cures for ALS, they are treatment options.
That being said, we are concerned with the potential that NSI-566 or the surgery potentially accelerated the decline in some patients in the Phase 2a study, but that is pure speculation at this point. We are withholding any grand predictions about the success of failure of NSI-566 until after the full presentation of the data in April or May 2015.
SUBSCRIBE TO ZACKS SMALL CAP RESEARCH to receive our articles and reports emailed directly to you each morning. Please visit our website for additional information on Zacks SCR and to view our disclaimer.