Posted on July 17th, 2014
by Gerald Commissiong
We have just completed what was an extremely successful Alzheimer’s Association International Conference in Copenhagen Denmark for Amarantus. The Company made 3 poster presentations, the first at international conferences for the Company, on our proprietary LymPro Test® blood diagnostic for Alzheimer’s disease. We presented data that corroborated earlier findings on the relevance of cell cycle deficiencies in peripheral blood lymphocytes, and publicly outlined the conditions of Version 2 of the assay that we believe will be a major leap forward for LymPro. At the conference, the top-line data presented showed an accuracy of 78% in this Version 2, and many current shareholders have asked exactly how we can feel so good about that data as compared to previously published Work. To answer this question, there are a number of things that much first be put into context to inform the underlying understanding of how blood tests are developed, the statistical modeling employed to arrive at accuracy numbers, and most critically, the relevance of correlation of each individual marker within a panel to an ultimate diagnosis. Many of these points were not addressed in our posters because our audience consisted of scientific teams from Big Pharma companies and Key Opinion Leaders (KOLs) who fundamentally understand the process. However, to those not involved in the field, I have an appreciation as to why things as described may not seem to add up. So let me give some background context first and make some comparisons so as to demonstrate how LymPro data, that we will be submitting for publication upon completion of the data sets to be presented at #C4CT, measures up against other blood tests that have recently seen significant press in recent weeks and months. Let’s go through a step by step understanding of what is happening
GW Medical licensed the intellectual property underpinning these concepts from the University of Leipzig in 2004, and set about confirming the data after setting up the assay at Becton Dickinson Biosciences (“BD”) in San Diego, which is the same group Amarantus is currently working with. BD is the contract lab that produced the raw data that underpins the 2012 Stieler et al publication, and in fact most of the scientists who worked on the LymPro project in the early days are back working on LymPro today. BD was able to replicate the findings, and it was hypothesized that upon changing the conditions of the assay (the concentration of mitogen used, and the period of time during which the cells were incubated prior to measuring CD69 levels) a greater separation (signal to noise ratio) could be established.
During the time the work was conducted at BD, GW Medical merged with another small company to create Provista Life Sciences (PLS), and, as part of the merger, the GW team was gaining access to its own CLIA lab in which they believed their a new group of scientists being assembled would be able to re-establish the assay in-house and commercialize LymPro. After the results were established in 2005, the newly created PLS decided to move the assay in-house in order to commercialize it in their own CLIA, and as part of that process set about conducting a time and dose study to establish optimal parameters for a Version 2 of LymPro. Initially in late 2007, they began to see good results that generally reproduced the findings, furthering the work, and thought they had discovered optimal conditions using samples obtained from Banner Sun Health Research Institute in Phoenix to generate even better data for LymPro. However, what the PLS team did not do at that time, was transfer in-house all of the analytical methods for quality control and analysis that were the hallmark of reliability that had underpinned the reputation established at BD within the industry. Shortly after the early findings, they began to see significant analytical performance problems, including batch to batch variability in performance of the assay, which ultimately led to completely inconclusive results in the summer of 2008 in the midst of their CLIA-enabling study. At the time, they were in discussions with several large pharmaceutical companies who were interested in the scientific concepts behind LymPro and wanted to see the data corroborated in larger studies (no PET imaging at the time to confirm diagnosis, therefore initial Version 2 studies were too small). When financial markets collapsed in 2008, PLS was unable to raise the funds necessary to implement the strong analytical controls required to properly validate LymPro, and therefore they made a strategic decision to focus on other assets at the expense of LymPro.No further work was conducted.
In 2011, shortly after Amarantus went public around MANF and concepts around neurodegeneration, we were becoming increasingly interested in ER , and found one primary cause could be CCD. We hypothesized that if MANF could somehow address this problem at the cellular level in the brain (intracranial injection in Parkinson’s), and MANF indications could expand to Alzheimer’s. In reviewing the landscape, we believed that in order for this approach to work, we would need early diagnosis tests for ER stress diseases, and therefore acquired NuroPro from Power 3 Medical Products and LymPro from PLS in 2012 to cement these scientific principles in corporate action. We focused on the Alzheimer’s asset because of the significant market need, and the fact that drugs were already in late-stage as disease modifying treatments.
Once we acquired LymPro, we decided to correct the deficiencies of the 2008 work by re-engaging BD to run LymPro, and ensure that we would establish proper analytical methods necessary for tech transfer to a commercial-grade CLIA lab because we had seen the challenges in establishing our own. In 2013, we were extremely happy to learn that new data had emerged by Seward, et al demonstrating a clear link between amyloid aggregation and tau phosphorylation to CCD in Alzheimer’s, a finding that further validated CCD in Alzheimer’s and gave much greater scientific and commercial relevance to Dr. Arendt’s original findings.
Now we come to the most important discussion of all, the data presented at AAIC 2014. We presented the analytical data generated at BD’s labs over the last year demonstrating very tight controls in the running of the LymPro assay. With this, we had re-established the underlying measuring capabilities that led to the 2012 Stieler, et al publication. This was a very key step in our path to developing a commercially relevant diagnostic assay. This is poster #1, the analytical performance poster, presented on July 14th, 2014 and that was extremely welcome news to those in Big Pharma who had previously been aware of the LymPro program in 2008 timeframe.
In poster #2, we showed the initial studies conducted in the labs of PLS for Version 2 LymPro, using a total number of 44 patients. The data showed that using the best cell type response (CD4 cells), we achieved an accuracy of 81%. Beyond that single data point, accuracy readings for the next best responses had accuracy ratings individually between 70% and 80%. These are univariate results, not using any statistical modeling methods at all. The reason why we are so happy with this, beyond the specific 81% number in the topline section, is because what this is showing, without any shadow of a doubt, is that CCD is clearly a critical component of the disease, and we are able to identify it with high statistical significance across all cell types tested of interest in the blood.
Because we are measuring cellular response of a fundamental disease process, the most important thing to know is that the underlying patient diagnosis is correct; that we are measuring accuracy against the appropriate standard. To that end, we conducted a patient record study with Banner Sun Health’s Dr. Marwan Sabbagh to establish whether the original diagnosis held up to scrutiny 7 years later, during which time we would typically see changes that would either confirm or disprove the diagnosis. We were pleased to see that Dr. Sabbagh’s group did an excellent job in maintaining the diagnosis over time with only 4 real changes of diagnosis (in addition, 1 patient moved from normal to MCI and was therefore excluded, 1 patient moved from MCI, remains MCI and was LymPro negative, and 2 patients were MCI and remain MCI).
Based on these revised diagnoses of the 41 patients, we are seeing high accuracy in the response to mitogenic stimulation of each individual cell type measured in the assay. What this tells us is that despite the fact that the underlying raw data was produced in sub-optimal analytical conditions that ultimately led to the destruction of the program while at PLS, the fundamental biology is shining through without the need for fancy math that is the hallmark of multivariate analyses in diagnostics. Below is the table with a summary of the univariate data. This compares very favorably with the significance of recently published data sets from other groups measuring protein markers, on an individual marker significance basis. Please see the data below and reference significant against significance of individual markers public in others’ work.
With the raw data in hand, we believe we have more than sufficient signal to develop proper predictive statistical models that will ultimately be highly relevant without the need to ”over fit” algorithms with fancy math to relatively unrelated events, but rather will fit good biology together. We chose not to conduct a huge amount to statistical modeling because we have clean data being generated at BD that will be available in the very near future that will give us strong raw analysis that will much improve the multivariate modeling process using individual markers that have already demonstrated to have statistically significant differences in AD vs. other dimentias under sub-optimal analytical conditions.. From our standpoint, starting with 5 different cell types from 4 lymphocyte families (CD8 and CD4 and derivatives of CD3), and generally measuring the data using an actual raw statistic of median fluorescence intensity as opposed to a derivative of such a measure, we are starting with the truest and best raw data that will form the basis of our model building. We know that we will be able to get much better and cleaner raw data from BD that will give us greater confidence ahead of spending the time to establish the important statistical models needed.
Of note, when we compared only the Alzheimer’s and cognitively intact groups, we had an accuracy of 90% in this small data set. In the upcoming Bridging Study, we will be comparing only Alzheimer’s and cognitively intact controls in order to give us a true clean measure of Version 1 vs. Version 2 using analytically strong data from BD. We believe this step will give us all the tools necessary to support the transition of the assay from BD into the target commercial CLIA lab partner and then restart the work PLS had conducted, and add other dementias into the cohort group (distinguishing Alzheimer’s from other dementias and cognitively intact controls using biomarkers such as Amyvid). That study will be conducted in the partner lab and will support commercial launch.
Some may ask, “How will you get another clinical study completed by the end of the year in the CLIA lab, as is required to support submission?” The answer is that we are proving right now that we can execute under such timelines with the current study. This is where the relationships and expertise and experiences of Dr. Kirby’s over 400 clinical trials are a huge advantage for Amarantus. We are establishing a well-oiled clinical organization capable of translating protocols to practice, and ultimately to data at a very rapid pace.
We remain on track to complete the partnering process with our CLIA lab by the end of the summer, transfer the analytical process from BD to the target partner (a process that these organizations have completed extremely successfully several times), and achieve CLIA submission within our targeted timeframes. We believe the raw data for LymPro overwhelming demonstrates its validity even in suboptimal conditions, and now we are poised to move it to the main stage. We believe that by the time AAIC comes around in 2015, the data that we will be presenting will be the subject of significant oral presentations. We are excited with the progress made at this year and the credibility we have injected back into LymPro after rescuing it from PLS, and believe that we are very well positioned for the second half of 2014. LymPro has a bright future and we are happy to have brought it back to life for the benefit of patients and shareholders.
Forward Looking Statements
Certain statements, other than purely historical information, including estimates, projections, statements relating to our business plans, objectives, and expected operating results, and the assumptions upon which those statements are based, are forward-looking statements." These forward-looking statements generally are identified by the words believes," project," expects," anticipates," estimates," intends," strategy," plan," may," will," would," will be," will continue," will likely result," and similar expressions. Forward-looking statements are based on current expectations and assumptions that are subject to risks and uncertainties which may cause actual results to differ materially from the forward-looking statements. Our ability to predict results or the actual effect of future plans or strategies is inherently uncertain. Factors which could have a material adverse effect on our operations and future prospects on a consolidated basis include, but are not limited to: changes in economic conditions, legislative/regulatory changes, availability of capital, interest rates, competition, and generally accepted accounting principles. These risks and uncertainties should also be considered in evaluating forward-looking statements and undue reliance should not be placed on such statements.