In the last two days I was involved in lively twitter discussions about the results of Phase 2 PreSERVE-AMI clinical trial, sponsored by NeoStem. Unfortunately, interpretation of results divided professionals for two camps – (1) trial is failed and (2) trial is successful. Why did it happen? Maybe the truth somewhere in the middle? I’d like to share some of my thoughts and post some questions for discussion here.


Definition of trial failure
Clinical trials can fail for many reasons, not only because investigational product does not work. For example, bad trial design, wrong endpoints, unanticipated slow enrollment and logistical difficulties, sponsor out of money, changing business strategy, termination due to complications and safety issues, termination due to lack of feasibility and technical difficulties – all these and many other reasons could lead (or contribute to) trial failure. In case of NeoStem trial it was quite simple – according company’s press release and trial record on NCT database, primary endpoints were not met. Failure = endpoints not met! It was… until…

Changing of trial record in registry
Company described trial results as “positive”, because some measures were significantly different from placebo. However, those were other than primary endpoints readout measures – serious adverse events (SAEs), major adverse cardiac events (MACE) and survival benefit. In the night of news release, some analysts started to point out that trial is failed because primary endpoints were not met. And the company reacted to it by changing trial record on NCT database! I noticed that record was changed after press release, not any time before. You can see changes before and after Nov. 17, 2014 (night of press release) on this screenshot:

Thanks NCT for archiving all changes! So, now, based on the current record, endpoints are met! Magic! Some questions for discussion:

  1. How is it legitimate to change trial record as reflection to public reaction on the night of press release? If it’s ok to do, is it a common practice for companies?
  2. Were “new endpoints” approved by FDA as “acceptable changes” on the middle (or close to the end) of trial? If we assume (I really doubt it) it was, is it a common practice? If we assume NeoStem got FDA blessing to change endpoints (to fit the data) on a middle of the trial, why not update record right away instead of holding it until press release night?

Digging into endpoints
“Initial primary endpoints” of trial, which failed, were myocardial perfusion by SPECT imaging. So, there was no difference in blood perfusion of affected hearts. Was it because of lack of method sensitivity (as company put in press release) or because of other reasons? Maybe CD34+ cells did not impact angiogenesis/ vasculogenesis in any way? Maybe the mechanism of action was different from anticipated? We don’t know for sure.
Now, “new endpoints” – SAE, MACE and mortality were statistically significant. Results for MACE were dose-dependent. One year mortality was less in “CD34+” group compared to placebo. These are good news! Good for patients. These “new endpoints” were born in discussions with FDA in order to move forward:

… which is consistent with U.S. Food and Drug Administration (FDA) guidance that mortality and MACE are the appropriate approvable endpoints to determine efficacy of a cellular therapy for cardiac disease as opposed to imaging endpoints.”

Some questions:

  1. Is mortality readout at 1 year the best time point? Why not at 3 months or 6 months or 3 years? Are 3 deaths in placebo group versus 0 in “cells” group enough for confident conclusions?
  2. What was a number of patients in each cell dose group? If we look at only “>20 million cells” group with significant difference in MACE, was number of patients strong enough for confident conclusions? If every 5 million of cells make a difference, was the good positive trend in patients, received 25M, 30M, 35M…. of cells? How much cells were more than 20 – could be 21 or 50 or 100?

Heating up anticipation of results release and investors reaction
NeoStem stock was going up like a rocket few days before press release (+40% in last 2 days of the last week). The company did a good job to heat up anticipation of results. First, a week or so before was news circulation about upcoming release. Second, AHA released abstracts from its annual meeting (where results were reported to public) a week or so before and we were able to read NeoStem’s abstract. There were no any results in abstract. My question is – what was a point of this abstract?
So, investors rewarded the company in anticipation of results, but dumped it in next day of release. It was -29% on Tuesday, November 18 and -9% today.

Positive spin
It seem to me that NeoStem got the same problem as many other public “stem cell companies” in the field – hyping press releases. Let me just tell you how I see “normal” / balanced/ sober/ not hyped press release:

  1. Please start from endpoints, listed on trial record and say (a) met or (b) not met. There is nothing more important than that. I’d have a respect for the company, which can say “We failed to meet primary endpoints” in press release.
  2. Please linked to trial record on database, so reader can click on it, compare with press release language and trace possible discrepancies.
  3. Please avoid use “hype words”. It makes some readers suspicious.
  4. Press release should be crystal clear, should leave no doubts and no different interpretations.

The NeoStem’s release started from “positive” in title and continued with all good things – SAEs, MACE, mortality, but not “initial primary endpoints” (perfusion by imaging). Important “primary endpoints” came up somewhere in the middle and were “blurred” by reference to FDA, which says “appropriate approvable endpoints are mortality and MACE”. So, don’t worry guys, we don’t need these perfusion data, anyway.

The future
As we can see some results are promising and, I think, company on a good track with it, because:

  • they now know what cell dose to pick for future trial
  • they now know how to choose endpoints wisely
  • they now know how to design the trial correctly in order to expect good results
  • they in constant communication with FDA and in agreement with agency on “optimal” endpoints for future
  • they probably will not have any problems with FDA in submission for Phase 3 trial…

But, but, but it does not cancel the failure. It does not nullify the “unintended mistakes”, which they made with picking primary “imaging endpoints” (who knew that it is not going to work at the beginning of trial?). It does not nullify “intended mistake” with changing database record and twisting results for public as entirely positive. I think, mistakes and failures should always be acknowledged.

Despite company’s optimism, there will be challenges ahead. For example, manufacturing issues, related to purification of >20M CD34+ cell number from single bone marrow aspirate. As Robert Preti said in release – they are still “refining the methods for collecting and purifying CD34″. So, by the end of Phase 2, the manufacturing process is “not locked”. The other challenge is figure out the mechanism of action and finalize the potency assay. The company positioned mechanism of actions of NBS10 as pro-angiogenic. But lack of any detectable changes in myocardium perfusion (SPECT imaging) does not support this mechanism. We also have to remember that results of Phase 3 must be very very good in order to beat competitors on the market.

I’d like to point out that released results are preliminary. It was nicely highlighted by NeoStem in the title as “initial data”. The final results could be different – in positive or negative way. As of now, we cannot learn a lot of information from press release. We have to wait for publication and learn more about the trial and results.

To conclude: The positive spin of preliminary results of Phase 2 PreSERVE-AMI trial caused unanticipated confusions. Changing of trial record to fit the data in the night of news release is not good practice and will undermine company’s credibility.

{ 1 comment }

I’ve written two years ago about autopsy findings from ALS patients, treated by neural stem cell product candidate, commercialized by NeuralStem. Recently, company published an update on pathology of spinal cords and donor cell fate after transplantation. The article is freely available online.

The authors analyze 6 postmortem cases of transplantation investigational product NSI-566 into spinal cord of ALS patients. Knowing DNA sequence of donor cell line, they were able to trace cells in the sites of implantation. In all 6 cases donor’s DNA was detected (0.67–5.4% of total tissue DNA) in patients survived up to 3 years after surgery. Based on donor’s DNA detection, the authors made a conclusion that transplanted cells survived long-term, without any correlation with course of immunosuppression. From press release:

… researchers found no correlation of DNA content to survival period after immunosuppressant medications had been discontinued. These data demonstrate that transplanted HSSCs can survive for a prolonged period, even in the absence of immunosuppression.

The question is that whether detection of donor DNA equivalent of cell survival? The authors said YES, and, most likely they are right. However, we cannot exclude horizontal transfer of donor’s DNA with apoptotic bodies, described before as phenomenon of chimerism.

The real gem of the study is the case of sex-mismatch transplant. If male neural cell line was transplanted in female recipient, we can detect donor cells by difference in X and Y chromosomes, using fluorescence in situ hybridization (FISH analysis). Detection of XY+ (male) cells (64% in injection sites and nothing in other regions) was a definitive proof of donor’s cell survival and persistence (for 196 days after transplant in this particular patient). By adding immunohistochemistry to FISH analysis, researchers able to show neuronal differentiation (by coexpression of NeuN marker) of XY+ cells. The area of injections were stained negatively for glial markers. So, there was no evidence for glial differentiation (read multipotentiality) of transplanted cells. Undifferentiated SOX2+ cells were co-located with XY+ cells in areas of injections. Morphologically similar “nests of non-native” non-glial cells were identified in all other cases. Based on these observations, the authors “suspect” long-term survival of donor cells in all cases.

The biggest outstanding question remain: “Is donor cell long-term engraftment linked to therapeutic benefit”? It’s unclear from the pathology studies, but company believes that cell survival and efficacy are linked:

“The success of our therapy is predicated upon our cells ability to survive long-term and differentiate, providing neurotrophic support in the spinal cord and acting as ‘nurse’ cells for the patients’ own motor neurons that are attacked by the disease,” said Karl Johe, PhD, Neuralstem Chairman and Chief Scientific Officer.

I’d definitely avoid some strong claims in press release at this point.


It has been awhile since I called to collaborate and proposed to track results of clinical trials and studies. Unfortunately, nobody expressed an interest. So, starting from this year, I’ve attempted to do it myself. Today, I’m sharing results of the first 3 months of this experiment.

I track results of:

  1. published clinical trials;
  2. published not registered (as trials) clinical case studies (where number of patients >3);
  3. interim and final results of unpublished industry trials via press releases or/and conference reports.

In order to capture all trials and studies, I set “very loose” filter – PubMed RSS feed of “stem cell” term. I do “hand coding” and sort out all results manually.


  1. to determine usability of the proposed filter;
  2. to identify advantages/ disadvantages of of standard searching tools and find the best searching strategy;
  3. to determine a frequency of reporting results in cell therapy clinical studies;
  4. to determine a value of registered trials with ID among all reports;
  5. to set an example for potential crowdsourcing and collaboration.

I routinely use PubMed RSS feeds in order to capture everything in stem cell/ cell therapy/ regenerative medicine field. I found that “stem cell” feed is the best option to capture all published reports. Unexpectedly, this filter also captures almost all cell therapy trials (even without indication to stem cells), for example, cellular immunotherapy. Disadvantage of this filter is low selectivity – every week there is about 500-700 articles. It’s very time consuming.

To capture press releases from companies, I was using simple Google search for “news” by keywords: “stem cell”, “cell therapy”, “regenerative medicine”.

I’ve tried standard searching tools, used by many data miners, such as “clinical trial” filter in PubMed. It didn’t work. For the first quarter of 2014, search in PubMed by a query:

  • “stem cell” with filter “clinical trial” yielded only 1 result
  • “cell therapy” with filter “clinical trial” yielded 2 results
  • “mesenchymal” with filter “clinical trial” yielded 2 results

Using my RSS filter + Google search, I was able to track 41 clinical studies reports (Q1 of 2014), including:

  • 38 published trials and studies and
  • 3 companies press releases

24 of 41 (59%) studies had trial IDs and were registered in databases. 7 of 41 studies were categorized as “case series”. I was not able to identify trial ID in 10 of 41 (24%) reports.

As an example of tracking, I’m sharing raw data for the first quarter of 2014 – you can see it here (you can view, but not edit this spreadsheet).


  1. PubMed RSS feed for “stem cell” is the best tool to capture all reports for clinical studies results in cell therapy/ regenerative medicine. It’s not automated, required “hand coding” and very time consuming.
  2. Standard PubMed search, using “clinical trial” filter misses about 95% of reports.
  3. The frequency of reporting results for Q1 of 2014 was 1 report every 2.2 days. Basically, every 3 days somebody around the world reports cell therapy clinical studies results.
  4. More than half reports (~60%) were associated with defined trial ID and registered in international databases.

Future development:

  • Capturing cell therapy trials and clinical studies results is a huge task and required collaborative efforts. This post is a snapshot of the first 3 months of 2014 and call for collaboration and crowdsourcing of the data.
  • The best available tools for capturing reports are manual and not sensitive. Therefore, there is a big demand for improving strategy for capturing cell therapy reports. Crowdsourcing can provide a common strategy or an algorithm set by consensus.
  • Analysis of results will allow to define: failure/ success rates by phase of trial, rate and time of results reporting, trends in cell types, trial type, geographical distribution and so on.

I’d be happy to hear any opinions and suggestions for improvement and collaboration on this project.


Trends in cell therapy clinical trials 2011 – 2013

March 2, 2014

Since last year, I’m sharing some trends in cell therapy clinical trials. Real time tracking of all registered trials allows to identify trends in multiple dimensions. Today I”m posting a snapshot of trending for the last 3 years – from 2011 to 2013. There could be some minimal discrepancies in numbers between this and previous […]

Read the full article →

Role of monocytes in regenerative cell therapies

February 16, 2014

As I’ve written before, we largely don’t know what cell population underlies the mechanism of action of bone marrow or cord blood transplantation in regenerative medicine. Before infusion in neurological, cardiac, hepatic and other conditions, cord blood or bone marrow usually undergo minimal processing – red blood cells and platelet depletion. What remains before infusion […]

Read the full article →

Cell therapy clinical trials – 2013 Report

February 1, 2014

This is 2013 report of registered clinical trials in cell therapy. Every year I overview some tracked data from international clinical trials databases. You can see previous annual reports here. Definitions and criteria I tracked clinical trials which fall in definition of cell therapy: administration of living cells in human with therapeutic purpose. Besides “traditional […]

Read the full article →

Cell therapy trials failures in 2013

December 25, 2013

At the end of the year, I want to bring your attention to major failures and setbacks for cell therapy field in 2013. This is an attempt to capture the vast majority of failed clinical trials. This overview is based on (i) results reported in literature (via PubMed) or (ii) companies press releases and (iii) […]

Read the full article →

Trends in cord blood enhancement 2013

November 10, 2013

I’m following with a big interest the latest development in technologies for cord blood (CB) enhancement in hematology. “Enhancement” means boosting a therapeutic value of cord blood graft via its stimulation/ modification (homing, migration, engraftment) or increase cell number (expansion, processing optimization). Today, I’m going to overview very recent trends and give update on methodologies […]

Read the full article →

Hurdles in conducting of cell therapy trials for acute neurological conditions

October 20, 2013

Acute neurological conditions, such as stroke and traumatic brain injury (TBI) is an attractive target for cell therapy. Success of pre-clinical models allowed to move to clinical trials relatively quickly. Industry also sees a great opportunity in this direction. For example, Athersys and Cytomedix have active ongoing Phase 2 trials. However, the trials did not […]

Read the full article →

Failure of autologous cord blood transplant in diabetes

October 13, 2013

I’ve written about trials, assessing autologous cord blood transplantation (CBT) in pediatric type 1 diabetes. The results of first such trial, conducted by University of Florida in collaboration with NIH, were disappointing. The results of another trial, conducted in Germany and utilized the same protocol, have been published very recently. And as US trial, this […]

Read the full article →