How to Avoid Heartbreak

Celyad’s Chart-1 trial has failed.

The failure is unambiguous*.


With that out of the way, we can dissect some of what went wrong and how investors can avoid future bouts of Cardio3/Celyad-induced portfolio decompensation.


In the heart failure area, mouse models are not serious efforts at validating the platform. End of story. Companies such as Cardio3 / Celyad, and their academic collaborators, move therapies into the clinic based on efficacy in mouse models without repeating the observations in large animal models. Mouse preclinical models of heart failure designed to rationalize a treatment’s transition to the clinic are simply not useful, not reliable, and not reflective of human pathophysiology. They are necessary as a precursor to testing in a large animal model where the platform’s efficacy should be reproduced, albeit at smaller effect sizes. However, they are not sufficient.

And upon reflection, this point has a very obvious but important element: the company and its collaborators know this as well!

I can assure you the in-house scientists and academic collaborators at companies such as Cardio3 / Celyad know full well efficacy in mouse models is not a serious hurdle for the development of heart failure therapies. They know full well that multiple cell therapy efforts with stunning results in mouse models moved into the clinic and walked away with an HR of ~1. They know full well the importance of the therapy showing reproducible efficacy in large animal models.

Investors can debate the reasons and motivations behind companies and collaborators running through preclinical studies at such a record pace. However, many of the descriptors are unflattering.


Track record matters. Not a novel insight by any means. However, companies based primarily on the work of academic collaborators can offer investors an additional edge in appraising the prospects. Over the last 12-18 months, both Celladon and Cardio3/Celyad have had significant failures in the clinic. Each company relied heavily on work derived from academic collaborators. It is wholly appropriate for investors to examine future start-up efforts from those academics with skepticism and take track records into consideration when doing their due diligence.


Never trust a phase 2 trial that succeeded based on subgroup analyses. This, also, is not novel and may have exceptions. However, in the heart failure field, it is almost axiomatic. Cardio3/Celyad’s phase 2 trial leading up to Chart-1 did not focus on the intent-to-treat (ITT) analysis but kicked out some participants for whom the cell therapy could not be procured / manufactured. That was always a red flag. But it is compounded in heart failure where participants appear to be a much more heterogenous group at all stages. Small (n < 100) trials in heart failure tend to pop up positive even with proper control arms. Perhaps it is because we don’t understand the disease all that well, and we’re still learning how to classify stage / progression. Regardless, small, singular phase 2 clinical trials in this indication are simply poor predictors of success, especially when subgroups begin to pop up.


Never trust companies where the CEO corresponds with investors, directly or indirectly, via a message board. The internet is a big and wonderful place. Much is written, much is fabricated, much is conspiratorial. But if you notice the CEO of your company taking time out to gossip about the merits of a blog post, then s/he is not focused on the right questions. This can sound contradictory, because having the CEO or CMO of a company engage you directly can sometimes feel empowering and reassuring. But it’s a bad sign.


And lastly: Stay away from biotechs who use the word “cure” in their slogans or their treatment brand names. This one is self-explanatory.


And that’s that.



* One way to tell if a trial has failed is the Nouveau Analysis Rule: Did you hear about a specific analysis for the first time on the day of the data release? If so, the trial has failed.

Cardio3 and The Biotech Pivot

Precision curative care.

That was the early slogan delivered at the Cardio3 Investor Day on January 30, 2015. It can be hard to limit cynicism when buzzwords (precision), dubious promises (curative) and supposed empathy (care) are trotted out and put on display for prospective investors.

Cardio3’s lead treatment is C-Cure, an autologous cell therapy for heart failure. The method depends on a bone marrow harvest, followed by isolation of a limited progenitor cell population called mesenchymal stem cells. These cells are subsequently treated with a cocktail of factors to induce them into a “cardiopoietic” cell population that, per preclinical results, more effectively benefits the heart by presumably differentiating into cardiomyocytes and using paracrine effects to promote differentiation of endogenous stem cell populations in the heart. This is a more limited population compared to the crude bone marrow mononuclear cell preparations that have convincingly failed in human clinical trials.

This cell therapy preparation is planned to be tested in two phase III trials, named Chart-1 and Chart-2, that are primarily based in the EU and the US, respectively. The precursor phase II trial, the C-Cure trial, ostensibly showed an improvement in left ventricular ejection fraction (LVEF) in a randomized setting, rationalizing the company’s decision to move forward. That said, there are some factors to consider when appraising Cardio3’s chance of success in heart failure.


Perhaps the largest factor in assessing the chance of phase III success is to examine the underlying phase II data. In the case of the C-Cure trial, there were some concerns in the publication expressed by third parties, culminating in a formal reply by the authors. But regardless of reporting inconsistencies, one factor about the C-Cure trial is consistent: the company made clear in the Investor Day presentation that the outcome of the trial was successful. Perhaps a number was mistakenly reported in one location or an aspect of the protocol was not immediately clear when reported, but when all is clarified, the trial is presented as a successful phase II. This is an important point because the C-Cure method is labour intensive, and success with such a complex product should be respected. Each patient randomized into the treatment arm undergoes a bone marrow aspiration to start the process of producing the 6-700x10E6 cells that is re-injected into the patient’s heart. 

Unfortunately, this is perhaps the biggest flag when appraising the phase II data. The results reported by Cardio3 were not presented on an intent-to-treat (ITT) basis. In other words, patients for whom the cell preparation could not be successfully completed were removed from the analysis, effectively making it a per-protocol analysis. This type of analysis can be meaningful for early stages of product development as efficacy signs are looked for in a process that continues to be streamlined and perfected. Nonetheless, for subsequent phase III trials, an ITT population will be the major focus of regulators. For the C-Cure trial, of the 32 patients randomized into the cell therapy arm, 7 patients were subsequently excluded due to the cell collection / expansion process being suboptimal. Along with a few other exclusions for different reasons, only 21 of the 32 patients on the cell therapy arm made it to the per-protocol endpoint.

In short, a phase II trial wherein only 21 of the 32 ITT patients were included in the efficacy analysis on a per-protocol basis ends up on weaker footing than if the efficacy analysis was completed on an ITT basis. This argument isn’t novel, and is in fact noted by an accompanying editorial on the C-Cure trial results. Given the complexity of the cell harvesting and expansion process, it is not unreasonable to expect that Cardio3 will face similar challenges in their registrational phase III trials. However, they are unlikely to convince regulators if they deviate away from an ITT trial analysis. All told, there is reason to be highly cautious of extrapolating the C-Cure results into expectations for Chart-1 and -2.

And perhaps the company feels the same way. Chart-1, the EU-focused trial, has practically completed enrollment. An interim analysis is due in March, 2015, wherein the DSMB will render one of three opinions:

  1. Continue the trial as is
  2. Continue the trial but expand enrollment, as the DSMB signals that an aspect of the trial may need more patients to achieve adequate power
  3. Stop the trial due to clear safety signals or clear lack of efficacy

Outcome #3 is clearly negative. Outcome #2 is very likely a negative sign as well. Expansion of enrollment suggests that the assumptions going into the trial are incorrect, and history has shown that this is usually not a positive occurrence for phase III trials. Outcome #1 is likely neutral and doesn’t denote likely success or failure, simply ruling out overwhelming efficacy or overwhelming toxicity.


With Chart-1 setting itself up for an interim analysis in March, 2015, Chart-2 then stands in a curious place. The company was on record, and the CEO confirmed, that initial guidance had Chart-2 enrolling its first patient by the end of calendar year 2014. That has been pushed back. The claim is that the interactions with the FDA are the cause, specifically a change in the primary endpoint of the trial. Curious. Perhaps this timing gives outsiders a view into the confidence expressed by the company. By pushing back patient enrollment, Cardio3 will likely take in the outcome of the March 2015 DSMB recommendation for Chart-1 prior to actual patient enrollment in Chart-2. It would be reasonable to say that this is a prudent move; however, it would also be fair to say that the confidence in the phase II data is not unwavering.

Coincidentally, approximately half of the Investor’s Day presentation was focused on the company’s recent foray into the CAR-T space. This is definitely a burgeoning area for oncology, with multiple companies and academic institutions taking dedicated steps to develop these treatments. Therefore, it does come as a surprise to see a company called Cardio3 slowly morph into an equal part oncology company. Again, perhaps this is prudent by management. However, long term followers of biotechnology companies will recognize this as the possible first move in a classic Biotech Pivot (Tm). The rules of the Biotech Pivot state that as the day of judgment nears on your lead product, mentions of other pipeline efforts shall increase.


So is Cardio3 executing the Biotech Pivot? Or is this simply prudent management by an experienced team? A prediction is that the March 2015 analysis is benign and that the company continues Chart-1 as is and that Chart-2 enrollment is extremely slow for the first half of 2015. And, luckily for investors, there will be plenty of talk on the CAR-T side to keep them busy... or distracted.

Information Overload

Investors in biotechnology companies follow their investments closely. Therefore, they appreciate the knowledge that a clinical trial, previously announced by management to be in planning, is now actually up and running. That’s a tangible milestone and allows investors to better formulate their perception of a management team’s ability to execute.


But then, sometimes, one meaningful press release can turn into a dozen. Or can it? Consider the time line of Cardio3 and the opening of their European phase 3 trial, Chart 1.


November 22, 2012 - Cardio3 announces opening of Chart 1.

And there was much rejoicing.


June 10, 2013 - Cardio3 announces treatment of first patient in Chart 1.

Ok. It is one thing to establish the protocol and organization of the clinical trial, and now we get confirmation that an actual patient has been enrolled. It is appreciated. Thank you. Now go out there and execute execute execute!


September 30, 2013 - Cardio3 announces world’s first phase 3 trial in regenerative medicine for heart failure… in Spain.

Hmm… wait, is this another trial? In this short time did they manage to open another trial apart from Chart 1? Well, no. Chart 1 now has the occasion to enroll eligible patients in Spain.


October 23, 2013 - Cardio3 announces that Chart 1 can now enroll patients in Italy.

Surely they’re not going to announce the opening of a center in each and every new country under the jurisdiction of the EMA?


November 12, 2013 - Cardio3 announces, among other things, that Chart 1 can now enroll patients in Poland.

Maybe we are. And don’t call us Shirley.


April 10, 2014 - Cardio3 announces that Chart 1 can now enroll patients in Ireland.

Fair play. These lads are sound.


June 10, 2014 - Cardio3 announces that Chart 1 can now enroll patients in Sweden.

Heja Sverige! *


October 13, 2013 - Cardio3 announces that Chart 1 can now enroll patients in Switzerland.

Watch your back, Novartis.


One can reasonably argue whether all of these PRs were needed. Cardio3 do make tangible announcements of the trial being 50% enrolled as well as 100% enrolled. But it is reasonable to ask the company what the significance was of each of these territorial openings? Maybe. Maybe not.


And sometimes, fluffy PRs aren’t limited to trial site openings. To wit:


March 26, 2014 - Cardio3 announces that their investigational therapy was mentioned in a review published in Nature Reviews Cardiology.

This is great. You always want independent recognition of your work within the community. But wait. Could it be that one of those authors once worked at Cardio3? Could it be that two of those authors, at one point in time, were noted to have financial interest tied to Cardio3? To be fair, one of the authors of that review appears to have appreciated the work of Cardio3 so much that he has worked hard to take the horns and help the company execute another of its trials, this time in the US (warning, pdf link!). The authors of the manuscript explicitly state that they have no competing interests. I take them at their word.


April 7, 2014 - Cardio3 announces that their regenerative technology was recently highlighted in an editorial.

Again, very good news. But… huh. Those names look familiar.


October 8, 2014 - Cardio3 announce a research and development collaboration.

Details here (again, pdf link). Something in there sounds familiar.


In short, it is always interesting to watch the operation of companies from the outside. It can be like a game of Clue, and some companies put out more red herrings than material pieces of information. And sometimes, the pieces of information lack full context. Nonetheless, it’s a fun game for investors to play.


* Apologies to native Swedish speakers. I stole this from the internet in the hopes that it is in correct context.


Cardiomyocyte Regeneration And Heart Failure Therapies

A significant number of clinical cardiologists have been pursuing the use of various progenitor / precursor / pluripotent cell types to heal / recover / restore damaged and infarcted myocardium. Unlike the methods focusing on implanting cardiomyocytes following in vitro differentiation, a competing school of thought is attempting to inject progenitor cells into the blood stream. For the latter approach, there are a series of steps presumed for efficacy to be demonstrated. The general premise is that the precursor cells, one in the bloodstream, will:


Step 1. Migrate to the vessel proximal to the area of the injury

Step 2. Travel through the layers of the vasculature

Step 3. Migrate further to the specific area of ischemic / stunned / infarcted myocardium

Step 4. Accept unique local signaling cues to differentiate into mature myocardium

Step 5. (Optional) Secrete signaling cues to mobilize resident precursor cells to differentiate and ameliorate the damaged area


In the clinic, bone marrow-derived mononuclear cells used in this manner flopped spectacularly (1), although various efforts (C-Cure from Cardio3 to Capricore’s CDCs) continue to use populations of precursor cells that are expected to complete Steps 1 through 5 in order to achieve efficacy. Generally, I’m highly skeptical of these latter methods for various reasons. There are limited to data to describe the signaling that controls Step 1. Further, it is unclear how these precursor cells, which are not expected to be migratory, are able to travel through the layers of endothelium to exit the lumen of the vessels and capillaries that they’re in. For example, leukocytes have specific mechanisms that dictate their adhesion to, and migration out, of vessels at specific sites (2). For proponents of precursor-based myocardial repair, no meaningful mechanisms are provided to explain the mode of migration used to exit the blood stream. The limitations of Step 3 overlap with Step 1, and presumably a solution for the latter will apply to the former. The localized signaling cues required to achieve Step 4 largely remain mysterious in the context of cells injected into the blood stream.


Step 5 is similarly ambiguous. A recent paper by Tomasetti and Vogelstein (3) presents a significant challenge to the assumptions behind Step 5. In the paper, the authors test a possible link between organ-specific cancer risk and the effects associated with lifetime number of stem cell divisions (Figure 1, which can be seen at the following link). The note a correlation of 0.81 between lifetime risk and total stem cell divisions. So how does a figure linking lifetime cancer risk to total stem cell divisions relate at all to healing the heart?


One of the clues may be apparent in what Tomasetti and Vogelstein do not show: the heart is absent in their figure. The reason is largely because the heart is not a hot spot for cancers. Myxomas may be readily prevalent in some people with Carney’s complex (4), but there is a strong genetic basis related to endocrine overactivity and modulated cAMP-dependent protein kinase function. So if the prevalence of cancer is low, what about the regenerative capacity of the heart? 


Generally speaking, the heart is not remarkable for its regenerative capacity. A recent paper noted that a subpopulation of c-Kit expressing cells in the heart may produce new cardiomyocytes, comprising up to 0.016% when stimulated by injury (5). Yes, 0.016%. This finding should not be too surprising. Regeneration in the heart, despite the hopes and desires of some, has always been very low. Even further, an interesting study from 2009 suggested that cardiomyocytes renew at ~1% annually until the age of 25, and then decline to ~0.5% by the age of 75 (6).


Interestingly, the lack of regenerative capacity in the heart is very much at odds with companies that are publicizing a pipeline product that hopes to capitalize on this inherent regenerative capacity. In effect, acceptance that such a regenerative capacity may be insignificant is a considerable challenge to these programs. In that respect, I think the paper by Tomasetti and Vogelstein adds yet another nail to the coffin of the “heart-is-a-regenerative-organ” proponents. If the heart is readily regenerative, lifetime total stem cell divisions would be higher *and* the lifetime risk of neoplasia would be higher… or at least figure measurably in Tomasetti and Vogelstein’s data.


At this point, methodologies claiming to utilize the inherent regenerative capacity of the heart are based more on data-free optimism than a rational view of the literature.