Science Integrity Digest

Cassava Sciences: Of stocks and blots

A group of anonymous persons filed a Statement of Concern with the FDA regarding the integrity of research papers about Simufilam, an Alzheimer’s Disease treatment drug candidate developed by Cassava Sciences (Nasdaq: $SAVA). The petition raises concerns about Western blots and methodology. I took a look at the problematic photos included in the report, and agree with most of those concerns. I also found some additional problems.

Disclosures

Simufilam

Simufilam is a drug candidate for the treatment of Alzheimer’s disease (AD). Initially called PTI-125, it is a small molecule that binds to Filamin A, a scaffolding protein that is involved in cell shape and movement. By binding to Filamin A, it can prevent the molecular events that increase tau phosphorylation, which is associated with neurofibrillary lesions found in AD.

According to Cassava Sciences, Simufilam treatment in Phase 2 clinical trials improved cognition in Alzheimer’s patients. This would be great news because there is no current treatment for AD. Cassava therefore hopes to quickly initiate Phase 3 trials, and that the drug will be approved by the FDA.

Statement of Concern

But last week, on August 18, 2021, Jordan A Thomas from Labaton Sucharow LLP (LS), a law firm in New York, posted a Statement of Concern report, asking the FDA to halt Cassava’s clinical trials because of “grave concerns about the quality and integrity of the laboratory-based studies surrounding this drug candidate.” The petition was filed on behalf of as yet anonymous persons.

The pre-clinical PTI-125 experiments have been conducted in the lab of Dr. Hoau-Yan Wang, an Associate Professor at City University of New York (CUNY). The work has been funded by Pain Therapeutics, Inc (PTI), the old name of Cassava Sciences. Wang currently serves on Cassava’s Scientific Advisory Board.

The LS report raised three types of concerns regarding the Simufilam/PTI-125 papers.

  1. Concerns about the integrity of the clinical biomarker data
  2. Concerns about the integrity of Western blot photos
  3. Concerns about the analysis of frozen human brain tissue samples

Here, I will focus on the Western blot concerns as raised in the petition, both in the papers supporting Simufilam, as well as in some other papers from the lab of Wang et al.

Screenshots from Labaton Sucharow Statement of Concern report. Source: https://www.regulations.gov/docket/FDA-2021-P-0930/document

Western blot problems in Simufilam papers

The Cassava Sciences’ Publications page lists three Simufilam research papers and one review paper. According to the LS report, two of the three research papers contain Western blots with potential problems. I agree with many of these concerns and posted them on PubPeer – and also found some additional problems.

The three Simufilam research papers featured on the Cassava Sciences website are:

According to the LS report, several Western blots in the 2012 J Neurosci paper – in particular those with control proteins – show oversaturated bands with little background detail, irregularly spaced bands, and bands that look remarkably similar. I agree with these concerns and posted them on PubPeer. The blots are compressed and of relative low resolution, so I hope the authors can respond by showing the high-resolution scans to take away any concerns.

A collage of Western blots that might show irregularly spaced bands, saturated bands, and bands that look unexpectedly similar as shown with arrows of the same color. See: https://pubpeer.com/publications/F91E0D22B887598445BB1F908393EE

An additional concern in this paper, not mentioned in the report, is a potential duplication I found between two panels in Figure 8. Two panels displaying different immunostainings and treatment appear to share an area that looks similar. It’s indicated below with cyan boxes. These might be two consecutive tissue sections, perhaps stained for different proteins, although some features are remarkably similar in the right hand part of the image. However, the images are representing differently treated mice, so the samples should not look so similar.

Cyan boxes highlight two panels that appear to show an overlap. See: https://pubpeer.com/publications/F91E0D22B887598445BB1F908393EE

In the 2017 Neurobiol Aging paper, the LS report noticed some potential problems in Figure 12. In the top part, the NR1 blot contains only 12 bands, while 13 different conditions and lanes are shown in all the other blots. In addition, the NR1 blot appears to show a splice after lane 8, while the PLCgamma1 blot and most other blots show a splice after lane 10. This suggests that not all blots were derived from the same samples.

I agree with this concern and posted it on PubPeer. Of course, the authors could remove these concerns by showing the original, high resolution blots, and by explaining if and why the blots were spliced.

Screenshot from the “
Statement of Concern”. Source: https://www.regulations.gov/docket/FDA-2021-P-0930/document

An additional finding by me – not mentioned in the LS report – is the presence of a dark rectangle around a band in Figure 3B. This could mean that the band shown in that lane may have been derived from another blot, making it hard to know exposure or size of the band relative to the other bands in that photo.

Concern about Figure 3: A band in one of the 10-month lanes appears to be surrounded by a darker rectangle, as shown with green arrows. See: https://pubpeer.com/publications/80DD10169D3C375C5828BC2711A49B

The third and most recent paper, JPAD 2020, was mentioned in the LS report because it was submitted to the journal on October 31, 2019, and accepted on November 6, 2019. Usually peer review takes anything from one to many months, so a peer review turned around in a week is remarkably fast.

Screenshot from https://link.springer.com/article/10.14283/jpad.2020.6, which appears to show that the paper got accepted a week after submission.

The LS report did not mention any image problems, but Figure 3A of the paper appears to show a rectangular box around one of the pT231Tau bands. It is shown below with pink arrows. Some other bands, shown with blue arrows, appear to be surrounded by a white halo and to float above the background. The latter concern could just be a compression artifact, although it is not seen in all bands, just in a couple. And the authors can take away any concerns by showing the original blots in high resolution.

Figure 3A of doi: 10.14283/jpad.2020.6 appears to show a band surrounded by a lighter box. Source: https://pubpeer.com/publications/A8DD7059A8A7F13D4899049A83F61E

Other papers with image concerns

At least five other articles from the Wang lab at CUNY appear to show image concerns as well. These papers might not be directly related to Simufilam research, but they are still indirectly connected to Cassava Sciences or its drug candidates. Some articles are also about Alzheimer’s Disease or filamin A binding; some were funded by Pain Therapeutics, the Cassava Sciences precursor; and some are authored by Lindsay H Burns, the current Senior VP Neuroscience and lead scientist of Cassava Sciences. And of course, as mentioned above, Dr. Wang is one of Cassava’s Scientific Advisors. Together, the problems in these additional articles raise concerns about Western blots and perhaps also other data from this lab, spanning a period of 15 years.

The five additional papers are:

The Neuroscience (2005) paper has two blots that look similar; several bands that look similar; a rectangular box around a band; and a blot that appears to have been reused in the Biological Psychiatry (2010) paper, where it represented a different experiment. The collage below combines all findings, some of which were copied from the Report, plus some additional concerns that I identified.

The Behavioural Pharmacology (2007) paper was not included in the LS report, but Figure 4a appears to contain some duplicated bands. As always, making the original blots available might prove these bands are not the same.

Source: https://pubpeer.com/publications/5CA0279B6C960FB98610385BB2AE5C

Figure 7A of the PLOS ONE 2008 paper contains two blots that appear to look identical. In addition, some sharp background transitions suggestive of splicing may be visible.

Source: https://pubpeer.com/publications/0B03A2B682AAAD6A8E9D7C2C49DD22

An Actin blot in Figure 1a of the Biological Psychiatry (2010) paper looked, as mentioned above, remarkably similar to a blot panel from Neuroscience (2005), where it represented a different experiment. In addition, Figure 1A appears to show boxes around some lanes in the alpha7nAChR panel.

Finally, the PLOS ONE 2014 paper appears to show some problems in Figures 1B and 2A. These concerns were not part of the LS report, but are my own findings. Here, some blot bands appear to be surrounded by sharp transitions, suggesting that not all bands shown together in the same photo were derived from the same blot.

Source: https://pubpeer.com/publications/7D632D5EDFCE01EBBB7BDF55A59F36

A rebuttal by Cassava Sciences

After the LS report was filed with the FDA on August 18, 2021, and publicly posted on August 23, Cassava Sciences wrote a response on August 25. From the press release:

“As a science company, we champion facts that can be evaluated and verified,” said Remi Barbier, President & CEO. “This helps people make informed choices. It is important for stakeholders to separate fact from fiction, which is why we wish to address allegations head-on.”

The parts of the response dealing with the problematic Western blots are not very convincing.

Rather than trying to dismiss the valid concerns with vague statements that “control bands are supposed to be highly similar” and calling the concerns “fiction“, I would like to see the authors present high-quality scans of the uncropped Western blot films. Showing original blots could take away some of the concerns, and might put the discussion to an end.