FUDsters will object to the article, because there's no mention of "fumbled data" in this well elucidated review...
Anavex Alzheimer's And Parkinson's Drug Underappreciated By Investors
Nov. 13, 2023 11:41 PM ET Anavex Life Sciences Corp. (AVXL)BIIB9 Comments
The Political Economist
524 Followers
Summary
Anavex's Alzheimer's drug Blarcamesine demonstrates strong statistical significance in improving cognition among dementia patients as measured by one co-primary endpoint and one secondary endpoint.
For the other co-primary endpoint, it seems the drug may not demonstrate statistical significance using Least-Squares Mean methodology.
Most recently published data shows Anavex's drug demonstrated statistical significance in manipulating crucial biomarkers: preservation of brain mass and lowering amyloid accumulation.
Memory lapses, forgetting things, degenerative disease. Brain problems. Parkinson and alzheimer desease. Mental health. Face side view
Naeblys
Here is the crucial paragraph from Anavex's (NASDAQ:AVXL) September news release regarding its apparently groundbreaking new Alzheimer's drug, Blarcamesine:
The [Phase 2b/3] trial is successful in meeting the co-primary endpoints if the significance of each endpoint is P < 0.05, or if the significance of only one co-primary endpoint is P < 0.025. If only one primary endpoint is significant at an a level of 0.025, then the secondary endpoint will be evaluated at the same level of 0.025. The trial was successful, since the differences in the least-squares mean (LSM) change from baseline to 48 weeks between the blarcamesine and placebo groups were -1.783 [95% CI, -3.314 to -0.251]; (P = 0.0226) for ADAS-Cog13, and -0.456 [95% CI, -0.831 to -0.080]; (P = 0.0175) for CDR-SB in patients with early Alzheimer’s disease.
So, this is what Anavex is telling you: we have great results on ONE co-primary endpoint and ONE secondary endpoint. BOTH of these endpoints are fantastic measures of the severity of dementia in a patient. They are the ADAS and the CDR-SB.
Regarding the OTHER co-primary endpoint, the ADCS: Anavex has not released the least-squares mean data for it. Nor have they released the difference in the mean change for this endpoint. In the original press release and presentation from December 2022, Anavex stated that both co-primary endpoints were successfully met. If that is indeed true, then, it may be assumed that the ADCS endpoint was met successfully with statistical significance at .05.
The information we have regarding Blarcamesine's performance as measured by the ADCS comes from their December 2022 press release regarding the same Phase 2b/3 trial:
ANAVEX®2-73 treatment was 167% more likely to improve function compared with placebo, at a clinically meaningful improvement of ADCS-ADL score change of +3.5 points or better, Odds Ratio = 2.67 (p=0.0255). This reflects a robust improved and clinically meaningful outcome in cognition and function from baseline.
This sounds great, but it's not clear to me exactly what this means. At face value, it means that, if there are 100 people on Blarcamesine and 100 people on placebo, if 10 people IMPROVED by 3.5 points while on PLACEBO, then 27 people on Blarcamesine improved by 3.5 points. I would not be surprised by 9 people on placebo improving by 3.5 points on a scale that involves structured interviews with their caregiver. I am highly encouraged by 24 people out of 100 people taking Blarcamesine improving by 3.5 points. And I am even more highly encouraged that the Odds Ratio is 2.67 at a p of .0255. It seems, with this 2.67 Odds Ratio, that the worst case scenario would be that only 3 people on placebo improved by 3.5 points and 7 people on Blarcamesine improved by 3.5 points. Please note that this is not the only time Anavex has utilized the Odds Ratio of clinical improvement as a measure; they used it as their primary measure in their Phase 3 Rett Syndrome press release.
The September 2023 press release seems to imply that statistical significance at the p of .05 level was not met in the least-squares mean change from baseline to 48 weeks for the ADCS. In contrast, statistical significance was met in the mean change in scores, as implied by the December 2022 press release.
To continue with Anavex's assertions: if we had two statistically significant findings at p=.05 with our two co-primary endpoints, our study would be easily accepted as proof of drug efficacy. Instead, we have one co-primary endpoint, the ADAS-Cog, with an effect at a very statistically significant level, a P of less than .023. We also used a measure that is similar to the ADCS; this is the CDR-SB. And using the CDR-SB as a metric and an endpoint, we uncovered another very statistically significant result, at a P of .0175.
Two endpoints being met at less than a P of .025 and one that is not met at a P of .05 is just as good or better than two co-primary endpoints being met at .05. This is the statistical conclusion of Anavex.
So, according to the statisticians at Anavex, this study was a success based on the great success of the ADAS-Cog and the CDR-SB COMBINED with the failure of the ADCS. The statistical non-significance of the ADCS data is being taken into account by forcing the other two metrics to meet an extremely high statistical standard, that P must be less than .025.
How much higher of a standard is a P of .025 than a P of .05? A finding at P of .025 is 20 times less likely to be due to chance than a finding at .05. If two measures are statistically significant at a p of less than .025, and if those two measures both measure the same thing, then it may be estimated that it is 400 times less likely to be due to chance than a single finding at a p of .05.
According to the statisticians at Anavex, this outweighs, so to speak, the fact that one measure did not find a statistically significant difference between the effects of Blarcamesine and placebo at 48 weeks.
Another matter to take into consideration: the three tests do not all measure the same thing.
The ADAS is a direct measure of cognitive ability. It is a long and arduous written and oral exam that involves numbers, words, figures/diagrams, drawing, speaking, and remembering. The ADAS-Cog13 includes 13 different categories and the patient herself must complete the tasks on her own.
The ADCS-ADL is an indirect estimation of behavior. It involves a caretaker or monitor answering questions about the behaviors and habits of the patient. The caregiver may say something like, "The patient has been having difficulty using his phone this month."
Not only do the ADAS and the ADCS-ADL measure different aspects of the patient's condition, but the placebo effect is far more likely to play a role in the scores of the ADCS-ADL. An optimistic patient who believes she is getting better because of a placebo pill may put more energy into daily routines like dressing, eating, speaking, and leaving the home. But the placebo effect may play little or no role in helping the patient to correctly draw a clock.
In my eyes, the ADAS has the least room for any "fudge factor" of the three tests in question. And the ADCS has the most "fudge factor."
The Magnitude of Effect
The power of the findings was impressive, maybe incredible. The least squares mean of the differences in the ADAS-Cog13 scores was -1.783. Considering that one would expect the placebo group to INCREASE (worsen) their ADAS-Cog13 score by about 1 point in a year the fact that the group that took Blarcamesine had an average score that was almost 1.8 points LOWER than the placebo seems to indicate that the typical Blarcamesine-taker may have actually reversed the progression of the disease and IMPROVED their cognition.
Let's compare the least-squares mean (LSM) difference of -1.783 to the first mean difference calculation presented on December 2, 2022:
Additionally, treatment with ANAVEX®2-73 statistically significantly reduced cognitive decline, measured with ADAS-Cog, compared to placebo at end of treatment by 45%, representing a treatment difference in mean score change of -1.85 points (p=0.033).
The least-squares mean difference is smaller but at a much lower level of P. I prefer the least-squares mean difference over the original mean difference. The two numbers are very similar, as -1.85 is only .067 points lower than -1.783. So these two numbers are similarly impressive, but the least-squares mean offers a much greater statistical significance.
Another clue as to the shape of the distribution curve was given in the December 2022 press release:
On average, patients, who improved cognitively with ANAVEX®2-73 treatment, improved by ADAS-Cog cognition score of -4.03 points.
I did not accept this statement at face value when I originally read it. It says that the average patient in the treatment arm actually IMPROVED their ADAS-Cog13 score by over 4 points. I could not believe it, so my brain automatically interpreted the finding as, "On average, the patients who DID improve their ADAS-Cog13 scores improved the scores by over 4 points."
But having looked at the information again, it looks like it is supposed to be read as meaning that those in the treatment arm improved their ADAS-Cog13 scores by 4.03 points, and those in the placebo arm improved their ADAS-Cog13 by 2.18 points.
That's one hell of a placebo effect.
And that's one reason why I could not believe it.
If we assume that what they meant to say was, "those who did improve improved by over 4 points on average" then we can conclude that there is a big hump in the distribution curve, say, at around -6 points (by my guesstimate) and perhaps another hump maybe around (+0.5). This suggests a large group of "super-responders." Improving by 6 points on the ADAS-Cog13 may even warrant some patients to say that they have been cured of Alzheimer's Disease after a year of treatment.
Perhaps there were some miracle stories that circulated, and perhaps that may have contributed to a massive placebo effect, and the face-value interpretation is true.
What does the FDA do with the possibility that some people may respond fantastically well to the medication, while others do not respond much at all? It could be said that most drugs work differently for different people and under different circumstances. Some cancer treatments work wonders for some people and do not help others at all, for instance.
In the original December 2022 press release, Anavex stated that Blarcamesine slowed cognitive decline by 45% as measured by the ADAS-Cog13; this statement seems to indicate that the typical medicated patient did decline in cognition, just at a rate much slower than those on placebo.
That gives greater credence to the theory that there were good number of super-responders. If the average person on the drug only slowed their decline and the average person improved by a massive 4 points, then that means there is a cluster of super-responders.
A human can only subtract so many points from their ADAS-Cog13 score; the score can only be as low as 0. If patients are starting off at an ADAS-Cog13 score of 21, could they roll back their score to 3 in a matter of 3 years (6 points per year)? This would make Blarcamesine a true miracle drug.
The CDR-SB
The CDR-SB is a measure of both cognition and behavior for patients suffering from neurological degeneration.
The CDR is obtained through semistructured interviews of patients and informants, and cognitive functioning is rated in 6 domains of functioning: memory, orientation, judgment and problem solving, community affairs, home and hobbies, and personal care.
So, unlike the ADCS, the CDR includes input from the patients themselves, not just from the caregivers. But there are no real "tests" or "exams" like those conducted for the ADAS.
In the September press release, Anavex states that those medicated with Blarcamesine slowed cognitive decline by .456 points on the CDR-SB over the course of only 48 weeks. This slowing of decline is certain at a P of .0175. Just how much is an approximately half-point slowing in CDR-SB? In one study,
CDR-SB levels of 0, 0.5, 1.0–2.5, 2.5–4.0, and =4.5 were defined as NC, SCD [Subjective Cognitive Decline], MCI [Mild Cognitive Impairment], very mild dementia (VMD), and dementia, respectively.
From the above categories, you can see that avoiding the addition of .5 points to your CDR-SB score can mean remaining at 0.5, in the Subjective Cognitive Decline group, rather than worsening into the Mild Cognitive Impairment category, at 1.0.
Or it may mean staying at 2.5 (Mild Cognitive Impairment) rather than worsening to a 3.0 and being solidly in the Very Mild Dementia category.
Let us consider that on the CDR-SB, there is only a 4.5-point difference between perfection and dementia. If it takes a person 20 years to go from 0 to 4.5, then Blarcamesine could on average give you an extra two years before reaching dementia. And that's a result of taking the drug for only 48 weeks.
What happens when the drug is taken for two or three years? How much is cognitive decline slowed? Again, the power of the drug seems to indicate that its effects are real and likely cumulative over the course of years.
Similar to the ADAS-Cog13 results, some of these patients must have actually improved their cognition and behavior, not just slowed their decline. We will have to see the final results to understand the degree of improvement on the CDR-SB scale.
Multiple Tests and Magnitude of Effect
Will the FDA find Anavex's findings persuasive? Here is some "draft guidance" published by the FDA; I could not find the "official" guidance, but it all makes sense to me:
Given the panoply of available neuropsychological tests, a pattern of putatively beneficial effects demonstrated across multiple individual tests would increase the persuasiveness of the finding; conversely, a finding on a single test unsupported by consistent findings on other tests would be less persuasive. A large magnitude of effect on sensitive measures of neuropsychological performance may also increase their persuasiveness.
Blarcamesine, based on the neuropsychological tests only, checks a couple of boxes. It demonstrates effects across two major Alzheimer's tests. It also demonstrates "a large magnitude of effect on sensitive measures" such as the ADAS and CDR.
Biomarkers
It all sounds too good to be true. Do we have any hard proof that this drug alters the actual biology and chemistry of the brains of patients?
My mother has Alzheimer's Disease. When I examined an MRI of her brain, it was obvious that she suffered from a degenerative brain ailment. Her brain had shrunk and was eroding in various places. At that moment, I knew for certain she had Alzheimer's Disease and there was no going back.
Blarcamesine can apparently slow down the pace at which the brain erodes and loses mass. "Brain volume" is the term that Anavex used in the September 2023 press release:
In addition, validated biomarkers of amyloid beta pathology, plasma Aß42/40 ratio increased significantly (P = 0.048), demonstrating strong anti-amyloid effects of blarcamesine in Alzheimer’s disease patients, while MRI revealed significant reduction in brain volume loss, including whole brain (P = 0.0005), comparing treatment to placebo.
For me, the important biomarker here is the slowing of brain loss, as there is no returning from that, and it absolutely directly affects brain activity and ability. The anti-amyloid effects are important too, as amyloid may cause dementia; research has not shown exactly how, though.
Perhaps just as importantly, the FDA has approved previous Alzheimer's Disease drugs for emergency approval based on their ability to reduce amyloid in the brain.
What is completely new about Blarcamesine is that it may be the first drug ever to demonstrate any reduction in brain volume loss.
Biomarker data has become more and more important, it seems, over the years. In essence, researchers can match the biomarker changes caused by disease and the biomarker amelioration caused by individual medications. If, for instance, a medication such as Blarcamesine reduces brain loss, researchers can investigate if Blarcamesine can be useful in treating various diseases that cause brain loss.
One question may be this: if brain loss is typical as we age, could Blarcamesine be a treatment for aging-related brain loss?
Processing the biomarker data apparently is one of the reasons for the slow publication of data from the 2b/3 trial.
According to Anavex, Blarcamesine may work by improving or correcting the body's own self-regulation system (autophagy). And if that is the case, Blarcamesine may be useful in treating a number of different diseases.
Two Successful Parkinson's Disease Trials
While Anavex initially proposed that Blarcamesine treat Parkinson's Disease Dementia, it turned out that Blarcamesine treats Parkinson's Disease in general, not just the dementia associated with it. In a placebo-controlled, double-blind, randomized Phase 2 study, researchers found the following improvements among those treated with the drug:
MDS-UPDRS[1] Total score improved significantly by -14.51 (p=0.034) for patients treated with ANAVEX®2-73 high oral once-daily dose compared to placebo. The improvement is clinically relevant corresponding to a relative improvement of 18.9% over 14 weeks.
Balanced and global improvements were observed within all MDS-UPDRS sub-scores Part I-IV:
• MDS-UPDRS Part I: 92.23% items improved (12 items out of 13)
• MDS-UPDRS Part II: 76.92% items improved (10 items out of 13)
• MDS-UPDRS Part III: 88.23% items improved (30 items out of 34)
• MDS-UPDRS Part IV: 71.42% items improved (5 items out of 7)
Considering that the standard of care for Parkinson's Disease only results in a slowing of the progression of disease, or perhaps only an amelioration of the symptoms, the result that, over the course of only 14 weeks, Blarcamesine significantly improved patients scores along the entire spectrum of the many symptoms of Parkinson's Disease, is kind of amazing.
The Phase 2 study also found changes in the targeted biomarker (SIGMAR1) from the use of Blarcamesine. The changes in SIGMAR1 levels in those who took the drug were statistically significant at a p of 0.035, and those changes in SIGMAR1 actually corresponded with improvements in performance on the MDRS-UPDS test.
An open-label extension study was scheduled, but Covid shutdowns postponed access to patients, causing a delay of 41 weeks between the original Phase 2 study and the extension study.
This long hiatus allowed researchers to see what happens when those treated by Blarcamesine were then denied access to Blarcamesine. As might be expected, they got worse, and so did those who took the placebo.
The "drug holiday" (time without Blarcamesine) involved 41 weeks during which only the "standard of care" was utilized for the Parkinson's patients, and under the "standard of care," which typically involves certain well-known Parkinson's drugs, patients got worse by 4.4 points as measured by one Parkinson's severity test.
Once the “drug holiday” was finished, all of the extension study patients took Blarcamesine. This included both those who had taken Blarcamesine in the original study and those who had taken the placebo in the original study.
From OLE baseline to week 48, investigators recorded a change of –2.25 (SE, 6.656) in MDS-UPDRS total score and change of –0.7 (SE, 0.309) in CGI-I. Additionally, at this time point, there were mean changes of –1.2 (SE, 0.537) in Montreal Cognitive Assessment and mean changes of –0.524 (SE, 0.620) in REM Sleep Behavior Disorder Screening Questionnaire. On specific MDS-UPDRS scores, patients on blarcamesine recorded mean changes of –3.95 (SE, 4.067) in Part III scores and mean changes of –2.20 (SE, 5.314) in part II and III scores combined.
To sum up, the patients in the OLE study all took Blarcamesine and they averaged a substantial improvement in the total MDS-UPDRS score; improvements in two cognitive test scores; and an improvement in REM sleep.
"Standard of care" caused these patients to worsen by 4.4 points over the course of 41 weeks, but the use of Blarcamesine over 48 weeks apparently caused the patients to claw back 2.25 points by improving.
The 2.25 points of improvement does not compare well to the nearly 11 points of improvement seen in the original 14-week study. The second study had far fewer participants, was done during Covid, and started 55 weeks after the original study, during which time the patients' condition generally worsened under "standard of care." So they started at a more difficult point in their disease progression.
Blarcamesine will now enter a pivotal Phase 3 trial, in which the primary endpoints will include Parts II and III of the MDS-UPDRS. Examining their results so far, Blarcamesine is reversing the effects of Parkinson's in these two sections, which are "Motor Experiences of Daily Living" and "Motor Examination" respectively. Comparing the means to the medians on these two sections, it seems that there are again some super-responders.
Is AVXL a BUY? And what catalysts lie ahead?
The phrase, "high risk, high reward," may apply here. But in this case, are the risks really that high? While Anavex has not given us all of the information, they may have given us enough.
The most near-term catalyst for the stock price is the upcoming press release regarding their Rett Syndrome trial for minors suffering from this rare disease; this will happen this year. After this trial data is published, Anavex, which has already completed what it says is a successful trial of Blarcamesine on adults with Rett, will meet with the FDA to discuss approval.
A Rett treatment that does not cause diarrhea and vomiting in this wheelchair-bound population (like the competition does) may be worth in the hundreds of millions.
The next catalyst is publication of a scientific journal article on the Phase 2b/3 Alzheimer's Disease trial. Publication will require highly qualified peer reviewers to analyze and critique Anavex's research and statistics before publication. Anavex will then meet with the FDA upon publication to discuss emergency approval for Blarcamesine for the treatment of Alzheimer's Disease.
A treatment for Alzheimer's Disease that does not involve brain swelling and brain bleeding like the kind Leqembi causes is worth the billions of dollars in sales per year. Just how many billions may depend simply on how much Anavex charges for the medication.
An estimated 6.7 million Americans struggle with Alzheimer's, requiring about $340 billion in unpaid caregiving each year. If one excludes some of those with Alzheimer's and includes some of those who have yet to be diagnosed with outright dementia, and if you go with around 6 million figure as the U.S. market for Blarcamesine; and if you assume that Anavex receives on average $2000 from each of those 6 million (some buy it for $8000 and most don't buy it at all, for instance); then you are looking at $12 billion in annual revenues. If you value Anavex at 1x revenues, then the stock price is $146.
For context, Biogen's Leqembi is priced at $26,500 per year and requires consistent brain MRIs for Leqembi-caused brain-bleed monitoring and regular in-office infusions.
That is my calculation for the value of the drug if it is only approved by the U.S. FDA. There are certainly many more millions of people suffering from Alzheimer's Disease in the rest of the world, representing tens of billions in revenue each year.
If the world can access Blarcamesine, millions will find some relief and joy, and the drug will act as a healthcare "inflation reduction act."
Wait. There's more. It may take another year before we find out the results of the Phase 3 trial for Parkinson's Disease, but it is a large market too, with few treatments. Around 10 million people worldwide are diagnosed with Parkinson's Disease and 10% of them are under the age of 50.
Between 500,000 and 1 million of the afflicted reside in the United States. If 750,000 people pay an average of $2000 per year for Blarcamesine, that's a $1.5 billion revenue stream. The "rest of world" revenues can be counted in the billions.
Add the other treatments Anavex has in the pipeline, including Rett, schizophrenia, and Fragile X, and the "reward" side of the balance gets only heavier.
I foresee FDA approval of Blarcamesine for Rett Syndrome treatment in 2024, and I also foresee FDA emergency approval for Blarcamesine for Alzheimer's treatment in 2024, even if the FDA requests another drug trial. I also believe the pivotal Phase 3 trial for Blarcamesine as a Parkinson's treatment will be completed in 2024.
WILDCARD: Anavex is talking with larger pharmaceutical companies to work together (perhaps as a joint venture) to market Blarcamesine. This would spike the stock price immediately. I expect some comment on this during the Q3 conference call.
THE RISK is that Blarcamesine does not work at all. However, as evidenced by the cratering of Anavex's market cap to around $500 million, this has largely been priced in.
Addendum 1: Anavex's Current Chief Statistician
When, in December 2022, Anavex first published some of the data from their Alzheimer's Phase 2b/3 study, some writers claimed that Anavex's statistics were wrong. Their claims were inaccurate, but because there was no political support for Anavex (the FDA was working secretly with competitor Biogen, and the Alzheimer's Association had openly thrown their support behind Biogen), the stock was destroyed.
Since that day, all of Anavex's data has been questioned.
Part of this was Anavex's fault, as they did not release some of the data that investors hoped to see, such as the ADCS data.
In March 2023, Anavex hired a former FDA top-level statistician as the Vice-President and chief of biostatistics. According to Anavex, these are his qualifications:
Dr. [Kun] Jin will draw on his extensive experience, including recently as the Statistical Team Leader at the U.S. Food and Drug Administration (FDA). Dr. Jin provided statistical review coverage and expertise for neurological drug products for the Center for Drug Evaluation and Research (CDER), and performed timely and quality reviews of marketing applications, including New Drug Applications (NDA), Biologic License Applications (BLA), and Investigational New Drug (IND) applications. Under the leadership of Dr. Jin, the neuropharmacological statistical team has completed several hundred statistical reviews of NDAs, BLAs, and efficacy supplements.
The September press release declaring again the AD trial a "success" was published six months after the hiring of Dr. Jin, so if there is any question about the veracity of the "successful" Phase 2b/3 trial, it is on Dr. Jin's name. And it is up to you, the individual investor, to decide whether you trust Dr. Jin or if you trust certain writers who claimed that Anavex could not even subtract numbers accurately.
My hunch is that Dr. Jin will not risk his reputation by writing a press release that does not meet strict statistical standards. So when Anavex says that a p<.025 on both the ADAS-Cog and the CDR-SB combined with statistically significant findings in two important biomarkers spells a successful trial, I expect it to be true.
Addendum 2: Use of Least Squares Mean
In their September news release, the company stated that they were using the Least Squares Mean method of figuring the differences between treatment and placebo groups, from baseline to end of treatment.
The Least Squares Mean has a number of advantages, according to statisticians, in more accurately discerning the actual effects of a drug on trial. And in fact, Biogen and Eisai utilized the same method when they published the results of the successful Leqembi Phase 3 drug trial in a major medical journal.
Leqembi has been approved by the FDA in both an emergency-use capacity and also in a final decision for approval.
The Lease Squares Mean is the best computation for drug trials because it can take into account missing information, such as patients who did not complete the study or who had missing measurements.
