found_it2
18 hours ago
"Anavex: Looking At Different Statistical Approaches Before EMA Opinion"
Feb. 14, 2025 6:02 PM ET
Seeking Alpha
Avisol Capital Partners, Investing Group Leader
https://seekingalpha.com/article/4758654-anavex-looking-at-different-statistical-approaches-before-ema-opinion
(It's a long one folks, but pretty comprehensive reporting....positive in the conclusion)
Summary
Anavex Life Sciences Corp.'s stock surged over 50% due to EMA's acceptance of blarcamesine for Alzheimer's review, marking a pivotal moment for the company.
Despite mixed trial results, blarcamesine showed significant cognitive improvements, meeting key endpoints, though it failed in functional assessments.
Financially, Anavex has a market cap of $720M and a cash runway of 8–10 quarters, with ongoing R&D expenses.
Given the EMA review's potential impact, I am considering a small investment, though not right now, acknowledging AVXL stock's high volatility and associated risks.
We are Avisol Capital Partners, a team of medical/biotech experts and finance professionals. We lead the investing group Total Pharma Tracker, where we aim to make the science of biopharma investing easily understandable to regular investors.
I covered Anavex Life Sciences Corp. (NASDAQ:AVXL) in December, after which, I note with some surprise (given the stock’s generally lackluster performance last year), that the stock jumped over 50% between late December and early January. The basis for that surprise positive performance was the company’s announcement that the European Medicines Agency (“EMA”) has accepted for review the application for blarcamesine, the company’s Alzheimer's disease (“AD”) drug candidate. The review period is 7 months, or July 2025.
This is major good news for the beleaguered company, which has been toiling with trial data and confusing market perception for nearly a decade. An even more major news would be an FDA review, but the EMA is a good second place.
Background
In my last coverage, I noted how the company “started” an MAA (the European application for approval over there) in January 2024, and submitted the application in November. It was accepted in December, and this I called a “precipitous” moment. This is because AVXL has a history of confusing and/or poor data in multiple targeted indications.
In the world of small-cap biotech, there is, arguably, no other company which grabs so much investor attention — both bullish and bearish — over its every announcement and every real or perceived problem. All of that long history comes to a head with a regulatory event like an MAA approval, and acceptance of an application is the first unchangeable step towards such clarity. There is no turning back from here. This will end in either an approval or a rejection — there can be no confusion about that.
The MAA was based on a Phase 2a and a Phase 2b/3 trial in Alzheimer’s. The phase 2a trial had two co-primary endpoints, in one of which the company succeeded, and in the other, it failed. According to a particular statistical analysis which I discussed at length earlier, such an outcome has to be considered an overall trial failure — so the phase 2a trial failed.
As to the phase 2b/3 trials, the data announcement was fraught with many errors, including basic Math errors and so on which, when corrected, could have an impact on the endpoint analysis. While the company said that the trial met the primary endpoint of improvements in cognitive function as measured by ADAS-Cog (and ADCS-ADL, see below), these errors I alluded to make the claims confusing. The company claims, among other things, that a) as to efficacy, they have numerical clinical efficacy compared to approved therapies, b) as to safety, the treatment does not need MRI monitoring and is safe and well-tolerated, and c), as to dosing, blarcamesine is an oral small molecule with a convenient route of administration. The CHMP will be the final arbiter of these claims. Recall that both trials had two co-primary endpoints and in both trials, the molecule failed the ADCS-ADL endpoint while succeeding with the ADAS-Cog endpoint.
Alzheimer’s: what clinical trials measure
There are three basic impairments in Alzheimer’s disease — cognition, function, and behavior. If you draw a line with these three metrics, and you mark the 6 stages of AD researchers have identified, you will see there is a progression in the stages according to those three impairments. Thus, preclinical AD, or stage 1/2 AD, does not produce obvious impairments in any of these three metrics, but can be detected through biomarkers in a research setting. Stage 3 is MCI (Mild Cognitive Impairment) due to AD, where the first early signs of cognitive impairment occur, but the patient can compensate for functional and behavioral impairments. In Stage 4, which is early-stage AD with mild dementia, patients can still compensate for functional and behavioral impairments as long as their cognitive decline is mild. Therefore, it is more difficult to measure functional impairment in early-stage AD than it is to measure cognitive decline.
Blarcamesine targets early Alzheimer’s. Its clinical trials have used both the ADAS-Cog13 and the ADCS-ADL scores. Now, the ADAS-Cog13 score covers all cognitive areas in dementia, while the ADCS-ADL is designed to evaluate daily functioning. It considers both BADL and iADL, or basic activities of daily living and instrumental activities of daily living. However, since there is progression from cognitive to functional/behavioral decline in AD patients through the 6 stages of AD, functional scoring may not be suitable to detect early AD. As research says:
Patients with early AD have subtle cognitive deficits and do not present with functional impairment; patients who are closer to the onset of dementia may have noticeable functional deficits that progress slowly, creating sensitivity issues with currently available scales. Furthermore, the extent to which an individual compensates for cognitive deficits and adjusts for daily activities is very variable. It is thus challenging to establish a clinically meaningful effect during a trial of reasonable duration.
In its draft guidance for trials in early AD, the FDA notes the following:
Historically, studies to support approval for drugs in the overt dementia stages of AD (Stages 4 through 6) have used an approach which required the assessment of both cognitive and functional (or global) measures as co-primary endpoints. The co-primary endpoint approach was used, in part, because the cognitive assessments used in the studies were not considered inherently clinically meaningful.
Recall that the AVXL phase 2b/3 study used this co-primary endpoint approach, although the trial was in early AD patients.
Thus, as the FDA concludes:
Additionally, many of the assessment tools typically used to measure functional impairment in patients with later dementia stages of AD (Stages 4 through 6) would not be sensitive to detect subtle functional changes in early AD. Therefore, FDA may consider other approaches, including endpoints based on cognitive assessments or surrogate endpoints, which may allow for shorter trial durations as a basis for approval in the earliest stages of AD (i.e., Stages 1, 2, and early 3).
It is under this light that Blarcamesine’s failure to succeed with ADCS-ADL despite consistently doing well with ADAS-Cog13 in a pair of trials which ultimately aims to treat early AD patients becomes interesting. Despite my complete neutrality to the debate surrounding this company, the above discussion, I am sure, will produce some doubts in honest AVXL bears. Maybe functional assessment is not a metric where Blarcamesine can succeed with early AD patients, despite having visible treatment effect in halting cognitive decline? That is a debate only the EMA and the FDA can settle.
Multiple statistical approaches to the same data
Thus, as we see in recently published data from the phase 2b/3 study:
Among 462 randomized participants in the intent-to-treat population (mean age, 73.7 years; 225 [48.7%] women), 338 (73.2%) completed the trial. The co-primary outcome was met under the multiplicity control rule, since the differences in the least-squares mean (LSM) change from baseline to 48 weeks between the prespecified blarcamesine and placebo groups for ADAS-Cog13 was significant at a level of P < 0.025 and for CDR-SB was significant at a level of P < 0.025, while ADCS-ADL did not reach significance at Week 48 (ADAS-Cog13 difference of -2.027 [95% CI -3.522 to -0.533]; P = 0.008; CDR-SB difference of -0.483 [95% CI -0.853 to -0.114]; P = 0.010; ADCS-ADL difference of 0.775 [95%CI -0.874 to 2.423]; P = 0.357).
The ADCS-ADL endpoint was not met, while the ADAS-Cog13 endpoint was met with statistical significance (P < 0.025). Now, the company also brings in a third endpoint, CDR-SB (Clinical Dementia Rating – Sum of Boxes), and invokes the multiplicity control rule. There are various types of multiplicity adjustments, and in prior discussions, I have cited two of the more conservative ones which are designed to reduce Type I errors (false positives). These are Bonferroni and Hochberg. While Hochberg's is less conservative, it can still preserve control of Type I errors while allowing some non-significant results. Using these conservative tests, the phase 2b/3 trial does not become successful.
However, if we use a less conservative Hierarchical Testing (Gatekeeping), we get to order the endpoints by their degree of significance. Here, the common-sense concept we have discussed about the relative importance of cognitive metrics over functional ones in early AD becomes relevant. Thus, if we order the 3 endpoints in the order of importance for early AD, we develop the following scenario:
1. Pre-specified Order: Let us assume the endpoints were ordered as follows given the early stage of the disease:
1st: ADAS-Cog13
2nd: CDR-SB
3rd: ADCS-ADL
2.Now, using Sequential Testing:
ADAS-Cog13: Significant (P = 0.008 < 0.025) ? Allows testing of the next endpoint.
CDR-SB: Significant (P = 0.010 < 0.025) ? Allows testing of the next endpoint.
ADCS-ADL: Not significant (P = 0.357 > 0.025).
So we can say that since the first two endpoints (ADAS-Cog13 and CDR-SB) were significant, the trial is considered successful, demonstrating efficacy on cognitive and clinical outcomes. Although the ADCS-ADL results were non-significant, they do not invalidate the positive findings for the other two endpoints, as they were tested in sequence and met their significance levels. This is just a big worded scientific way of saying that in early AD patients, detecting cognitive decline gets precedence over detecting functional decline, and hence a trial which halts cognitive impairment is successful.
Financials
AVXL released earnings recently, and their market cap is now $720mn, while the cash balance is $120mn. Research and development expenses were $10.4 million for the quarter, while general and administrative expenses were $3.1 million. That $13.5 per quarter cash burn gives them an 8 quarter cash runway.
Looking at an FCF model, we have CFO of ($12.1mn) while there was no capital expenditure, which now gives them a 10 quarter runway.
Risks
AVXL has always been a super risky stock whose price movement follows market sentiment like no other. An EMA approval will take this stock way up beyond what logic should indicate, while any indication of rejection, even if tenuous and temporary, will similarly destroy it more than it may probably deserve. I am very wary of investing in such stocks.
Bottom line
It is tempting to take a small position in Anavex Life Sciences Corp. based on the EMA outcome later this year. The stock is down considerably from its latest peaks, and while it was way more down in October, the MAA filing in November has derisked it from October, so the added expense may be worth it. I am strongly considering taking a position, albeit a very small one.