2023-08-10 14:45:13 ET
Summary
- Cassava Sciences' core AD treatment candidate simufilam slowed cognitive decline compared to placebo (albeit non-significantly) and stabilized cognition over 18 months in mild patients, which is unprecedented compared to historical data.
- However, I am concerned about lack of statistical significance, red flags in SAVA reports of top-line results, questionable validity of comparisons with historical data, ongoing litigations and limited cash runaway.
- Despite the promising data, I give SAVA a "Hold" recommendation considering uncertainty about the outcome of phase 3 studies, several red flags, and likely absence of short-term catalysts.
Editor's note: Seeking Alpha is proud to welcome Biotech Pharma Investor as a new contributor. It's easy to become a Seeking Alpha contributor and earn money for your best investment ideas. Active contributors also get free access to SA Premium. Click here to find out more »
Investment thesis
Cassava Sciences ( SAVA ) is a high-risk, high-reward speculative investment. SAVA develops simufilam, a novel treatment for Alzheimer's disease ((AD)) with promising phase 2 results and two ongoing phase 3 studies which are expected to complete enrolment by the end of 2023 (top-line results late 2024/ early 2025). Although I tend to be bullish on SAVA based on phase 2 results, I give it a "Hold" recommendation because of the high risk, some red flags, and potentially better entry points due to lack of short-term catalysts. A small "watchlist" position may be worth it, but invest only what you can afford to lose.
Overview of SAVA's pipeline
SAVA has two assets under development; a therapeutic product candidate, called simufilam (a novel oral treatment for AD) and a diagnostic product candidate, called SavaDx (for diagnosis of AD).
Simufilam is a small molecule that restores the normal shape and function of altered filamin A ((FLNA)) in the brain, thus preventing aberrant association of altered FLNA that result in neuronal dysfunction/ degeneration and neuroinflammation (additional insight in the mechanism of action in this recent publication). SavaDx is currently designed as an antibody-based detection system for altered FLNA. However, as stated by SAVA, SavaDx is an early-stage, low-priority program. Therefore, the rest of the article focuses on simufilam.
The phase 2 study design
Understanding the phase 2 study design is necessary to be able to correctly interpret the results. One can find details about the design in clinicaltrials.gov and 10-Q of the company. History of changes in the protocol can also be found in clinicaltrials.gov .
The study, enrolling mild-to-moderate (MMSE 16-26) AD patients, includes 3 phases:
- A 12-month open-label phase during which all patients received simufilam (top-line results reported on January 2023 ).
- A 6-month double-blind randomized placebo-controlled phase. Patients completing the prior phase (i.e., 12-month open label simufilam) were randomized 1:1 to either placebo or simufilam for 6 months (topline results reported on July 5, 2023 ).
- Another 6-month open-label extension following completion of the CMS. Considering that SAVA announced completion of the CMS study on May 2023 this last phase should be completed by November this year.
It is very important to highlight that the CMS study is not the typical placebo-controlled RCT, because all patients were on simufilam for 12 months before being randomized to either continuing simufilam or being withdrawn from simufilam (placebo arm), hence the characterization of CMS as a "randomized withdrawal trial". As reported in SAVA's 10-Q "The CMS study design is intended to evaluate simufilam's effects on cognition and health outcomes in Alzheimer's patients who continue with drug treatment versus patients who discontinue drug treatment."
Overview of results of the open label part
Results (Table 1) from the 12-month open label simufilam treatment demonstrate much slower cognitive decline than expected for patients with similar baseline AD severity (as quantified by baseline ADAS-cog score). However, this result is driven by the mild (MMSE 21-26) AD subgroup, while effect in the moderate (MMSE 16-20) AD subgroup appears to be limited (if any).
Table 1 : Summary of results from 12-month open-label study
mean baseline ADAS-cog | mean ADAS-cog at 12 months | change at 12 months (95% CI) | expected change at 12 months (95% CI) 1 | mean baseline MMSE | expected MMSE 2 | |
Total patients (n=216 3 ) | 19.1 | 19.6 | 0.5 (-1.7 to 2.7) | 4.6 (4.0-5.2) | 21.5 | 22.6 |
Mild patient (n=133 3 ) | 15 | 12.6 | -2.4 (-4.1 to -0.7) | 3.9(3.3-4.5) | Not reported | 24.8 |
Moderate patients (n=83 3 ) | 25.7 | 30.1 | 4.4 (0.9-7.7) | 5.6(5.0-6.2) | Not reported | 19.0 |
Data extracted from SAVA press release. The higher the ADAS-cog score the higher the cognitive impairment. 1 Based on the formula by Ito et al ; 5.49*(baseline ADAS-cog/25)0.669. 2 Based on the formula by Ito et al; ADAS-cog=60.9-1.85*MMSE. 3 These numbers represent the "full analysis set", i.e., includes drop-outs / protocol deviations. However, it is unclear if the reported ADAS-cog scores represent the "full analysis set" given discrepancies between numbers reported in the text and ADAS-cog changes depicted in the figures (discussed later).
Overview of results of CMS study
The CMS study shows more cognitive decline in placebo vs simufilam-treated patients, but the difference was not statistically significant. Results were especially promising in mild AD patients, with simufilam-treated patients showing improved cognition, while placebo patients declined, but the difference was not statistically significant in the mild subgroup either. Finally, simufilam stabilized cognition in mild AD patients over 18 months, which is unprecedented compared to historical placebo groups of prior studies (Figure 1).
Below (Table 2) is a summary of CMS topline results. Of note SAVA did not report results for "moderate" AD patients. However, these can be calculated based on results for the total population and the mild AD subgroup based on the formula for calculating the mean of two means: ?= (?1*n1 + ?2*n2)/(n1+n2), where ? is the mean of the total population, ?1 is the mean of the mild AD subgroup, n1 is the number of mild AD patients, ?2 is the mean of the "moderate" AD subgroup, and n2 is the number of the "moderate" (total minus mild) patients.
Table 2 : Summary of data from 6-month CMS study
mean ADAS-cog from 12 to 18 months (change) | simufilam-placebo (95% CI, p value 1 ) | expected change at 6 months (95% CI) 2 | mean baseline MMSE | expected MMSE 3 | |
Total patients (n=155) | 20.6 -> 21.8 (+1.2) | 2.4 (1.8-3.0) | 18.4 | 21.8 | |
Simufilam arm (n=78) | 19.3 -> 20.2 (+0.9) | -0.6 (- 2.1 to 1.0, p=0.44) | 2.3(1.7-2.9) | 18.6 | 22.5 |
Placebo arm (n=77) | 21.9->23.4 (+1.5) | 2.5 (1.9-3.1) | 18.1 | 21.1 | |
Mild patients (n=76) | 11.09->11.06 (-0.03) | 1.6(1.0-2.2) | 24.0 | 26.9 | |
Simufilam arm (n=40) | 11.0 -> 10.4 (-0.6) | -1.1 4 (- 2.6 to 0.4, p=0.15) | 1.6(1.0-2.2) | 24.0 | 27.0 |
Placebo arm (n=36) | 11.2 -> 11.8 (+0.6) | 1.6(1.0-2.2) | 24.1 | 26.9 | |
Moderate patients (n=79) 4 | 29.7 -> 32.1 (+2.4) | 3.1(2.5-3.7) | 12.9 | 16.9 | |
Simufilam arm (n=38) | 28.0 -> 30.5 (+2.5) | 0.2 | 3.0(2.4-3.6) | 12.9 | 17.8 |
Placebo arm (n=41) | 31.3 -> 31.6 (+2.3) | 3.2(2.6-3.8) | 12.8 | 16.0 |
1 SAVA PR does not report p values, but these can be estimated from 95% confidence intervals based on the method described here . 2 Based on the formula derived by Ito et al; 5.49*(baseline ADAS-cog/25)0.669, divided by 2 considering 6 (vs 12) months study duration. 3 Based on the formula derived by Ito et al; ADAS-cog=60.9-1.85*MMSE. 4 Based on -0.6 change in simufilam arm and +0.6 in placebo arm the difference seems to be -1.2. However, this discordance likely has to do with reported numbers being rounded.
Notable is that decline in placebo groups during the 6-month CMS study was much lower than expected, especially for mild AD patients. As explained in this paper, "While it is important to compare the treatment group versus placebo group obtained within the double-blind study, evaluation of the placebo control arm data relative to response for historical placebo response provides an additional mechanism to evaluate a study's results, and to guard against Type I or false positive error. If the placebo response is different from the expected historical placebo response, and the overall disease progression is quantitatively discordant with the model predicted values after adjustment for baseline disease severity, then it is important to investigate the reasons for the placebo difference prior to interpreting treatment effects".
There are 2 potential explanations for the slower than expected cognitive decline in placebo; (1) The bullish explanation: Simufilam might have a disease modifying activity with beneficial effects persisting for some time even after withdrawal of the drug. Therefore, the 6-month duration of the CMS study was insufficient to demonstrate separation between placebo and simufilam. As stated by SAVA; "It is not known how long a washout period may be needed to remove lingering drug effects, if any, from prior treatment with open-label simufilam for 12 months". "The placebo group in the CMS has started to decline again but continues to maintain benefit over historical placebo groups.". (2) The bearish explanation: There is no lingering effect and patients enrolled in the phase 2 study are not comparable to historical placebo data. Considering simufilam's mechanism of action and impressive results so far in mild patients, I tend to favor the first explanation. However, no one really knows pending results from phase 3 trials.
What about lack of statistical significance in CMS study and what does it mean for the ongoing phase 3 studies?
It has been pointed out repeatedly, especially by SAVA bears, that results of the CMS study were non-significant. However, I must emphasize that readers should be careful about how to interpret p values. As explained in this publication; "Lack of statistical significance does not mean that no difference exists; this is one of the most basic misinterpretations of significance testing". In my opinion you should consider the following when interpreting the statistical significance of the CMS results; (1) The CMS study was NOT powered (small study size) to show statistical significance between simufilam and placebo. (2) Not only was the CMS study relatively small, but potential lingering effects of simufilam after withdrawal combined with the short (6 months) study duration could have blunted differences between placebo and simufilam.
Nevertheless, lack of statistical significance increases uncertainty about outcome in phase 3 trials. Will phase 3 studies be powered to detect a difference between placebo and simufilam? One of the phase 3 is aiming to enroll n=750 patients. This sample should be powered enough (80%) to detect a statistically significant difference between placebo and simufilam of -0.71 (estimated based on this calculator. Variance was estimated as 23.85 from confidence intervals of CMS study based on this guidance). Note that assessment of ADAS-cog change in phase 3 will be done at 12 months of treatment (vs 6 months in the CMS study), therefore separation between placebo and simufilam should be clearer at that time point if simufilam is indeed effective. Also note that phase 3 trials have another primary endpoint; ADCS-ADL. As reported in 10-Q it is a "co-primary" endpoint, which means that efficacy in any of the 2 endpoints should be sufficient (i.e., SAVA does not need to show efficacy in both endpoints).
SAVA's results vs historical placebo and prior phase 3 trials; a valid comparison?
In my view the bullish thesis on SAVA is currently predominantly based on indirect comparisons with historical placebo controls (see Figure 2 here ). Therefore, I believe it is very important to understand what the expected cognitive decline would be without treatment. Importantly, the slope of cognitive decline is dependent on baseline cognitive impairment (the higher the baseline impairment the higher the rate of decline). Ito et al based on a meta-analysis of then available data derived the following formula; 5.49*(baseline ADAS-cog/25)0.669, which calculates expected decline (change in ADAS-cog score) over 12 months, adjusted for baseline ADAS-cog, in untreated patients or patients on stable AChEI dose (considering lack of disease modifying effect, the slope of decline is the same following the time point of max symptomatic relief by AChEI). Patients participating in SAVA's ph2 study were allowed to take memantine, rivastigmine, galantamine or an AChEI, as long as they were receiving a stable treatment regimen for at least 90 days before screening. According to Ito et al, time to reaching maximum symptomatic relief ranges from 20 days (donepezil) to 131 days (rivastigmine) and 182 days (galantamine), i.e. might have been enrolled before reaching maximum benefit from prior treatments. Details about background memantine/AChEI treatment of participants in the OL study are not available. However, this is unlikely to account for the impressive results seen in mild patients.
SAVA bulls also keep reproducing various figures (like this one) comparing simufilam's results to other AD drugs recently approved or pending approval. However, I want to emphasize that in my opinion these figures have serious limitations; (1) They do not account for baseline ADAS-cog scores (the higher the baseline the higher the expected decline). See table below. (2) In SAVA's phase 2 trial ADAS-cog11 (range 0-70) was used vs ADAS-cog13 (range 0-85) in aducanumab trials ("EMERGE" and "ENGAGE") and donanemab trial ("TRAILBLAZER-ALZ 2"), and ADAS-cog14 (range 0-90) in lecanemab ("Clarity AD") trial. (3) In aducanumab, lecanemab and donanemab trials, reported results refer to all randomized patients which corresponds to the "Full Analysis Set" in SAVA's ph2 topline results. The various circulating figures appear to have missed this important detail, therefore overestimating SAVA's results. (4) Plotting OL and CMS data as a continuous line has pitfalls because only part of OL patients (most likely those that were doing better in my opinion) proceeded to the CMS study.
Table 3 ; Baseline AD severity in SAVA's ph2 trial vs ph3 trials of aducanumab, lecanemab and donanemab.
AD severity | Baseline ADAS-Cog | Change at 18 months | |
SAVA phase 2 (simufilam-treated arm) | mild-to-moderate (MMSE 16-26) | 19.1 | 2.44* |
SAVA phase 2 mild subgroup (simufilam-treated arm) | MMSE 21-26 | 15.0 | 0.11 # |
EMERGE (high-dose arm) | MCI or mild (MMSE 24-30) | 22.3 | 3.76 |
ENGAGE (high-dose arm) | MCI or mild (MMSE 24-30) | 22.4 | 4.55 |
Clarity AD | MCI or mild (MMSE 22-30) | 24.5 | 4.14 |
TRAILBLAZER-ALZ 2 (low-medium tau population) | MCI or mild (MMSE 20-28) | 27.4 | 3.17 |
* =1.54+0.9. Mean Adas-cog change (+ 1.54) during the first 12 months is extracted from Figure 1 from here , which I believe represents the Full Analysis Set. # Based on Figure 1 from here .
Based on this table I can make the following observations; (1) Despite enrolling patients with lower (worse) MMSE scores, baseline cognitive impairment (baseline ADAS-cog) in SAVA mild-to-moderate AD patients appears to be similar or lower compared to other trials which enrolled MCI/mild AD patients. This is a discrepancy, because based on worse baseline MMSE scores I would expect worse baseline ADAS-cog scores in SAVA trial.(2) "Mild" SAVA patients appear to have much less serious cognitive impairment (lower ADAS-cog) at baseline compared to "MCI/mild" patients enrolled in other trials. Although I have to acknowledge the different ADAS-cog versions used in each trial, I doubt the different scales fully explain the discrepancies in baseline scores. Despite these limitations, I believe SAVA results are comparably good. Specifically, the stabilization of ADAS-cog in simufilam-treated mild patients over 18 months is impressive, but the lack of a placebo control group for the same period is a major limitation.
Red flags about SAVA data
The following issues are concerning.
Selective reporting: In the press release for the 12-month open label study SAVA reports results for both mild and moderate subgroups of patients. On the contrary, SAVA reports results only for the mild subgroup in the press release for the CMS study. As explained in the legend of Table 2, it is possible to calculate results for the "moderate" patients and you can see in the Table that these are not good. In my opinion this is a major red flag.
Misleading reporting: SAVA reported that "simufilam treatment for 6 months slowed cognitive decline > 200% compared to placebo in mild Alzheimer's disease". While this percentage is technically correct, it is misleading as explained here . Let me explain this; assume the following hypothetical ADAS-cog changes for simufilam vs placebo; -1 vs +1, -0.5 vs +0.5, -0.1 vs +0.1, -0.001 vs +0.001, -x vs +x. In all these hypothetical examples the relative improvement compared to placebo is 200%; [x-(-x)]/x=2. In other words the absolute difference is more clinically relevant. As I hope is clear from the examples above, a big relative change may be clinically meaningless (no perceptible benefit for the patients or caregivers) if the absolute difference is small. Nevertheless, I consider this to be a minor red flag.
Some numbers in the CMS study don't add up.
- In the open label study, there were n=133 (62%) mild AD patients and n=83 (38%) moderate patients. However, in the CMS study there were n=76 (49%) mild patients and n=79 (51%) moderate patients. In other words, the phase 2 study started with 62% of the patients being mild, while in the CMS part only 49% of the patients had mild AD. There are several speculative explanations about this, but I am not going through them here since it's just speculations. It would be much better if SAVA was more transparent about enrolment details.
- In the open label study, mild AD patients entered the study with a mean ADAS-cog score of 15 and ended the 12-month period with a mean ADAS-cog of 12.6. However, mild AD patients enrolled in the CMS study had a baseline ADAS-cog of 11.1 (i.e., quite lower).
- As has been highlighted by Seeking Alpha author Lane Simonian , there appears to be a discordance between some ADAS-cog scores and respective MMSE scores. As explained in the meta-analysis by Ito et al, there is good correlation between ADAS-cog and MMSE, which can be expressed by the following formula: ADAS-cog=60.9-1.85*MMSE. Indeed, mean MMSE score at enrolment in the open label study was 21.5, which is very close to the expected 22.6 (Table 1). However, mean MMSE scores at enrollment in CMS study are far from what expected (Table 2). Of course, it is irrational to expect perfect performance of the above formula, and the smaller the study size the larger the possibility of discordance. However, discordances (especially in moderate patients) are too large to ignore. Considering enrolment criteria (MMSE 16-26) the calculated mean MMSE score of 12.9 for the moderate group at enrolment in the CMS study is improbable. The most reasonable explanation is that the reported MMSE scores for the full study population are either wrong or refer to moderate patients. Given history SAVA history with mistakes in publications .
Discrepancies between ADAS-cog changes reported in the text and ADAS-cog changes depicted in figures (open-label topline PR); In text "ADAS-Cog11 scores changed from 19.1 (±9.2) to 19.6 (±13.3)" (which corresponds to a change of +0.5), yet in Figure 1 the change is depicted as +1.54. Same discrepancy for mild patients; change "from 15.0 (±6.3) to 12.6 (±7.8)" (which corresponds to a change of -2.4) in the text vs -0.73 in Figure 2. Such discrepancies could represent full analysis set in figures vs per-protocol data in the text (however, this is just a speculation, since it's not clarified in the PR).
Post-hoc analyses can be manipulated: Based on available evidence (clinicaltrials.gov protocols and 10-Q) the subgroup analysis on mild AD patients does not seem to have been predefined. Nevertheless, it's a standard subgroup to look at in AD trials and SAVA seems to be consistent on the definition of mild AD based on MMSE score.
To be fair, as SAVA clarifies: "'Top-line data' is a summary of the clinical data prior to the completion of a full and final audit or quality-control of the clinical database... Final data may change from today's top-line data." Also, a press release is not a paper and should not be judged as such (i.e., it is reasonable not being too detailed in a press release).
Brief overview of research fraud allegations
Another red flag about SAVA are ongoing litigations regarding potential research fraud with several experts questioning SAVA's data. Notably, 5 publications (none related to simufilam) have been retracted , but all from PLOS One journal. On the other hand, despite allegations, at least 5 other journals ( Journal of Prevention of Alzheimer's Disease , Journal of Neuroscience , Neuroscience , Neurobiology of Aging , The Journal of Clinical Investigations ) failed to find convincing evidence of misconduct and none of the simufilam papers have been retracted. Nevertheless, both Neurobiology of Aging and Journal of Neuroscience have expressed concern pending results of ongoing investigation by academic authorities at The City University of New York.
From a different viewpoint , one should take into account financial conflicts of "whistleblowers" and there is reasonable concern that allegations against SAVA were made for the purposes of stock price manipulation, with at least 2 cases proven to have profited from shorting SAVA. Personally, I would consider SAVA innocent until proven guilty, as I am not aware of convincing evidence of misconduct, but I can't say I am not concerned. Notably, SAVA has filed a lawsuit against the "Short and Distort" Campaign, which could be an important catalyst if successful.
Simufilam's potential advantages over currently approved treatments
Based on its mechanism of action simufilam is expected to have a neuroprotective effect, thus preventing/delaying further cognitive decline. As discussed above, I believe Phase 2 results are compatible with this hypothesis. If such disease-modifying activity is proven, combined with the very good so far safety profile, this would be a major commercial advantage over current competition and would significantly improve the market share that SAVA can capture.
As explained in an UpToDate article , there is no convincing evidence that cholinesterase inhibitors are neuroprotective or have the ability to alter the underlying disease trajectory. Memantine may have a neuroprotective effect based on its mechanism of action, but both cholinesterase inhibitors and memantine result in only " clinically marginal improvement ". Furthermore, there is little, if any, evidence that patients with milder AD benefit from memantine. Similarly, recently approved aducanumab and lecanemab are associated with only small beneficial effects (imperceptible to a majority of patients and caregivers), substantial harms (most notable risk of intracranial bleeding) and unprecedented costs.
Finally, given simufilam's different mechanism of action there might be potential for combination therapy with already approved treatments. In other words, I expect simufilam to be used as an add-on to other treatments rather than compete with them. Notably, in ongoing phase 3 trials, patients on background AD medications are allowed.
SAVA's target market and market share potential
According to the Alzheimer's Association 2023 report an estimated 6.7 million Americans aged 65 and older have AD (projected to steadily increase to 8.5 million by 2030, 11.2 million by 2040, 12.7 million by 2050, and 13.8 million by 2060). About half of those are estimated to have mild AD (corresponding to more than 3m patients), 30% moderate AD and 19% severe AD. Additionally, about 5-7m of Americans aged 65 and older are estimated to have mild cognitive impairment ((MCI)). Globally , more than 50m people are estimated to have AD. Furthermore, with new and more sensitive diagnostic criteria being adopted, accurate and earlier diagnosis of AD is likely, resulting in an even larger target market in the future.
Market share of simufilam and associated income for SAVA will depend on several (unpredictable) factors:
- Efficacy in mild only (target population in the US; 3.3m) vs both mild and moderate patients (target population in the US; 5.4m). In my opinion, and based on available phase 2 data, I expect simufilam to be effective only in mild patients.
- Safety profile in phase 3 studies; The much larger phase 3 studies may identify side effects that the smaller phase 2 studies were underpowered to detect (but I find this unlikely).
- Symptomatic vs disease-modifying activity (discussed above).
- Competition and timing to reaching commercialization; e.g. phase 3 results from BIVI are expected in October 2023 and if positive could reach commercialization much faster than simufilam.
- pricing and insurance coverage. Pricing (and competitiveness) of simufilam will depend on all above factors. For comparison, estimated annual costs of currently approved treatments are: memantine $936, galantamine $1900, donepezil $2000 ($7900 for Aricept), rivastigmine $5700, lecanemab $26,500 , aduhelm $28,000 .
- Commercialization and marketing by SAVA alone vs with partners (probably the safest option for SAVA and SAVA investors). The structure (upfront payment, developmental-commercial milestone payments, royalties) of such a partnership would depend on numerous factors, including above considerations, timing (before/after phase 3 results or before/after FDA approval) and geography (US vs ex-US). Average royalty rates at the ph3/ pre-approval stage are in the range of 10-20%, but there is considerable variation.
Assuming an annual cost in the low-to-mid-range (and this is conservative in my view- BIVI for example is predicting annual sales of 30K per patient for a its candidate) of above costs, i.e. $2.5- 10K , this would translate to a market opportunity of $7.5-30B for just mild AD US patients (3.3 million target population). Furthermore, if simufilam is proven to have disease modifying activity then label expansion to MCI patients is very likely to follow (but would need a new clinical trial to prove it), increasing the target population by 5-7m patients.
Based on above estimate ($7.5-30 billion market potential just for mild AD, just in US) and assuming a 50% penetration (if simufilam proves disease-modifying potential with same efficacy as in ph2 study this should be achievable) peak sales may be in the range of $ 3.75-15B. Based on a 7.1x EV/revenue multiple this would translate to $ 26.7-106 billion enterprise value. Assuming a partnership with 15% royalty rates, potential peak revenue for SAVA would be in the range of $ 0.6-2.3B, corresponding to a fair enterprise value of $ 4.26-16.3 billion.
Based on $168 million of cash ( as of June 30 ) and current market cap ($871 million at the time of writing) SAVA's enterprise value is currently about $700 million. Therefore, under above estimations (which are very conservative if simufilam proves same efficacy in phase 3 trials) SAVA is very undervalued. Considering fierce competition in the field and that SAVA has no infrastructure for large-scale manufacturing, distribution and marketing of simufilam I would prefer the partnership scenario. However, even if SAVA elects to proceed alone there is significant room for dilution considering upside potential, but there is also significant risk SAVA might be outpaced by competition. Cash needed, should SAVA elect to proceed alone, is dependent on several unknown variables. But for some perspective here are sales and marketing expense for ADUHELM (approved in June 2021 ); $353 million in 2020, $562 million in 2021, and $134 million in 2022.
Financials
Based on last quarterly filing SAVA had $168 million cash and cash equivalents (as of June 30, 2023), with no debt and a net loss of 26.4 million in Q2 2023 (vs $19.3 million for the same period in 2022). More specifically, R&D expenses were $24.97 million (vs $16.95 million for the same period in 2022) and G&A expenses were $3.8 million (vs $2.97 million for the same period in 2022). As a pre-revenue clinical-stage biotech without currently any partnership or research funding, SAVA does not have much income (interest income was $2.2 million, while income from leasing office space was 203K). SAVA has not officially given a guidance on the cash runaway. This is another red flag in my opinion. Nevertheless, considering ongoing phase 3 trials and expenses associated with ongoing legal procedures SAVA expects both R&D and G&A to increase in 2023. Increases in expenses are in my opinion likely to continue through 2024, for the same reasons.
Based on expected completion of enrollment of ph3 trials by the end of the year and 12-month study duration we should expect topline results by the end of 2024/early 2025. Considering $168 million in cash and assuming the same burn rate ($26 million net loss per quarter) this cash should be enough for about 6 more quarters (i.e., Q3-Q4 2023 and Q1-Q4 2024). However, I believe cash runaway will be shorter and SAVA will raise cash before that (and most likely before ph3 topline results) because; (1) As discussed above both R&D and G&A expenses are expected to remain high, possibly higher, going forward. (2) Without any income it is unlikely SAVA would risk waiting for ph3 topline results (at which point SAVA would have very limited if any remaining cash), (3) Without any partnership SAVA will need significant cash to support further development of simufilam, including regulatory and commercialization costs.
Above estimations do not account for potential income by partnership or from compensation in the context of above-mentioned lawsuit. However, I wouldn't count on a good partnership scenario before ph3 results based on available data, multiple red flags and ongoing litigations.
Risks
SAVA is a single-candidate company currently only being developed for AD. Therefore, the main risk is that SAVA might fail in phase 3 trials. Currently, the bullish thesis on SAVA is based on indirect comparisons of simufilam-treated patients with expected cognitive decline in historical placebo groups, but the validity of such comparisons is questionable in my opinion, as discussed above.
Furthermore, considering the lack of near-term catalysts there might be better entry points in SAVA stock (at the least I expect the stock to trade sideways for a while). In the meantime, announcement of big pharma partnership or positive outcome in ongoing litigations or breakthrough therapy designation could be positive surprises.
Another risk is positive results from competitors. For example, topline results of the phase 3 study by BioVie ((BIVI)) are expected by November this year. If positive, BioVie's candidate could reach commercialization sooner than simufilam.
My recommendation
Do not misinterpret my "Hold" recommendation. Personally, I tend to be bullish on SAVA. There are enough data to confirm safety and some good signals that it also might be effective, especially in mild AD patients (but likely not in moderate AD). However, it's not an investment for everyone considering risks and red flags discussed above. Invest only what you can afford to lose.
For further details see:
Cassava Sciences: CMS Data Promising But Many Red Flags Remain; Hold