Description

RLMD offers a binary outcome in 2024 with high asymmetry to the upside and a decent probability of working out. It feels odd to post a 1-year, ~5X OTM call option to VIC, but I believe I make a convincing case that the options are severely underpriced and do not reflect the reality of the scenario at hand. OTM calls often make more sense than shares in binary situations with extreme asymmetry, and I’ve found them useful in gaining meaningful exposure to biotech trial readouts while having very little capital at risk – especially after a company has been left for dead. Michael Lewis elaborates on this concept in The Big Short (Cornwall Capital approach).

RLMD is developing an antidepressant, REL-1017, and has two phase 3 trials ongoing (first readout expected MY2024). The drug has a validated mechanism of action in major depressive disorder (MDD) (NMDAR antagonism; same mechanism as Spravato and Auvelity – both recently approved in MDD – and racemic ketamine which is used off-label), solid preclinical data, and fantastic phase 2 data that gave rise to a $1.1B valuation upon its announcement. Since the Ph2 readout, however, the company has failed two phase 3 trials and the stock has been left for dead, now trading at an $80M market cap.

The company claims that through post-hoc analysis, they’ve identified structural issues with their failed phase 3 trials that resulted in an unusually high placebo response, and show a reasonable amount of evidence that these issues prevented a positive trial outcome. RLMD brought in Cedric O’Gorman as CMO, who led the successful clinical development of Auvelity in major depressive disorder (MDD) (at Axsome; approved 2022), and now says that these issues have been corrected and should not affect the two ongoing phase 3 trials. Though post-hoc analysis of clinical trial data is often spurious, the data seem to support their claim, and were convincing enough to attract Cedric (and compel him to make a $31K open-market share purchase – a binary bet on REL-1017’s future).

If data are positive, I expect the market to trade RLMD up to at least where it traded after positive phase 2 data – $1B. This would yield a share price of $30. In reality, I think that there is likely upside from $30. It’s not hard to get to $1-2 billion in NPV for a third-line MDD drug, as I will lay out, and comps corroborate this valuation range. 

Otherwise, the stock should go <$1 (the other pipeline asset is worthless in my view. Happy to expand on that). 

Call options offer exposure to the binary with greatly reduced risk. With shares, one risks $2.6 to make $30. The 15-strike Jan 2025 calls are marked-to-market at $0.65. At call-breakeven, RLMD would trade at a $470M market cap, well below the valuation generated by RLMD’s Ph2 data. At a share price of $30, these calls offer 25X upside compared to 10X offered by shares.

This writeup is organized as follows (Section 2 is dense but worth the read if you are inclined):

1. MDD – background and opportunity

   1. Etiology

   2. Standard of care

   3. Unmet need

   4. Market opportunity

2. REL-1017 – mechanistic and preclinical validation

   1. Mechanistic validation

   2. NMDAR antagonism

   3. REL-1017 preclinical data

3. REL-1017 in clinical study – promising data and setbacks

   1. Phase 2 readout

   2. Phase 3 readouts (RELIANCE III AND I)

   3. A closer look at RELIANCE I

4. RELIANCE II and RELIGHT phase 3 trials are improved shots on goal

   1. Cendric O’Gorman brought in as CMO

   2. Trial amendments

5. Conclusion/valuation

1. MDD – Background and Market Opportunity

Etiology

Despite decades of investigation into genetics, neurology, and biochemistry, the underpinnings of depression remain elusive. The monoamine hypothesis had long held that depression stems from a deficit of serotonin, norepinephrine, and dopamine in the brain. Most approved antidepressants increase concentrations of these neurotransmitters through inhibition of reuptake from the synapse. Recently, however, evidence in dispute of the monoamine hypothesis has been been accumulating:

1. There is no conclusive difference between depressed patients and healthy controls in: serotonin levels, binding activity/abundance of serotonin receptors, or the level of serotonin reuptake.

2. Serotonin depletion (by dietary exclusion of its precursor – tryptophan) lowers serotonin levels but does not conclusively lower one's mood.

3. A once-intriguing finding, that there is an association between depression and a polymorphism in the promoter of the SERT (serotonin reuptake) gene, has since been debunked by two very large association studies.

4. SSRIs inhibit reuptake immediately, but it takes 4-6 weeks to observe an antidepressant effect.

Standard of Care

The debunking of the monoamine hypothesis does not necessarily mean that raising monoamine levels in the brain can’t be beneficial.

Generic monoamine reuptake inhibitors (SSRIs and SNRIs) are mainstays of treatment in MDD. Patients will usually start on an SSRI. About one-third will achieve remission, and half will have a response within 6 weeks. Those who experience an incomplete response will typically stay on their current SSRI but add on an SNRI or bupropion (NDRI). Those who do not respond at all will typically switch to a new SSRI. Those who do not remit in the first two lines of therapy have a few options: 1) switch to branded Trintellix (SSRI that serotonergic modulation activity), 2) switch to branded Auvelity (NDMAR channel blocker), 3) add on branded or generic second-generation-antipsychotics (SGAs; antagonists of dopamine and 5HT-2A receptors), or 4) cycle through a number of agents with questionable efficacy/tolerability including tricyclics, T3 (thyroid hormone), or monoamine oxidase inhibitors. Beyond 3L patients will continue trialing new therapies, including IV ketamine, IN esketamine (Spravato), or transcranial magnetic stimulation (TMS).

Unfortunately, for those patients that fail each line of therapy, the odds of remitting in the next line diminish. Fig. 1 shows remission rates at each step, and cumulative remission rates.

Figure 1. Remission rates at each therapy - at each LoT and cumulative.

Source: STAR*D trial. Plotted in GHRS investor deck.

Unmet Need

Because insurance will not cover a branded agent until a patient has failed two lines of therapy with generics, any new branded drug will be used almost exclusively in 3L+. This is also where remission rates fall to their lowest, and where a drug with good relative efficacy would have the greatest value-add.

Those that do achieve remission with SSRI/SNRIs in 1L/2L have to live with side effects including weight gain and sexual dysfunction. Auvelity may have a cleaner side effect profile, but patients will have to wait until its patent expires (likely 2034; still subject to challenge) to access it early in the treatment algorithm.

Market opportunity

A new antidepressant that works well, with limited side effects and a reasonably rapid onset of action (1-3 weeks) should sell between $1-2 billion per year at peak. MDD is such a massive indication (8% prevalence) that any new drug that can capture a small slice of it in 3L can easily sell $1B+. As a comp, look to Trintellix – just a marginally better SSRI – peak sales of just under $1B. Shown below, it’s not hard to arrive at $1B/yr in peak sales while capturing only a 3% slice of the third line+ market, even without factoring ex-US sales. Slap a 2x or 3x NPV multiplier on peak sales, and you’re well in excess of the $470M breakeven on the Jan 2025 15-strike calls.

2. REL-1017 – Mechanistic and Preclinical Validation

Mechanistic validation

Currently, one of the most promising areas of development in MDD is the modulation of glutamatergic signaling (glutamine, like serotonin, is a neurotransmitter). This line of research kicked off in 2000 when a researcher at Mt. Sinai conducted the first ever clinical trial of ketamine, an NMDAR (glutamine receptor) channel blocker, in patients with MDD. His group observed a rapid and strong antidepressant effect, which reproduced in a larger trial in 2004. 

Since then, a cottage-industry of clinics administering IV ketamine to patients with MDD has arisen, and JNJ launched Spravato, an intranasal esketamine that is administered in an interventional psychiatrist’s office over a 2-hour period (ketamine and esketamine must be administered in a supervised environment because the majority of patients have a trip-like experience).

Last year, Axsome launched Auvelity, a fixed-dose-combination of dextromethorphan and bupropion. Like ketamine, dextromethorphan is an NMDAR channel blocker while bupropion, in addition to having NDRI activity, extends the PK of dextromethorphan by inhibiting CYP26 (which degrades dextromethorphan). Auvelity is a great drug in that onset of action is quick relative to SSRI’s (a few weeks), efficacy is good, and it does not induce weight gain and sexual dysfunction like SSRI/SNRIs. Moreover, it doesn’t cause trip-like side effects like ketamine does

REL-1017, like ketamine and dextromethorphan, is an NMDAR channel-blocker. As I lay out below, it appears that REL-1017 has the same effect as ketamine in mouse models of despair, and acts through the same pathway.

NMDAR antagonism

Consensus in the field suggests that the mechanism by which NMDAR channel blockage leads to an antidepressant effect is as follows: 

1. In the presence of a stimulus, glutamate binds to synaptic AMPA receptors, resulting in an influx of sodium and calcium ions, and depolarization of the cell. When the cell depolarizes, Mg+2 comes off of the NMDAR, which allows it to bind glutamate and open its channel to calcium ions.

2. However, NMDAR can also be activated tonically (in the absence of a stimulus). At resting membrane potential, there are a small number of NMDARs that are stochastically free of Mg+2 and open to calcium ions. The percentage of open NMDARs increases as the membrane depolarizes from -80mV to -50mV.

3. There is a low level of ambient glutamate in the synaptic cleft, which causes miniature presynaptic events (mPSEs) slightly reducing the membrane potential (but not enough to trigger an action potential). In stress-related circuits, repeated stressful events cause an increase in ambient glutamate in the synaptic cleft, and thus increase the proportion of open NMDARs at resting membrane potential.

4. The result of open NMDAR’s is an increased Ca+2 flux into the neuron. Ca+2 regulates CAMKIII phosphorylation of eEF2, which inhibits translation. Thus, high Ca2+ leads to decreased translation and decreased availability of synaptic proteins required for action-potential mediated neuroplasticity.

5. NDMAR blockers like ketamine, dextromethorphan, and REL-1017 can block tonic activation of NMDAR and have achieved success in clinical trials for MDD. Ketamine, unlike the latter two, has a high affinity for NMDAR and is thought to cause dissociative effects due to its interference with phasic activity (in response to a stimulus) in addition to tonic activity. MK-801 is an NMDAR channel blocker even more potent than ketamine. Its interference with phasic activity causes severe adverse events which preclude it from clinical study.

6. High potency is not thought to be required for an NMDAR antagonist to contribute to increased synaptic plasticity – after molecules come off the NMDAR channel, they are “trapped” within it for some time. The degree of trapping may be significant for the conferral of antidepressant effects, as NMDA channel blockers memantine and lanicemine are poor trappers and have no effect on depression. The degree of “trapping” between ketamine and REL-1017 is fairly similar: high, as far as NMDA channel blockers go. 

It’s clear that ketamine treatment results in an upregulation of synaptic protein expression (PSD95 and Synapsin I), and that this effect is blocked by pretreatment with rapamycin, which inhibits the mTOR pathway (mTOR pathway activation is also required for eEF2 function) (Fig. 2). Furthermore, in three well-established rodent models for despair, the forced swim test (FST) learned helplessness in response to inescapable stress (LH), and the novelty suppressed feeding test (NSFT), ketamine induced a rapid antidepressant effect (Fig. 3). Ketamine’s effects in these behavioral models are also blocked by rapamycin pretreatment, suggesting that they may arise from the upregulation of synaptic protein expression. Ketamine’s synaptic and behavioral effects are the opposite of those caused by chronic stress.

Fig. 2: Ketamine results in rapid upregulation of synaptic proteins. Blockage of mTOR pathway by rapamycin inhibits effect

Fig. 3: Ketamine results in behavioral responses in FST (A), NSFT (B), and LH (C) models of despair. The effect is blocked by pretreatment with rapamycin, which inhibits mTOR pathway. (Saline and DMSO = vehicle controls = no ketamine)

REL-1017 preclinical data

REL-1017, AKA esmethadone, is also an NMDAR channel blocker. Its affinities to all four NMDAR subtypes are within the same order of magnitude as ketamine and dextromethorphan, as are its degrees of trapping (note that MK-801 is highly potent [low Kb], causes a major trip, and is unsuitable for use in the clinic) (Fig. 4).

Fig. 4: NMDAR channel blockers' affinities to the NMDAR. Note: dextromethorphan (Auvelity) and ketamine (esketamine/Spravato) are approved in MDD

In the same in vitro assays and rodent models described above, REL-1017 shows the same effects as ketamine: increased expression of synaptic proteins (Fig. 4) and behavioral changes in FST (Fig. 5). Like those of ketamine, these behavioral changes appear mTOR-dependent, suggesting they occur via the same pathway (Fig. 6).

Fig. 4: REL-1017 (d-methadone) induces expression of synaptic proteins 24h after treatment.

Fig. 5: REL-1017 treatment with FST in comparison to ketamine 

Fig. 6: Rapamycin appears to block RLMD-107’s anxiolytic effect observed in NSFT assay, suggesting mTOR pathway is involved.

In summary, it appears that REL-1017 has the same effect as ketamine on mouse synaptic protein expression, and acts through the same pathway, Moreover, it appears that REL-1017 has the same effect as ketamine in mouse behavioral assays.

3. REL-1017 in Clinical Study – Promising Data and Setbacks

Ph2 Trial

Following the establishment of safety and tolerability in Ph1, REL-1017 entered a Ph2 trial in which 62 patients with MDD were administered either placebo, 25mg, or 50mg of REL-1017 for 7 days and observed out to 21 days. The drug was given as an adjunctive, meaning that patients were allowed to continue on their current antidepressant therapy (common practice in MDD), as opposed to a monotherapy. 

In October of 2019, the Ph2 trial read out overwhelmingly positive with both dose groups separating from placebo within 4 days, and with large effect sizes (Fig. 7; MADRS score reduction indicates decreasing symptoms of MDD). In the context of the MDD landscape, these results were remarkable–both on speed of onset and magnitude of effect–positioning RLMD to deliver the best-in-class antidepressant if results replicated in phase 3. The stock reflected that, 3X-ing and reaching a market cap of about $1B. The company promptly initiated three Ph3 trials–one a monotherapy trial (RELIANCE III) and the other two adjunctive trials (RELIANCE I and II).

Fig. 7: MADRS score change from baseline (gray = placebo, green = 25mg, orange = 50mg)

Ph3 readouts (RELIANCE III and I)

In October 2022, RLMD read out disappointing results from their first Ph3 trial of REL-1017 (monotherapy). Despite the drug arm experiencing an impressive 14.8 point reduction in MADRS score at day 28, the placebo arm experienced an unusually high 13.9 point reduction in MADRS score, and the study was negative. Typically, the placebo group will experience a 5-10 point reduction in MADRS scores (Fig. 8; from a poster presented by SAGE Therapeutics on the impact of check-in frequency on placebo performance in MDD trials)

Fig. 8: Placebo group change from baseline in Ph3 MDD trials.

The company performed a post-hoc analysis which identified two study sites in which placebo drastically out-performed REL-1017. Upon further investigation, the company determined that these sites, which were the highest-enrolling sites in the study, were enrolling patients who weren’t actually depressed (they may have been feeling a bit down) as a cash grab. The CEO reported that, unfortunately, RELIANCE I (still ongoing at the time) had enrolled many patients from these sites as well, portending its failure. Importantly, however, only about 20 patients had been enrolled in RELIANCE II at these sites, and enrollment there was stopped immediately. 

The company also showed that if you removed a handful of sites, the data looked somewhat better. This earned a collective shoulder shrug from the investment community, and an 80% drop in the stock. It certainly didn’t help that the CEO does not give off an impression of credibility.

To nobody’s surprise, when the company read out the results from RELIANCE I in December 2022, these same sites behaved anomalously again and the study was negative. However, the data wasn’t all bad. Despite not achieving statistical significance on change-from-baseline MADRS score reduction (15.1 vs. 12.9, drug vs. placebo), the REL-1017 arm achieved significantly greater rate of response (39.8% vs. 27.2%; response defined as a 50% reduction in MADRS score). Nevertheless the stock took another 50% dive. 

A closer look at RELIANCE I 

A closer look at RELIANCE I paints a different picture than the headline does. 

First, when excluding the two highest-enrolling problematic sites, there was a statistically significant 4.1 point difference between REL-1017 and placebo arms. This is a great result in an MDD trial.

Second, though the trial failed in ITT analysis, when analyzed on a per-protocol basis, a clinically meaningful, nearly statistically-significant (p = 0.051), 3.1 point placebo-adjusted MADRS reduction was observed at day 28 in the REL-1017 group (ITT analysis [gold-standard] includes all patients who entered the trial; PP analysis uses only those patients who completed the drug/placebo dosing regimen).

Third, the company analyzed patients who came from “verifiable” and “unverifiable” sources (see below,) and found a statistically-significant separation between REL-1017 and placebo at day 28 in verifiable patients, which was not present in unverifiable patients (Fig. 9). It’s worth noting that this difference was observed by analyzing only 92 patients (RELIANCE II target enrollment is 300).

Fig. 9: MADRS cfb in patients sourced from “verifiable” vs. “unverifiable” sources.

In summary, on top of preclinical data showing that REL-1017 has the same effect as ketamine in animal models of depression, the clinical study of REL-1017 shows that the drug has at least some (I would argue a good) chance of having real efficacy in MDD. This chance is not reflected in the current stock or option prices.

4. RELIANCE II and RELIGHT are improved shots on goal.

Cedric O’Gorman brought in as CMO

So where are we today? In Jan 2023, one month after RELIANCE I read out, the company hired Cedric O’Gorman as CMO. Cedric comes from Axsome, where he led the successful clinical development of Auvelity. With >2 decades of experience in clinical neuropsychiatry, Cedric was tasked with fixing REL-1017’s clinical development program, and he’s made a $31K open-market purchase of RLMD since his hiring. The bet he’s made is a binary bet that the phase 3 program works, because if it doesn’t, the stock is worthless.

Trial amendments

Cedric knows how to run an MDD trial, and since his hiring, RLMD has made some amendments to the RELIANCE II trial. Some include: 1) removal of the problematic sites from RELIANCE I and III. 2) reduction in the degree of interaction between patient and doctor. It’s known that the level of interaction between MDD patients and healthcare professionals correlates with the magnitude of the placebo response, which makes sense logically. Cedric reports that RELIANCE I’s protocol included way too much interaction, they’ve cut that down. 3) verified medical records are required for enrollment into the study. 4) limiting the number of enrollees allowed per-site. At the time of change-implementation, the trial had enrolled about 80 patients, and will enroll 300. Keep in mind that even without these improvements, the mere exclusion of the two highest-enrolling sites from the RELIANCE I trial would have resulted in a statistically significant 4.1 point difference between REL-1017 and placebo arms.

A sell-side analyst recently reported that the baseline MADRS score for enrollees in RELIANCE II so far is 35. I don’t know where he got this from, but if real it indicates that the company truly is enrolling severely depressed patients. It’s worth noting that more severely depressed patients tend to be less susceptible to the placebo effect. 

A quick note on safey/tolerablity

I’m not spending a ton of time here, but happy to discuss further. REL-1017 appears safe and tolerable. There have been some concerns over “likability” (does this drug have abuse potential?), but the company ran a trial which alleviated them.

5. Conclusion/Valuation

Maybe REL-1017 works, maybe it doesn’t. Given the mechanism and clinical data, I’m giving it some chance of having real efficacy in MDD.  If it does work, I believe RLMD has done everything it possibly can to produce a good result in Ph3’s next year.  

In the days before the first of these three Ph3’s read out (RELIANCE III), the stock traded at a market cap of ~$1.1B. Today, the stock trades at a market cap of $80 million and we have essentially the same shot on goal, but with tighter trial controland with more data suggesting efficacy. The $1.1B valuation made sense at the time and it makes sense now – NMRA recently announced positive Ph2 data for their MDD candidate and trades at a $1.8B valuation (they have other assets in the pipeline so it’s not a perfect comp, but MDD is the lead and focus). AXSM traded at a market cap of $3.5B after positive Ph3 data for Auvelity. Aside from looking at comps, $2-3B is a reasonable NPV for a good MDD drug in 3L+ (see market opportunity section above). 

That RLMD trades at $80M ($2.6/sh) is, I believe, a function of repeated Ph3 setbacks, a CEO who lacks credibility (which, I believe, prevents most investors from diving deep into the post-hoc analysis and rationale for the drug in MDD), and a down-market in biotech. However it’s not reflective of the potential value of REL-1017, or the probability that the value is realized.

If data are positive, I expect the market to trade RLMD up to at least where it traded to after positive phase 2 data – $1B. This would yield a share price of $30. In reality, I think the company is worth between $1-2B, so there is likely upside from $30.

Otherwise, the stock should go <$1 (other pipeline assets are worthless in my view. Happy to expand on that). 

Call options offer exposure to the binary with greatly reduced risk. With shares, one risks $2.6 to make $30. The 15-strike Jan 2025 calls are marked-to-market at $0.65. At call-breakeven, RLMD would trade at a $470M market cap, well below the valuation generated by RLMD’s Ph2 data. At a share price of $30, these calls offer 25X upside compared to 10X offered by shares.

Catalyst

RELIANCE II readout, expected MY2024

RELIGHT readout, possibly in late 2024