The share price for Tonix (NASDAQ:TNXP) has risen from its close at $.35 at the beginning of December 2016 (The intraday price was as low as $.31) to its close at $.63 during the first week of 2017 (The intraday price reached as high as $.65). Although when the intraday prices are taken into account the stock price doubled in a five-week period. While it has since slid back (it went above $.60 on February 13 and closed just above $.58), the title of this article does NOT refer to the recent trajectory of the stock price. Rather, it refers to the steady flow of developments that, when properly interpreted, has been steadily encouraging. The purpose of this article is to demonstrate how, collectively, these developments provide strong support for four claims about Tonix's future trajectory..
Of course, for more than a year, the headlines about Tonix's clinical trials have been disastrous. The company's products have missed four out of four primary endpoints. In addition, the number of outstanding shares has markedly increased. Many investors have either abandoned Tonix or reluctantly concluded that given the low stock price, if they held on, there was a chance that they could recover some of their losses. This article looks beyond the headlines.
One of the missed primary endpoints was for a headache product that Tonix has discontinued. However, while the fate of TNX-201 may spook investors, it has no statistical relevance for the TNX-102 SL, the sublingual formulation of Cyclobenzaprine (CBP) that was tested in the other three trials. Two of these were with people who have fibromyalgia (FM), and one was with people who have Post Traumatic Stress Syndrome (PTSD). As discussed below, when all of the trial findings are examined, there is a strong basis for optimism.
The remainder of the article contains the following sections:
- A Summary of Seventeen Upward Trajectory Findings & Developments
- TNX-102 SL is Effective with FM, & the FM Findings Bode Well for PTSD's Phase III Trial
- Differences Between How FM & PTSD are Viewed
- Why Subjects Appear to Do Better if Their Baseline Symptoms are Relatively Severe
The Disconnect Between My Estimate of Tonix's NPV/Share and Its Share Price
A Summary of Seventeen Upward
Trajectory Findings & Developments
This section briefly describes 17 findings and developments in Tonix's upward trajectory, grouped under four headings: FM Findings, PTSD Findings, Corporate Strategy, and Changes in the External Context.
FM Findings. Below are four out of several favorable findings from the two FM trials.
- Phase II findings re: two outcome measures proposed by the (OMERACT) subcommittee. The Outcome Measures in Rheumatology subcommittee proposed various composite responder outcome criteria. Two of them had to do with the extent of changes in three separate dimensions. For both of the multidimensional outcome measures, the first two dimensions were the same: (1) (1) a pain reduction of at least 30%, and (2) an improvement of at least 10% in the extent to which subjects' FM limits their ability to carry out ten everyday activities. For one of these composite outcome measures, the third dimension is at least a 10% improvement in either sleep or energy/fatigue; for the other measure the third dimension is at least a 10% improvement in sleep, energy, depression, anxiety, or cognition. As shown in Figure 5 in my article about Tonix's future, the difference between Tonix's treatment and a placebo was strongly statistically significant: The p values were .015 and .011, respectively (Researchers take notice whenever p is lower than .050).
- Phase III finding re: 30% responder analysis in which when data are missing for subjects, their last known data points are used. The primary endpoint, which was missed, was a 30% responder analysis in which when data were missing for subjects, their initial (baseline) scores was used. In other words, they were recorded as not having improved even if they had improved. However, Tonix reported an analysis in which subject baseline data was used (as the "after" score) only for subjects who dropped out because of "lack of efficacy" or adverse events, but for subjects who dropped out for any other reason, their last known pain rating was used as their "after" score. When this approach to the 30% responder analysis was used, the difference between Tonix's treatment and a placebo was strongly statistically significant: The p values was .012.
- Phase III finding re: a 50% responder analysis. The primary endpoint was an analysis in which the proportion of subjects who received the treatment who reported a reduction of at least 30% were compared to those who received the placebo who reported a reduction of at least 30%. But Tonix reported an analysis with a higher threshold: Subjects had to report a reduction of at least 50%. For this analysis (even with baseline scores used for all subjects who dropped out), the difference between the proportion of those who received the treatment and that of subjects who received the placebo was statistically significant: p = .035.
- Phase III finding re: average change in pain from baseline. In Phase II, the primary endpoint involved comparing subjects who received treatment to those receiving the placebo with respect to the average difference between their baseline pain scores and their final pain scores. In Phase II, this difference was not statistically significant. However, in Phase III (where this difference was no longer the primary endpoint), it was strongly statistically significant: p = .001.
PTSD findings. Below are five out of several favorable findings from the PTSD trial.
- Findings for subjects whose treatment dose was 5.6 mg. In the PTSD trial, the primary endpoint (which was missed) was comparing subjects who received 2.8 mg of the treatment with those who received the placebo with respect to their improvement in their CAPS-5 scores. However, when the improvement in CAPS-5 scores of subjects who received 5.6 mg of treatment was compared to the improvement of those who received the placebo, the difference was statistically significant: p = .037.
- The finding re: the exaggerated startle response. Compared to subjects taking the placebo, subjects taking 5.6 mg of treatment meaningfully reduced their exaggerated startle response (Item E4). The difference was statistically significant, with p = .015. This finding is especially noteworthy because no other medication for PTSD has successfully reduced the exaggerated startle response.
- Findings for subjects who suffered from combat-related PTSD and had higher baseline CAPS-5 scores. Tonix's clinical trial of TNX-102 SL for PTSD was the first clinical trial using the new CAPS-5 system. As a result, it was not clear what score should be used as the minimal baseline score for inclusion in the study. Tonix decided to use 29 as the minimal baseline score. However, in a retrospective analysis, Tonix determined that what has been accepted as the appropriate minimal baseline score when using the previous CAPS-4 system was actually equivalent to a 33 minimal baseline score. So Tonix analysts examined the data for various comparisons of subjects receiving treatment with those receiving the treatment for subjects who had a baseline score of at least 33. When carried out with subjects whose PTSD was combat-related, two of these comparisons are especially noteworthy (the second of which is reported immediately below). When the improvement in CAPS-5 scores of subjects taking 5.6 mg was compared to that of subjects taking the placebo, the difference was now strongly statistically significant: p < .013.
- The finding re: reckless or self-destructive behavior. The CAPS-5 hyperarousal cluster contains a new item (E2) that can include dangerous driving, high-risk thrill seeking, excessive substance use, and injurious behaviors to self or others, including or suicidal behaviors. On this item, the difference in improvement between those taking 5.6 mg of treatment and those taking the placebo was highly statistically significant: p = .012.
- Dose-related effects. The average improvement for those taking 2.8 mg is consistently between the average improvement for those taking the placebo and those taking 5.6 mg of the treatment.
Corporate Strategy. Of the four items described under this heading, only the first two stem from corporate announcements. The third and fourth fifth should be viewed solely as ideas management is considering.
- Tonix's pivot from a dual focus to a single focus. Tonix management has declared that even though it has documented statistically significant evidence (such as that reported above) demonstrating that TNX-102 SL is effective at reducing FM symptoms, it is discontinuing its work on Fibromyalgia and focusing its resources on demonstrating that TNX-102 SL is substantially reduces symptoms for those with PTSD. Note the minor tweak to this point immediately below.
- Two drugs in the pipeline. Tonix recently announced that it is also studying two additional products, one of which is for PTSD, and the other is a vaccine for smallpox. In this article, I will not discuss their claims and plans for either of these products, except to make two points: (1) Tonix does not intend to expend more than 5% of its financial resources on R&D related to these products for the next one to two years, and (2) their main intent with these products is to develop them far enough so that they can sell or license the rights in order to increase the company's capitalization, thereby reducing the need for diluting shares.
- TNXP pricing may be based on a market analysis. In a telephone interview, Dr. Lederman estimated that other pharmaceutical companies charge about $30/day for a basket of similar psychiatric drugs.
- TNXP may do its own marketing. Investors have assumed that Tonix might have to merge with or be acquired by a larger company. However, Slide 30 of Tonix's 2/7/17 presentation said that the company "is exploring a variety of options to commercialize TNX-102 SL, including commercializing on our own or partnering all or some indications in specific regions of the world." The slide points out that the target physician audience is well defined (about 30,000 psychiatrists in the U.S.) and that a small specialty sales force would be "sufficient for coverage."
Changes in the External Context. Of the four items described under this heading, the first three relate directly to Tonix. The fourth is more general, relevant to many pharmaceutical companies.
- FDA's breakthrough therapy designation for TNX-102 SL as a treatment for PTSD. In its 12/19/16 press release, Tonix described the benefits of a Breakthrough Therapy designation as including "the eligibility for priority review of the New Drug Application (NDA) within six months instead of ten months and rolling submission of portions of the NDA, in addition to an organizational commitment involving FDA's senior managers contributing significant guidance."
- FDA's approval of a two-stage adaptive design for the Phase III PTSD trial. While Tonix has announced that the FDA has approved an adaptive design, to the best of my knowledge, the only details made public is that (1) the primary endpoint will be the difference between subjects taking 5.6 mg and those taking the placebo with respect to the average reduction in CAPS-5 scores, (2) there will be a total of 550 subjects whose baseline CAPS-5 score equals or exceeds 33, and (3) there will be one interim analysis after half the subjects (275) have completed the 12-week period. What we don't yet know is what the maximum p value will be for the interim analysis, and the maximum p value will be if there is a need to enroll 550 subjects. It is my understanding that (1) the maximum interim analysis p value will be lower than .05, (2) if there is a need to enroll 550 subjects, the maximum p value will be still lower, and (3) if the interim analysis is statistically significant, any subjects who are in the pipeline will continue to receive treatment or placebos until they have completed the 12-week program, but their outcomes won't count against the fact that the interim analysis came out positively.
- Army activity related to Tonix. Tonix has entered into a Cooperative Research and Development Agreement (CRADA) with a branch of the Department of Defense called the U.S. Army Medical Materiel Development Activity (USAMMDA). One benefit is that the agreement makes it possible for Tonix to gain access to and study military personnel who are on active duty. At one point, I was under the impression that USAMMDA could conduct studies of TNX-102 SL, but could not award Tonix a grant or contract. However, I have learned that the army can indeed award a grant or contract, and if it were to do so, a logical first step would be to announce that any organization that was studying the effectiveness of treatments for PTSD should inform the army about its work. The army has taken such a step. Tonix intends to start enrolling subjects this quarter in its Phase III clinical trial (which Tonix calls "HONOR"), and, as I understand it, beginning the trials will not affect the army's decision as to whether to provide financial support. However, as a federal entity, USAMMDA cannot decide to spend money that has not yet been allocated. That is why the next item is relevant.
- The 2016 national elections. Hilary Clinton promised to try to set upper limits on how much patients (and presumably their insurance companies) would have to pay per year for chronic conditions. Although Donald Trump also attacked pharmaceutical companies for the high price of drugs, most observers believe that both his administration and Congress will favor pharmaceutical companies over patients. For example, Trump has argued that the FDA approval process should be streamlined or somehow shortened. He has also expressed strong support and interest in both the military and veterans. For example, he has repeatedly asserted that he will increase the Pentagon's budget. On the other hand, it is very difficult to predict what President Trump will actually do.
TNX-102 SL is Effective with FM,
& the FM Findings Bode Well for PTSD's Phase III Trial
Above, I described four findings from the FM clinical trials. Very briefly, here are three that view FM as a multi-symptom condition:
- Patient Global Impression of Change (p = .025)
- FM Impact Questionnaire, Revised, Symptom domain (p = .001)
- FM Impact Questionnaire, Revised, Total score (p = 001)
And here are some that focus on specific FM symptoms other than pain:
- Level of sensitivity (p = .0017)
- Level of anxiety (p = .0015)
- Four different sleep or fatigue scores (all have p = .001).
In a later section, I will talk about some difference between the two conditions that also have implications for Tonix investors. However, The reason I have spent time reviewing the evidence that TNX-102 SL is effective with FM is that there are some important similarities between FM and PTSD that also have implications for investors. Below are three types of similarities either between the two conditions or between the patterns of how TNX-102 SL affects subjects.
Both conditions have multiple symptoms. I have already described several of FM's symptoms. With respect to PTSD, the CAPS-5 consists of four symptom clusters: Intrusions, Avoidance, Mood/Cognition, and Arousal/Reactivity.
The two conditions have common symptoms. They include sleep disturbance, sensitivity/arousal, and anxiety, and both interfere with home, work, and social functioning.
TNX-102 SL affects patients with FM and PTSD in similar ways. Here are four patterns that highlight the similarity:
- TNX-102 SL reduces multiple symptoms;
- The more severe the symptoms, the clearer TNX-102 SL's effects (when compared to a placebo);
- Better sleep precedes other symptoms; and
- The greater the improvement in sleeping, the greater the overall symptom improvement.
Differences Between How FM & PTSD are Viewed
This section explores how the two conditions are viewed differently by how three populations: physicians and medical researchers, the FDA branches that address the conditions, and investors.
Differences in how FM and PTSD are viewed by physicians and medical researchers. Here are two differences:
- The extent of confidence regarding the reality of the conditions. There is considerable skepticism whether FM is real. The consensus regarding the reality and nature of PTSD keeps growing, with each revision of the CAPS diagnostic system.
- The willingness to accept the outcome measures. While the branch of FDA dealing with FM is clear that assessing the improvement in pain is the way to go (see the next bulleted item), researchers in at least both the U.S. and Europe recognize that genuine improvement has only occurred when the severity of several of the other symptoms mentioned above has been reduced. In addition, assessing pain is highly subjective. For various ailments I have had in the last two years, I keep getting asked to rate my pain on a scale from 1 to 10. Personally, I find it very difficult, and don't feel that I have given my doctors useful information. Then there is the notion of interrater reliability, that is, the notion that your "7" is similar to my "7." In contrast, not only has the consensus about the nature and proper way of assessing the severity of PTSD steadily grown, but also, the criteria for making the ratings for each of the four clusters are substantially more objective, and they are based on probing interviews.
Differences in how FM and PTSD are viewed by the FDA branches that address them. Below are three differences that mattered or will matter.
- Familiarity with CBP. Since the FDA division that addresses FM was thoroughly familiar with CBP, it saw no reason to require more than one dose. Accordingly, the Phase II and Phase III studies were done with 2.8 mg of TNX-102 SL. However, since the FDA division that addresses PTSD had less familiarity with CBP, it followed its standard practice of requiring that the Phase II trial be done with two different dosages: 2.8 mg and 5.6 mg. This division's decision saved Tonix's bacon. It seems that though 2.8 mg is sufficient for FM, 5.6 mg is a more optimal dose for PTSD, possibly because the higher dose is needed for both REM and non-REM restorative sleep.
- Willingness to accept the best indicator as the primary endpoint. The FDA division that addresses FM insisted on using a measure of pain by itself as the primary endpoint, although a better primary endpoint would have been one of the OMERACT measures, both because it takes into account that FM is a multiple-symptom condition, and because as a result of its research program, OMERACT's outcome measures successfully differentiate between treatments and placebos (See the next section about why subjects appear to do better if their baseline symptoms are relatively severe). Had either OMERACT outcome measure been selected, TNX-102 SL would by now have received FDA approval. In contrast, the FDA division that addresses PTSD seeks to use the CAPS-5, which takes into account the latest research and thinking about the nature of PTSD, and how its severity should be assessed.
- Treatment of missing data. The FDA division that addresses FM insists upon treating missing data (for subjects who drop out for reasons other than adverse events or lack of efficacy) by using their baseline score as their 12-week score. In contrast, the FDA division that addresses PTSD allows pharmaceutical companies to use their last recorded score as their 12-week score. Had the FDA division that addresses FM operated by the same policy as the division that addresses PTSD, TNX-102 SL would by now have received FDA approval.
Differences in how FM and PTSD are and will be viewed by investors. Fund managers express greater interest in promising medications for PTSD than for promising medications for FM. Below are five differences between the investment contexts for the two conditions that explain why fund managers view (or will view) PTSD as more attractive:
- Political and institutional support. Above, the possibility of military funding was mentioned. In assessing the probability of the military providing financial support, keep in mind that the lack of an approved effective treatment negatively affects the military's recruitment efforts. According to a 2012 Cleveland Clinic Journal of Medicine article, "During the wars in Iraq and Afghanistan, the US Army's suicide rate has increased from 12.4 per 100,000 in 2003 to 18.1 per 100,000 in 2008. . . . And PTSD has been shown to be a risk factor for suicidal ideation in American veterans of the wars in Iraq and Afghanistan. In a survey of 407 veterans, those who screened positive for PTSD (n = 202) were more than four times as likely to endorse having suicidal ideation compared with veterans who screened negative for PTSD."
- Size of target population. The potential target market for successful PTSD treatments is larger, especially when Europe and Asia are taken into account. In the last section, my estimate for Tonix's Net Present Value/share takes the European market into account, but not the Asian market.
- The FDA's Breakthrough Therapy Designation. In addition to the direct benefits mentioned above, the FDA's Breakthrough Designation has two important implications for investors. First, the shortened timeline for priority review of the NDA and the opportunity to submit the NDA in installments means that approval could come quicker, thereby slowing down the cash burn. Furthermore, the FDA only confers a Breakthrough designation when it has determined not only the treatment is for a serious medical condition, that treatment for the condition is an unmet need, and, most important, that the pattern of data relating to the performance of the treatment for the condition is indeed promising. To make the last determination, the FDA is given access to the entire set of Tonix's data related to PTSD, including data not accessible to the general public. The fact that the FDA conferred the designation after examining the entire data set should provide some reassurance to investors.
- The FDA approval of an adaptive design for HONOR. Being able to analyze the data when only half the subjects have completed the 12-week trial means that approval could come quicker, thereby slowing down the cash burn.
- Pricing of medication. While Tonix will not decide how to price TNX-102 SL until after FDA approval, investors gain from the fact that the treatment doesn't have to be competitive with FM treatments. As mentioned, the selling price of a basket of comparable psychiatric medications is around $30.
Why Subjects Appear to Do Better if Their
Baseline Symptoms are Relatively Severe
In this section, three explanations will be explored. The first one is important because it suggests that TNX-102 is not an effective treatment for PTSD. If this explanation is correct, then HONOR will fail, and soon thereafter, so will the company. However, I shall vigorously argue that this explanation doesn't fit the entire set of data reviewed above. At first glance, the second explanation sounds reasonable, but on closer examination, no one has offered any physiological reasons why it might be the correct explanation. Since I believe that there are several reasons and ample evidence to support the proposition that the third explanation is correct, most of what appears below examines these supportive reasons and evidence
Explanation 1. The positive findings related to subjects whose baseline CAPS-5 score was at least 33 is due to a sampling fluke, or to a related statistical phenomenon known as regression toward the mean (Some of the information below was introduced in one of my earlier articles, and therefore will only be summarized here). Pharmaceutical researchers hope that when they obtain findings that are statistically significant, it means that variation in the treatment variable is a cause of variation in an effectiveness variable (whether treatment subjects reduce their symptoms more than placebo subjects). But if the actual state of nature is that variation in treatments is actually not associated with variation in effectiveness, then it would follow that any finding that appeared to demonstrate that a real association existed would be mere a sampling fluke, meaning that it was just luck that caused the researchers to draw a sample in which such a relationship existed. As a result, when researchers drew new samples, the probability would be high that they would not find a statistically significant relationship suggesting that the treatment was effective.
Regression toward the mean (that is, toward the average) is the idea that the more severe the symptoms are at the start of a clinical trial, the larger you should expect the reduction in symptoms to be. This is related to the sampling fluke explanation because both depend upon principles related to how random variation works (although I won't explain why here).
Below are two arguments against the two explanations that depend upon random variation (that is, "It's just a sampling fluke." or "It's only regression toward the mean").
- Related evidence. Although there are techniques such as analysis of variance and analysis of covariance, conventional statisticians tend to estimate the probability that a finding will be replicated with the same or larger sample using the finding itself, but not any related findings. But Bayesian statisticians recognize that related evidence that is consistent with a finding increases the probability that the finding will be replicated. As reported above, (1) the entire set of data for both conditions (FM and PTSD) is internally consistent, (2) the dose-related effect is consistent, and (3) the symptoms most dramatically affected (sleep disturbance, intrusions, and hypersensitivity/reactivity fit Tonix's theory that restoring REM and non-REM sleep reduce intrusions and hypersensitivity/reactivity..
- Placebos. Regression toward the mean occurs because extreme initial scores tend to become more moderate during the next assessment, thereby giving the illusion that the treatment reduced the symptoms. However, placebos provide protection against being fooled by statistical regression, because as long as the average baseline scores for subjects receiving the placebo are roughly the same as those receiving the treatment, the average end score for placebo subjects will also move toward the average. So to control for the statistical regression effect, researchers look at how much more the symptoms of treatment subjects decreases that those of the placebo subjects.
Explanation 2. The medicine works better with people who have severe symptoms. While this explanation might be true, no one has put forth a theory about the physiology involved or provided corroborative evidence to suggest that it is true. Since it is logically possible for both Explanations 2 and 3 to be correct, no more will be said about Explanation 2.
Explanation 3. The effectiveness of placebos in reducing the severity of symptoms depends upon the conditions. The table below both compares the conditions under which the placebo effect is maximized to those in which it is minimized, and provides examples that demonstrate:
- Why the placebo effect was maximized when Tonix tried to FM's primary endpoints according to the rules of the FDA division that addresses FM, and
- Why the placebo effect will be minimized for PTSD's Phase III HONOR trial.
The table not only explains why subjects appear to do better if their baseline symptoms are relatively severe, but also explains why (1) clinical trials involving FM are risky when a pain measure is the primary endpoint, and, more important, (2) why the stars are aligned for PTSD's Phase III HONOR clinical trial.
The Disconnect Between My Estimate of
Tonix's NPV/Share and Its Share Price
For all the reasons above, and based on market information contained in Tonix's February investor presentation, I have made the assumptions listed below in order to calculate the NPV/Share model.
- The Probability for Success for PTSD is 75%.
- 80 Million Shares (including warrants, options, etc.), which is more than double the 39.2 million outstanding shares as of 2/10/17 (See Slide 39).
- Peak annual sales to 100,000 U.S. veterans, 175,000 U.S. civilians, & 165,000 Europeans.
- Fraction of peak sales achieved in Years 1‒6: 17%, 42%,60%, 75%, 88%, & 100%.
- 300 tablets/year/patient @ $30/tablet for U.S. patients = $9,000/year/U.S. patient.
- 300 tablets/year/patient @ $15/tablet for European patients = $4,500/year/European patient.
- 18% discount rate.
- Net Present Value/Share (NPV/share) is based on 1.0 time sales.
- The NPV/Share calculation does NOT take into account the value of the two new drugs in the pipeline.
Note a few changes in assumptions since my most recent estimate:
- The probability of PTSD success has been scaled back from 90% to 75% (even though HONOR is "powered" at 90%, and I thinkg the chances of success are indeed 90%).
- The number of shares in the divisor has been increased from 50 million to 80 million.
- The number of tablets per year per patient has been reduced from 360 to 300.
- Rather than market penetration rates, the peak annual sales estimates are shown as subpopulation sizes for U.S. veterans, U.S. civilians, and Europeans.
- The sales yield per patient per day has increased from $15 to $30 for U.S. patients, but remained at $15 for European patients.
If sales began in 2020 or 2019, the NPV/share would be $42.60 or $50.27, respectively. While the size of the NPV/share under conservative assumptions and the cumulative weight of impressive evidence that HONOR is highly likely to meet its primary endpoints has led me to increase the size of my Tonix holdings in the last two months, I recognize that the stock price will not make spectacular gains until HONOR does in fact meet its primary endpoint, which could come in the first half of 2018, but may not come until the second half of 2018.
Four missed primary endpoints, substantial past and likely future dilution, and a steep decline in TNXP have caused investors to feel burned and to react to my analysis skeptically. I welcome comments challenging my arguments and assumptions.
One catalyst that could boost the price of the stock is an announcement that the army will provide financial support. Although the size of the boost would depend upon the amount of financial support, I consider it unlikely that TNXP will exceed $1.50 until TONIX announces both that the army has stepped forward, and that since there have been at least 275 completers, the interim analysis will begin. This could happen within a year, and would represent an increase in TNXP of more than 150%. Of course, if Tonix announces that HONOR has met its primary endpoint, which it could do within 15 months, TNXP will jump considerably. However, it will not reach double digits until the FDA has approved TNX-102 SL and the company has (1) set forth a credible manufacturing and marketing plan, and (2) explained how it will be financed.
If you are considering whether to invest or increase your holdings, think about what more needs to happen before you are comfortable taking the risk.
Disclosure: I am/we are long TNXP.
I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.
Editor's Note: This article covers one or more stocks trading at less than $1 per share and/or with less than a $100 million market cap. Please be aware of the risks associated with these stocks.