Tonix (NASDAQ:TNXP) has just sold 5 million shares, and possibly 750,000 more. This article puts this dilutive sale of new shares into perspective by pointing out that the company's recently published findings from two clinical Phase II studies have provided an extremely solid basis to justify going forward, and, to make that possible, to raise the additional money that will be needed under more favorable terms.
I introduce a framework for assessing the persuasiveness of various kinds of therapeutic outcome data, which I then use to examine the findings from these the two studies. Both of the studies tested the same compound for two serious conditions that have somewhat overlapping symptoms. On June 21, TNXP closed at $2.05. My claim, which I justify below, is that Tonix's Net Present Value (NPV) per share is at minimum, ten times higher.
Below are the article's five sections:
- Updated Background Information
- Tonix's Money Raise: Its Timing and Scope
- A Framework for Assessing the Persuasiveness of Types of Outcome Data
- The Persuasiveness of the Phase II Data
- My Updated Valuation Assuming 50 Million Shares
Updated Background Information
Tonix has an attractive strategy: Investigating potential new uses and new formulations for FDA-approved drugs that have long clinical track records. Part of this strategy includes developing ways to minimize known side effects. It used this strategy to develop new formulations for cyclobenzaprine (CBP) for both fibromyalgia (FM) and Post Traumatic Stress Disorder (PTSD).
Results have been reported for two recent Phase II trials of Tonix's CBP formulation (known as TNX-102 SL): one for FM and the other for PTSD. Neither trial met its primary endpoint, but both trials demonstrated that Tonix was on the right track. A Phase III trial for FM is now underway, and its results will probably be reported during the first two weeks of September. Three more clinical trials are planned:
- A second Phase III trial for subjects with FM. This trial is expected to start during July 2016.
- A Phase III trial for military subjects with PTSD. This trial is expected to start during January 2017.
- A Phase III trial for civilian subjects with PTSD. This trial is expected to start during 2017, a few months after the military trial is launched.
Oral CBP is swallowed, absorbed into the small intestine, and sent directly to the liver where it metabolizes into a problematic byproduct that (1) causes patients to feel tired, and (2) lingers in the body long enough to reduce the effectiveness of subsequent CBP doses, thereby causing diminishing returns after a few weeks. TNX-102 SL improves on oral CBP in two important ways:
- While the standard dose is between 15 and 30 milligrams per day, Tonix's versions for FM and PTSD, respectively, are for 2.8 and 5.6 mg. Patients take it just once a day, at bedtime.
- Oral CBP takes almost two hours to reach its maximum effectiveness. Patients take Tonix's patented eutectic formulation sublingually (under their tongues), so that the CBP is rapidly absorbed across the mucous membrane into their bloodstreams, and bypasses their livers on its first journey through their bodies (Placing crushed oral CBP tablets under the tongue does not produce a rapid effective dose or initially bypass the liver).
Although the CBP eventually travels to the liver and produces the problematic byproduct, these two improvements mean that its therapeutic effect begins within 15 minutes, and produces too little of the problematic byproduct to either cause drowsiness, or linger long enough to interfere with the therapeutic effect of subsequent doses.
Tonix's Money Raise: Its Timing and Scope
In April, Tonix's management announced that it was making preparations to sell shares to raise $15 million. It indicated that its burn rate would be substantially lower in the remaining three quarters of 2016, and that it had enough cash to last through March 2017. Some Tonix investors argued that its cash level was so low that unless it raised money quickly, the company's financial situation and apparent health would be rapidly compromised. Others argued that there was enough cash to postpone dilution until after the results were announced for Phase III trial for FM, in order to increase the revenue from the new shares and therefore minimize the extent of dilution.
Tonix's management decided to respond to the concerns of the former group and announced its offering of 5,000,000 shares immediately after the market closed on June 15, which, according to its June 21 press release, is now completed. Together, the June 15 announcement and June 16 clarification precipitated an immediate 20% reduction in TNXP (For other factors and events that have driven down the share price to a fraction of its value, look here).
The company has not yet announced whether all of the overallotment of 750,000 were also sold, but in the final section (which provides estimates for the NPV/share), I shall assume that they have been and make allowances for substantial additional share dilution.
A Framework for Assessing the Persuasiveness of Types of Therapeutic Outcome Data
In order to assess the persuasiveness of therapeutic outcome data, investors need to grasp and take into account of a variety of distinctions.
Statistical significance vs. clinical significance. A finding is statistically significant if it is unlikely that the finding could be attributable to a sampling fluke. In other words, repeating the study is likely to yield similar results. Using study data, statisticians calculate a p value, which is an estimate of the chances that a particular study finding can be attributable to a sampling fluke. The lower the p value, the better.
For example, if p = ≤ 0.05, it means that the chances are 5 out of a hundred (5% or 1 out of 20) that the finding can be attributable to a sampling fluke, or in other words that the chances are 100% minus 5% or 95% that repeating the study is likely to lead to similar results. If p = 0.0001 (as it does in a finding reported in Figure 1 and Table 2), the chances are only 1 out of 10,000 that the study finding can be attributed to a sampling fluke, and therefore 9,999 out of 10,000 that repeating the study is likely to lead to the same results.
However, even if a finding is statistically significant (and therefore repeatable), it does not mean that it is clinically significant. For a therapeutic effect to be clinically significant, it has to make a noticeable difference to the patient. You can have very small effects that are statistically significant, and you can have sizable and noticeable effects that are merely attributable to sampling flukes (not repeatable). For informed people to take notice, outcome data should be both statistically and clinically significant.
One-dimensional measures vs. multidimensional measures. Measures are one-dimensional or multidimensional if they only measure symptoms in one domain or assess symptoms in multiple domains. When dealing with a condition that produces symptoms in multiple domains, a multidimensional measure tends to be more persuasive than a one-dimensional measure, unless the domain assessed by the one-dimensional measure is overriding in its importance.
Improving average scores vs. meeting clinically significant thresholds. In order to point out how misleading averages can be, Former Labor Secretary Robert Reich (who is 5'11") likes to say that the average height of him and Shaquille O'Neal (who is 7'1") is 6 feet. Outliers can distort averages and there may be few people in a population who whose value is close to the average. So even if you know that an average improvement is both statistically and clinically significant, you can't tell from looking at the average what proportion of the population actually achieved a clinically significant improvement. That is the advantage of using a "responder analysis," which tells you exactly what proportion of the population met or exceeded a specified clinically significant threshold.
Meeting a single clinically significant multidimensional threshold vs. meeting multiple clinically significant thresholds simultaneously. This distinction is the most subtle. If you discover that 20% of a population improved their score by at least 30% on a measure that assesses multiple domains, that doesn't tell you as much as discovering that 20% of population met multiple clinically significant thresholds simultaneously. That is because some of the patients using the single multidimensional threshold might have improved meaningfully in symptoms in only one or two of the condition's key domains, whereas all of the patients using the multiple-threshold construct will have improved meaningfully is symptom in all of the condition's key domains. If all other things are equal, when dealing with a condition that produces symptoms in multiple domains, multiple-threshold construct outcome data are more persuasive than that based on any other kind of single therapeutic outcome measure.
Primary endpoints vs. total patterns of data. Many investors in the pharmaceutical industry believe that the only thing that the FDA cares about is whether a clinical trial meets its primary endpoint. Without doubt, whether or not a trial meets its primary endpoint is of paramount importance to the FDA, but its ultimate and more elusive standard is the extent to which the total pattern of data is persuasive. Because of the supreme importance of meeting a trial's primary endpoint, pharmaceutical companies tend to choose relatively simple measures that they also believe, on the basis of Phase II data, that their product is likely to meet. But they also collect a lot of additional data that they intend to use to build a case based on the total pattern of data. FDA decision makers fully grasp the distinction described here and appreciate the wisdom of selecting a reasonably simple, attainable, and clinically meaningful measure as a trial's primary endpoint, but this is often not true for headline writers and those who make investment decisions rapidly and superficially.
Persuasiveness to the FDA vs. persuasiveness to insurance companies and prescribing physicians. Both the FDA and prescribing physicians have overlapping goals. Both want therapies that result in clinically significant improvements with minimal side effects, and both want to be sure that new medicines have a better profile than that of already existing therapies. However, compared to the FDA, prescribing physicians care even more about whether new therapies have noticeably better profiles. That is because they prefer to rely on therapies that have accumulated a large number of patient years (to guard against as-yet-unidentified rare side effects), as well as those for which they already know what to expect (with respect to when to use it, how it relates to other medicines, how to stop using it, etc.). Accordingly, investors need to focus not only upon whether the FDA will approve a new therapy, but also how physicians will view its comparative advantages (partly so that they can persuade insurance companies that it is necessary or count on insurance companies covering it without a hassle), and the ease with which they can become proficient in its proper use.
Table 1 uses the first five of these distinctions, especially the first four to provide a framework for assessing the persuasiveness of therapeutic data for conditions that produce symptoms in multiple domains. The closer that you get to the bottom of the table, the more persuasive the outcome data type.
The Persuasiveness of the Phase II Data
This section is divided into three sections: one for Fibromyalgia alone, one for PTSD alone, and the third looking at the totality of the Phase II outcome data for TNX-102 SL. Each uses the framework explicated in Table 1.
The Persuasiveness of the BESTFIT data related to Fibromyalgia. Figures 1‒5 display BESTFIT's findings related to FM. The findings in each successive figure use a more persuasive outcome data type than those used in the previous figure. For most of the figures, information about the level of statistical significance had previously been announced, but not relative difference between the averages or proportions had not been announced. In several cases, the information about statistical significance is more precise. For example, in Figure 1, rather than saying that the p value of the finding related to the FIQ-R sleep score was ≤ 0.001 (which was what Tonix had previously said), the newly released data say that p = .0001, which is substantially more impressive.
Most of the data displayed in Figures 4 and 5 were only revealed this week. Please take a little time reviewing Figure 5, since it is both the most complicated as well as the most persuasive of the five graphs.
The Persuasiveness of the AT-EASE data related to PTSD. Figures 6‒8 display AT-EASE's findings related to PTSD. Because they were released on the same day that I published my most recent article about Tonix, they were not analyzed in the article (although some of us discussed them in the comment section). Just as was the case with Figures 1‒5, the Table 1's persuasiveness framework provides you with the tools to interpret the findings.
In Figure 6, note that the two bar pairs on the right. The reduction in the both of these scales for those taking TNX-102 SL is double that of those taking the placebo. Tonix's CEO stated in the conference call that he suspects that a 5.6 mg. dose is needed to reduce the disturbance in REM sleep (Apparently, in FM patients, sleep disturbances tend to occur in non-REM sleep).
Figure 6's other bar pair has to do with reducing the exaggerated startle response. In addition to this finding being statistically and clinically significant, it is socially significant because none of the other treatments for PTSD have successfully dampened patients' exaggerated startle responses. This is one key factor (in addition to a substantially lower side effect profile) that differentiates TNX-102 SL from the other FDA-approved treatments.
Figure 7 focuses upon the CAPS-5, which was AT-EASE's primary endpoint (for 2.8 mg.), and will be for the forthcoming PTSD Phase III trials (for 5. mg.). What makes it multidimensional are the four domains used to make a diagnosis: arousal and reactivity, intrusions, mood and cognition, and avoidance. What is especially interesting about this measure is that it requires an extensive clinical interview to obtain a score. Ratings are made for 20 items. While this is time-consuming, the fact that the judgments are made by trained clinicians increases the measure's reliability.
Figure 8 focuses on two measures. The first of these is different from all of the other threshold measures addressed in this article. While all the others have to achieve an improvement of a specified size (usually 30%), achieving a remission doesn't involve achieving an improvement of a specified size, but rather requires a reduction below a specified cutoff point (in this case, below 11).
AT-EASE subjects had an average CAPS-5 baseline score of 39.5, which means that, on average, achieving remission means an average reduction of more than 70%. However, unlike the other threshold measures, because some subjects started with a higher CAPS-5 score and some others started with lower CAPS-5 scores, we don't know what proportion achieved reductions as large as 70%. (Note that this is the only finding presented in this article that is not statistically significant, presumably because remission was achieved by such a small number of subjects. I included it not because of its persuasiveness but because of its clinical and social significance.)
Figure 8's other measure is the patients global impression of the extent of their improvement. Collectively, although their level of statistical significance was more impressive for the improvements in sleep disturbance and the exaggerated startle response, the data types displayed in Figures 7 and 8 are typically considered to be more persuasive.
The Persuasiveness of the total pattern of Phase II TNX-102 SL-related data. Table 2 provides a synthesis of Table 1 and Figures 1‒8.
In trying to grasp the total pattern of data, note that:
- Tonix's key premise is that getting a small but sufficient sublingual dose of CBP rapidly into the patient's bloodstream sat bedtime will help patients get both more sleep and sleep that is more restorative, thereby reducing the frequency and intensity of most or all of the condition's other symptoms. The effect on sleep is clear, and in the bottom (miscellaneous) section of Table 3, there are several findings that suggest the validity of the hypothesis that better sleep is in fact associated with such a comprehensive set of consequences. Perhaps the most important piece of evidence is that the patients with the most improved sleep are also those report the greatest amount of change in the other symptoms (I explained why this makes sense in my previously cited May 2016 article).
- Reductions in symptoms occur in several domains in both conditions.
- Most of the improvements in individual averages for Tonmya subjects were highly statistically significant. At first glance, the levels of statistical significance for PTSD subjects were less impressive, but when you take into account the fact that the population of PTSD subjects who took 5.6 mg. was slightly less than half that of the population of FM subjects who took 2.8 mg., it doesn't raise a red flag (Typically, the smaller the sample, the less impressive the level of statistical significance.
- By definition, all of the findings in the three threshold rows are clinically significant.
- There is a myth that male veterans with PTSD will not seek treatment (partly based on the aversive sexual side effects of SSRIs), and if they do seek treatment, they find that it is not effective. However, 84% of those taking 5.6 mg. completed the study, and from the threshold (or responder) findings, it is clear that many responded favorably, at least enough to justify the level of market penetration described in the next section.
- In both BESTFIT and AT-EASE, sleep quality continued to improve during the last time period, while the gains that had initially been reported by those taking the placebo started to reverse. This implies that going longer than 12 weeks might result in further improvement.
- The overall pattern is one that patients, physicians, and insurers will view as highly convincing, and that is before seeing any Phase III results.
My Updated Valuation Assuming 50 Million Shares
Here are four criticisms that have appeared in the comment sections of my previous articles on Tonix:
Some have argued that I have overestimated the level of peak sales for both indications. I am somewhat bewildered by this assertion since I have disclosed and justified my assumptions. If you believe that the results of the analysis are too optimistic, please identify the assumptions that you question. Two assumptions leading to my peak sales estimates have been challenged, and I take each of them up below
Some have argued that I have overestimated the size of the target market. I have used Tonix's frequently publicized estimates for the number of patients receiving or seeking treatment. Its presentations to investors contain these incidents, as well as references to the research articles upon which they are based. One particular variation of this criticism is that on the one hand, Tonix gives population figures for the total number of American adults receiving or seeking treatment for PTSD, but that its just concluded Phase II trial was focused solely upon patients currently or formerly affiliated with the military. My response has been that it seeks a wider label and intend to market TNX-102 SL to both military and civilian populations. On this point, Tonix's June 16, 2016 prospectus is very explicit.
It is also important to remember that all of my target market estimates focus only on adults in the USA. Quite apart from the possibility that Americans in their late teens may develop PTSD, there are also markets outside the USA. Tonix is considering the option of including European sites in its second Phase III trial for FM patients. And recall that Tonix's secondary analysis of its Phase II data for subjects with FM used composite metrics developed and used in Europe.
Some have argued that I have overestimated the extent of market penetration. I have addressed this criticism in two ways.
- In my most recent article, I offered only what I consider conservative projections, based no more than a 6% share of patients actually receiving or seeking treatment (whereas I believe that both indications might actually attain 10% of these markets).
- Instead of displaying separate tables for conservative and aggressive valuations, I have issues a single table that varies the percentage of market penetration for both FM and PTSD from 3% to 6%. This allows readers to look up NPV/share based on their own estimates of the extent of market penetration at peak sales.
Some have argued that I have underestimated the number of shares among which future profits must be distributed. I have consistently taken into account not only the shares on the open market, but also warrants, etc., and projections of the size of future sales of new shares. Thus, in my last article, even though Tonix said that there were 18.9 million shares outstanding, my valuation estimates/share were based upon 26 million shares.
In this article, I am taking into account the following information listed in the prospectus:
- 20,079,475 shares outstanding as of June 15, 2016, including 5,000,00 shares as part of the offering that closed June 21;
- 750,000 shares of common stock issuable upon the exercise of the underwriters' over-allotment option;
- 1,729,217 shares of common stock issuable upon the exercise of warrants outstanding as of June 15, 2016, with a weighted average exercise price of $10.54 per share;
- 2,333,721 shares of common stock issuable upon the exercise of options outstanding as of June 15, 2016, with a weighted average exercise price of $8.89 per share;
- 112,500 shares of common stock issuable upon the vesting of restricted stock units outstanding as of June 15, 2016;
- 1,966,496 shares of common stock reserved for future grants, awards and issuances under our equity compensation plans as of June 15, 2016; and
- 250,406 shares of common stock reserved for future purchases under our employee stock purchase plan as of June 15, 2016.
These amounts to a total of 32,221,815 shares, assuming that the over-allotment is fully exercised. Since we all know that unless both Phase III trials fail, there will need to be additional shares sold, I decided to base my revised estimates for NPV/share on a much larger number of shares. In Table 3, which shows my latest valuation, I increased the 32 million share total by more than 50% to 50 million diluted shares. In my view, by the time the company needs to go beyond issuing another 18 million shares, the chances for commercial success will have increased and the time until peak sales will be shorter, both increasing the NPV/share.
The upper left hand corner shows all of my assumptions. The chances for success are estimated at 80% for both conditions because the Phase III studies are "powered" to 80%, meaning that given all of the data collected so far, the sample size is selected so that the chances that each study will meet its primary endpoint is 80%
As mentioned at the beginning of this section, I have provided you with four conservative levels of market penetration for each diagnosis so that you can see how the NPV/share works out under various scenarios.
The Tonix story keeps on getting better. The company is now in better financial shape, and will be even more so after the next round(s) of money raises that Table 3 anticipates. Tonix's secondary analysis of BESTFIT and its last month's publication of its AT-EASE data make it clear that conditions are highly favorable. I hope that my reorganization and interpretation of these findings provides you with a clearer picture of this company's future prospects.
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.