- Prana is an Australian biotech with a potential treatment for Alzheimer's disease.
- Potential future revenues of at least $5 billion, but market cap is only $295 million.
- Lead compound backed by spectacularly good science.
- Imminent catalyst: results from Phase 2b IMAGINE trial due in March.
- Conservative pre-result risk-adjusted net present value is at least $9.60/share.
Prana Biotechnology Limited (PRAN) is a tiny clinical-stage company nearing completion of a pivotal Phase 2b trial of its compound PBT2 for the treatment of Alzheimer's disease. The company has also completed a Phase 2 study of PBT2 for the treatment of Huntington's disease, with results due at any time. The company's price history has been volatile to say the least, ranging from $2.15 a year ago to a peak of $13.29 on January 30th. A negative note from Summer Street Capital triggered a panic, with consecutive 5-10% daily declines to the current price.
In this note, we address the extensive scientific background supporting PBT2. The peer-reviewed research published in numerous scientific journals strongly predicts a positive result in the IMAGINE trial. In this note, I do not discuss the REACH2HD trial, which is of secondary importance and does not contribute to my valuation estimate.
Prana is chasing what is arguably the biggest prize in pharmacology: Alzheimer's disease. Alzheimer's disease is the sixth leading cause of death in the United States. It is the leading cause of dementia, affects 5.2 million Americans, and is present in about 1/9th of Americans over the age of 65 and 1/3rd of Americans over the age of 85. The total cost of treating Americans with Alzheimer's disease is $203 billion, with some $5.3 billion spent on drug therapies. Five drugs are currently approved for Alzheimer's disease: galantamine, memantine, donepezil, rivastigmine, and tacrine. However, these drugs are minimally effective and only mildly slow the progression of the disease. The approved drugs act on NMDA and cholinergic neurotransmission, but do not affect the underlying pathophysiology of the disease that leads to progressive neurological injury. They are stop-gap measures, and leave much indeed to be desired. No new drugs for Alzheimer's disease have been approved since 2003.
A new drug that altered the course of Alzheimer's disease would have immediate and massive adoption; it would most definitely have sales that would be at least the total sales of all existing Alzheimer's drugs and would quite likely be substantially greater. I use the $5.3 billion current drug expenditure as an extreme lower limit for the potential retail market for a disease-modifying Alzheimer's therapy. Using a 3.5 price/sales rule of thumb as a plausible future valuation, Prana could achieve a market capitalization of $18.5 billion by 2017. When adjusting for time discounting and future dilution, the current market capitalization of $295 million implies that the market pegs the chances of success in the IMAGINE trial at only 7.5%. Institutional ownership is a paltry 3%.
Prana's intellectual property has a long life ahead:
In August 2009, a key patent protecting our clinical drug asset PBT2 was granted by the European Patent Office, or the EPO. The patent entitled '8-Hydroxyquinoline derivatives' covers the composition of matter of selected families of 8-Hydroxyquinoline compounds, including PBT2, and the uses of such compounds for the treatment of neurological diseases, including Alzheimer's disease and Huntington's disease. The European patent has a 20-year term expiring on July 16, 2023, with a possible extension of the term of up to five additional years under supplementary protection provisions. Also in August 2009, we received a notice of allowance from the United States Patent and Trade Mark Office, or USPTO, for our key patent protecting our clinical drug asset PBT2. The patent was granted in November 2009. The U.S. patent, which is also entitled '8-Hydroxyquinoline derivatives,' covers the composition of matter of selected families of 8-Hydroxyquinoline compounds, including PBT2, and will expire on December 21, 2025. It is possible that the patent may be further extended in the future under the pharmaceutical extension of term provisions that apply in the United States. In April 2011, the Japanese Patent Office had granted the same patent, also entitled '8-Hydroxyquinoline derivatives', with the claimed subject matter encompassing compounds and pharmaceutical compositions containing PBT2 and the use of the compounds for the treatment of Alzheimer's disease. The Japanese patent will expire on July 2023 and may be eligible for pharmaceutical extension of patent term for up to a further five years. In November 2011, we received a notice of allowance from the USPTO, for our key patent protecting our product candidate for Parkinson's disease, PBT434. The patent is entitled 'Neurologically Active Compounds' and covers the composition of matter and pharmaceutical compositions of selected families of 8-hydroxyquinazolinone compounds, including PBT434. In March and April 2013, we also received a Notice of Grant from the Canadian Patent Office and European Patent Office, respectively, for our key patent protecting PBT434. The patents, which are entitled, 'Neurologically Active Derivatives' cover the composition of matter of selected quinazolinone compounds, including PBT434. These two cases also included additional granted claims to the use of the compounds for the treatment of neurodegenerative diseases.
Based on the scientific evidence reviewed below, I hold these conservative beliefs:
- There is an 80% chance that the data will support advancing to a Phase 3 clinical trial.
- There is a 20% chance that the Phase 2 IMAGINE trial data will be a home run, with incontrovertible and large improvement in cognitive scores and physiologic data. It is difficult to estimate valuations in this best-case scenario, but I would guess at greater than $2 billion.
- Intermediate results consisting of positive imaging and physiologic data without a measurable improvement in cognition should result in an immediate valuation of $782 million or more. Likelihood: 60%.
- Total failure probability: 20%.
- The true pre-result value is $391 million ($9.60/share), assuming a 25% discount rate, 20% likelihood of a "home run" Phase 2 trial, zero value to all outcomes other than the best-case scenario, and an additional 50% fudge factor to discount for extra conservatism.
There are numerous technical articles on Alzheimer's disease, cerebral amyloid angiopathy, Huntington's disease, and metal dyshomeostasis. This note attempts to provide background that should be sufficient for most investors. Wikipedia has several reasonably decent articles on Alzheimer's disease, Huntington's disease, and the tauopathies for those who need to catch up on the absolute basics. Those interested in a more sophisticated level of understanding should review the following:
Alzheimer's disease is a relentless neurodegenerative disorder predominantly affecting older individuals as an acquired disease. It appears in much younger individuals with Down syndrome (who have 3 copies of chromosome 21), usually striking by the time patients reach their mid-30s. Familial forms due to mutations in presenilin 1 and 2, amyloid beta precursor protein, and apolipoprotein E are well-described in the literature. Symptoms include progressive declines in memory and executive function. Personality changes, paranoia, and hallucinations are common. Superimposed on the gradual neurological decline, many patients experience episodic changes in behavior, often having relatively lucid periods during which they can converse somewhat normally followed by intervals of confusion, agitation, and delirium. The acutely worse symptoms frequently appear at night, with changes in routine, or after sedating medications such as benzodiazepines. Death from aspiration, falls, bedsores, urinary tract infections, or withdrawal of care is the inevitable result.
The triggering event underlying Alzheimer's disease is unknown. There are several hallmarks of the disease in pathology specimens. Grossly, patients have shriveled brains, with volume loss and hypometabolism often most prominent in the temporal and parietal lobes.
(Image source: Prana)
There are extra-cellular "senile plaques" that consist of organized protein fragments of 38-43 amino acids. These plaques have a characteristic appearance under the microscope, and have an apple-green birefringence when stained with Congo Red.
(Modified open-source image from here)
There are also intracellular accumulations of a protein known as tau, which normally stabilizes a structural unit of the cell. The tau protein in these "neurofibrillary tangles" is excessively phosphorylated, with glycogen synthase kinase 3 catalyzing much of the hyperphosphorylation.
(Open-source microscopic image of a neurofibrillary tangle)
The amyloid plaques are comprised of a protein known as beta amyloid (Aβ). Aβ is a group of small polypeptides cleaved from the amyloid precursor protein (APP) by beta- and gamma-secretase. APP is normally a transmembrane protein found near synapses. The normal function of APP is not totally clear, but it appears to be involved in iron regulation, recovery from brain trauma, and in the metabolism of catecholamines. When APP is cleaved to yield Aβ, the fragments range in size from 38-43 amino acids; the 42-amino acid fragment is most commonly found in the insoluble amyloid plaques.
The metals hypothesis of Alzheimer's disease
Some researchers believe that the amyloid plaques are toxic; others suggest that soluble Aβ fragments injure neurons; still others believe that the toxic effect is mediated by neurofibrillary tangles of hyperphosphorylated tau. These multiple hypotheses are summarized here.
Unifying all of these abnormalities is the "metal hypothesis," which is the working diagnosis at Prana. In 1994, Bush, Tanzi, and colleagues published in Science a finding that zinc ions stabilize human beta-amyloid aggregates. Zinc is found in high levels within senile plaques, which also demonstrate high levels of zinc transporters such as the ZnT3 protein. ZnT3, a transporter that pumps zinc into presynaptic vesicles for release during neurotransmission, is concentrated in zinc-enriched terminals and near cortical capillaries. Furthermore, in mice with mutations in the ZnT3 gene, formation of senile plaques is inhibited. Transgenic mice expressing human amyloid precursor protein and mutant human mutant presenilin 1 develop amyloid plaques by 6-7 months. These APP/PS1 mice also demonstrate increased expression of almost all zinc transporters in the ZnT family, especially ZnT3.
A paper in the Journal of Biological Chemistry explored the mechanism relating zinc, Aβ, and neuronal toxicity. Oligomeric spheres of Aβ1-40 or Aβ1-42 demonstrate potent toxicity to neurons at concentrations below 1 x 10-9 molar. Using electron microscopy and circular dichroism measurements, researchers demonstrated that zinc stabilized an oligomeric form of Aβ. Zinc-Aβ complexes suppressed neuron action potentials and caused neuronal death in cell culture. The fibrillar form of Aβ, which predominates in senile plaques, binds zinc but is significantly less toxic than the soluble, oligomeric forms.
Exactly how the biochemistry of zinc, copper, and Aβ work to create neuronal toxicity in vivo has not been fully elucidated. Ex vivo experiments demonstrate that Aβ complexed with copper generates highly reactive oxygen species such as hydroxyl radical in a Fenton-type reaction. In particular, Guilloreau 2007 demonstrated that copper complexed with Aβ peptides generated hydrogen peroxide, which then splits to yield the even more reactive hydroxyl radical. The reducing agent in this particular experiment was ascorbate, which is found at high concentrations in neurons, cerebrospinal fluid, and the interstitial spaces of the brain; multiple other common, biologically relevant reducing agents can furnish the electrons for the metal-Aβ-catalyzed generation of reactive oxygen species. The 42-amino-acid form of Aβ was more effective at generating reactive oxygen species than the 40-amino acid form. Furthermore, using size-exclusion chromatography, the researchers found that intermediate-sized aggregates, rather than monomers or large aggregates, were responsible for the generation of reactive oxygen species.
Sequestration of zinc as physiologically unavailable zinc-amyloid plaques may also play a significant role in the pathophysiology of Alzheimer's disease. Synaptic zinc release, which requires the function of ZnT3, is essential for cognition and promotes dendritic spine growth. When zinc is sequestered, calcineurin activity decreases, glycogen synthase kinase 3 becomes disinhibited, tau becomes hyperphosphorylated, and zinc-dependent microtubules become unstable. In addition, Aβ rises as matrix-metalloprotease-dependent clearance mechanisms fail. The rising Aβ oligomerizes; although it usually binds zinc, high levels of Aβ in the presence of low zinc levels may bind copper instead due to the lack of competition, generating reactive oxygen species. Finally, the high levels of uncleared Aβ precipitate as senile plaques, which sequester more zinc, creating a pathological positive feedback loop.
PBT2: Success in mice and men
PBT2 is a chlorinated hydroxyquinoline derivative that functions by binding sequestered zinc and/or copper and making the metals physiologically available. PBT2 also probably pulls zinc from Aβ oligomers and polymers, reducing the toxic effects of the higher-order Aβ structures and rendering them amenable to clearance by physiologic pathways. PBT2 is the second-generation metal-protein attenuation compound put forward by Prana, the first being clioquinol. Clioquinol was a promising beginning, but was abandoned for a variety of reasons. (Clioquinol was previously withdrawn from the market due to its association with subacute myelo-optic neuropathy.) From a medicinal chemistry perspective, PBT2 is near-ideal as a drug prospect. With a low molecular weight of only 271.14, PBT2 is trivial to synthesize chemically. There is no cis-trans isomerism or chirality to complicate synthesis or purification. PBT2 will be cheap and easy to manufacture at large scale. It is a highly stable, minimally reactive molecule with excellent penetration into the brain. To date, no clinically significant toxicities have been identified with PBT2.
(Structure of PBT2)
PBT2's mechanism of action was explored at the biochemical level in a superb paper by Crouch et al. Recall that glycogen synthase kinase 3 (GSK3) contributes to the hyperphosphorylation of tau. Glycogen synthase kinase 3 itself is regulated through phosphorylation, with phosphorylation of particular serine residues causing a reduction in GSK hyperphosphorylation of tau. Crouch demonstrated that treatment of cell cultures with PBT2 dramatically increases phosphorylation of GSK. When cultured cells were washed with zinc-free media, the addition of free zinc ions + PBT2 induced phosphorylation of GSK, whereas the addition of zinc alone or PBT2 alone had no effect. PBT2 increased intracellular zinc levels, as demonstrated by fluorescence measurements using a zinc-sensitive dye. Phosphorylation of GSK3 could be achieved by treating cells with Aβ:zinc aggregates and PBT2. In other words, PBT2 can deliver intracellular zinc using Aβ:zinc aggregates as a zinc source. In fact, Aβ:zinc aggregates were as good as free zinc ions in terms of serving as a source for intracellular zinc delivery by PBT2.
Crouch had one other key finding: PBT2 can trigger the destruction of Aβ:zinc aggregates by matrix metalloproteases. Aggregates of Aβ and zinc were resistant to lysis by matrix metalloproteases in the absence of PBT2. In the presence of PBT2, Aβ:zinc aggregates became susceptible to cleavage by matrix metalloproteases. This biochemical work gives a clear mechanism for the clearance of senile plaques by PBT2 seen in vivo.
The in vivo experimental data for PBT2 is compelling. Unlike clioquinol, PBT2 has excellent penetration across the blood-brain barrier. In mice, the brain:plasma concentration ratio is 3.6:1. Assuming similar absorption in humans and mice, a 250 mg dose should achieve brain concentrations of around 0.1 micromolar. Data from cell cultures indicate saturation of biochemical activity at concentrations of less than 2.5 micromolar. Adlard and colleagues made the following critical findings:
- PBT2 reduces reactive oxygen species generation from Aβ:metal complexes.
- PBT2 dissolves precipitated Aβ:metal complexes.
- PBT2 reverses Aβ's inhibition of long-term potentiation in hippocampal slices.
- In mice over-expressing amyloid precursor protein (Tg2576 mice), PBT2 reduces soluble Aβ 37%, insoluble Aβ 30%, and senile plaques by 80%.
- PBT2 reduces oligomeric Aβ.
- In APP/PS1 mixed transgenic mice, PBT2 decreased Aβ in the interstitial fluid by 81% at 20 hours after a single oral dose. In Tg2576 mice, PBT2 decreased Aβ in the interstitial fluid by 32% at 16 hours after a single oral dose.
- After 6 days of treatment with PBT2, APP/PS1 mice demonstrated faster learning (p < 0.0001) and better memory in a water maze model (p < 0.01). There was no effect on normal mice treated with PBT2.
The bottom line is that biochemical, histopathological, and behavioral effects were seen after treatment, even relatively short-term treatment. The researchers also examined clioquinol and found it to be universally less effective than PBT2 in the above models.
In separate experiments, PBT2 increased dendritic spine density, an effect that was abolished with the cell-impermeant caged polyamine metal chelator diaminosarcophagine (diamsar). Animal data not discussed in this article also shows improved behavioral responses, larger brain volume, and decreased abnormal weight loss in a mouse model of Huntington's disease.
In his 2008 review, Rudy Tanzi writes:
With regard to mechanism, in mice, CQ [cliquinol] is understood to enter the brain and to combine with metallated Aβ in plaques and possibly, soluble pools. CQ treatment of transgenic mice modestly increased brain zinc and copper levels, and in the Phase 2 clinical trial in AD patients the plasma zinc levels significantly increased (normalized from a baseline of deficiency), therefore CQ (and PBT2) do not act as chelators. In cell culture, CQ-Cu complexes enter cells where they markedly inhibit the secretion of Aβ by a mechanism where the peptide is degraded through up-regulation of matrix metalloprotease MMP-2 and MMP-3. MMP activity was increased through activation of phosphoinositol 3-kinase and JNK. CQ-Cu also promoted phosphorylation of glycogen synthase kinase-3, and this potentiated activation of JNK and degradation of Aβ1-40.
We propose a mechanism of action for treatment of AD where CQ or PBT2 enters the brain and is attracted to the extracellular pool of metals that are in a dissociable equilibrium with Aβ; e.g., in senile plaques and oligomers CQ and PBT2 then bind zinc and copper in the Aβ deposits, possibly forming a ternary complex with Aβ. We have previously seen that stripping metals away from Aβ leads to dissolution of Aβ aggregates back down to monomer. Aβ monomer can then be readily cleared from the brain or degraded. Along these lines, an alternative mechanism of action involves the drug-metal complex entering the cell. This then activates MMPs and facilitates the clearance of Aβ; e.g., in the synapse. In both cases, CQ and PBT2 would also attenuate oxidative cross-linking of Aβ oligomers into covalently bonded species, and reduce the neurotoxic redox activity of Aβ oligomers. Thus, in essence, the MPACS, CQ and PBT2, most likely block Aβ oligomerization and aggregation, dissolve non-cross-linked Aβ aggregates, induce peptidolytic degradation of Aβ, and neutralize Aβ redox activity.
Prior clinical trial results
The data from Prana's Phase 2a trial is excellent. The randomized, double-blind, placebo-controlled trial recruited 78 patients with early Alzheimer's disease and assigned them to 0, 50, or 250 mg of PBT2 for a period of 12 weeks. At 250 mg per day, statistically significant improvements in executive function (as measured by a neuropsychological test battery) but not memory were seen. Importantly, results were diluted by a relatively large placebo effect on the neuropsychological testing. CSF Aβ42values decreased; CSF Aβ40 demonstrated a non-significant trend toward decrease. There was no change in CSF tau levels.
A supplementary analysis of the data was performed in 2010 using a method from the Phase 2 trials in donepezil's path to FDA approval. The full quotation:
To apply a statistical analysis to the distribution of improvements induced by PBT2 250 mg treatment, we performed a ROC analysis adapted to clinical trials. This test analyzed whether treatment with PBT2 250 mg was more likely than placebo to induce an improvement of any dimension in the performance readouts. In this analysis, all the 12-week responses (differences from pretreatment) for a given z-score test were ranked, and then the probability of the response coming from the PBT2 250 mg treatment group (True positive) or the placebo group (False positive) was calculated. This revealed that improvement of any dimension on the NTB Composite z-score was significantly more likely when on PBT2 than on placebo (AUC = 0.76, p = 0.0007, Fig. 2A). The curve also revealed that the greatest likelihood of differences between the PBT2 treatment group and the placebo group were in the range of a z-score of 0.49-0.71 above the pretreatment performance score (Fig. 2A). The ROC for the NTB Executive Factor z-scores revealed that improvement of any dimension was also more likely in thePBT2 treatment group than in the placebo group (AUC = 0.93, p = 1.3 × 10-9), with the greatest likelihood of differences between the PBT2 treatment group and the placebo group at a z-score of 0.2 above the pretreatment performance score (Fig. 2C). The ROC of the Memory Factor z-scores found no significant effects of either PBT2 or placebo: there was a trend for improvement in the z-score range of 0.99-1.31, the larger end of the response rank, to be more likely in the PBT2-treated group (Fig. 2B).
In graphical form, it is visually apparent that a true treatment effect is present in the executive function. Please see Figure 1C of the supplementary analysis.
The results are all the more impressive when one considers the small sample size (29 patients on placebo, 20 patients on 50 mg PBT2, and 29 patients on 250 mg PBT2) and very short study duration compared to the protracted course typical of Alzheimer's disease. The rapid improvement seen in humans in the Phase 2a trial is concordant with the rapid responses seen in mouse models of Alzheimer's disease in response to PBT2 treatment.
The Phase 2a trial data for PBT2 should be interpreted in the context of previous data from a randomized, double-blind, placebo-controlled clinical trial of clioquinol published in 2003. In the clioquinol study, 36 patients were randomized to clioquinol or placebo, and were studied for a period of 36 weeks. As in the PBT2 study, patients had early Alzheimer's disease and were evaluated using neuropsychological testing and physiologic measurements. In the clioquinol study, a significant reduction in CSF Aβ42 levels was observed (p < 0.05). A trend that did not achieve significance was seen in slowing cognitive deterioration on the ADAS-cog. Given that we now know that PBT2 is much more potent than clioquinol at the biochemical level and has much higher penetration into the brain, the clioquinol data imply that we are very likely to see a decrease in CSF Aβ42 levels in the IMAGINE trial.
Based on the results from the clioquinol and Phase 2a trials, it is possible that cognitive improvement will be seen in the Phase 2b PBT2 IMAGINE trial. However, the relatively small sample size and short duration again increase the likelihood that the study will not be adequately powered to detect the small changes between treatment and control groups early in the course of Alzheimer's disease.
What does success look like?
Before the clinical trial results are returned, it is important to make a judgment about what outcomes are possible and how they prognosticate PBT2's prospects.
My predictions for the most likely outcome of the IMAGINE trial:
- No change in memory function on neuropsychological assessment
- Possible modest improvement in executive function
- Statistically significant decrease in CSF Aβ levels
- Decrease in amyloid burden on Pittsburgh B PET imaging and possible partial resolution of hypometabolism on fluorodeoxyglucose PET imaging
- No change in brain volume by MRI
My predictions are based on the Phase 2a result and our knowledge of PBT2's interaction with Aβ physiology. It is my belief that the decline in memory function seen with Alzheimer's occurs slowly over time and 1-year clinical trial is probably too short to capture any difference between treatment and placebo.
My best guess about the improvement in executive function seen in the Phase 2a trial is that it partially reflects an improvement in the acute neurotoxic effect (perceived clinically as delirium superimposed on baseline cognitive decline later in the disease course) that is reduced by PBT2. Therefore, we are reasonably likely to capture a benefit in a 1-year trial, just as seen in the 12-week Phase 2a trial.
The decrease in CSF Aβ levels is attributable to better handling of Aβ by matrix metalloprotease-dependent pathways, well-documented in prior experiments and in the Phase 2a trial. We should also see lower amyloid burden with Pittsburgh B compound PET imaging to correspond to the changes seen in animal models. I anticipate that the formation of new senile plaques will be reduced with PBT2. There should also be clearing of some pre-existing amyloid plaques as the stabilizing divalent metal ions are stripped away. Incomplete clearing should be our expectation, however, as chronic senile plaques develop covalent crosslinks, that will probably render them resistant to proteases.
Finally, it is unlikely that any measurable difference in brain volume will be seen. I look at brain MRIs on a regular basis and believe that the brain volume loss seen in Alzheimer's disease progresses too slowly to be measurable in a 1-year trial. Likewise, finding a difference in fluorodeoxyglucose PET imaging between treatment and control groups is unlikely given the expected slow progression hypometabolism and likely irreversible nature of the underlying neuronal injury.
The FDA has given important guidance on what is needed for a new drug approval in Alzheimer's disease. The FDA demands a cognitive or functional benefit for a new drug, but also acknowledges that the goal is to develop drugs that forestall the development of dementia. Early in the disease, cognitive or functional metrics are inherently similar to healthy controls, limiting our ability to measure differences due to interventions. According to the FDA:
We are open to considering the argument that a positive biomarker result (generally included as a secondary outcome measure in a trial) in combination with a positive finding on a primary clinical outcome measure may support a claim of disease modification in AD. For this to be the case, however, there should be widespread evidence-based agreement in the research community that the chosen biomarker reflects a pathophysiologic entity that is fundamental to the underlying disease process.
Reading between the lines, the FDA will accept a very modest cognitive or functional improvement if it is accompanied by compelling physiologic data. I believe that any improvement in executive function scores on neuropsychological testing, in addition to evidence of altered Aβ physiology (either by CSF Aβ concentration measurements, decreased plaque burden with Pittsburgh B PET imaging, or restoration of normal metabolism on FDG PET imaging) will be adequate for initial drug approval.
The critical finding in the Phase 2b trial will be some proof that PBT2 alters Aβ pathophysiology. Neuropsychological testing is inherently challenging due to large placebo effects and the ill-defined nature of the functions being tested. On a relatively short-term clinical trial with fewer than 100 experimental subjects, the differences in disease progression between the treated and untreated patients may not be adequate to escape from wide error bars in cognitive testing. Conversely, placebo effects are unlikely in CSF Aβ measurements or PET imaging findings. The questions that we want answers to are:
- Is PBT2 doing something to the pathophysiology?
- If we measure differences in neuropsychological testing that are less than statistically significant, do the placebo effects, trends, and error bars give us practical guidance about how large a sample we would need in the Phase 3 trial to identify a difference between treated patients and controls, if a difference truly exists?
Hence, in this small, short-term trial, the real weight will be on the physiologic parameters. Even through we can't measure Aβ oligomers directly, PET imaging is probably a reasonable proxy for Aβ clearance. Although it is less than ideal, it is the best we have at present. I speculate that copper-62, a positron emitter which has been described in the nuclear medicine literature, may be a helpful imaging agent in the future. Alternatively, zinc-62, other divalent positron/gamma ray emitters, or bromine-76 hydroxyquinoline derivatives might be useful to image the response to PBT2 or other ionophores.
The outcomes we could see in the IMAGINE trial can be stratified:
- Best outcome (20% likelihood): decreased CSF Aβ, decreased plaques and hypometabolism on PET imaging, and improved neuropsychological scores --> proceed to Phase 3 with certainty of success;
- Next best (60% likelihood): decreased CSF Aβ, decrease in plaques on PET imaging, no measurable change in cognition --> proceed to Phase 3, with intermediate level of confidence;
- Disaster (20% likelihood): no measurable change in physiology or neuropsychological testing --> game over.
If Prana moves forward with Phase 3 trials, we are very likely to see positive cognitive results in a large 2-3 year trial even if no change is detectable in the small, short-duration trials performed thus far.
Risks and mistakes
The risks of playing a binary event in biotech are enormous. If Prana's Phase 2 trial has a totally negative result, the stock will have zero value. For it to be worthwhile to play this risk, the rewards have to be commensurately enormous. Should the Phase 2 trial be unambiguously positive, Prana could easily rise 500-1000% in a day. By far, the largest risk in Prana is that the clinical trials will fail.
Alzheimer's disease has been a graveyard for promising drug prospects. Seeking Alpha user Julia Skripka-Serry of BioAssociate Innovative Consulting documents the implosion of Big Pharma's Alzheimer's pipeline in "The great neuro-pipeline 'brain drain." Recent failures include anti-amyloid immunotherapies Gammagard, bapineuzumab, and solanezumab; gamma-secretase inhibitors/modulators semagacestat and flurizan; and multi-action dimebon. The issue behind these failures is that they have focused on clearing the senile plaques, which are not the primary toxic species injuring neurons in Alzheimer's disease.
The competitive landscape is complex. Existing Alzheimer's therapies are not relevant to Prana because of their dramatically poor efficacy and failure to modify the fundamental pathophysiology. Taurx, a privately held company, has a methylene blue derivative that is currently in Phase 3 clinical trial. The drug apparently reduces the hyperphosphorylation of tau protein, but so far the company has failed to publish any Phase 2 results. Taurx's refusal to submit its results to peer review strongly suggests that the company's results are too weak for public scrutiny. Merck (MRK) has a beta-secretase inhibitor MK-8931 that is in Phase 2/3 trial targeted for results in April 2017. MK-8931 recently completed the Phase 2 portion of the trial (results not publicly available) and is in the Phase 3 portion of the study. The drug is known to reduce CSF Aβ concentrations, and avoids the liver injury that torpedoed Lilly's beta-secretase inhibitor. MK-8931 is probably the biggest competitive threat. At present, none of Prana's competitors address the metal dyshomeostasis that appears to be at the heart of Alzheimer's disease pathophysiology.
As if the scientific risk were not enough, Prana itself has managed to amplify the business risk through at least one and possibly two execution errors during clinical trials. During Prana's Phase 2a trial data analysis, erroneous scoring of the neuropsychological tests was discovered after publication:
In this study, which measured the effect of two doses of PBT2 (50 mg and 250 mg) against placebo in patients with mild Alzheimer's disease, cognition was principally assessed using a neuropsychological test battery (NTB). Grouping all nine tests of the NTB using Z statistics generates a composite Z score, and grouping the five executive tests generates an executive factor Z score. There was an error in how the composite and executive factor Z scores were calculated. The controlled word association test (COWAT) is a constituent test of the executive factor Z score in which the patient is asked to list as many words beginning with a certain letter as possible in 60 s; this is repeated for three different letters. In the published paper, each letter score contributed to the Z score, whereas the correct analysis should have included only the sum of the three letters in the generation of the Z statistic. In the composite Z score, the difference in least squares mean change from baseline at week 12 for PBT2 50 mg compared with placebo was 0・07 (95% CI -0・13 to 0・28; p=0・485) and for PBT2 250 mg compared with placebo was 0・15 (-0・08 to 0・37; p=0・193). In the executive factor Z score, the difference in least squares mean change from baseline at week 12 for PBT2 50 mg compared with placebo was 0・18 (-0・06 to 0・42; p=0・137) and for PBT2 250 mg compared with placebo was 0・27 (0・01 to 0・53; p=0・042). Parts C and E of Figure 3 of the main paper should be as shown here. Data are least squares means (+/- SE).
Taken in total, especially in context of the data re-analysis published in the Journal of Alzheimer's Disease (2010), I don't think that this mistake calls into question the fundamental findings in the Phase 2a trial.
In the REACH2HD trial, a mysterious delay in releasing the data calls into question whether a major data collection, processing, or reporting error occurred. Instead of introducing my own biases here, I offer a press release and an SEC filing with full context, so that you may make your own inferences:
The Reach 2HD trial is a six month double-blind placebo controlled Phase 2 trial on 109 early-to-mid stage Huntington's disease patients. The trial was successfully completed at the end of July 2013 with 95% of participants completing the entire six months of treatment. There has been a delay in finalising the database to achieve 'database lock', required before statistical analysis of the data may begin. The results, originally anticipated in the last quarter of 2013, are now expected to be reported early in 2014.
"Apart from the timing delay, which is disappointing, nothing has changed. The trial was conducted and completed to protocol, and will provide the robust data needed to meet with the FDA in 2014 as we prepare for the next PBT2 trial," said Geoffrey Kempler, Chairman and Chief Executive Officer.
SEC filing (page 23):
In addition to the current activities to initiate an imaging trial in Alzheimer's patients, in late 2012 we finalized the enrolment to a Phase II trial to test PBT2 in patients with Huntington's disease. The trial, known as "Reach2HD", is being undertaken under an open IND application through the FDA and is being conducted in clinical sites across the United States and Australia. The Phase IIa trial design entails a double blind placebo controlled study of 109 patients with early to mid-stage Huntington Disease. The trial will investigate the effect of PBT2 on cognition, behaviour, functional capacity, motor effects and safety and tolerability measures. In addition, an exploratory arm of the study, under the guidance of the co-Principal Investigator of the study, Professor Diana Rosas, will involve MRI brain imaging to undertake iron mapping in a patient's brain. Professor Rosas has published that iron and other metals change in concentration and distribution in the brain with increasing severity of the condition. This study is the first clinical trial with PBT2 in this patient population. We completed the study at the end of July 2013 and project reporting out in early 2014, a delay from the fourth quarter reporting of results that was previously anticipated. The decision to delay reporting permits additional time for us to reconcile data inconsistencies between the source data and the database prior to database lock. This 'cleaning' of clinical trial data is a normal and necessary process to ensure database integrity ahead of executing the statistical analysis on the data contained in the locked database. One of the steps to ensure database integrity being undertaken is the re-entry of the original source data from all of the sites into a database and checking the veracity of the data within the database. We believe this is a prudent step as we prepare for an end of Phase II meeting with the FDA during 2014.
At present, although there is significant animal and in vitro data supporting the use of PBT2 in Huntington's disease, I do not include the possibility that PBT2 will be successful in the REACH2HD trial in my valuation of Prana.
While investors would prefer near-flawless clinical trials, imperfections in clinical trials are exceedingly common. They are not fatal unless they truly alter the experimental results, and it is definitely better that the company be forthright about errors rather than try to obscure mistakes. If I did not include discussion of these mistakes in this article, my objectivity would be in serious doubt.
Prana does not currently have the financial resources to prosecute Phase 3 trials. With $10 million or less on hand, it is not physically possible for the company to move forward without a secondary (or being acquired). The company is currently running a roadshow in anticipation of an offering after the REACH2HD and IMAGINE trial results come back. I have included this dilution in my valuation estimates.
The last risk that I wish to highlight is the very real possibility that I am way off. Although I have made every effort to present the scientific facts as I see them in the literature, I accept the notion that I could have interpreted them incorrectly, read the literature incompletely, or made other mistakes in my information-gathering. It is also possible that even if my reading and experimental interpretation are correct, the clinical trials simply may not work out. If we knew in advance with certainty that the clinical trials would succeed, we would not need to perform them. The best policy is to practice humility in the face of capital markets and scientific uncertainty.
If you identify any factual errors in this article, please have the kindness to call them to my attention. The topic's complexity all but guarantees I made some mistakes, hopefully none of which will prove to be material. Acknowledging my limitations and failings, I hope that I have landed not entirely on the wrong side of Plato's maxim:
Wise men speak because they have something to say; Fools because they have to say something.
Prana's quest is a difficult one, yet the company is pursuing drugs that could easily yield annual revenues in excess of $5 billion. Prana has by far the best science backing its lead compound of all the companies trying to treat Alzheimer's disease, and for good reason: many of the most important academic discoveries in Alzheimer's disease were made by Prana's scientists. The mechanism of action behind PBT2 is not completely understood, but the in vitro data, animal models, and human trials have coalesced into a powerful story that favors eventual success with a high probability. Because the IMAGINE trial was relatively small and quite short in duration relative to the typical course of Alzheimer's disease, I assign a modest probability to the "home run" result-measurable changes in physiology combined with incontrovertible and clinically significant cognitive improvement. However, it is very likely (~80% likelihood) that the data will support advancement into Phase 3 clinical trials. Now that some investors have panicked and the stock price has cratered, those who take a serious and critical look at the science should perceive massive, unappreciated value in Prana. PBT2 is not the fountain of youth, but it might very well prove to be the elixir of a more humane old age.
Additional disclosure: This is a volatile stock and I may change my positions at any time. Rely on your own due diligence for all of your investment decisions.