Prana Biotechnology (stock ticker PRAN) are a small-cap Australian biotech firm focused on neurodegenerative diseases. Co-founded in 1997 by Professor Rudy Tanzi of Harvard Medical School (who as of 21 February 2014 remained a paid consultant and shareholder), the company is focused on exploiting his "metal hypothesis of neurodegenerative disease" . This theory postulates that metals such as iron, copper, and zinc are implicated in the pathogenesis of neurodegenerative diseases, including Alzheimer's. It is hence proposed that by targeting these metals the course of several diseases could potentially be slowed or reversed.
For those interested in the scientific background, comprehensive coverage was provided by a January 2013 New York Academy of Sciences eBriefing  which was co-sponsored by Prana and at which Prof. Tanzi was a co-organiser and speaker. In brief, Prana's lead compound PBT2 is believed to act as an ionophore that mobilises extracellular metal ions, facilitating neuronal uptake.
Clinical Background in Alzheimer's Disease
On the 26th February 2008 Prana reported Phase 2a results for PBT2 in Alzheimer's. In this disease, copper and zinc are thought to be involved in the formation of toxic Abeta (amyloid) aggregates in the brain. 78 participants were randomly assigned (29 to placebo, 20 to PBT2 50 mg, and 29 to PBT2 250 mg) with 74 completing the study . There were no serious adverse events and the drug was declared safe in this trial population. Statistically significant reductions in Abeta were seen in the Cerebrospinal fluid (NASDAQ:CSF) at 250mg PBT2. No effect was observed on the industry standard ADAS-Cog and MMSE neuropsychological scores of cognitive function. Analysis of a less rigorous/more sensitive neuropsychological test battery (NYSE:NTB) showed statistically significant benefits in two executive function tests at 250mg PBT2. These were the Category Fluency (p=0.041) and Trail Making Part B tests (p=0.009).
Much was made of these NTB data in a catchily-titled, follow-on analysis "PBT2 rapidly improves cognition in Alzheimer's Disease" . Those who are excited by this statistical reanalysis would, however, do well to also read the brutally worded Cochrane Summary on the efficacy of PBT2 and its precursor PBT1. In truth the trial simply wasn't powered for efficacy. The fair conclusion was that PBT2 was safe in Alzheimer's patients with a much larger Phase 3 trial being needed to confirm drug benefits. It is worth noting that the data were insufficient to secure Big Pharma interest at a time when questionable anti-amyloid drugs such as flurizan were managing to find partners.
Reach2HD: Huntington's Disease Phase 2a
Fast-forward to 18th February 2014 and Prana had another Phase 2a to report. This time it was the turn of the Reach2HD trial in Huntington's disease - the very definition of an unmet medical need. The sample size was slightly larger than the previous Alzheimer's trial with 104 participants completing. 34 were randomised to placebo, 38 to 100mg PBT2 and 32 to 250mg PBT2. My own summary of the study data is available for download in pdf format.
The primary endpoint of the trial was safety in Huntington's patients. Prana stated that this had been met successfully before passing to the analysis of secondary endpoints. Given the previous Phase 2a in Alzheimer's, confirmation of safety in a further disease population was never going to excite observers. I would, however, have liked to have heard a little more. There were ten serious adverse events (SAEs) reported. 1 occurred in the placebo group, 3 with 100mg PBT2 and 6 with 250mg PBT2. To an uninformed observer this unfortunately looked rather like a dose-response. In fact, only one patient (on 250mg PBT2) was deemed by the site investigator to have suffered an SAE as a result of drug treatment. Taking that at face value, even the single SAE is still potentially worrying. A drug-related SAE rate of 1% would equate to 10,000 additional SAEs per million patients treated. Nevertheless, the Reach2HD trial was stated to have met its primary endpoint and was hence a success.
Inevitably though, the audience's main interest was in efficacy. This was always going to be a difficult debate for two reasons. Firstly, the trial was again underpowered. A much larger Phase 3 study will be needed to conclusively demonstrate any positive effects of PBT2. Secondly, the secondary efficacy endpoints were simply bewildering in their complexity. The main secondaries focused on cognition. Three composite scores were tested comprising differing groupings of eight component tests. There was also a pre-specified sub-analysis in patients with mild Huntington's disease. As deficits in Huntington's are arguably more about motor function than cognition, Prana also tested this along with behaviour and functional abilities. They carried out Global Assessments and assayed biomarkers. Finally, there was a tiny exploratory imaging sub study (2 on placebo vs. 4 on drug) looking at MRI brain volumes and function.
Ignoring encouraging, but statistically insignificant, trends only two readouts from this slew of data demonstrated improvements. The Executive Function Composite score, comprising the Category Fluency and Trail Making Part B tests highlighted in the Alzheimer's Phase 2a, showed improvement in the pre-specified mild Huntington's disease sub-group analysis only (p=0.038, 250mg PBT2). Unfortunately, as Prana chose not to reveal the number of patients in that sub-group the utility of these results is somewhat limited. The second positive was that, in isolation, the Trail Making Part B test showed improvement across the whole trial population (p=0.042, 250mg PBT2). It is hence assumed that this test drove the benefits seen in the mild Huntington's sub-group.
One might argue that if Prana hadn't tested so many variables they would have had little chance of detecting efficacy. Not to mention seeing results in one of the same tests, and at the same PBT2 dose, as suggested by the the Alzheimer's Phase 2a. Unfortunately things aren't quite that simple. I apologise to any statisticians for what follows - you'll already know it all. But, for the armchair reader, the analysis of clinical trials may seem a somewhat opaque art.
The best trial is one that uses a single readout to compare a single dose of a drug against placebo. If the probability of the null hypothesis (that effects interpreted as drug efficacy can naturally occur by chance) is then less than 5% (p < 0.05) we can argue that the drug worked. Things get a lot more complicated when multiple drug doses, and multiple efficacy measures are employed. Intuitively, even if a given experimental outcome is very unlikely to occur, the fact that the experiment is repeated multiple times will increase the probability that it crops up at least once. If you test several independent null hypotheses and leave the significance threshold at 0.05 for each comparison, the chance of obtaining at least one "statistically significant" result is greater than 5% (even if all null hypotheses are true). Failure to correct for this phenomenon, whilst still claiming significance, can result in a Type I error, or the failure to prove the null hypothesis.
This is an old problem in clinical trials with the solutions normally being focused on interpreting primary endpoints (which in Reach2HD was actually safety). Generally if one does not require all efficacy measures to be positive to demonstrate drug effectiveness (as is the case here) each endpoint should be tested with a significance level corrected for multiplicity. Bonferroni correction (with various improvements to sensitivity) is a widely used technique . This is, however, rather a blunt statistical tool and can lead to subtle efficacy being suppressed by the process of sanitising the statistics. One solution to the multiplicity problem is hence to use composite scores - bundling several related measures together into one readout.
Measures of cognitive executive function generally move in step to some degree or another , hence the usage of composite executive scores rather than single tests. This makes Prana's findings a little unusual as they saw benefit on only a single measure of executive function. Whether this will be clinically meaningful if born out by larger trials remains to be determined. Regardless, it is not generally a good idea to use several closely-related composites (as Prana did) because it simply reintroduces the multiplicity problem. It also seems fairly futile to analyse the components of the composite scores separately - otherwise why bother with the composites?
It is tempting to gloss over this mathematical argument and focus simply on the headline results. Sadly the casual audience are not really the ones who matter. Drug regulatory agencies examining Prana's data would demand rigorous statistical correction before accepting any claims of efficacy. The bottom line is that, in order to claim statistical significance on the Trail Making Part B test, Prana must have done some pretty fancy stats. They're certainly capable of it - Prof. Tanzi routinely runs genome-wide association studies which by necessity require correction for multiplicity. Unfortunately Prana chose not to discuss their methods. This may have been somewhat of a public relations error.
The experienced clinicians Prof. Paul Aisen (Director, Alzheimer's Disease Cooperative Study) and Prof. Rachelle Doody (dimebon clinical lead) were quoted by Alzforum as non-committal:
"Paul Aisen, University of California, San Diego, pointed out that these results cannot be interpreted as suggesting efficacy, however, because the Trail Making Test was the only one of eight cognitive tests administered that showed a positive effect, and also because Prana did not control for false positives on this outcome. 'Analyses of anything having to do with cognition or function did not meet the pre-specified objectives,' added Doody. 'The tertiary analyses of various subcomponents are inconclusive and uninterpretable.'"
Researchers involved with Prana unsurprisingly hold different views. Frustratingly, they didn't provide us with sufficient methodological information for independent judgement.
It's not like we haven't been here before. Medivation (stock ticker MDVN), for whom both Professors Aisen and Doody both consulted, were responsible for bringing the experimental drug dimebon to tender. They had excellent Phase 2 results in Alzheimer's disease - greatly superior to those obtained by Prana - with a statistically significant benefit (p=0.0005) in the ADAS-Cog gold-standard . They also had weaker but still impressive Phase 2 data in Huntington's disease with a significant improvement in MMSE (p=0.03)  - again superior to Prana. Medivation subsequently secured a $725m deal with Pfizer involving an impressive $225 million up-front cash payment. I don't need to tell you what happened next. Dimebon failed in every Phase 3 trial it was entered into. Whilst obviously not indicative of Prana's future chances of success this should be a cautionary tale for any patient wishing to believe that Prana have found a miracle cure. It's certainly a cautionary tale for Big Pharma.
Given the borderline statistical significance of the Reach2HD findings, acceptance of Prana's efficacy claims is a matter of trust. Trust that their statistical analysis was rigorous and not simply cleverly slanted at maintaining some level of significance in the data. Faith in the company must inevitably be balanced against the commercial necessity for their results to be well received. Prana has two problems going forward - money and time.
As of the end of 2013 Prana had enough cash to last around a year at their current rate of expenditure. Certainly not enough to go anywhere near a large Phase 3 trial in any neurodegenerative disease indication. And if a large Phase 3 is indeed required, they need to hurry up. The US patent on PBT2 expires in 2025. 11 years may seem like a long time but it really isn't in drug discovery. It took the amyloid-targeting, Alzheimer's drug bapineuzumab four years to progress from the Phase 2 announcement to the point at which the Phase 3 trials were declared a failure - and that was with the efficient backing of some of the biggest names in pharmaceuticals. Prana will not be that quick. Add in the time required for regulatory approval and 2020 could easily tick round well before the first pill hits the pharmacy shelves. Of course, patent extensions gained by hook or by crook are not unheard of. This narrow window of profitability will, however, be very much be on the minds of anyone looking to make a deal with Prana.
IMAGINE: Alzheimer's Disease Phase 2b
Prana may yet have one more ace up their sleeves - the IMAGINE Phase 2b trial in Alzheimer's disease which is due to report March 2014. This wasn't the trial that Prana wanted or needed given their results of 2008. That would have been a large Phase 3. Instead, cost constraints drove them to look at a primary endpoint of amyloid load in the brains of patients with mild Alzheimer's (MMSE >= 20) - a biomarker measure believed to correlate, at least early in disease, with changes to Abeta in the CSF .
At first glance running almost the same Phase 2 all over again might just be inviting repetition failure. After all, Prana already identified that PBT2 lowers Abeta in the CSF in their previous trial. Amyloid brain imaging could also prove to have been a risky choice. Many healthy elderly individuals test positive for binding of the amyloid PET imaging tracer 11C-PiB yet never develop Alzheimer's disease . Others display Alzheimer's-like cognitive impairment without any PiB binding at all . It is hence important to note that PiB positivity was a necessary inclusion criteria for the IMAGINE trial.
What is the probability of PBT2 meeting the primary endpoint? This is a tough question to answer. As a reference point, we have the previous example of an experimental Alzheimer's drug, bapineuzumab, which successfully demonstrated statistically significant reductions in brain amyloid with a lower n number (28 patients in total as opposed to the 40 completing the IMAGINE trial) .
One could hence compare preclinical data for the two drugs and guesstimate Prana's chances. The difficulty is, the data just aren't publicly available. A chronic-administration PBT1 study was carried out in a mouse model of brain amyloidosis  but the only PBT2 publication deals with acute dosing . Certainly 3D6 (the preclinical label for non-humanised bapineuzumab) seems a little superior to PBT1  but differences in the animal model used, duration of treatment and other factors preclude a real comparison. Gut instinct tells me it will be a tough ask for PBT2 to meet the primary endpoint of the IMAGINE trial. They do, however, undeniably have a chance.
For their secondary endpoints Prana have measured cognition via the Neuropsychological Test Battery (NTB) questionnaires and the Mini-mental State Examination (MMSE). Functional ability was also tested via the ADCS-ADL-23 questionnaire. Hope for significant changes in the NTB measures to match previous studies should again be tempered by sample size. Even given that only a single drug dose of 250mg PBT2 was tested, a completer number of 40 still gives the trial a lower power than the original Alzheimer's study reported back in 2008. Hence, yet again, the IMAGINE trial is not powered to detect cognitive efficacy. Nevertheless it will be interesting to see whether previous, suggestive hints of efficacy from the Category Fluency and Trail Making Part B tests are repeated.
Let's assume, and it's a big assumption, that Prana get the ideal trial result: PBT2 meets the primary endpoint of brain amyloid reduction with hints of cognitive efficacy from NTB measures. What would this mean for the drug?
Well what I don't think it would mean is an immediate application for drug approval. There seems to be much confusion around this issue - generally brought about by the FDA deciding to lower the bar somewhat with regard to drug development in early Alzheimer's disease . Previously, claims of improved cognition needed to be married to evidence of functional improvement. This is obviously very challenging in mild cognitive impairment where deficits are limited. With their changed stance, the FDA may be prepared to recommend approval on the basis of sensitive cognitive measures with a stipulation that post-approval trials are carried out to prove clinical benefit.
The FDA have spoken encouragingly about biomarker measures such as PiB binding, but are not yet prepared to accept these as surrogate endpoints. This is a critical point. Bapineuzumab demonstrated a statistically significant reduction in PiB binding but later failed against cognitive endpoints in Phase 3 trials.
The FDA's focus hence remains firmly on cognition. Prana have never run a PBT2 trial in which cognition was the primary endpoint. It is my understanding that they cannot file on the basis of weaker secondary cognitive outcomes that their trials were not powered to properly detect. In any case, all cognitive data would be subject to an intensive scrutiny of statistical methods which has thus far been avoided.
So what would Prana gain from a successful study result? Possibly a deal with Big Pharma which would secure the Phase 3 trial they badly need. Possibly. The Alzheimer's acquisition market is not what it was and many companies will be looking hard at both the bapineuzumab example and the residual PBT2 patent life. Alternatively Prana may well be able to raise future funds off the back of a positive result. Can they go it alone? The clue here may be in future positioning. If the FDA were, for some reason, to accept brain amyloid load as a surrogate primary endpoint in Alzheimer's disease, Prana could find themselves in a very different scenario. Is that likely to happen?
The Dominantly Inherited Alzheimer Network (DIAN) are currently running drug trials with two amyloid-lowering antibody therapies (solanezumab from Eli Lilly and gantenerumab from Roche). What makes these trials unique is that participants are at risk of developing familial, "early-onset" Alzheimer's disease. This is the rare (<1%) autosomal-dominantly inherited form of the disease caused by genetic mutations which increase the relative amounts of 'toxic' Abeta42. It is indisputable that the familial form of the disease is driven by amyloid, generally over shorter time-scales than the much more common, sporadic form (in which amyloid may be necessary but not sufficient to cause disease).
If these trials show real cognitive benefit then pressure may well mount on the FDA to accept brain amyloid load as a surrogate endpoint for the more complicated, sporadic form of Alzheimer's disease. Would this come soon enough for Prana? Perhaps. The DIAN trial is scheduled to complete end of 2016. A more pertinent question would be, will the DIAN trials actually show benefit?
Rightly or wrongly, the amyloid hypothesis of Alzheimer's disease has been the prevalent dogma fuelling drug research efforts for the last 15 years. It was founded on the rare familial forms of the disease and extrapolated those findings to the more prevalent sporadic form. Unfortunately the staging of amyloid influence in Alzheimer's disease remains poorly understood.
It is likely that the second brain pathology of the disease - malformed tau protein - is responsible for the observed neuronal death and cognitive impairment as it correlates better with these phenotypes than amyloid load . Tau PET imaging ligands are hence an area of great future excitement in Alzheimer's research. Quite how, and when, amyloid triggers tau-related neurodegeneration still remains unclear. It is sobering, therefore, that DIAN's own biomarker data indicate that even in familial Alzheimer's disease, changes in CSF amyloid occur decades in advance of clinical symptoms. It is hence quite possible that even in the DIAN trial, treatment with anti-amyloid agents will be too late in the day for real benefit. Only time will tell.
If the DIAN trials succeed, and the FDA approve brain amyloid load as a surrogate market of Alzheimer's disease progression, an almost empty market will likely become quite crowded. On one hand Prana may struggle to compete with much larger competitors with better-characterised therapeutics. On the other hand they have a unique mechanism of action in the metal hypothesis which could potentially allow co-treatment with other marketed agents.
Regardless of Prana's chances it is to be desperately hoped that all of the years invested in the amyloid hypothesis result in at least some viable treatment options for Alzheimer's patients. Otherwise they may have quite a long wait for the next best opportunity.
 Science. 1994 Sep 2;265(5177):1464-7. Rapid induction of Alzheimer A beta amyloid formation by zinc. Bush AI, Pettingell WH, Multhaup G, d Paradis M, Vonsattel JP, Gusella JF, Beyreuther K, Masters CL, Tanzi RE.
 Targeting Metals in Alzheimer's and Other Neurodegenerative Diseases. New York Academy of Sciences eBriefing, January 17, 2013
 Lancet Neurol. 2008 Sep;7(9):779-86. Safety, efficacy, and biomarker findings of PBT2 in targeting Abeta as a modifying therapy for Alzheimer's disease: a phase IIa, double-blind, randomised, placebo-controlled trial. Lannfelt L, Blennow K, Zetterberg H, Batsman S, Ames D, Harrison J, Masters CL, Targum S, Bush AI, Murdoch R, Wilson J, Ritchie CW; PBT2-201-EURO study group.
. J Alzheimers Dis. 2010;20(2):509-16. PBT2 rapidly improves cognition in Alzheimer's Disease: additional phase II analyses. Faux NG, Ritchie CW, Gunn A, Rembach A, Tsatsanis A, Bedo J, Harrison J, Lannfelt L, Blennow K, Zetterberg H, Ingelsson M, Masters CL, Tanzi RE, Cummings JL, Herd CM, Bush AI.
 Pain. 2008 Oct 31;139(3):485-93. Analyzing multiple endpoints in clinical trials of pain treatments: IMMPACT recommendations. Initiative on Methods, Measurement, and Pain Assessment in Clinical Trials. Turk DC, Dworkin RH, McDermott MP, Bellamy N, Burke LB, Chandler JM, Cleeland CS, Cowan P, Dimitrova R, Farrar JT, Hertz S, Heyse JF, Iyengar S, Jadad AR, Jay GW, Jermano JA, Katz NP, Manning DC, Martin S, Max MB, McGrath P, McQuay HJ, Quessy S, Rappaport BA, Revicki DA, Rothman M, Stauffer JW, Svensson O, White RE, Witter J.
 J Int Neuropsychol Soc. 2014 Jan;20(1):11-9. NIH EXAMINER: conceptualization and development of an executive function battery. Kramer JH, Mungas D, Possin KL, Rankin KP, Boxer AL, Rosen HJ, Bostrom A, Sinha L, Berhel A, Widmeyer M.
 Lancet. 2008 Jul 19;372(9634):207-15. Effect of dimebon on cognition, activities of daily living, behaviour, and global function in patients with mild-to-moderate Alzheimer's disease: a randomised, double-blind, placebo-controlled study. Doody RS, Gavrilova SI, Sano M, Thomas RG, Aisen PS, Bachurin SO, Seely L, Hung D; dimebon investigators.
 Arch Neurol. 2010 Feb;67(2):154-60. A randomized, placebo-controlled trial of latrepirdine in Huntington disease. Kieburtz K, McDermott MP, Voss TS, Corey-Bloom J, Deuel LM, Dorsey ER, Factor S, Geschwind MD, Hodgeman K, Kayson E, Noonberg S, Pourfar M, Rabinowitz K, Ravina B, Sanchez-Ramos J, Seely L, Walker F, Feigin A; Huntington Disease Study Group DIMOND Investigators.
 Curr Alzheimer Res. 2010 Feb;7(1):56-66. High PIB retention in Alzheimer's disease is an early event with complex relationship with CSF biomarkers and functional parameters. Forsberg A, Almkvist O, Engler H, Wall A, Långström B, Nordberg A.
 Ann Neurol. 2013 Jun;73(6):751-61. In vivo assessment of amyloid-β deposition in nondemented very elderly subjects. Mathis CA, Kuller LH, Klunk WE, Snitz BE, Price JC, Weissfeld LA, Rosario BL, Lopresti BJ, Saxton JA, Aizenstein HJ, McDade EM, Kamboh MI, DeKosky ST, Lopez OL.
 Dement Geriatr Cogn Disord. 2012;34(2):112-20. Clinical features of Pittsburgh compound-B-negative dementia. Takeuchi J, Shimada H, Ataka S, Kawabe J, Mori H, Mizuno K, Wada Y, Shiomi S, Watanabe Y, Miki T.
 Lancet Neurol. 2010 Apr;9(4):363-72. 11C-PiB PET assessment of change in fibrillar amyloid-beta load in patients with Alzheimer's disease treated with bapineuzumab: a phase 2, double-blind, placebo-controlled, ascending-dose study. Rinne JO, Brooks DJ, Rossor MN, Fox NC, Bullock R, Klunk WE, Mathis CA, Blennow K, Barakos J, Okello AA, Rodriguez Martinez de Liano S, Liu E, Koller M, Gregg KM, Schenk D, Black R, Grundman M.
 Neuron. 2001 Jun;30(3):665-76. Treatment with a copper-zinc chelator markedly and rapidly inhibits beta-amyloid accumulation in Alzheimer's disease transgenic mice. Cherny RA, Atwood CS, Xilinas ME, Gray DN, Jones WD, McLean CA, Barnham KJ, Volitakis I, Fraser FW, Kim Y, Huang X, Goldstein LE, Moir RD, Lim JT, Beyreuther K, Zheng H, Tanzi RE, Masters CL, Bush AI.
 Neuron. 2008 Jul 10;59(1):43-55. Rapid restoration of cognition in Alzheimer's transgenic mice with 8-hydroxy quinoline analogs is associated with decreased interstitial Abeta. Adlard PA, Cherny RA, Finkelstein DI, Gautier E, Robb E, Cortes M, Volitakis I, Liu X, Smith JP, Perez K, Laughton K, Li QX, Charman SA, Nicolazzo JA, Wilkins S, Deleva K, Lynch T, Kok G, Ritchie CW, Tanzi RE, Cappai R, Masters CL, Barnham KJ, Bush AI.
 Nat Med. 2000 Aug;6(8):916-9. Peripherally administered antibodies against amyloid beta-peptide enter the central nervous system and reduce pathology in a mouse model of Alzheimer disease. Bard F, Cannon C, Barbour R, Burke RL, Games D, Grajeda H, Guido T, Hu K, Huang J, Johnson-Wood K, Khan K, Kholodenko D, Lee M, Lieberburg I, Motter R, Nguyen M, Soriano F, Vasquez N, Weiss K, Welch B, Seubert P, Schenk D, Yednock T.
 N Engl J Med. 2013 Mar 28;368(13):1169-71. Regulatory innovation and drug development for early-stage Alzheimer's disease. Kozauer N, Katz R.
 Acta Neuropathol. 1991;82(4):239-59. Neuropathological stageing of Alzheimer-related changes. Braak H, Braak E.
Disclosure: I have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.
Additional disclosure: As of writing I am not, and have never been, a shareholder in Prana or any other competitor organisation. I have not been paid for this, or any other publicly available blog post or social media communication. The above article reflects my personal opinions only and should not be construed as medical or investment advice in any way