Business Presentation Predicting the future progression of dementia Ashish Raj, PhD Founder and CEO Associate Professor of Computer Science in Radiology Director, IDEAL Laboratory Associate Professor of Neuroscience Weill Cornell Medical College, New York Technology funded by: Research highlighted in: Neuron, Nature Reviews, Science Daily, Future Neurology, Neurology Today
Problem Alzheimer s disease and other dementias are debilitating, incurable diseases Affects 1 in 5 elderly, 10 million in US Current diagnosis, prognosis highly subjective, cannot predict future course Alzheimer Association Based on cognitive test, visual assessment of MRI Ø When/whether will I get Alzheimer s? Ø Which brain regions will I lose? Ø Will I lose memory, attention, behavior or executive? Ø Life/financial planning Ø Targeted brain stimulation Ø Therapeutic options (aggressive or not?) Ø Cognitive reserve, memory exercises
Enter BrainWire Technology A diagnostic and predictive software tool Based on cutting edge graph theoretic models Barrier to entry: new, emerging technology, IP Not possible 2 years ago baseline MRI Computer algorithms, machine learning The Brainwire product: Prognostics on likely future disease patterns Risk of conversion assessment Best anatomic targets for therapy Full neuroradiologist report
How does it work Stereotyped atrophy patterns, not random (but considerable variations amongst patients) Apostolova, Arch Neurol 2007 Caused by stereotyped spread of misfolded proteins (A-beta, tau) Alzheimer Association BrainWire captures the latest basic science of trans- neuronal transmission of misfolded proteins
The Science
Diffusion on Graphs and Rela2onship to Demen2as Signal x(t): amount of disease agent (Ab, tau) in brain regions We mathema<cally model neurodegenera<on as a diffusive process Raj, Kuceyeski, Weiner. Neuron 2012 R2 x 2 c 12 R1 x 1 Spread of tau pathology from entorhinal cortex traces classic Braak stages (left). We model the trans-neuronal spread from region 2 to 1 as a diffusion process whose rate is regulated by network connectivity (right). Ini<al AD pafern Future spread pafern Closed form solu<on gives x(0) à x(t), for any future <me t, star<ng at baseline pathology/atrophy pafern x(0) 6
How to get the connectome? High Angular Resolu2on Diffusion Imaging (dmri) credit: P Mukherjee 7
Whole brain tractogram from dmri Connect the dots to draw tracts between dmri voxels Raj, Kuceyeski, Weiner. Neuron 2012 Graphic designed by Eve LoCastro 8
Neuroimaging pipeline to obtain atrophy paferns and connectome from human brains MRI scans are atlased, parcellated and volumed diffusion MRI scans are reconstructed, tractograms drawn Datasets are combined to achieve atrophy and connectivity de?ined on the same set of nodes
From DTI to tractography to network or graph 10
the modeled disease spread to mesial temporal lobe (Fig 3 middle) before continuing frontal-temporal-parietal association areas (Fig 3 right). ND Model seeded at Hippocampus recapitulates classic AD progression: ND Model eigen- modes match AD atrophy topography: The eigen- mode of ND model Raj, Kuceyeski, Weiner. Neuron 2012 Measured atrophy pafern from Alzheimer s subjects 11
Sta2s2cal correla2on results on 18 AD, 18 HC, 19 FTD pa2ents Raj, Kuceyeski, Weiner. Neuron 2012 Diagonal plots show strong correla<ons, off diagonals show weak correla<ons This supports the conclusion that there is a one- to- one rela<onship between the Brainwire model and various demen<as
Parkinson s Disease Parkinson s disease is ini<ated at a brainstem region called substan<a nigra, then spreads outwards Network diffusion was seeded at SNpc è predicted future PD pathology Spread of Parkinson s pathology traces 6 Braak stages (left). This spread is modeled as a network diffusion process whose rate is regulated by network connectivity. SNpc- seeded Network diffusion 13 captures PD topography
Predicted spa<al pafern from the Network Diffusion Model applied to (direc<onal) mouse brain network accurately predicts the regional distribu<on of tau deposi<on, early stage APP concentra<on and tau- induced atrophy 1 st network eigenvector versus regional increase in tau burden (Clavaguera, 2009) 1 st network eigenvector versus rela<ve regional APP level (Harris, 2010) 1 st network eigenvector versus future regional atrophy (de Calignon, 2012) Measured Slope of Tau Burden Increase 0.35 0.3 0.25 0.2 0.15 0.1 0.05 Hippocampal Formation Thalamic Visual Cortex Hypothalamus Amygdala Caudoputamen Somatosensory Cortex Superior Colliculus Substantia Nigra Entorhinal Area Pontine Gray 1.4 entorhinal area r 2 = 0.56, p < 0.01 r 2 = 0.86, p < 0.01 Entorhinal Area dentate gyrus 0.5 r 2 = 0.88, p < 0.01 Dentate Gyrus 1.2 CA1 CA3 CA2 0.4 CA2 CA3 1 CA1 perirhinal area Subiculum 0.3 0.8 0 0.05 0.1 0.15 0.2 0.25 0.3 Estimated Node Centrality Measured APP Levels 0.6 0.4 0.2 0 0.2 0.4 0 0.05 0.1 0.15 0.2 0.25 0.3 Estimated Node Centrality Measured Neuronal Atrophy 0.2 0.1 0 0.1 0.2 0 0.05 0.1 0.15 0.2 0.25 0.3 Estimated Node Centrality MODEL DATA
Clinical Implica2ons One of the first predic<ve, testable, mathema<cal models of spread of neurodegenera<on Model is fully predic<ve Can use baseline MRI to predict future atrophy Just play out the diffusion equa<on U<lity in monitoring of clinical trials Preliminary results promising: Alzheimer Parkinson Frontotemporal demen<a 15
Predic2ng future atrophy paferns Alzheimer s Disease Patient Baseline Projected: 5 years Raj, et al. Cell Reports 2015 10 years
Predic2ng future atrophy paferns MCI subject who converted to AD Baseline 10 years 15 years MCI subject who did not convert to AD Baseline Raj, et al. Cell Reports 2015ç Projected: 5 years Temporal lobe 10 years
Predic2ng future atrophy paferns 687-subject ADNI T1 volumetrics using Freesurfer, 2-4 year longitudinal follow up Baseline was applied to ND model, future prediction tested against follow up scans Significant improvement in correla2on between measured and predicted out- year atrophy 18 Raj, et al. Cell Reports 2015
Back to the Business
The Clinical Market Opportunity, Based on Dementia Research Roadmap Key focus areas in AD research includes biologic / imaging The biomarker use of biomarkers in both Alzheimer s dementia and MCI due to Alzheimer s disease is intended only for research at this time. However, some biomarkers, especially those using advanced imaging techniques, could enter clinical practice in the near future, though much remains to be learned about their utility in this setting. * Early detection is key to altering the disease and associated costs a hypothetical intervention that delayed the onset of Alzheimer s dementia by five years would result in a nearly 45 percent reduction in the number of people with Alzheimer s by 2050, and reduce the projected Medicare costs of Alzheimer s from $627 billion to $344 billion dollars. * * Source: Alzheimer Association publication, NEW CRITERIA AND GUIDELINES FOR THE DIAGNOSIS OF ALZHEIMER S DISEASE PUBLISHED FOR FIRST TIME IN 27 YEARS, April 2011
Market Very large market, rapidly growing 10m+ dementia patients in US Growth drivers: aging, baby boomers better diagnostic technology 16740 neurologists in the US, 3628 neuroradiologists 3000 primary care in dementia 330 memory clinics 600 500 400 300 200 100 0 US spending on AD (billion $)* 2012 2020 2030 2050 * Total disease-related expenses Source: AHRQ report, NEJM 2013 2500 clinics to reach Plan: reach 10% each year post-launch Requires sales force ramp up, 7-10 sales professionals required 2% penetration ratio target by year 10
Products and Markets Product Market PACS-integrated neuro reporting tool Full integration into EHR/PACS Workstation model Customized reports based on requirements patient Academic center, large hospital Brainwire Radiology PACS Neuro report tool for clinicians Cloud-served, SaaS Neuro report sent automatically, Push to clinic EHR Trial cohort selection, stratification tool patient Neurology clinic Brainwire cloud Upload Cognitive testing MRI If cog is diagnostic Cloud-served SaaS or custom install No report, but quantitative tables and decision points Brainwire Clinical trials organization, big pharma
Clinical Trials Market Earliest commercial opportunity Does not require FDA clearance Clinical trials applications: a) screening: cohort stratification b) monitoring: improve ability to detect small changes from untreated disease trajectory, reduce sample size, hence reduce $$ at any given point of time there are 150 clinical studies on AD/dementia underway in the US, need at least 50,000 participants. To reach that goal, researchers will need to screen at least half a million potential participants ** **Source: Alzheimer Association Public Health Roadmap 2013 ( http://www.alz.org/publichealth/2013-report/index.html)
Competition Players in dementia diagnostics * Venture backed ** post-ipo! Cognitive Imaging high ç Cost low GPCOG, MoCA, ADAS-cog free, screen Cognistat, Screen*, Cantab** $300-$500, screen Cognoptix * >$400 invasive Spinal tap (CSF amyloid) $3000 Only available prognostic tool Genomics+imaging i2dx * NeuroQuant * ~$1000 Glucose PET $2000 $?? Amyvid * PET $5000 $200 BrainWire Closest comparable MRI-based atrophy measurement Only meant for Dx low Diagnostic specificity è high
Financial Projections 2015 2016 2017 2018 2019 2020 2021 2022 CAPITAL $150,000 $1000,000 $3000,000 - - REVENUE $0 $0 $0 $2.1mil $6.2mil $10.5mil $15.2mil $20.0mil EXPENDITURES $149800 $1.0mil $1.65mil $2.12mil $2.7mil $2.85mil $3.0mil $3.05mil NET ($149800) ($1.0mil) ($1.65mil) ($20000) $3.5mil $7.65mil $12.2mil $16.95mil Use of funds by category Seed investment was obtained from ADDF Plan NIH SBIR funding (Phase I, II) Break even in year 5 Require VC investment of $3m in 2018 Year Year 1-3 Year 4- R&D 40% 10% Sales/ marketing 10% 60% Regulatory 40% 10% G&A 10% 20%
Management Team Ashish Raj, PhD (Founder, CEO) Faculty at Weill Cornell 3 NIH grants Director of Advanced imaging Lab Development team Michael Dayan, PhD (chief s/w architect) 5 years experience designing image processing s/w for brain scans Eve LoCastro (engineer) 4 years experience designing in image processing Systems administrator Clinical / business advisors Manish Raj, MBA (Business Dev) 15 years experience in IT consulting, finance, I-banking MBA from Wharton B-school Gloria Chiang, MD Faculty in neuroradiology at Weill Cornell Expert in reading MRI exams in dementia John Tsiouris, MD Faculty in neuroradiology at Weill Cornell Expert in reading MRI/PET exams in dementia
Business Presentation Predicting the future progression of dementia Ashish Raj, PhD (Founder, CEO) Brain imaging-based prognostic software technology Technology, validation data and team in place Seeking: Seed and venture capital Developers, marketing/business advisors FDA regulatory, s/w guidelines experts Clinical advisors Contact: Ashish Raj (ashish.raj@gmail.com) THANK YOU!!