Every Cure

Patients around the world suffer every day from diseases with no approved treatments while potential cures sit on the local pharmacy shelf. In many cases, data linking these drugs and diseases hide in plain sight, but doctors and patients aren’t aware that these drugs might work because no systematic efforts have been made to unlock the full potential of their use across diseases. There are currently ~3,000 FDA-approved treatments that are approved for ~3,000 diseases, leaving ~9,000 diseases without a single approved therapy. Millions are affected and a shocking one in ten people have a disease with no treatments. Given that many diseases share the same underlying mechanisms and can thus be treated with the same drug, many of the 3,000 existing medicines can likely treat many of the 9,000 diseases with no treatments. In many cases, there is already data supporting their potential role in additional diseases. Unfortunately, these data sit in silos and opportunities aren’t being pursued due to several systemic barriers, including:

• No organization has the mission of ensuring that medicines are used to treat every disease they possibly can. Instead, disparate organizations pursue drugs and diseases that are aligned with their particular interests. 
• No central database exists of all potential uses for every medicine. Data sit in siloes and there’s no process to systematically identify and confirm new drug uses even though modern data analytics is now capable of identifying promising candidates for clinical trials. 
• Drug companies have insufficient financial incentives to run expensive trials or bring new risks to existing brands, especially for rare diseases and generic drugs. 

“Sadly, no one is responsible for making sure that drugs are fully utilized across diseases. Paradoxically, once a drug is approved…all hope is lost that it will be studied and used in all diseases it can treat.” Janet Woodcock, MD, Principal Deputy Commissioner, FDA

We estimate that over $3 trillion has been spent on pharmaceutical R&D to arrive at a precious 3,000 FDA approved drugs. Even though it may not be profitable for traditional developers, it is both inefficient and unethical for our system to not invest the additional marginal effort to maximize their full potential for society’s benefit. These drugs are safe and could be used to save lives immediately. What is a more important use of research funding than identifying new uses for existing drugs that have already been developed and can reach patients for a fraction of the time and costs?

Instead of repurposing existing medicines, the biomedical system pursues new drug development which requires $1-2B and 10-15 years for one FDA-approved drug. This focus on new, expensive drugs has a disproportionately negative impact on individuals with reduced access to care. Repurposing safe, available drugs for new diseases is 5X faster and costs less than 1% of new drug development to save equivalent numbers of lives. There is no greater return on investment for saving human lives than unlocking the full potential of every existing drug to treat every disease it possibly can.  

Every Cure’s Approach:

In 2022, we launched Every Cure, a non-profit organization designed to address these issues by connecting and supporting the drug repurposing ecosystem. Every Cure is developing the world’s most comprehensive open-source AI engine, which will be a foundational platform for linking biological changes to human disease and therapies at scale. The final product is a 36 million cell (3,000 drug X 12,000 disease) heatmap with linkage scores for every possible drug-disease combination based on the strength of evidence. This will be open to all physicians, researchers, and patients to provide hope and spur new hypotheses. This “all vs all” query is unprecedented and will significantly advance the science of drug repurposing and discovery.    

Four Critical Steps

First, we centralize the world’s medical knowledge to identify all possible drug-disease links. Knowledge graphs of curated public information, such as the federally-funded ROBOKOP knowledge graph act as the base layer and are integrated with public and private datasets and expert insights. This master biomedical knowledge graph includes over 70 data sets including LLM literature scrapes, drug target and pathway data, -omics, and more. We also engage pharmaceutical leaders to gain insights into additional uses that were considered but never pursued.

Second, we use advanced analytics and AI ranking algorithms to grade and predict the most promising drug-disease links. Several ranking algorithms, are optimized and combined. Currently we use random forrest and neural net algos to categorize and models that mimic electric circuits to identify potential mechanisms of action via paths through the graph. We are also exploring using LLMs to "read" the graph as language. After promising hits are identified, we evaluate the potential mechanism through which the drug may interact with a given disease and the likelihood of its effect. We also perform in vivo and in vitro studies to further investigate how promising a drug-disease link may be and evaluate signals of clinical effectiveness. After identifying the most compelling drug-disease pairs, we prioritize these opportunities based on scientific probability of success, amenability to low-cost trials, and impact on patients for inclusion in patient-centric clinical trials.

Third, we launch efficient clinical trials to evaluate the effectiveness of the most promising treatments. Recent innovations in clinical trial design allow for less expensive, more streamlined, decentralized, pragmatic, and even platform clinical trials. Every Cure has partnered with a leading decentralized trial company, Medable, to execute efficient trials designed with patients in mind. We will perform trials with the express goal of demonstrating effectiveness to change clinical practice, which may or may not involve changing the label. By combining efficient clinical trials with real world off-label use data, Every Cure can advance promising treatments to patients for a fraction of the cost of a novel therapy. 

Finally, we work to ensure the treatment is accessible to every patient that can benefit from it. Once a drug has been shown to be effective in a disease, a variety of stakeholders must be engaged.

See the four-step flow below:

Every Cure is on a mission to alleviate suffering for the 300M people globally battling diseases with no approved treatments. 

Sadly, access to medicines is highly dependent on geographic location and socioeconomic status. In fact, novel therapies for diseases go first to those with privileged access in wealthy nations. Existing drugs, particularly those that are generic, are more widely available, more affordable, and have the potential to treat many more people worldwide compared to novel drugs. However, no organizations are working to systematically identify additional diseases that can be treated with these widely-accessible drugs.

As a nonprofit organization focused on maximizing patient impact regardless of the drug, the disease, or profit, our approach eliminates the commercial bias in drug development decision making that leaves some diseases neglected. In the current paradigm, the most likely way that a disease receives attention and funding is if there is a clearly profitable path for a new drug. Every Cure disregards this typical approach by looking across all diseases and all drugs, elevating those that impact patients the most. 

We first learned about these barriers when we repurposed a drug to save the life of one of our co-founders, Dr. David Fajgenbaum. When Dr. Fajgenbaum was diagnosed with a deadly, rare disease with no approved treatments, the only hope to save his life would be to find a drug that already existed for another disease that could be repurposed for him. Though Castleman Disease (CD) nearly killed him five times over 3 years, he has been in remission for over nine years since identifying a mechanism underlying his disease (mTOR activation) and testing the mTOR inhibitor sirolimus on himself. Amazingly, this drug was sitting at his neighborhood pharmacy the whole time. It was approved for the treatment of transplant rejection but had never been used for CD. 

Over the last decade, Dr. Fajgenbaum has become a pioneer in the rare disease, hematology/oncology, and drug repurposing space through his work as the Founding Director at the Center for Cytokine Storm Treatment & Laboratory at the University of Pennsylvania and the Castleman Disease Collaborative Network (CDCN). Since 2015, our team has discovered and advanced 15 additional repurposed drugs for CD and cancer. For example, we uncovered a drug for angiosarcoma that had never reported to be used when we began treating a patient in 2016. This patient has been in remission for over 7 years, and this drug is now widely used for angiosarcoma. During the COVID-19 pandemic, the Center’s work to apply cutting-edge computational approaches to systematically rank all COVID-19 treatments played an important role in identifying the most promising treatments for large clinical trials, such as the NIH’s ACTIV-6 study. 

In parallel to this work being done, another co-founder, Dr. Grant Mitchell, was leading teams at McKinsey in its advanced analytics group, QuantumBlack, focused on harnessing large EHR databases and machine learning models to identify subpopulations where existing drugs performed better in efficacy and safety for specific pharmaceutical company clients. It became apparent that the technologies being used by industry to expand market share of their assets could be used to look across all drugs and diseases to benefit society. A key insight through this experience is that much of this process is automatable and insights are discoverable through data. We are now on a mission to build this at scale and apply our approach across all drugs and diseases.

In 2022, we partnered with the Chan Zuckerberg Initiative to launch the ROADMAP Project to establish a roadmap for repurposing drugs and offer solutions to the many roadblocks along the way (www.everycure.org/roadmap). The roadmap was informed by surveys and interviews with over 500 members of the rare disease and drug repurposing communities. Based on this research, we have been able to map out and engage with all key organizations across the drug repurposing landscape. 

These disparate efforts need to be linked and accelerated. That is the mission of Every Cure as it works to coordinate across the ecosystem and elevate all efforts. 

Every Cure is changing drug repurposing from serendipitous chance to systematic discovery. No organization is looking methodically across all drugs and all diseases to find the most impactful repurposed cures. Some are focused on finding a cure for a specific disease, while others are focused on commercializing a specific drug for profit. We upend this paradigm by looking across every drug and every disease simultaneously - millions of potential possibilities. Then, after systematically evaluating and ranking each opportunity, we choose the most promising ones to advance to clinical trials and save lives, regardless of profit. Furthermore, we integrate proprietary datasets (generated or sometimes donated at no cost due to our non-profit structure) with publicly-available data and share all findings, encouraging additional research. Lastly, Every Cure will continue to lead the promotion of drug repurposing as a concept, encouraging policy and legislative changes to incentivize others. Systematic drug repurposing provides the greatest return on investment for saving lives. 

We are defining a new field of research that we have termed "pharmaco-phenotyping". This effort effectively builds a comprehensive map of known human biology, but also predicts the gaps that are unknown - the space inbetween we call the "ignorome". Given the scale of that effort, it is not possible to do alone. That is why we endeavor to be the glue of the drug repurposing ecosystem by taking full responsibility of a drug actually reaching patients. That means not just generating hypothesis, but validating them, ushering them through clinical trials, and ensuring accessibility to patients. Taking the full "soup to nuts" approach ensures that discoveries do not sit idle once enough information is compiled to help patients. 

A great example of our impact on the market is an event we held last month in conjunction with the Chan Zuckerberg Initiative where we invited the top 70 researchers in data driven drug development to a summit designed to share learnings and improve collaboration. Attendees included luminaries in the field like Eric Horvitz, MD, PhD, the Chief Scientific Officer at Microsoft. A proud moment came when one of the top researchers stated that even though he dedicated his career to data driven drug repurposing, 80% of the content and technical advances he learned from others during the summit were new to him. 

The outcome of that event has been tremendous momentum ensuring entities work together to not reinvent the wheel. In addition to already planning for another event next year, a Slack group was created and joined by attendees (and still growing) to share ideas and a monthly Zoom has been initiated for presentation of the latest breakthroughs. 

We strongly believe that the build out and public publication of both our master knowledge graph, the ranking algorithms, and the resulting normalized scores will have a significant impact on drug repurposing and drug development generally. Even a 5% increase in the rate of success of repurposing efforts compounded over time will have a dramatic impact on humanity. 

Identifying and accessing the appropriate treatment for a disease is one of the biggest drivers of achieving health and wellbeing. We like to say that rare diseases aren't rare, in fact nearly 90% of the 9,000 diseases that have no approved treatment fit that category, and in aggregate they impact millions of people. In fact, 1 in 10 people will develop a disease during their lifetime that has no approved treatment. Sadly, many of those people are children who never even make it past kindergarten. 

Our mission starts from when our co-founder, Dr. David Fajgenbaum, was told by his doctor, "I'm sorry...we've tried everything, there's nothing else we can do." No one should hear those words when a life saving therapy sits idly on the local pharmacy shelf. 

One reason we think globally and in alignment with the UN Sustainable Development Goal for Good Health and Well-Being is that 80-90% of currently FDA approved drugs are already generic. This means that there is no intellectual property protections on the molecules because the patent exclusivity period of the drug has lapsed. These drugs then are dramatically cheaper and therefore more accessible to patients, not just in the United States but around the world. 

The leading human biology knowledge graphs including NCATS Translator Projects ARAX, ROBOKOP, MediKanren, SPOKE, and BTE (who are all subcontractors of Every Cure), PrimeKG, and others are being evaluated to determine the strengths and weaknesses of each in the context of identifying drug repurposing opportunities. A list of additional target datasets is being generated and reviewed by a cross-disciplinary team of experts to ensure they're best suited for computational drug repurposing and are as complete as possible. Datasets include sources such as UMLS, SemMedDB, ChEMBL, DrugBank, SMPDB, Reactome, KEGG, and UniProtKB that provide data on chemical substances, genes, biological pathways or processes, cell types, anatomical structures or systems, and diseases. Underlying data sets that are already included in existing knowledge graphs (70+ data sets in each KG) will be categorized into type and structure, and gaps identified. Known data gaps in the existing knowledge graphs that we are looking to fill include patient level and clinical data in addition to privately generated -omic analyses that support nodes and edges of the graph. An example of a dataset type that we are seeking that could help to fill those gaps is EHRs data linked with genomic and other -omic data such as the the BioVU dataset at Vanderbilt, the Million Veterans Program dataset at the VA (the former Secretary of the VA has joined our board), Genomics England, and others. Source of truth and validation datasets also need to be identified from multiple sources including clinical trials. 

In the build out of the master data set we will have experts review the quality of each dataset we’ve identified. Factors considered will include quality/accuracy of NLP and LLM algorithms (if the databases are derived from raw text sources such as literature), confidence in underlying methodology used to generate the data, or statistical validity of sets of results (e.g.  p-values of analysis). The evaluation criteria and results will be documented for analysis and inclusion of future data sets. 

We will then perform manual curation of the datasets to determine what classes of data can be extracted into the knowledge graph. This may include concepts such as: (Chemical, Treats, Disease), (Gene, is_related_to, Disease), (Chemical, Causes, Gene Mutation). Specifically, data analysts will go through each data set and identify what types of data, and metadata (such as dates, dosage, strength of connection, etc) can be extracted from each data source. Part of the work will be to classify what categories of data are able to be found in each data source. This will be a set of concepts that are biomedically important, such as genes, diseases, chemicals, anatomical entities, etc. We will also perform cross-validation to determine the consistency of overlapping data provided across databases. We will build upon the Monarch BioLink ontology developed by NCATS, which contains a very strong starting point for these classifications, and work directly with the teams that built that ontology. The processes of identifying and categorizing data classes will be documented and turned into protocols for future use. 

The two primary risks we see resulting from the completion of our project are the misuse of the predicted results by uninformed parties and lack of adoption of the findings once proven out. 

As shown in the schematic in the "Solution" section above, the final result of the build out of the master biomedical knowledge graph and the application of the predictive algorithms is a web-based visualization tool that will be made available to the public. This visualization is a "heat map" that shows over 36 million cells, with ~3,000 drugs on one axis and ~12,000 diseases on the other - all possible combinations of drug and disease pairs. The ability to manipulate the ordering of the axes by disease and drug type results in a dynamic tool that helps generate new hypotheses for researchers, especially when they are able to dive into the underlying data. Patients will also be able to zoom in to see an ordered efficacy ranking of all ~3,000 drugs in their particular disease. In the best-case-scenario these results will give hope to patients and guidance to physicians. In the worst-case-scenario, patients may dangerously misinterpret the data and attempt treatment on their own. We are working with design experts and focus groups to ensure that the tool is built in a way that serves each population of users in the most effective way. Misuse can be mitigated by creating permissioned access to certain sections and providing proper education and context to the results. We also partner directly with patient advocacy groups to help them interpret the heat maps for best use by their organization. 

The other risk we see is the lack of uptake of a life saving discovery due to unawareness or lack of access (typically due to payor reimbursement hurdles). The surest way to combat lack of awareness is to ensure the adoption of the drug's use in clinical guidelines that are published by physician groups in a particular specialty even when it is "off-label". We work directly with the publishers of those guidelines to share compelling data and ensure proper placement of a new discovery in the treatment protocols. 

The other driver of non-adoption is lack of insurer reimbursement. This is especially true if the label change has not occurred and the new indication is still off-label for the drug. Important legislative advances have been made in cancer on this front, where adoption in guidelines or confirmatory publication in peer reviewed journals will compel an insurer to reimburse. We are working with lobby groups right now to expand that approach to non-cancer indications as well. Additionally, the FDA is turning a corner in the acceptance of real world data for evaluation of efficacy for drugs. We are pushing hard on both of these fronts. Lastly, since most of the drugs we repurpose will be generic, we hope that the drug cost burden for society is actually reduced by our efforts and insurers will welcome collaboration to increase use of cheaper drugs. 

Our impact goals:

End of year 1: Master biomedical knowledge graph built and released to the public

End of year 2: Release of all 36+ million drug-disease pair predictive scores in a publicly accessible, dynamic visualization tool

End of year 3: 20 drug-disease pairs validated for advancement to clinical trials

End of year 4: 5 clinical trials launched

End of year 5: 10 clinical trials launched

Steady state: Launch of 5-10 clinical trials per year and engagement with support of third party efforts on drug-disease pairs that Every Cure cannot pursue

It's hard to calculate upfront the impact in terms of number of lives since our approach is agnostic to both drug and disease. Because of that we do not yet know which patient population will be impacted. However, in addition to probability of clinical trial success and amenability to a fast, efficient trial, our top prioritization criteria is patient impact. We have an entire work stream dedicated to calculating a patient impact score, and we will be advancing drugs that target the populations most in need. At a steady state of 5-10 clinical trials per year with much higher clinical trial success rates than novel drug development, our organization's impact will be significant for its size considering the pharmaceutical industry as a whole approves approximately 40 new drugs each year. 

Current full time team: 7

Planned full time hires over the next 6 months: ~30

Request access to the planned organizational chart for 2024 at the following link: https://docs.google.com/presen...

Dr. Mitchell and Dr. Fajgenbaum were roommates in medical school in 2011 at the UPenn when Dr. Fajgenbaum first fell ill with iMCD. They have since dedicated their careers to understanding how to identify new drug repurposing opportunities and scale the process to relieve patient suffering. 

While the organization is only 1 year old, it is the culmination of over 10 years of work, and the collective expertise of the team and board includes decades of experience. Please see the link below for the announcement of the founding board hand picked to ensure the completion of our mission: https://everycure.org/every-cure-announces-full-board-of-directors

Every Cure’s leadership team represents several of the populations that we focus our work on. Our Project Lead is a patient battling a neglected, deadly disease who is alive thanks to a repurposed drug. As a first-generation American whose family lives in Trinidad & Tobago, Dr. Fajgenbaum has witnessed the challenges patients face globally and has worked closely with physicians in the Caribbean and around the world to make repurposed drugs accessible. Our Co-Founder, Tracey Sikora, who has a mixed-race background, has seen the inequities in healthcare based on racial prejudices and biases, witnessing first-hand the disparate care received by family members of Black and Asian backgrounds. We also recognize that there are many additional lived experiences that we do not understand and are engaging a community panel of individuals with diverse lived experiences to assist with selecting promising drug repurposing opportunities for further advancement.

We are also committed to recruiting team members with diverse lived experiences who can identify with the populations our work benefits. Of course, human disease affects everyone – directly or indirectly – and we have assembled an initial team that is connected to our mission and the patients we help. We also work to ensure equitable representation in our data by integrating multiple datasets (e.g., EHR and insurance claims data) outside of clinical trial data, which is often biased towards privileged populations. We will also directly focus on addressing inequities in clinical trial representation by working with community members and performing clinical trials at smaller institutions where these patients are. Finally, just as we’ve done for Castleman disease and COVID, we will work directly with patients and patient advocacy groups to ensure our work is informed by and centered on those who will benefit from it and make sure treatments make it to patients most in need.

We strongly believe that humanity’s greatest lifesaving resource is the drugs we already have. Unlocking their full potential is the fastest, most efficient way to advance cures and save human lives. Since launching in Fall 2022 (video: everycure.org/cgi), we have rapidly canvassed the drug repurposing landscape and made key partnerships and relationships with all known major stakeholders. We have partnered with the Renaissance Computing Institute (RENCI) and the University of North Carolina to leverage their federally-funded knowledge graph (ROBO-KOP) as a foundational resource to identify and prioritize drug-disease candidates. In parallel, we have established collaborative partnerships with Dr. Evidence, Eversana, Elsevier, Medable, Clinton Global Initiative, and other leading tech, philanthropic, and pharmaceutical companies to obtain additional highly valuable donated data sets to integrate into the system and strengthen the insights that are generated. Through a beta/pilot, we have already identified 103 additional promising treatments in 73 diseases and are continuing our unique, systematic approach to drug repurposing. We’re are also excited to partner with ARPA-H for funding to scale this approach and expect to receive funding in early Q1 2024.

Key collaborators engaged at each step:

Step 1:

• Drug and disease ontology standardization (Monarch Initiative)
• In-silico drug-target binding predictions (SandboxAQ, AlphaFold)
• Sentiment analysis and co-mentions from natural language processing of published literature (Dr. Evidence, Elsevier)
• Medical record and claims data that provide insights into real-world drug disease linkages (Eversana, Atropos Health)
• EHR+genomics databases that provide a link between empiric clinical data and human biology (Regeneron, Nashville Biosciences, Million Veterans Program)
• Clinical trial results that can provide a gold standard confirmation of known drug-disease links (clinicaltrials.gov)
• Disease-specific registries and -omics data (CZI’s Rare as One cohort)
• Pharmome discovery via high throughput drug screens (EvE)
• Phenotype model organism data from knockout projects (JAX Labs)
• Untapped insights from the pharmaceutical industry, investors, patient advocacy organizations, and key opinion leaders about additional uses for existing medicines

Step 2:

• Public knowledge graph providers
• SPOKE (Sui Huang and Sergio Baranzini)
• MediKanren (Matt Might and William Byrd)
• BTE/BioThings Explorer (Andrew Su)
• ROBOKOP (Chris Bizon and Alexander Tropsha)
• ARAX (David Koslicki)
• Knowledge graph data science experts to improve prediction accuracy (RENCI)
• Labs, academic institutions (UPenn) and CROS (World Wide Clinical Trials) to validate findings in vitro and in vivo 
• Validate findings with synthetic trials (Eversana, Atropos)

 Step 3:

• Distributed trial operators (Medable, PLS)
• CROs and hospital systems (Apollo Health, World Wide Clinical Trials)

Step 4:

• Policy influencers (Friends of Cancer Research, Reagan Udall Foundation
• Regulators (high ranking FDA officials)
• Insurance companies (BCBS of America)

As a non-profit our efforts are funded philanthropically. Since the announcement of our launch in Sept 2022 at the Clinton Global Initiative at the behest of President Clinton we have been building out the initial team, board and funding. We are happy to announce our foundational funding partners have been secured and we continue to pursue funding from philanthropies and government agencies such as ARPA-H. You can see our foundational funding partner announcement here: https://www.prnewswire.com/news-releases/every-cure-announces-foundational-funding-partners-and-major-milestones-at-clinton-global-initiative-2023-meeting-301930456.html

We are grateful for support from leading philanthropic institutions including the Chan Zuckerberg Initiative, Emerson Collective, Schmidt Family Foundation, Arnold Ventures, Elevate Prize, and Flagship Pioneering. We hope their support grows over the years as the organization expands and continues to successfully repurpose existing medicines. 

These partners have provided us a runway for at least 3-5 years, so our fundraising efforts have now shifted to endowment funding for the sustainability of the organization. 

Additionally, in years 3-5 we plan to also begin pursuing IP-generating efforts that allow for out-licensing of new discoveries. Specifically, some generic drugs may be identified to work in a new indication but require modification in dosing, formulation, or administration routes. Every Cure can patent these modifications for capture of IP and generate royalty opportunities that also support the overall mission of the organization. 

Please request access to the budget for 2024-2027 at the following link:

https://docs.google.com/spread...

We are seeking $50,000 as an award that we plan to go to marketing and PR of our expansion and efforts in NYC at Cure. While the Cure Xchange Challenge funding is a small part of our budget, every dollar counts to accomplish our mission and raise awareness. That being said, we are extremely eager to become members of the ecosystem being built at Cure and hope to build our NY headquarters there. I have personally toured the building and met many of the leaders there, and I'm so excited that this exists in NYC. Every Cure is unique in that its efforts span the entire discovery process of data-driven drug repurposing, from data base building, to algorithm building, lab validation and clinical trial management. Many of the organizations already housed at Cure touch these efforts, and we are eager to collaborate and learn from each of them. 

We plan to be very active members of the Cure ecosystem and to house our NYC based employees there. We are most excited about the mentorship, networking and office space. The mentorship and networking are extremely important to the success of Every Cure, as we seek to be the glue of the drug repurposing landscape. The Cure building is absolutely spectacular, and we believe that having a presence there will support our recruiting efforts as top talent will thrive in an environment of similarly ambitious organizations.