BETA
This is a BETA experience. You may opt-out by clicking here

More From Forbes

Edit Story

Can We Build A Kickstarter For Cancer?

This article is more than 10 years old.

The University of Texas M. D. Anderson Cancer Center (Photo credit: Wikipedia)

Starting you own band, writing your first novel, or re-publishing your favorite ‘80s tabletop RPG are all cool goals. You can do them all on Kickstarter. What would be cooler?

How about funding a virtual biotech company with one goal: Saving or extending the life of a cancer patient who doesn’t respond to “standard of care” treatments.

Building large analytical databases to mine clinical and molecular data, and scan the scientific literature to identify better treatments for cancer patients is happening today. But what about patients who fall outside what we already know – whose cancer subtypes haven’t been discovered yet, and who don’t have access to the technologies that could make a difference in unraveling the aberrations driving their cancers?

The technology to unravel the molecular drivers of cancer is, for the most part, available today: “-omics” technologies for screening tumor samples from patients and comparing them to healthy tissue samples to pick out cancer-specific mutations; diagnostics that can track patients’ response to treatment in real time at a molecular level; and Web-based tools and apps (e.g., from CollabRx.com) that can mine expert-curated molecular models of specific cancers (like melanoma) that patients and community oncologists can use to guide treatment decisions (and feed those outcomes, good and bad, back into the research process).

Thousands of cancer patients with rare and hard-to-treat cancers receiving care in the community oncology setting would benefit from research focused on their individual diseases – in real time – if we set up the right funding structures, regulatory framework, and financial incentives to encourage drug and biotech companies to open up their compound libraries and experimental drugs to the broader oncology community.

The resulting knowledge gained – connected on the Web and analyzed via machine learning and predictive analytic programs (like IBM’s Watson program) – would improve cancer outcomes for all cancer patients and help de-risk cancer drug-development programs.  Cancer drug development timeframes could be slashed from one to two decades to two to five years—fast enough to make a difference in the lives of tens of thousands of patients.

The Importance Of Learning From Failure

“Failure” in cancer treatment is really an opportunity for real-world learning in real time. For instance, Memorial Sloan Kettering Cancer Center (MSKCC) ran a bladder cancer trial with an FDA-approved drug for kidney cancer (Affinitor/everolimus). This strategy of “off-label treatment” is widely used by oncologists, who experiment with novel drug combinations to (ideally) keep a patient’s cancer in check.

The everolimus trial was an abysmal failure, according to traditional standards. Everolimus didn’t help the vast majority of patients, but a single patient – a 73-year-old woman – saw her cancer respond to the drug. Two years after being treated, researchers report, “all signs of her disease are gone.” Other patients in the trial saw their disease worsen within months. Not long ago, researchers and drug companies might’ve shrugged off the case as an anecdote and moved on, shelving the product.

But MSKCC researchers didn’t stop there. They sequenced her tumor against known cancer gene abnormalities – but didn’t find any. Then they went a step further and did a whole genome scan of the tumor and found a rare mutation in the TSC1 gene, known to be involved in the mTOR pathway, which made her cancer more sensitive to the drug (evorlimus is an mTor inhibitor).

They confirmed this result by comparing tumor samples from other, partial responders (who only showed a short-term response), and found that they had a mutation in the same gene, while the non-responders didn’t. MSKCC then launched a new trial, enrolling patients with just this specific mutation – about 1 in 10 bladder-cancer patients. Whether this discovery will turn out to be an important therapeutic advance for this subset of bladder cancer patients remains to be seen.

But what is truly interesting about the trial is that the patient’s biology (which revealed a potential new biomarker for personalized cancer care) wouldn’t have been uncovered unless and until it was tested against a targeted drug. It is precisely through these unexpected discoveries – when otherwise targeted drugs fail – that we learn how to slice and dice cancer into increasingly small, molecularly defined populations that give us a better chance of treating it successfully.

Typically, FDA regulators want to approve drugs where the “benefit vs. risk” is balanced across an entire population of patients, with, say, bladder cancer. But what MSKCC’s experience shows is that cancer researchers need the broadest armamentarium possible of molecular-based treatments to disassemble cancers into ever smaller slivers of patients (as the failed bladder cancer trial and others suggest).

While the FDA is often concerned about validating biomarkers before we use them in trials, biomarkers can also be discovered through an adaptive trial process designed to elicit these types of serendipitous discoveries – as well as improve how we can tailor safer and more effective cancer treatments in the “real world”.

Faster, Please

Patients with cancer face an ever widening gap between the exponential rate at which technology improves and the linear rate at which these advances are translated into clinical practice. Closing this gap will require the establishment of learning loops that intimately link lab and clinic and enable the immediate transfer of knowledge, thereby engaging highly motivated patients with cancer as true partners in research.

Can we deconstruct cancer, one patient at a time?

In 2011, Intel founder and Parkinson’s patient Andy Grove suggested that we could quicken the pace of drug development by “[creating] a system that does more with fewer patients”:

Once safety is proven [in FDA approved Phase I trials], patients could access the medicine in question through qualified physicians. Patients’ responses to a drug would be stored in a database, along with their medical histories. Patient identity would be protected by biometric identifiers, and the database would be open to qualified medical researchers as a “commons.” The response of any patient or group of patients to a drug or treatment would be tracked and compared to those of others in the database who were treated in a different manner or not at all. These comparisons would provide insights into the factors that determine real-life efficacy: how individuals or subgroups respond to the drug. This would liberate drugs from the tyranny of the averages that characterize trial information today. The technology would facilitate such comparisons at incredible speeds and could quickly highlight negative results. As the patient population in the database grows and time passes, analysis of the data would also provide the information needed to conduct postmarketing studies and comparative effectiveness research.

Critics responded that drug development – and the human body – is far more complex than software or semiconductors, and that the pharmaceutical industry labors under a far more stringent regulatory environment (the FDA) than any other industry.  All that is true but misses Grove’s central contention that classical clinical trials lead to a “tyranny of the averages,” rather than helping us to – as in the case of cancer – disassemble complex diseases that might share the same clinical symptoms (and which we happen to call cancer or diabetes) but which are really molecularly distinct and thus require different treatment approaches.

The success of the high-tech industry hinges on a “learn as you go” approach to every step of the product-development process, including real-world use, generating information that leads to new products and even entire new industries. Wal-Mart, Amazon, Microsoft, and Google couldn’t exist without these technologies, and they’ve helped revolutionize the U.S. economy.

The advent of massive datasets, sophisticated predictive analytic platforms, and increasingly inexpensive molecular diagnostic tests is creating the same opportunities for “learn as you go” health care research, particularly for cancer, which is, after all, an information-driven disease in the sense that errors (mutations, deletions/copy number variations, epigenetic changes) in the body’s “software” (genes and DNA) lead to malignancies.

Researchers recognize today that cancer isn’t one disease, but likely hundreds of orphan diseases, and as a result, there will never be one silver bullet to cure it. It will instead require combination therapies tailored to an individual patients’ tumor biology. Big data Initiatives, like the National Cancer Institute’s (NCI) Cancer Genome Atlas and the American Society of Clinical Oncology’s (ASCO) CancerLinQ, as well as joint ventures between top cancer centers and leading IT vendors, like IBM and Oracle, are already exploring this space to improve how we treat the disease.

But there’s still much more information to be gleaned from real-time clinical oncology practice, discoveries that aren’t (today) codified in any database, journal article, or registered clinical trial that “Big Data” efforts can scan.

With the right tools and data-sharing infrastructure, the thousands of ‘experiments’ that take place daily in community oncology practices can be aggregated into a massive database that Bayesian analysis can mine to identify new subtypes of cancer and identify better treatment regiments for patients today.

If you think about cancer in those terms – as an information-collection/analysis/distribution problem – Grove’s proposal makes sense, because it outlines a conceptual pathway for developing  treatments tailored to the unique biochemical profiles of individual patients.

Cancer Is Too Smart And Too Fast For “Business as Usual”

Our current research approach – one drug, one clinical trial, one cancer type at a time – won’t generate enough of the information we need to unravel cancer’s molecular mysteries at the patient level. And it is too slow, too bureaucratic, and too expensive to be sustainable, given the number of compounds we have to test and the limited pool of patients who participate in clinical trials.

Only about 3% of all cancer patients participate in cancer clinical trials, and those patients – because of restrictive inclusion/exclusion criteria – are often very different (i.e., healthier) than the average cancer patient, who is likely to be in poorer health and have one or more co-morbidities (obesity, diabetes, etc.). This limits the applicability of even the best drug guidelines based on classical trials for real-world patients, and underscores Grove’s complaint about the “tyranny of the averages.”

In short, we won’t develop the drugs or complex treatment regimens we need to for truly personalized cancer treatment regimens for patients if we keep doing business as usual.

Liberating Us From The Tyranny Of The Average Patient

In 2013, 1.6 million Americans will be diagnosed with cancer, with 580,000 succumbing to it. Only 48,000 cancer patients will enter clinical trials. The remaining 1.55 million will receive the “standard of care” – and while this may be curative for patients with early stage or highly responsive cancers, for far too many patients, it will buy them little time at the cost of serious side effects. Indeed, it is these patients who would most benefit from participating in a movement focused on distributed research and learning.

Thankfully, that vision may be coming to fruition. For instance, in March, ASCO released its prototype version of CancerLinQ for breast cancer. CancerLinQ is an analytic program that can extract all information from electronic medical records (regardless of format, and including physician notes), provide clinical decision support, measure physician quality against benchmarks, identify treatment trends, measure prospective decisions against guidelines, and (it is hoped) eventually lead to improved treatment guidelines and generate novel hypotheses for formal follow up testing.

CancerLinQ is starting with over 100,000 de-identified breast cancer patient records, but will later be expanded to other cancers.  Still, CancerLinQ is primarily designed as a resource for practicing oncologists. Another promising approach would be to create more patient-facing resources and tools.

Building A Kickstarter For Cancer

Jen Williams’ husband was diagnosed with a rare type of kidney cancer and eventually sought treatment at MSKCC. She also started a Web site, 100 Million Prayers, to collect donations (over $64,000 to date) to fund research related to his type of cancer – a type so rare that it’s not going to be a major source of interest for drug developers.

MSKCC has all the infrastructure for developing new personalized therapies for kidney cancer, including a tumor-tissue bank (the largest and most comprehensive in the world), state-of-the-art gene sequencing technology, and a high-throughput screening facility that gives researchers the ability to prescribe more personalized treatments for cancers. This includes screening tumor samples against FDA-approved cancer drugs (all small molecules) and pathway-discovery cancer drugs (i.e., experimental or shelved drugs that are known to inhibit a specific cancer pathway) to see if any signal a “hit” and show efficacy against the disease.

We can hope and pray that Jen’s husband beats his disease. But what would help his cause, and that of thousands of other patients, is distributing this type of effort across many other patients who don’t have access to MSKCC or participate in clinical trials (and remember, only 3% of patients do) – in other words, building a Kickstarter for cancer.

The key insight is that funding a “virtual biotech” for a single patient – like Jen’s husband - could uncover new cancer sub-types and perhaps also help identify an effective drug or drug combination for that sub-type.

Given that the same molecular aberrations are likely to be found in other patients and cancers, a successful “N of 1” virtual biotech effort could rapidly scale into hundreds or even thousands of lives saved on a global basis.

If, that is, we have the right tools for aggregating and validating the information we gain from such efforts.

Rapid-Learning Communities For Cancer

With a large enough database, matching more information about patient phenotypes and genotypes and treatment regimens, researchers can create even better molecular models of cancer – going even further than scanning “Big Data” looking for correlations.

One novel experiment in this space is Cancer Commons, which defines itself as a rapid-learning community (RLC) focused on cancer. RLCs focus on providing patients and their physicians with the knowledge and tools needed to select the best available therapies and trials, and to continuously update that knowledge based on each patient’s response (both in terms of the choice they make, and how they respond to treatment). Framed as “biomedicine in the Internet age”, Cancer Commons founders hope that “by tightly integrating research and care around individual patients”

…RLCs can dramatically reduce delays in getting promising investigational drugs into the clinic. Developers can get early validation by testing new drugs on patients with the right mutations who are otherwise out of options; physicians can share and learn from the thousands of unreported clinical experiments that take place daily in community oncology practices, involving the off-label use of approved drugs alone and in cocktails; scientists can use preclinical experiments on a patient’s cell line or xenograft to inform that patient’s treatment.

The Cancer Commons approach – a distributed framework for empowering patients and learning from every patient/treatment combination – breaks down traditional distinctions between clinical trials and patient treatment in the “real world.” Instead of developing treatments in a lab and then testing them on randomized patients in clinical trials (designed to benefit future patients), researchers would apply the latest scientific knowledge and tools to help each patient achieve the best possible outcome today based on what we know – or think we can predict – about a molecular subtype of cancer.

Everything learned across the Commons is immediately shared to benefit other patients and focus research on cancer sub-types with the greatest unmet medical need.  Along the way, the Commons is also assembling a network that can put patients and their physicians in contact with specialized firms offering high quality drug screenings and other specialized oncology services that would normally only be available at cutting-edge hospitals like MSKCC.

Family and friends, rather than raising money for the next bike race or half-marathon, could put their fundraising efforts into services designed to help their loved ones – with the knowledge that both successful and “failed” efforts could glean important information that would benefit other cancer patients. In fact, Cancer Commons has plans to build a crowdfunding cancer research platform by teaming up with sites like Kickstarter.

Let’s “Connect The Dots” Faster

The Web has the potential to revolutionize how quickly discoveries in cancer research can change the lives of patients. Take the story of Gleevec, the wunderkind of targeted cancer drugs. Gleevec’s developers realized early on that in addition to blocking the BCR-ABL enzyme, the drug also blocked a similar enzyme called c-Kit, active in a cancer called GIST. In fact, one researcher had noted in 1991 that c-Kit seemed to be active in some melanomas.

Gleevec turned out to be an incredibly powerful drug for CML, and researchers quickly tested it in other cancers, including a 2003 trial for melanoma. The trial failed. But some patients did respond, and were in fact positive for the c-Kit mutation, in addition to having tumors that weren’t located on their skin. No one at the time connected the dots, and it took until 2008 before there was a definitive paper published in the Journal of Clinical Oncology showing that Gleevec was effective in this sliver of melanoma patients. In other words, it took 17 years for researchers to definitively test the thesis connecting c-Kit and melanoma.

Finding those slivers of patients, and rapidly testing targeted drugs in those populations – distributed over many different community-based sites and using rigorous statistics and treatment protocols in a Web-based environment – can ensure that those dots are connected much sooner.

The world is really just one big clinical trial. We’re just not collecting, analyzing, and testing the knowledge we’re generating every day very effectively.

Powering a Kickstarter for Cancer

We’ll need more than money to power a Kickstarter-for-cancer movement. We’ll need to encourage companies – from Big Pharma to “small” biotechs – to participate in distributed, Bayesian trials where new biomarkers or combinations of biomarkers are tested in patients with particular molecular profiles. And the FDA is going to have to be convinced that the system is going to generate high quality data that benefits patients, not sell them snake-oil cures.

In return for companies making their compound libraries and experimental drugs available for the “virtual biotechs” launched by cancer patients and their families, there should be a regulatory path established to take the most promising drugs and drug combinations to market.

One way to do this would be to create a novel clinical trial format – let’s call it Phase 1.5 or 2.5. Companies whose products showed strong efficacy signals in a “Kickstarter” framework could launch drugs in targeted (biomarker selected) populations in these “Phase 1.5” trials, and if they showed significant response (exceptional tumor response or disease control, and acceptable toxicity compared to standard treatments) the product could be eligible to receive Breakthrough Therapy designation and marketing approval  from the FDA. If there was any question about the products long term safety and efficacy, confirmatory trials could be required after marketing approval, which would have to be completed expeditiously.

Companies might find this attractive, because the Kickstarter framework could provide very early data on proof of concept as well as a ready pool of patients to enroll in targeted trials, lowering the cost and time needed to validate new cancer treatments.

The FDA should (as it can today) revoke approvals for drugs whose benefits don’t pan out after confirmatory trials. The key should be to test drugs quickly, succeed or fail quickly, and get that knowledge into the hands of the oncology community quickly.

The Risks and the Benefits

It may seem counterintuitive for patients to bankroll research into their own cancers. Critics could argue that wealthy, sympathetic, or particularly motivated cancer patients would reap the benefits of the system, while the poor or less educated would be left behind. Patients could be fleeced by companies offering shoddy research tools. Poorly vetted research programs could send patients, physicians, and companies scrambling down dead ends, wasting valuable time and effort – and lives.

These are all legitimate concerns.

But wealth, education, and motivation can work to our advantage. “Early adopters” in any technology typically pay more, and bear more risks than later adopters, but they’re also critical to shaking out the bugs and helping to “consumerize” new technologies. There’s no reason this should be any different in a “Kickstarter-for-cancer” movement.

Snake-oil cures for cancer also abound today, and patients seek them out of desperation. But it should be possible to create trusted Web tools, apps, and research services – validated by trusted intermediaries like the American Cancer Society, ASCO, NCI, etc.

Developing the statistical tools and protocols to learn from “real-world” experiments will be another significant challenge – but these should be surmountable by companies like Google and Microsoft, IBM and Intel, who already analyze massive amounts of data on a daily basis.

Finally, we shouldn’t ignore the benefits of wider access to cancer research tools in terms of cost. A true cancer RLC can help de-risk cancer drug discovery, sharply lowering the economic costs of drug discovery and development and (eventually) even cancer drug prices.

The patients who have the most to gain from this approach are those who have the most to lose today – patients with rare or hard-to-treat cancers, who fail rapidly on standard or even targeted treatments. And it’s exactly these patients who will, in all likelihood, be most eager to embrace the risks and promise of Kickstarting their own cancer research.

INVESTORS' NOTE: Foundation Medicine, a Cambridge based genomic testing firm recently announced its plans for an IPO.  Companies invested in the IPO include industry giants like Johnson & Johnson (NYSE:JNJ), Novartis (NASDAQ:NVS), and Sanofi (NYSE:SNY). FDA's new Breakthrough Therapy designation should increase access to the most promising targeted cancer therapies identified in early-stage clinical trials.