Several years ago, Stanford University researchers stumbled across something unexpected. In the course of examining the autopsied brain tissue of people with multiple sclerosis (MS), a team in Lawrence Steinman’s laboratory noticed significantly elevated levels of angiotensin enzymes and receptors, better known for their role in hypertension. This led them to believe that perhaps an angiotensin inhibitor, such as lisinopril, a common, inexpensive, and relatively safe high blood pressure medication, might effectively reduce the damaging flareups of MS. In fact, when they administered the drug to mice crippled by MS-like inflammatory nerve damage, it reversed their paralysis.
What happened next may well be a harbinger of future drug development: Building upon Steinman’s findings, Transparency Life Sciences, LLC (TLS), a drug company cofounded by Steinman, constructed a Web site to encourage researchers and patients to help design clinical trials to assess drugs such as lisinopril. In 2012, the U.S. Food and Drug Administration (FDA) approved the lisinopril protocol as the first investigational new drug application developed with the aid of crowdsourcing.
Since its early days, the Internet has been a popular forum for sharing health information. But the rise of Web 2.0 technologies, allowing individuals to measure, collect, and share their own health data, is beginning to transform the closed and competitive world of biomedical innovation and clinical research. Crowdsourcing, the solicitation of ideas from a large group of people through the Internet, is increasingly being used as a means of conducting research. Proponents say it can be a quick, cheap, and effective way of gathering study participants, ideas, and data. Many observers believe the large groundswell of collective knowledge provided through crowdsourcing will ultimately provide key insights into diseases and help get new, better, affordable drugs and devices to the right patients faster.
“We’re not quite at the point where patients are walking into the operating theater to take out their own appendix,” says Paul Wicks, director of research and development for PatientsLikeMe, a clinical research platform that hosts an online patient network, “but there is a big culture change,” he adds. As of October 2013, PatientsLikeMe has more than 220,000 members sharing health data on more than 2,000 diseases and conditions. Patients with serious chronic illnesses—especially those without effective or reasonably priced treatments—are beginning to take research into their own hands. Scientists, companies, and others with an interest in advancing health care are learning how to tap into that phenomenon.
“A fundamental aspect of crowdsourced research is that it is something you do with patients,” Wicks says. “You don’t do research to them, and you don’t do research on them.”
“Crowdsourced research studies are emerging as an important new means of investigation in a multitier ecosystem that could include self-experimentation, participant-organized crowdsourced studies, and researcher-organized crowdsourced studies that are a complement and precursor to traditional randomized controlled trials,” wrote Melanie Swan, cofounder of the nonprofit health research organization DIYgenomics, in a narrative review of crowdsourced health research studies published in 2012 in the Journal of Medical Internet Research. “There are exciting opportunities for researcher-organized study activity in health social networks to move beyond survey-based methods toward active intervention-testing.”
Patients Hack Their Own Diseases
The seeds of the patient revolution—and its potential to speed treatments and transform biomedical research—began in the early 1980s with HIV, when people with a rapidly lethal and infectious disease protested the maddeningly slow pace of testing the first promising agents. These days, instead of protesting in the streets, patients are channeling their energies online by organizing their own clinical trials, donating data, developing outcomes measures, designing interventions, reporting side effects of existing drugs, and experimenting with treatments off label.
In one striking example, people with amyotrophic lateral sclerosis (ALS) tuned into promising results presented at a 2007 conference of an Italian association dedicated to finding a cure for Parkinson’s disease, movement disorders, and dementia. During the conference, researchers from the University of Pisa reported that lithium, a drug for bipolar disease, seemed to delay ALS progression in 16 patients, compared to 28 who did not receive the drug. Even before the paper was published officially in the Proceedings of the National Academy of Sciences, patients with ALS used Google Translate to convert the Italian abstract into English and share it with others online. Soon, these patients took it upon themselves to post links to a Google Spreadsheet to collect their own data to see whether they could replicate the findings. About 160 patients tested lithium on themselves in an off-label ad hoc unapproved study in what may be the first patient-initiated study to identify a potential cure for their disease.
In response, the PatientsLikeMe research team modified their platform to accommodate more orderly structured data to ensure that outcomes and side effects could be entered more consistently than the spreadsheet would permit. Patients used the ALS Functional Rating Scale, a validated revised tool used in clinical trials that incorporates respiratory function. Some patients also developed their own personalized measures, such as lifting weights, and a few posted the video evidence on YouTube.
Sadly, lithium didn’t work—not even showing the expected placebo effect. Wicks and the trained research staff at PatientsLikeMe analyzed the findings, using a statistical method that incorporated multiple closely matched controls drawn from people with ALS on PatientsLikeMe who opted not to try lithium. The researchers published the results in Nature Biotechnology. Since the PatientsLikeMe study, four more traditional studies have been done using lithium on patients with ALS, but no one has replicated the refuted original promising findings, and one group found that lithium might make patients worse.
The ALS study on lithium is one of more than 40 studies published by PatientsLikeMe in peer-reviewed journals since its founding in 2004. Although in its first five years the company focused mostly on neurological conditions such as ALS, MS, Parkinson’s, and epilepsy, it has since expanded its pool of patient-reported data to include data from patients with any condition, including diabetes, cancer, and psoriasis.
PatientsLikeMe also connects members with more traditional clinical trials. Recently, the company linked its platform to ClinicalTrials.Gov to help members identify relevant trials of drugs, devices, or therapies or noninterventional studies, such as genetics studies or questionnaires. Drug companies also pay PatientsLikeMe to help recruit study participants. The business model involves selling members’ aggregated and anonymized data for others to study. The company operates under a philosophical ideal of openness, and it encourages study results to be shared back with participants through open-access publications.
Failing “Faster and Better”
Crowdsourcing allows patients to participate in the search for a cure while offering support and a greater understanding of their diseases. For researchers, crowdsourcing offers an opportunity to “fail fast and fail better” in the search for better therapies, says Wicks. To entice more research professionals to consult its 220,000-plus members in one area well suited to the Internet platform, PatientsLikeMe has set up the Open Research Exchange, which it bills as the world’s first open-participation research platform for creating health-outcome measurements. Researchers can design, test, and share new self-reported measures of health impacts and issues that matter to the cross section of patients affected by a particular disease. For example, in the initial pilot phase of the Open Research Exchange, the first four measurement tools in development include surveys to measure how type 2 diabetes medications affect eating patterns, detect barriers in managing hypertension at home, advise doctors on discussing future palliative care with patients newly diagnosed with a serious illness, and investigate how people with multiple chronic conditions manage their medications and doctors’ visits.
“The measurements, in many diseases, are out of date,” Wicks says about the research need. He cites as an example how a screening instrument for the characteristics of autism or Asperger’s syndrome asks how well the person can program a VCR. Crowdsourced platforms, he says, allow for cheaper and quicker revision and calibration of new health outcomes measures.
While PatientsLikeMe may be the largest research-oriented patient network, Swan says that such participatory health initiatives are now firmly a part of the public health ecosystem and are expanding quickly. And participant-organized studies may hold particular promise in preventive health, as enthusiastic groups of people use self-tracking health devices, personal genome data, and medical testing results to assess and adjust their steps walked, nutrients consumed, hours slept, mental performance, blood sugar, blood pressure, blood lipids, and more. “Everyone is tracking stuff,” she says. Quantified Self, an international online community of self-trackers, has 5,000 participants, 70 meet-up groups, and 120 events as of October 2012, she notes. New devices are automating data collection and feedback; for example, the Scanadu turns a smartphone into a Star Trek-like tricorder. The next challenge is to harness these big data for better health outcomes, she says.
Swan defines crowdsourced health studies as overlapping but distinct from citizen science. Participants recruited by crowdsourcing, such as via the Internet, may or may not be acting as citizen scientists in a study. And citizen science extends to participation in research that may not involve a patient’s own health data. A citizen scientist, for example, might help fold proteins, trace retinal neurons, or count birds in the backyard.
Another leading operator of crowdsourced health research studies is 23andMe, a privately held personal genomics and biotech company based in Mountain View, California, providing rapid genetic testing directly to consumers, 75% of whom have indicated a willingness to participate in company studies. The company has published peer-reviewed papers on its crowdsourced cohort of subscribers. The company also has a community research effort, 23andWe, with ongoing open enrollment on their Web site for various conditions, such as Parkinson’s disease, sarcoma, myeloproliferative neoplasms, and genetic disease links in African Americans.
Bottom-up and Top-down
The patient-driven crowdsourced platforms might be called a bottom-up movement, but the same themes are resonating in top-down fashion along the biomedical innovation pathway. There is a growing recognition among corporate decision makers that certain types of data today can have greater value for a company when they are shared and serve as building blocks for collective knowledge generation than when they are tightly held as a proprietary asset, says Gigi Hirsch, executive director of the New Drug Development Paradigms program in the Center for Biomedical Innovation at Massachusetts Institute of Technology in Cambridge.
Fewer organizations are trying to tackle the challenges of the complex product-development pipeline alone, Hirsch noted in a recent paper in Science Translational Medicine. In the last decade, thousands of researchers, pharmaceutical and biotechnology companies, government regulators, payers, clinicians, and patients have formed more than 100 multistakeholder collaborations. In contrast to patient-powered collaborations, which started at the late-stage clinical end with outcomes measures and postmarket drug use, top-down multistakeholder collaborations began at the preclinical end with efforts to improve trial design, biomarkers, and animal models. They now reach beyond the early stages of the innovation lifecycle to include product development and postmarketing monitoring for safety and efficacy.
Collaboration may help fix the arduous, long, expensive, and risky business of biomedical innovation and has been prescribed as a treatment “to restore the ailing drug development community” by no less an authority than Janet Woodcock, director of the FDA Center for Drug Evaluation and Research (CDER). Related initiatives at the FDA focus on “patient-centeredness,” such as expanding patient-reported outcomes measurements to capture the impact of interventions. The FDA’s Elektra Papadopoulos, an endpoints reviewer at CDER, said, “I think social media can be used to complement traditional methods, so it’s not one or the other, but I think both can potentially be used together,” according to a 3 June 2013 report in Elsevier’s The Pink Sheet of her talk at an outcomes research meeting in New Orleans.
Leveraging Crowd Sourcing to Develop New Treatments
Among the pioneering companies that are putting crowdsourcing to work, Steinman’s TLS is the New York-based company that pushed through the first FDA protocol developed with crowdsourcing. TLS develops all of its protocols with the help of the crowd. The company has focused on new indications for compounds that have already won FDA approval for other conditions, also a priority of the U.S. National Institutes of Health. It also plans to test scientifically promising drug candidates that have stalled in development.
TLS works with both researchers and patients who use what TLS calls its “Protocol Builder,” which resembles an industrial-strength Web-based survey. Volunteers register with the site as researchers or patients and answer questions tailored to their overlapping expertise in the disease. For the lisinopril study, patients with MS answered extensive questions about how to meet specific endpoints throughtelemonitoring, and researchers were asked about the larger overall monitoring plan. “The goal here is to be specific enough to get answers to questions that are clearly formulated,” TLS Chief Executive Officer Tomasz Sablinski says. “The comments fields typically yield the more helpful input. Contributors supply explanations and suggestions that otherwise might not have surfaced.” Curation is crucial, he said, ideally by a researcher and a patient. The lisinopril protocol attracted 100 users, 30 researchers, and 70 patients. According to TLS, many valuable comments shaped the lisinopril protocol, for example, “real advances in one’s ability to walk, use hands, reduction in stiffness are key endpoints that I would like to see in studies” (patient), and “metabolic response from phase II would be required and useful for phase III response ranges” (researcher).
TLS is pushing even further into the patient-centered technological sphere to move clinical trial monitoring into the patient’s home through telemonitoring, where potential cost savings could be huge. TLS estimates it can drop the cost of clinical trials by half or more in this way. TLS is partnering with Advanced Monitored Caregiving Health, a New York-based company that provides remote patient monitoring at home for health-care organizations. In the proposed 12-month lisinopril study, patients will visit a clinical trial site at the start and end of the trial. All other study data will be collected at home by remote monitoring devices and home visits.
TLS is working on building a more robust platform for crowdsourcing input to protocols. Another crowdsourced protocol includes an interventional study to test the oral diabetes drug metformin in men with prostate cancer treated by chemotherapy or radiation treatment. The goal is to learn whether it can lower the recurrence of cancer and the short-term risk of death. The metformin protocol builder had 46 responses four days before the 31 October 2013 deadline to complete the protocol.
Crowdsourcing: the Future
Crowdsourcing may lower costs, speed up innovation, and generate products that better meet patient needs, but Sablinski warns that it is “not a freebie.” It takes hard work on the back end to ask the right questions that solicit the most helpful comments from patients and other researchers, he says.
And there are other shortcomings, Melanie Swan points out. Scientifically, participant-led studies do not always follow the rigorous protocols of randomized blinded controlled studies and often do not include a placebo arm. Alternative study designs keep costs low and are more feasible to run, but the studies may have flaws, including more pronounced selection bias and placebo effects. Swan argues that crowdsourced studies need creative approaches that give them the same degree of research and ethical oversight as more traditional studies, not only to protect people and identifiable data but also to guarantee patients’ rights to their own data and to ensure greater transparency and access to aggregated data and scientific papers.
These issues are likely to be overcome as crowdsourcing continues to shape the future not just for drug development but for biomedical product development in general. “Crowdsourcing augments but does not replace our internal judgment,” says Marc Foster, cofounder of TLS.
“We’re applying techniques used in other parts of the economy to drug development, and there’s no reason this couldn’t be applied to medical devices,” says Foster. He points to InnoCentive, a platform that matches “seekers” who have problems to solve with “solvers” who submit proposals, including those in biomedical innovation. The solvers can even earn money for their ideas.
“Crowdsourcing exists to fill a gap,” Wicks says. “In that sense, patients have been the disruptive innovators.”
For More Information
- Open Research Exchange
- Hugo Campos TEDx Cambrige talk (2011, Nov. 19)
- Society for Participatory Medicine
- Transparency Life Sciences
- J. H. Frost, M. P. Massagli, P. Wicks, and J. Heywood. (2008, Nov. 6). How the social Web supports patient experimentation with a new therapy: The demand for patient-controlled and patient-centered informatics. AMIA Annu. Symp. Proc., pp. 217–221. [Online].
- M. Papadaki and G. Hirsch, “Curing consortium fatigue,” Sci. Transl. Med., vol. 5, no. 200, p. 200fs35, Aug. 2013.
- M. Swan. (2012, Mar. 7). Crowdsourced health research studies: An important emerging complement to clinical trials in the public health research ecosystem. J. Med. Internet Res., 14(2), p. e46. [Online].
- M. Swan. (2013, June). The quantified self: Fundamental disruption in big data science and biological discovery. Big Data, 1(2), pp. 85–99. [Online].
- E. Vayena and J. Tasioulas. (2013). Adapting standards: Ethical oversight of participant-led health research. PLoS Med., 10(3), p. e1001402. [Online].
- P. Wicks and M. Little, “The virtuous circle of the quantified self: A human computational approach to improved health outcomes,” in Handbook of Human Computation, P. Michelucci, Ed. New York: Springer, 2013.
- P. Wicks, T. E. Vaughan, M. P. Massagli, and J. Heywood. (2011, May). Accelerated clinical discovery using self-reported patient data collected online and a patient-matching algorithm. Nat. Biotechnol., 29(5), pp. 411–414. [Online].
- J. Woodcock, “Precompetitive research: A new prescription for drug development?” Clin. Pharmacol. Ther., vol. 87, no. 5, pp. 521–523, May 2010.