As the knowledge of tumor genetics becomes more sophisticated, we continue to further stratify cancer into subtypes. Leukemia, for example, was defined as a single disease in the 1950s, and today it is comprised of more than 70 distinct histologic subtypes. A major challenge to drug developers is to produce drugs for disease subtypes that are poorly defined clinically. To address this challenge, companion diagnostic devices have been codeveloped and comarketed with drugs. However, developing companion diagnostics has several associated challenges related to cost, access to samples, and clinical trial recruitment. Below, we describe these challenges and report on survey data that suggest a pronounced willingness of patients to share tumor genetics data with pharmaceutical companies to make their contribution to the acceleration of drug development.
The Challenge of Companion Diagnostics
In 2011, the FDA issued a guidance document for the co-development of companion diagnostics,1 and later in 2012, a draft guidance document on enrichment strategies for clinical trials.2 These documents recommend that companion diagnostic assays be analytically and clinically validated prior to phase 3 trials. Effectively, this recommendation means that diagnostic assays need to be developed as early as phase 1 and be in use for phase 1b and phase 2 trials.
Many challenges exist for the codevelopment of diagnostics,3 and as we pursued the codevelopment of biomarkers for oncology programs at Merck, we assessed the cost of the dual process. The cost of developing a simple clinical assay is at least $1 million, but can be as high as $4 million for a complex assay such as the TP53 Amplichip.4 It would subsequently cost an additional $1200 to $1500 to analyze a single sample.
These costs may appear insignificant in comparison with the high overall costs associated with drug development. However, they pose significant challenges for executing clinical trials, especially for low-prevalence biomarkers. A biomarker with a 10% prevalence would require testing 200 eligible patients with available tumor biopsies to enroll 20 patients. For trials evaluating crizotinib, Pfizer had to assess 20 to 25 eligible lung cancer patients to enroll.1 In practice, the company probably examined more than 1600 patient samples to run the crizotinib trial that finally enrolled 82 patients.5 While such an effort can be justified in retrospect, similar efforts for all drugs going into phase 1b or 2 can be difficult to justify, since most drugs fail early in the pipeline due to safety or efficacy issues. Biomarkers are very important for running shorter and smaller trials, however, the current approach taken by most drug developers in inefficient financially and practically.
The Expected Solution
Molecular diagnostics have grown in sophistication through advancements in clinical next-generation sequencing (NGS). This technology offers the possibility of capturing all genetic alterations in a single assay and may eventually obviate the need for additional genetic diagnostic assays. Leading cancer hospitals already offer their patients exon sequencing of 100 to 300 genes. Companies like Foundation Medicine offer NGS services to patients and have reported delivering clinical results to 3752 patients in 2013. However, the FDA has been cautious about adopting NGS-based biomarkers since there are no accepted standards for data generation, validation, and computational analysis. Hence, pharmaceutical companies are not yet using NGS as a replacement for companion diagnostics.
To address the challenges and costs associated with developing companion diagnostics, we suggest an alternative model, based on a collaboration between patients and pharmaceutical researchers. Instead of developing clinical-grade assays early in the pipeline, pharmaceutical companies could offer to sponsor the NGS costs of patients who are otherwise eligible for one of their clinical trials. The results—owned by the patient—could be shared with the sponsors in a deidentified form, for research purposes only.
Patients would stand to gain with this approach. The number of trials in which they could likely enroll would no longer be limited by the amount of available tumor sample. In addition, they could receive valuable information that would enable them to make an informed decision when choosing to participate in a clinical trial. Finally, this approach reduces the out-of-pocket costs for patients. While privacy concerns exist and need to be addressed, results from our survey below suggest they may be significantly outweighed by perceived advantages.
For pharmaceutical companies, this approach might seem expensive since the current NGS cost is in the range of $5000 to $6000 per patient. However, given the cost of assay development, it would result in significant cost savings. In addition, knowledge generated from this effort would be extremely valuable—providing molecular information on tumors that do not respond to standards of care—since patients looking for clinical trials represent the unmet need. Finally, additional data would be available for retrospective analysis of clinical trials.
The Patient Perspective
To better understand patients’ perspectives on this potential approach, Merck collaborated with Smart Patients, an online community of cancer patients and caregivers. The staffs of Merck and Smart Patients designed a survey that was administered by Smart Patients, and the results were deidentified prior to being shared with Merck.
We first assessed the prevalence of tumor NGS within the existing patient population. Of the 92 cancer patients who answered the survey, 18% (17 patients) had previously had their tumor sequenced. Of these 17 patients, 9 reported that the results helped the make treatment decisions; 4 answered that the results had not helped, and 4 did not respond. For 8 of those 13 patients who responded, the results helped in identifying clinical trials in which they could participate. More than 70% of the patients who had not had their tumor sequenced cited lack of awareness about this option. Eight percent of patients did not think sequencing data would be useful, while the remaining patients indicated financial or technical reasons.
We then asked patients who had not had NGS done on their tumor if they would be willing to share the results of the sequencing with a pharmaceutical company for research purposes, in exchange for sponsoring the costs of the assay. An overwhelming 63 of 64 patients responded “Yes.” Even the majority of patients who had already had NGS responded that they would be willing to share their results. Finally, patients discussed their perception of benefits and concerns with this approach on the Smart Patients conversation platform. Most cited progress in research as the main benefit, while some patients were worried the information might be used against them by insurance companies to deny future coverage or by pharmaceutical companies to deny participation in clinical trials. These concerns need to be addressed by better educating patients on these issues and by more visible, regulation-driven patient protection.
As genetic testing and study become increasingly important to our understanding of cancer, researchers will need to develop scalable approaches to obtaining genetic information from patient tumors to accelerate and improve drug development. Our data suggest that patients may be willing to contribute data for these purposes if the cost of testing were covered. An important caveat of these results is that the Smart Patients survey population is biased toward patients who are more engaged and knowledgeable about their own disease. Nonetheless, the opportunity is real. It will be up to pharmaceutical researchers, regulators, patient advocacy organizations, and patients themselves, to collaborate and make the most of it. Merck and Company, Boston, 470 Atlantic Ave (YB, BM,); Smart Patients, Inc, Mountain View, CA (RZ); Merck and Company, Boston, 33 Avenue Louis Pasteur (SHJ, AB); Harvard Medical School, Boston, MA (SHJ); Boston VA Medical Center, MA (SHJ).
Address correspondence to: Sachin H. Jain, MD, MBA, 33 Avenue Louis Pasteur, BB2-402, Boston, MA 02115. E-mail: email@example.com.References
1. Draft guidance for industry and FDA staff—in vitro companion diagnostic devices. FDA website. http://www.fda.gov/MedicalDevices/%20DeviceRegulationandGuidanc/GuidanceDocuments/ucm262292.htm. Published July 14, 2011. Accessed June 10, 2014.
2. Draft guidance for industry-enrichment strategies for clinical trials to support approval of humandrugs and biological products. FDA website. http://www.fda.gov/downloads/%20Drugs/GuidanceComplianceRegulatoryInformation/%20Guidances/UCM332181.pdf. Published December 2012. Accessed June 10, 2014.
3. Fridlyand J, Simon RM, Walrath JC, et al. Considerations for the successful co-development of targeted cancer therapies and companion diagnostics. Nat Rev Drug Discov. 2013;12(10):743-755.
4. Chiaretti S, Tavolaro S, Marinelli M, et al. Evaluation of TP53 mutations with the AmpliChip p53 research test in chronic lymphocytic leukemia: correlation with clinical outcome and gene expression profiling. Genes Chromosomes Cancer. 2011;50(4):263-274.
5. Shaw AT, Yeap BY, Solomon BJ, et al. Effect of crizotinib on overall survival in patients with advanced non-small-cell lung cancer harbouring ALK gene rearrangement: a retrospective analysis. Lancet Oncol. 2011;12(11):1004-1112.