A decade ago, the publication of the first draft of an entire human genome marked the start of a new era for genomic medicine Citation[1,2], and the expectation for the discovery and clinical application of genetic variations in multiple disorders immediately soared. However, 10 years later there is still a feeling of underachievement among researchers and physicians, given that all this genomic information has not yet been translated into dramatic health improvements globally. In fact, only a handful of genetic tests have been implemented into routine cancer clinical practice, and such implementation has not been made swiftly across the world in many cases Citation[3]. This ‘apparent’ lack of success of genomic medicine is likely the result of a combination of multiple cumulative effects including the slow implementation of technological advances in clinical practice, different scientific and clinical research agendas, regulatory and quality issues, as well as practical implementation in the clinical setting.
First, we must consider that the sequencing of the first human genome took 15 years from conception in 1988 to the finished project in 2003, with hundreds of research groups and thousands of scientists involved in the project over the years. The project itself generated a boost in the biotechnology sector and we now have second/third generation of sequencing instruments that can individually generate the same amount of information in less than a week. However, the first ever cancer genome was only published less than 3 years ago Citation[4], and we are still waiting for entire cancer genomes to become fully annotated, as many studies focus only on the exome or transcriptome, rather than the genome. Furthermore, most cancer genomes published until now come from cell lines rather than primary tumors, and therefore they do not reflect the reality of tumor heterogeneity in human cancers, but this is a task that is now underway and we should find exciting results in the next few years Citation[5].
There are various concerns regarding the clinical value of this genomic revolution. One of the main reasons why these huge technological advances do not seem to be readily translated in cancer care is owing to the lack of linked clinical patient information on outcome, since most sequencing projects to date have been more focused on the biology of disease and, while this is essential to understand tumor behavior, it does not provide a translational benefit in the short term. There is also the issue that pharmaceutical industries want their products to be used in the largest possible patient population, while it has now been accepted that many treatments will only provide a significant benefit to a limited number of patients with a particular molecular profile. The increasing use of companion diagnostics will probably shift this trend and new trials could be specifically designed for genetically selected tumor types, providing positive results much faster than previous approaches. However, there is an inherent flaw to such a design, since there is no gold standard for many of these companion diagnostics at the time of the first trials being produced. As such, some diagnostic tests used on Phase I–III trials may not detect all possible mutations or have significant limitations owing to the lack of previous large series of patients studied. Since the information collected in those trials will be used by the regulatory authorities, some valuable data may be missed and this will be perpetuated in future trial design or routine clinical practice. Even in the best of cases, we are still faced with the reality that tumors are very heterogeneous genetically and in terms of clinical behavior, and the current trend of analyzing one single variable is clearly not enough. For example, only 60% of lung cancer patients with the p.L858R mutation in the epidermal growth factor receptor gene (EGFR) show a response to gefitinib, even though they all carry exactly the same mutation in the target gene, suggesting that multiple, as yet unidentified, genetic changes could be playing a crucial role in the response to targeted drugs.
There are also regulatory issues that affect the clinical implementation of molecular diagnostics in the clinic, including the fact that there are multiple regulatory bodies around the world that act somewhat independently of each other. These agencies and regulatory bodies can sometimes make different decisions using the same clinical evidence available, potentially resulting in the approval of different drugs or under different settings. In addition, the bodies that regulate the implementation and quality assurance of molecular tests are also different in each country and there is no single unifying structure that aligns all molecular oncology clinical practice worldwide. The methods of reimbursement are also clearly different between countries (from entirely public health systems to private owned) or even between regions of the same country. This situation inevitably leads to nonuniform access to genetic tests, and the quality of these tests can be severely compromised by trying to reduce the costs owing to increasing funding constraints in the current economic climate.
Finally, the practical implementation of genetic tests that have been validated and approved is not trivial and this is perhaps an area of molecular medicine that can be easily underestimated. First, who should be responsible for requesting a particular genetic test? The histopathologist diagnosing the biopsy or resection is the first to know the provisional diagnosis and is the one that can release the sample for further testing. However, currently the treating physicians are on many occasions responsible for requesting these tests, as they know the patient’s clinical history and whether they could potentially be best suited for one therapy or another. This issue could be potentially magnified when more targeted therapies and tests are available for the ‘same’ tumor, greatly depending on the adoption of a reflex (based on predesigned algorithms) versus a multiplexed approach to testing, which in turn will ultimately be dependent on the financial constrains in each country/institution. Another consequence of this dichotomy is the inevitable delay in testing, depending on who orders the test and when, an issue that is amplified when the testing laboratory is not colocated within or next to the histopathology laboratory where the sample is kept. There is also great variation in the quality of the samples received, which is caused by a lack of consensus protocols for preservation and fixation of the tissue, affecting the quality of the DNA/RNA and therefore the quality of the molecular test. In addition, routine biopsies (as opposed to clinical trials where certain conditions must be met for entry and centralization of testing is the norm) are largely variable not only in quality but also in size, tumor content and degree of necrosis, for example. Therefore, when analyzing such a variety of tissues, highly sensitive and specific technologies are required. Alternatively, macrodissection of the tissue is a relatively affordable alternative; however, this is not always possible, especially when analyzing cytological samples, tumors with diffuse infiltration or with high lymphocyte/stromal component. Furthermore, any added steps in the processing of a sample can lead to errors and this is especially relevant for manual processes such as macrodissection. Finally, the clinical reports for somatic mutations are not standardized and there are no clear guidelines for clinical interpretation. For example, the American Society of Clinical Oncology’s provisional opinion on Kirsten ras (KRAS) oncogene testing in metastatic colorectal cancer clearly states that only mutations in codons 12 and 13 of KRAS should be tested, owing to the available information from large randomized Phase III studies that have enough statistical power. By contrast, many companion diagnostic tests and laboratories offer testing of codons 61 and 146 on the basis of smaller studies, without the power of the other studies, and this is similar for any mutations occurring at low frequencies (e.g., EGFR exon 18 mutations or BRAF mutations other than p.V600E)
In this article, we have highlighted multiple challenges that are some of the factors delaying the widespread adoption of molecular diagnostics for patients’ stratification, with some of them being more critical than others. There are clearly multiple solutions to all these challenges, and in the long term, molecular diagnostics will become an essential tool for the correct diagnosis and patient stratification of targeted cancer therapies, but the question remains of what can we do in the short term? The speed at which molecular biomarker discovery is currently operating will bring answers to many scientific and clinical questions in the next few years that will allow for fine-tuning of molecular tests for the right therapies. The same technology, now only available for research purposes, will be made affordable and the quality used in the clinical diagnostic setting will be improved. Multiple simultaneous testing of different genetic markers for the same sample will offer not only a faster turnaround time but also a more comprehensive genetic picture and increased hit rate – potentially reducing the risk of false negatives. However, although the cost per test may be maintained or even reduced, the total cost per sample is likely to be higher, and the issue of reimbursement still remains. New trials will be designed to enroll only molecularly defined patients to increase the power of the analysis with fewer patients, which together with a better suited regulatory environment could lead to rapid implementation of new drugs in the clinic in a fraction of the time that it currently takes for such drugs to be approved. Finally, even though centralized molecular testing in comprehensive cancer centers would help to control the cost, the quality and the clinical interpretation of complex specialized testing, this will require worldwide collaboration to design common protocols for processing and transfer of the tissue, molecular testing and clinical report. Eventually, a virtual global molecular diagnostics laboratory comprised of multiple reference centers worldwide could be the answer to unrestricted patient access to stratified cancer medicine.
Financial & competing interests disclosure
The author has no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.
No writing assistance was utilized in the production of this manuscript.
References
- Lander ES, Linton LM, Birren B et al. Initial sequencing and analysis of the human genome. Nature409, 860–921 (2001).
- Venter JC, Adams MD, Myers EW et al. The sequence of the human genome. Science291, 1304–1351 (2001).
- Varmus H. Ten years on – the human genome and medicine. N. Engl. J. Med.362, 2028–2029 (2010).
- Ley TJ, Mardis ER, Ding L et al. DNA sequencing of a cytogenetically normal acute myeloid leukaemia genome. Nature456, 66–72 (2008).
- Hudson TJ, Anderson W, Artez A et al. International network of cancer genome projects. Nature464, 993–998 (2010).