The extraordinary scientific and technical advances in basic and translational cancer research, coupled with our increased knowledge of the nonmalignant and cancer genomes has changed our understanding of the multitude of diseases termed cancer, at every level: these advances include diagnosis, screening, prognosis, treatment, clinical trial design, and patient stratification. The pace and scale of the improvements combined with economic and societal challenges has influenced cancer management more profoundly than most therapeutic areas. Key among the many challenges in clinical oncology and other non-communicable diseases, however, is the juggernaut of how to implement precision medicine in the real world. At the Worldwide Innovative Networking (WIN) conference on personalized cancer medicine, held in Paris at the end of June, key challenges debated and discussed included the following: how can 'big data' from trials and real-world settings be used to define value-based care? How will global implementation of precision medicine be addressed in light of trial design challenges and our inability to prioritize the seemingly impossible number of testable drug combinations? Finally, how do we ensure increased patient access to trials and experimental drugs that are cost-efficient and provide true clinical benefit?

One of our biggest hurdles is how to translate the data captured from adaptive basket and umbrella trials to trials that are subsequently initiated in local hospital settings, in which the inherent infrastructures are not set up or scalable in a way that reflects the sophisticated coordinated approaches adopted at internationally recognized cancer centres in high-income countries. ASCO has expanded its TAPUR trial to over 100 sites in the USA, which is welcome news. TAPUR is a nonrandomized basket study that will investigate the safety and efficacy of commercially available cancer drugs when they are given off-label on the basis of the genomic characteristics of patients' tumours. Two of the largest challenges associated with TAPUR have been ensuring that patients received access to molecular testing, and that such testing was standardized. As part of this initiative, Cure-One, which runs an N1 Registry, aims to provide patients access to high-quality next-generation-sequencing testing, in order to help improve our understanding of genetic markers in cancer, and to track how genetic testing could be used in real-world settings, and ultimately impact patient outcomes.

Disturbingly, the paradox of our current drug regulatory system is that we design, study, and stratify patients to participate in adaptive trials according to the specific molecular profile of each patient's tumour, but then approve drugs by indication on the basis of the tumour site of origin. Encouraging progress was demonstrated by the tissue-agnostic approval of pembrolizumab by the FDA in May 2017 for adult and paediatric patients with unresectable or metastatic microsatellite-instability-high or mismatch-repair-deficient solid tumours. Crucially, it is important for patients that future 'adaptive' regulatory decisions reflect 'real-world' patients, and that care for such patients is not denied on the basis of optimal trial end points and restrictive inclusion and exclusion criteria. One travesty is our inability to embrace drug repurposing. We test agents in phase I trials in patients with limited or no treatment alternatives, but the lack of benefit observed in the adjuvant and neoadjuvant settings creates a warped cycle of inability to define and test true clinical benefit owing to financial disincentives. We must overcome this barrier.

Intriguingly, low and middle income countries benefit from the opposite approach: they focus on research using standard therapies with proven benefit, in which pragmatism and necessity can drive lower costs and innovation. In any economic setting, the complexities of biopsy acquisition, sequencing technology, and data-analysis algorithms, and the synthesis of genomic information to guide treatment decisions, creates a situation in which a patient can receive different recommendations from different centres. Many examples exist in which good treatments enter a wasteland between translation from preclinical research to the marketplace: too often even when preclinical evidence is promising, drugs entering late-stage clinical testing are dropped by biotechnology companies or industry because of cash constraints. In reality, a considerable number of neglected therapeutic agents and interventions arise because of a complex set of financial and motivational reasons.

...we need to avoid the 'valley of death' by adopting a long-term vision of curative and preventive healthcare

Although collaborative efforts are underway by governments, academia, and private industry who are working together to translate the latest scientific breakthroughs into advances in treatment, we need to avoid the 'valley of death' by adopting a long-term vision of curative and preventive health care. Leroy Hood's keynote lecture epitomised the radical steps required globally. We must deal with the complexity of complex adaptive systems and avoid the noise from 'irrelevant biology' using n-of-1 dynamic personalized data clouds — P4 medicine — to reveal deep insights into the processes that govern wellness and the slide into disease.