Rise of the ‘personalised medicine’ paradigm
The onset of high throughput data generation technology was a major enabler for the patient-centred ‘personalised medicine’ paradigm. Personalised medical practice requires research on patient stratification, leading to the identification of homogeneous patient clusters. Identification of patient subgroups based on a limited number of determinants (companion diagnostics) is now increasingly being replaced by patient stratification based on complex, multimodal profiling using biological (genomic, epigenomic, transcriptomic, proteomic, metabolomic etc.), clinical, imaging, environmental (microbiome, exposome), and real-world data (wearable sensors and life-style data).
Data-driven patient stratification
Data analysis through conventional methods, or through machine learning algorithms, then leads to data-driven patient stratification, independent of the understanding of the disease mechanism. Such stratification may be used to define a new disease taxonomy, to refine diagnostic procedures, or to propose more targeted treatments for each of the homogeneous patient clusters. However, linking this data-driven stratification to treatment options is, in most cases, a difficult exercise, except when the stratification clearly fits with a known mechanism of drug action. This is for instance the case where somatic mutations in cancer cells suggest testing treatment targeting the mutated transduction pathway. In other cases, an additional step of translational research based on animal, cellular, organoid or in-silico models is needed to link data-driven stratification with the pharmacodynamic effects of candidate treatments, suggesting which treatment options should be tested in the various clusters.
Current challenge: complex personalised medicine research programmes
As a consequence, most current research in ‘personalised medicine’ typically consists of complex research programmes including first a stratification cohort (in many cases a retrospective study reusing data and biosamples from existing cohorts) to generate multimodal data and run stratification algorithms. Then a prospective validation cohort assesses the reproducibility, robustness and validity of the clustering in another sufficiently large patient sample. A third (optional) translational step using animal, cellular, organoid, or in-silico models may be necessary to identify possible treatment options to be tested in the various patient clusters, as defined by the multi-omic profiling. Finally randomised clinical trials (including umbrella or basket designs) are needed to test these treatment options in the subgroups of patients (which is relevant to the medicine agencies delivering marketing authorisation for innovative or repurposed treatments), and to explore the added value and cost-effectiveness of the personalised approach versus a non-personalised strategy (or to define an optimal level of stratification), which is relevant for patients, and for the Health Technology Assessment bodies (HTAs) in charge of optimising the quality and cost of healthcare.
The need for established standards and the role of PERMIT
Both scientific excellence and acceptance by health authorities of results derived from such personalised medicine approaches require established standards. These standards must address issues such as high-throughput data generation techniques, data quality, security and traceability, cohort design and methodology, statistical power, use of algorithms, scoring and validation approaches, clinical trial design, choice of comparator, etc. The PERMIT project was designed to drive the development of established standards.
PERMIT coordination and objective
Coordinated by the European Clinical Research Infrastructure Network (ECRIN), PERMIT will be based on a series of workshops where the project participants and partners invite selected experts to address the various aspects of methodology, design, data management, analysis and interpretation in personalised medicine research programmes. The objective is to reach consensus and publish recommendations on methodological standards to ensure the scientific excellence, validity, robustness and reproducibility of results, and the acceptability of the results generated by personalised medicine programmes.
Expected deliverables include:
- Literature mapping on methods in personalised medicine and detailed gap analysis
- Recommendations on stratification cohort design and methodology, on data collection, quality, traceability, management
- Recommendations on the use of stratification algorithms
- Recommendations on translational research for treatment selection
- Recommendations on clinical trial design and methodology including innovative clinical trial designs
To learn more about PERMIT activities, click here