Measuring the chronology of the translational process of molecular genetic discoveries

Introduction

Since the completion of the Human Genome Project in 2001, and the HapMap Project in 2006, the pace of genetic discovery has been exponential [1], [2]. These projects have shown that the genome is a significant determinant of health [3] and response to healthcare [4], and that most diseases have a causal genetic component, thus creating expectations that genetic discoveries will strongly impact healthcare [5], [6]. Seventy percent of medical decisions rely on laboratory results [7], [8] and genetic innovations are foreseen as major sources of diagnostic and prognostic information during the present century [4], [9], [10]. However, developed countries do not efficiently deliver medical innovations to their population [11], [12], especially new gene-based biomarkers of disease derived from the human genome project [13], [14], [15]. According to the Foundation for Genomics and Population Health, although genomic science is in a “robust state”, “progress is dramatically slower in evaluating the clinical and public health relevance of these scientific advances and in developing systems for effective translation of validated tests and interventions into clinical practice” [16]. We therefore must improve our understanding of the process of transferring genetic and genomic innovations from “bench to bed side” [17], [18], [19], [20], [21]. It has been documented that it takes on average 17 years for a scientific discovery to reach routine clinical use [22]. In order to inform the translational process for molecular diagnostic tests, we have studied the chronology and pace of technology transfer for 29 different molecular genetic test discoveries with the objectives to (1) provide estimates of the timeframe between discovery and complete clinical implementation, and (2) compare the trajectories between different new tests to identify common patterns and bottlenecks.

Materials and methods

We first developed a list of 11 publicly available “timestamps” for the technology transfer process (Table 1) and identified 29 widely used molecular genetic tests with different characteristics of mode of inheritance and clinical indications (Table 2). Most of the tests concerned one specific gene involved in one or more pathologies (e.g. FMR1 and APOE), however, we also included two cytochromes P450 markers (CYP2D6 and CYP2C9) used to determine drug response, a panel of microsatellite instability (HNPCC MSI) associated with hereditary non-polyposis colorectal carcinoma as well as a genomic test for non-invasive prenatal screening for fetal aneuploidies (NIPS). In addition the markers could be categorized as follows: autosomal dominant (LDLR, RB1, F2, APC, DMPK, MSH2, RET, HTT, HNPCC MSI, BRCA1, F5, MLH1, BRCA2, MSH6); autosomal recessive (HBA1, HEXA, CFTR, LPL, SMN1, HFE, FXN); X-linked (DMD, FMR1); predictive (APOE, HTT); pharmacogenetics (CYP2D6, CYP2C9, MTHFR and TPMT); screening (NIPS). For each marker, a search was undertaken in the indexed databases and websites of Health Technology Assessment (HTA) agencies, professional associations, medical genetics information resources and biotechnology companies. The search covered the period anterior to December 31st 2017. A list of the databases searched is presented in Table 3.

Results

A long and heterogeneous translation process

Figure 1 shows the chronology of each available timestamp for the 29 markers in relation to time expressed in years. The figure is arranged so that the top marker was the latest discovered and the bottom marker was the first discovered. For 21 out of 29 cases, the discovery of the marker was almost coincident with the discovery of the marker-disease association (not shown), suggesting that the marker was discovered because of its pathogenic role. In six cases, at the end of data collection, there was still no FDA or CE approved marketed tests available, indicating that in the great majority of cases that we studied, the identification of the marker led to the development of a test for the condition. Figure 2 shows a box-and-whisker plot of the time taken for each timestamp where the dot indicates the mean. The graph shows that it takes a median of 4 years after the discovery of the marker-disease association for the patents to be awarded and guidelines to be published; a median of less than 5 years for quality control assessment programs to be available, 9 years for accessibility and 9.5 years for marketing of the test; a median of 10 years for publication of HTA studies and meta-analyses; availability of reference material takes 1 more year; and a median of 18 years for availability of FDA- or CE-approved test and 17 years for a certified reference material (although for this last timestamp only 10 tests have reached this stage). The test that was translated with the smallest time lag is NIPS, with almost half the time observed for other tests.

Figure 1:

Chronology of public timestamps for translation of genetic discoveries into the clinic.

Chronology of public timestamps for translation of genetic discoveries into the clinic. The x-axis represents the year of the timestamps events identified for each test evaluated (y-axis, see Table 2). Each timestamp observed is shown with a different symbol (see graphic legend).

Figure 2:

Box-and-whisker plot of the time taken for each test to reach specific timestamps.

Box-and-whisker plot of the time taken for each timestamp where the dot indicates the mean and the extremities of the box the upper and lower quartiles. The vertical line within the box represents the median, and the extreme lines the maximum and minimum of the values observed.

Discussion

The main observation emerging from our analysis is that there is heterogeneity in the process of translation from discovery to clinical application between the markers studied. Indeed, all timestamps were not present for all tests, they were also or exactly in the same order and the delays between each timestamp were different. Another observation is that in terms of timeline, the translation process remains very long, with nearly 10 years before a test is marketed, and more than 18 years for a FDA- or CE-approved test. Factors that may influence this timeline include: year of discovery of the marker, strength of the link between test result and clinical validity, potential for improved health care, demonstration of clinical utility (in jurisdictions where health technology assessment is mandatory prior to implementation and coverage), funding (or reimbursement) by the health care system or health care insurers, easy interpretation of test result by the physician, laboratory constraints (such as test complexity, increased regulation of laboratory developed tests), marketing of the test, ethical considerations such as social acceptability, and lobbying by patients associations. Figure 1 shows that the most recent discoveries have a shorter timeline than the oldest. Although the numbers presented here are not sufficient to establish the causality of such an observation, technology improvements in the field of molecular diagnostics and the fact that such assays are becoming more of a commodity may contribute to this phenomenon. The reasons for the much quicker translation of NIPS may have included rapid uptake, validation and implementation by industry accompanied by strong marketing of a test that has the potential of improving significantly the safety of prenatal testing and of generating huge revenues as it is a screening test [23], [24].

One additional problem in the translation of basic research results into clinical practice is the lack of specific research funding available from public funding agencies to perform the studies providing the evidence base informing the decisions about implementing or not a new diagnostic test. There are also alternate commercialization strategies and potential barriers to commercialization that may not be related to research funding [25]. These might include the challenges of protecting diagnostic IP, the generally poor profit margins on diagnostic testing (as compared to medication), and the potential disincentive of pharma to support diagnostic tests that could limit the pool of available patients eligible to treatment. Public health systems and payers (such as health insurance companies) might also be alternate sources of funding if the test has potential for economic return, system efficiency or health improvement.

Khoury and collaborators have proposed a model in which translational research is divided into four phases: (1) development of candidate health applications; (2) evaluation leading to recommendations and guidelines; (3) implementation and integration into clinical practice; and (4) health outcomes and population impacts [26]. A survey of research grants funded by the National Cancer Institute (USA) in 2010 indicates that only 1.7% of these grants supported the second phase or beyond [27]. An analysis of the biomedical literature between 2001 and 2006 showed that less than 3% of publications addressed phases 2–4 [26]. To further complicate the problem, due to more and more efficient technologies there is a dramatic increase in basic research findings that need replication and solid clinical validation, while some authors have raised issues concerning credibility and replication of these findings [28]. There is thus a need for a funding and research strategy to improve the systematic replication, evaluation, implementation and translation of molecular genetic markers, especially in the present context and high hopes for precision medicine based on genomic applications in health [24], [25], [29], [30].