A Review of the Challenges of Using Biomedical Big Data for Economic Evaluations of Precision Medicine

Evaluating the Value for Money of Precision Medicine from Early Cycle to Market Access: A Comprehensive Review of Approaches and Challenges

OBJECTIVES

This study aimed to perform a comprehensive review of modeling approaches and methodological and policy challenges in the economic evaluation (EE) of precision medicine (PM) across clinical stages.

METHODS

First, a systematic review was performed to assess the approaches of EEs in the past 10 years. Next, a targeted review of methodological articles was conducted for methodological and policy challenges in performing EEs of PM. All findings were synthesized into a structured framework that focused on patient population, Intervention, Comparator, Outcome, Time, Equity and ethics, Adaptability and Modeling aspects, named the "PICOTEAM" framework. Finally, a stakeholder consultation was conducted to understand the major determinants of decision making in PM investment.

RESULTS

In 39 methodological articles, we identified major challenges to the EE of PM. These challenges include that PM applications involve complex and evolving clinical decision space, that clinical evidence is sparse because of small subgroups and complex pathways in PM settings, a one-time PM application may have lifetime or intergenerational impacts but long-term evidence is often unavailable, and that equity and ethics concerns are exceptional. In 275 EEs of PM, current approaches did not sufficiently capture the value of PM compared with targeted therapies, nor did they differentiate Early EEs from Conventional EEs. Finally, policy makers perceived the budget impact, cost savings, and cost-effectiveness of PM as the most important determinants in decision making.

CONCLUSIONS

There is an urgent need to modify existing guidelines or develop a new reference case that fits into the new healthcare paradigm of PM to guide decision making in research and development and market access.

Defining a Core Data Set for the Economic Evaluation of Precision Oncology

OBJECTIVES

Precision oncology is generating vast amounts of multiomic data to improve human health and accelerate research. Existing clinical study designs and attendant data are unable to provide comparative evidence for economic evaluations. This lack of evidence can cause inconsistent and inappropriate reimbursement. Our study defines a core data set to facilitate economic evaluations of precision oncology.

METHODS

We conducted a literature review of economic evaluations of next-generation sequencing technologies, a common application of precision oncology, published between 2005 and 2018 and indexed in PubMed (MEDLINE). Based on this review, we developed a preliminary core data set for informal expert feedback. We then used a modified-Delphi approach with individuals involved in implementation and evaluation of precision medicine, including 2 survey rounds followed by a final voting conference to refine the data set.

RESULTS

Two authors determined that variation in published data elements was reached after abstraction of 20 economic evaluations. Expert consultation refined the data set to 83 unique data elements, and a multidisciplinary sample of 46 experts participated in the modified-Delphi process. A total of 68 elements (81%) were selected as required, spanning demographics and clinical characteristics, genomic data, cancer treatment, health and quality of life outcomes, and resource use.

CONCLUSIONS

Cost-effectiveness analyses will fail to reflect the real-world impacts of precision oncology without data to accurately characterize patient care trajectories and outcomes. Data collection in accordance with the proposed core data set will promote standardization and enable the generation of decision-grade evidence to inform reimbursement.

Making Biomedical Sciences publications more accessible for machines

With the rapidly expanding catalogue of scientific publications, especially within the Biomedical Sciences field, it is becoming increasingly difficult for researchers to search for, read or even interpret emerging scientific findings. PubMed, just one of the current biomedical data repositories, comprises over 33 million citations for biomedical research, and over 2500 publications are added each day. To further strengthen the impact biomedical research, we suggest that there should be more synergy between publications and machines. By bringing machines into the realm of research and publication, we can greatly augment the assessment, investigation and cataloging of the biomedical literary corpus. The effective application of machine-based manuscript assessment and interpretation is now crucial, and potentially stands as the most effective way for researchers to comprehend and process the tsunami of biomedical data and literature. Many biomedical manuscripts are currently published online in poorly searchable document types, with figures and data presented in formats that are partially inaccessible to machine-based approaches. The structure and format of biomedical manuscripts should be adapted to facilitate machine-assisted interrogation of this important literary corpus. In this context, it is important to embrace the concept that biomedical scientists should also write manuscripts that can be read by machines. It is likely that an enhanced human-machine synergy in reading biomedical publications will greatly enhance biomedical data retrieval and reveal novel insights into complex datasets.

Big Data and Real-World Data based Cost-Effectiveness Studies and Decision-making Models: A Systematic Review and Analysis

Background: Big data and real-world data (RWD) have been increasingly used to measure the effectiveness and costs in cost-effectiveness analysis (CEA). However, the characteristics and methodologies of CEA based on big data and RWD remain unknown. The objectives of this study were to review the characteristics and methodologies of the CEA studies based on big data and RWD and to compare the characteristics and methodologies between the CEA studies with or without decision-analytic models. Methods: The literature search was conducted in Medline (Pubmed), Embase, Web of Science, and Cochrane Library (as of June 2020). Full CEA studies with an incremental analysis that used big data and RWD for both effectiveness and costs written in English were included. There were no restrictions regarding publication date. Results: 70 studies on CEA using RWD (37 with decision-analytic models and 33 without) were included. The majority of the studies were published between 2011 and 2020, and the number of CEA based on RWD has been increasing over the years. Few CEA studies used big data. Pharmacological interventions were the most frequently studied intervention, and they were more frequently evaluated by the studies without decision-analytic models, while those with the model focused on treatment regimen. Compared to CEA studies using decision-analytic models, both effectiveness and costs of those using the model were more likely to be obtained from literature review. All the studies using decision-analytic models included sensitivity analyses, while four studies no using the model neither used sensitivity analysis nor controlled for confounders. Conclusion: The review shows that RWD has been increasingly applied in conducting the cost-effectiveness analysis. However, few CEA studies are based on big data. In future CEA studies using big data and RWD, it is encouraged to control confounders and to discount in long-term research when decision-analytic models are not used.

Improving personalized tumor growth predictions using a Bayesian combination of mechanistic modeling and machine learning

BACKGROUND

In clinical practice, a plethora of medical examinations are conducted to assess the state of a patient's pathology producing a variety of clinical data. However, investigation of these data faces two major challenges. Firstly, we lack the knowledge of the mechanisms involved in regulating these data variables, and secondly, data collection is sparse in time since it relies on patient's clinical presentation. The former limits the predictive accuracy of clinical outcomes for any mechanistic model. The latter restrains any machine learning algorithm to accurately infer the corresponding disease dynamics.

METHODS

Here, we propose a novel method, based on the Bayesian coupling of mathematical modeling and machine learning, aiming at improving individualized predictions by addressing the aforementioned challenges.

RESULTS

We evaluate the proposed method on a synthetic dataset for brain tumor growth and analyze its performance in predicting two relevant clinical outputs. The method results in improved predictions in almost all simulated patients, especially for those with a late clinical presentation (>95% patients show improvements compared to standard mathematical modeling). In addition, we test the methodology in two additional settings dealing with real patient cohorts. In both cases, namely cancer growth in chronic lymphocytic leukemia and ovarian cancer, predictions show excellent agreement with reported clinical outcomes (around 60% reduction of mean squared error).

CONCLUSIONS

We show that the combination of machine learning and mathematical modeling approaches can lead to accurate predictions of clinical outputs in the context of data sparsity and limited knowledge of disease mechanisms.

The concept of justifiable healthcare and how big data can help us to achieve it

Over the last decades, the face of health care has changed dramatically, with big improvements in what is technically feasible. However, there are indicators that the current approach to evaluating evidence in health care is not holistic and hence in the long run, health care will not be sustainable. New conceptual and normative frameworks for the evaluation of health care need to be developed and investigated. The current paper presents a novel framework of justifiable health care and explores how the use of artificial intelligence and big data can contribute to achieving the goals of this framework.

Clinical Molecular Marker Testing Data Capture to Promote Precision Medicine Research Within the Cancer Research Network

PURPOSE

To evaluate health care systems for the availability of population-level data on the frequency of use and results of clinical molecular marker tests to inform precision cancer care.

METHODS

We assessed cancer-related molecular marker test data availability across 12 US health care systems in the Cancer Research Network. Overall, these systems provide care to a diverse population of more than 12 million people in the United States. We performed qualitative analyses of test data availability for five blood-based protein, nine germline, and 14 tissue-based tumor marker tests in each health care system's electronic health record and tumor registry using key informants, test code lists, and manual review of data types and output. We then performed quantitative analyses to estimate the proportion of patients with cancer with test utilization data and results for specific molecular marker tests.

RESULTS

Health systems were able to systematically capture population-level data on all five blood protein markers, six of 14 tissue-based tumor markers, and none of the nine germline markers. Successful, systematic data capture was achievable for tests with electronic data feeds for test results (blood protein markers) or through prior manual abstraction by tumor registrars (select tumor-based markers). For test results stored in scanned image files (particularly germline and tumor marker tests), information on which test was performed and test results was not readily accessible in an electronic format.

CONCLUSION

Even in health care systems with sophisticated electronic health records, there were few codified data elements available for evaluating precision cancer medicine test use and results at the population level. Health care organizations should establish standards for electronic reporting of precision medicine tests to expedite cancer research and facilitate the implementation of precision medicine approaches.

Machine Learning for Precision Health Economics and Outcomes Research (P-HEOR): Conceptual Review of Applications and Next Steps

Precision health economics and outcomes research (P-HEOR) integrates economic and clinical value assessment by explicitly discovering distinct clinical and health care utilization phenotypes among patients. Through a conceptualized example, the objective of this review is to highlight the capabilities and limitations of machine learning (ML) applications to P-HEOR and to contextualize the potential opportunities and challenges for the wide adoption of ML for health economics. We outline a P-HEOR conceptual framework extending the ML methodology to comparatively assess the economic value of treatment regimens. Latest methodology developments on bias and confounding control in ML applications to precision medicine are also summarized.

Radiogenomics Based on PET Imaging

Radiogenomics or imaging genomics is a novel omics strategy of associating imaging data with genetic information, which has the potential to advance personalized medicine. Imaging features extracted from PET or PET/CT enable assessment of in vivo functional and physiological activity and provide comprehensive tumor information non-invasively. However, PET features are considered secondary to features on conventional imaging, and there has not yet been a review of the radiogenomic approach using PET features. This review article summarizes the current state of PET-based radiogenomic research for cancer, which discusses some of its limitations and directions for future study.

Quantifying Downstream Healthcare Utilization in Studies of Genomic Testing

OBJECTIVES

The challenges of understanding how interventions influence follow-up medical care are magnified during genomic testing because few patients have received it to date and because the scope of information it provides is complex and often unexpected. We tested a novel strategy for quantifying downstream healthcare utilization after genomic testing to more comprehensively and efficiently identify related services. We also evaluated the effectiveness of different methods for collecting these data.

METHODS

We developed a risk-based approach for a trial of newborn genomic sequencing in which we defined primary conditions based on existing diagnoses and family histories of disease and defined secondary conditions based on unexpected findings. We then created patient-specific lists of services associated with managing primary and secondary conditions. Services were quantified based on medical record reviews, surveys, and telephone check-ins with parents.

RESULTS

By focusing on services that genomic testing would most likely influence in the short-term, we reduced the number of services in our analyses by more than 90% compared with analyses of all observed services. We also identified the same services that were ordered in response to unexpected findings as were identified during expert review and by confirming whether recommendations were completed. Data also showed that quantifying healthcare utilization with surveys and telephone check-ins alone would have missed the majority of attributable services.

CONCLUSIONS

Our risk-based strategy provides an improved approach for assessing the short-term impact of genomic testing and other interventions on healthcare utilization while conforming as much as possible to existing best-practice recommendations.

Time to Review Authorisation and Funding for New Cancer Medicines in Europe? Inferences from the Case of Olaratumab

The potential benefits of early patient access to new medicines in areas of high unmet medical need are recognised, but uncertainties concerning effectiveness, safety and added value when new medicines are authorised, and subsequently funded based on initial preliminary data only, have important implications. In 2016 olaratumab received accelerated conditional approval from both the European Medicines Agency and the US Food and Drug Administration for the treatment of soft-tissue sarcoma, based on the claims of a substantial reduction in the risk of death with an 11.8-month improvement in median overall survival in a phase II trial in combination with doxorubicin vs. doxorubicin alone. The failure to confirm these benefits in the post-authorisation pivotal trial has highlighted key concerns regarding early access and conditional approvals for new medicines. Concerns include potentially considerable clinical and economic costs, so that patients may have received suboptimal treatment and any money spent has foregone the opportunity to improve access to effective treatments. As a result, it seems reasonable to reconsider current marketing authorisation models and approaches. Potential pathways forward include closer collaboration between regulators, pharmaceutical companies and payers to enhance the generation of rapid and comparative confirmatory trials in a safe and fair manner, with minimal patient exposure as required to achieve robust evidence. Additionally, it may be time to review early access systems, and to explore new avenues regarding who should pay or part pay for new treatments whilst information is being collected as part of any obligations for conditional marketing authorisation. Greater co-operation between countries regarding the collection of data in routine clinical care, and further research on post-marketing data analysis and interpretation, may also contribute to improved appraisal and continued access to new innovative cancer treatments.