The emerging role of real-world data in oncology care in Japan

Abstract

While the number of newly approved anticancer therapies has increased substantially in the past decade, there remains a need to bring effective treatments to patients more efficiently. In the United States and Europe, real-world data (RWD) are being utilized widely, including in regulatory submissions to assess the potential benefits of therapies in development and address the unmet needs of the patients with limited treatment options. In Japan, given the volume of annual new cancer cases and region-specific circumstances, oncology RWD are crucial for improving patient access to effective therapies. The development of such RWD is underway; a notable recent example is the use of the SCRUM-Japan Registry as a source of external control data in the regulatory approval of pertuzumab and trastuzumab for patients with HER2-positive metastatic colorectal cancer. In addition, electronic health records-based longitudinal, patient-level RWD in Japan are being curated by Flatiron Health K.K. (FHKK), the Japanese subsidiary of Flatiron Health, Inc., by leveraging both structured and unstructured data-processing methodologies developed in the United States over the past decade and tailoring the approach to local requirements. FHKK, in partnership with one of the flagship cancer hospitals in Japan, the National Cancer Center Hospital East (NCCHE), is constructing patient-level RWD in gastrointestinal and other cancers; the first datasets in gastrointestinal cancers will be provided in late 2023. As the curation of oncology RWD in Japan progresses, continuous expansion of the depth and breadth of the data is planned to improve and extend lives of people with cancer.

Highlights

• •The need for fit-for-purpose oncology RWD is growing in Japan.
• •NCCHE and FHKK are partnering to curate oncology RWD in Japan.
• •Established methods are adapted to process local electronic health records data.
• •The resultant RWD will enable the comprehensive evaluation of relevant endpoints.

Historically, the gold standard in evidence generation for advancing therapeutics and regulatory decision making in oncology has been the randomized controlled trial.^1,2 In the past decade, the rapid acceleration of innovation and drug development in oncology has created a unique challenge: how can we equitably, efficiently, and safely generate the data required to support cancer patients around the world? From assessing the potential clinical benefits of therapies in development, to addressing the unmet needs of subpopulations of patients with limited treatment options, the need for high-quality evidence has never been more urgent. Real-world data (RWD), defined as data relating to patient health status and/or the delivery of health care routinely collected from a variety of sources including electronic health records (EHRs),³ are an important part of generating real-world evidence (RWE) to address existing challenges.4, 5, 6

Cancer is a global challenge that requires global solutions. In 2020, there were >19 million incident cases of cancer and ∼10 million cancer deaths globally.⁷ While the number of new anticancer therapies approved annually has increased substantially in the past decade,⁸ there remains a need to bring effective treatments to patients more efficiently.⁹ The use of RWD and RWE as part of regulatory applications is a potential solution for accelerating the approval of anticancer therapies and improving the care and lives of people with cancer on a global scale. Technological advances and policy changes in the United States and Europe have enabled the use of RWD and RWE for preauthorization, postmarking monitoring, and regulatory approval of drug products. For example, applications utilizing RWD/RWE were approved for 43 oncology products by the United States Food and Drug Administration (FDA) between January 2019 and June 2021, and nine antineoplastic therapies by the European Medicines Agency (EMA) between January 2018 and December 2019.^10,11 The FDA, EMA, and other global agencies have developed frameworks and guidances on optimal methods for extracting and using RWD to generate robust RWE that can support regulatory decisions, including initiatives specific to oncology such as the Oncology Center of Excellence Real World Evidence program in the United States.12, 13, 14, 15, 16, 17 These advances have far-reaching implications for improving the efficiency of regulatory review and approval of oncology drugs in other areas of the world.

In Japan, around 400 000 deaths associated with cancer and nearly 1 million new cases of cancer are observed annually.¹⁸ The aforementioned need for accelerated development of effective anticancer therapies is also highly relevant to Japan, especially when considering region-specific circumstances such as the notably higher prevalence of gastric cancer, hepatocellular carcinoma, and esophageal squamous cell carcinoma in Japan.¹⁹ In Japan, there are currently three main categories of RWD commonly used in evidence generation in oncology and beyond: administrative claims databases, cancer registries or other prospectively collected data in observational studies, and EHR-derived RWD.^20,21 Each is provided by different stakeholders across the public and private sectors, and in recent years, the Japanese government has established publicly managed sources of RWD such as the National Database (NDB) and Medical Information Database Network (MID-NET).^22,23 It is widely acknowledged that the strengths and limitations of these RWD sources differ based on the nature of data collection and processing, such as in terms of availability of validated clinical outcomes, demographic representativeness, and the possibility of longitudinal follow-up of patients across medical institutions.^24,25 In particular, to address growing concerns surrounding the availability and quality of data to evaluate outcomes for accelerating research and to support regulatory decision making, government leaders and industry stakeholders have begun to promote the development and use of such RWD in Japan.²⁵

A notable recent case study is the use of RWD (e.g. the SCRUM-Japan Registry) as a source of external control data in the regulatory review of the label expansion of pertuzumab and trastuzumab in patients with HER2-positive unresectable advanced/recurrent colorectal cancer (approved on 28 March 2022).²⁶ In addition to results from 30 trial patients in the phase II TRIUMPH study, data from 6 patients in the SCRUM-Japan Registry were utilized as evaluation material, alongside referenced supplementary materials that included data from an additional 8 patients in the SCRUM-Japan Registry and 18 patients in the Flatiron Health-Foundation Medicine Clinico-Genomic Database.^26,27 Although precedence remains limited, this was one of the significant steps toward RWD applications in regulatory decision making in Japan, especially given the challenges surrounding the accuracy, traceability, consistency, and completeness of local RWD sources.^25,28

In light of the recent paradigm shift in the RWD space in Japan, the National Cancer Center Hospital East (NCCHE) and Flatiron Health K.K. (FHKK) announced a joint research partnership in January 2022, with the shared mission of addressing the critical need for oncology RWD in Japan. NCCHE is one of the flagship hospitals for cancer care and research in Japan. As a core cancer care hospital, NCCHE treats 12 000 new cancer patients annually and supports local cancer care systems. FHKK is a Japanese subsidiary of Flatiron Health, Inc. (FH). Headquartered in the United States, FH has generated USA-based nationwide EHR-derived de-identified oncology RWD at scale over the past decade by adopting a new approach of pairing deep clinical expertise with cutting-edge software engineering capabilities. FHKK was established as one of FH’s three international offices, which aim to create EHR-based oncology RWD in Japan, the UK, and Germany. Each country’s approach has been carefully tailored to the local context, including language, health systems, clinical context, data privacy, security requirements, and regulatory landscapes (Table 1).

Table 1.

Comparison of four markets (United States, Japan, United Kingdom, and Germany) where Flatiron Health operates in creating EHR-based oncology RWD

		United States	Japan	United Kingdom	Germany
Cancer incidence and associated mortality	Number of patients diagnosed with cancer per year	1 603 844 (2020)26	999 075 (2019)9	375 000 (average of 2016-2018)10	628 519 (2020)11
	Number of cancer deaths per year	602 347 (2020)26	381 505 (2021)9	167 000 (average of 2017-2019)10	252 065 (2020)11
RWD curation	FH’s RWD curation approach	•Abstraction, harmonization, and other structured and unstructured data-processing approach tailored to each market
		•Data sourced from cloud-based EHR with oncology-specific workflows used in the community oncology clinics •Integrated platform which ingests data from various EHR systems from academic medical centers	•Integrated platform which ingests data from various EHR systems from partner sites (academic medical institutions and health systems in all markets, and community oncology sites in Germany)
	Applicable law in handling personal medical information	HIPAA (Health Insurance Portability and Accountability Act)	APPI (Act on the Protection of Personal Information)	UK GDPR (Data Protection Act)	GDPR (General Data Protection Regulation)

EHR, electronic health record; FH, Flatiron Health, Inc.; RWD, real-world data.

NCCHE and FHKK are collaborating closely to construct longitudinal, de-identified patient-level RWD in gastrointestinal and other cancers, based on electronic medical records of eligible consented patients. The data-processing and data-transfer approaches in this collaboration leverage rigorous data curation methodologies established by FH in the United States, adjusted in adherence to the Ethical Guidelines for Medical and Biological Research Involving Human Subjects (hereafter referred to as ‘Ethical Guidelines’), the Act on the Protection of Personal Information (hereafter referred to as ‘APPI’), and security requirements in Japan, while taking into account NCCHE’s technical infrastructure and requirements. As a substantial proportion of important information about a cancer patient’s journey is located in unstructured data (e.g. scanned referral letters, pathology reports, clinician notes), both structured and unstructured data will be extracted from NCCHE’s EHR and other hospital information systems, and subsequently processed into research-ready RWD. In particular, FHKK’s unstructured data-processing approach in Japan is adapted from FH’s three-pronged approach: manual abstraction of data by clinical experts, automated extraction using machine learning (ML), and a hybrid approach where data curation is deferred to clinically trained humans when ML model confidence falls below a threshold.^29,30 In tailoring these established methods, proprietary software has been developed and localized to enable abstraction to be conducted locally in Japanese, and ML models and plausible hybrid approaches are concurrently being explored in Japan while accounting for potential transferability of fundamental model architectures across countries. This enables the evaluation of real-world clinical endpoints including real-world overall survival (rwOS), real-world progression-free-survival (rwPFS), real-world response rate (rwRR) as well as variables including line of therapy, biomarker-related information, and detailed treatment information. Through this partnership, NCCHE and FHKK endeavor to construct meaningful oncology RWD in Japan to fill existing gaps and enhance research, patient care, drug development, and regulatory decision making. With the upcoming inclusion of other Japanese medical institutions in FHKK’s network, this database will continue to evolve toward representativeness, and will be securely de-identified and made available for use by academia, regulatory bodies, and the industry to promote cancer research and improve care.

Realizing this vision is not without its challenges. Japan’s stringent data privacy regulation (the APPI) and Ethical Guidelines generally require obtaining informed consent from patients for the creation of research datasets containing sensitive medical information and the secondary use of such patient-level datasets by industry stakeholders for a wide range of research objectives.³¹ The implications of this include potential selection bias in the creation of a representative database, and operational challenges in obtaining informed consent. Policymakers continue to discuss the best approach to this topic,³² but it is expected that stringent data privacy and data storage requirements will remain a core design principle for FHKK’s creation of RWD in Japan. Operationalizing this vision also requires a tailored data curation approach that relies on the processing of a myriad types of documents in Japanese and underlying data formats, all of which vary by region, hospital, and EHR system. To this end, FHKK’s locally trained staff with appropriate clinical background are responsible for conducting abstraction in Japan. Simultaneously, NCCHE physicians provide clinical expertise to inform clinically meaningful data standardization and validation, in order to keep up the quality and relevance of RWD.

With the progress made over the past 2 years, and with plans for the first datasets in gastrointestinal cancers to be made available in late 2023, the potential for impact and bringing value to the lives of people with cancer is evident. Going forward, NCCHE and FH plan to expand the depth (i.e. additional data elements) and breadth (i.e. additional cancer types) of this collaboration, engaging key stakeholders including Japanese regulatory authorities to help shape the narrative around developing a strong RWD infrastructure, and ultimately, to improve and extend lives by learning from the experience of every person with cancer.