Chugai Pharmaceutical Co., Ltd. (4519) Earnings Call Transcript & Summary

Good afternoon, everyone. I'm Satoko Shisai, Vice President, Head of Digital and IT Supervisory Division at Chugai Pharmaceutical. Thank you very much for joining this meeting in person or via Zoom today. I would like to give you an overview of Chugai Digital Vision 2030. First of all, as you can see here, why does digital technology, digital transformation or Dx matter in the pharmaceutical industry right now? There is a recent trend for pharmaceutical companies like us focusing on drug discovery as a core business. You can see R&D productivity of mega pharma on the left side of this page. Productivity has been falling vis-a-vis costs required to generate one innovative new drug. It requires a lot of cost and time for very difficult drug discovery, resulting in lower productivity. The right side of this page shows the analysis by Dr. Okuno of Kyoto University. There have been 3 major challenges in conventional drug discovery. First, the duration required, from basic research to approval, has been more than 10 years. And costs have been rising, as I mentioned before. The success rate from basic research to approval has been very low. Leveraging recent digital and AI technologies could be useful in addressing and resolving these challenges according to expectations. In October last year, we launched an organization to promote Dx and developed Chugai Digital Vision 2030, which was announced in March this year. Under Chugai Digital Vision 2030, we aim to transform our business by using digital technologies to make Chugai a top innovator in the provision of society changing health care solutions. Originally, we have been aiming to become a top innovator according to our management vision, so we'd like to realize this through digital technologies. We'd like to provide innovative drug products continuously, quickly streamline all value chains for higher efficiency and productivity, and change employee awareness and organization structure and customs at Chugai, which will lead to society-changing health care solutions. We hope we can provide optimal personalized health care, enhance patients' QOL and bring about sustainable social assurance programs in the end. That is how we established our vision. Based on this, we started our company-wide Dx initiatives. Under Chugai Digital Vision, there are 3 basic strategies: the left in blue is to leverage Dx in drug discovery, which is our core business. This is what we call DxD3 internally. DxD3 stands for digital transformation for drug discovery and development. Another strategy is to optimize all value chains, from R&D to clinical development, manufacturing and commercial sales and marketing for higher efficiency at all divisions. To realize this, we need to strengthen the digital platform. Sorry for the busy slide, but this shows our plan to promote digital transformation with 3 focus areas for DxD3: digital plans and commercial for the optimization of all value chains, systems and frameworks to generate ideas, talent development, partnering and IT platforms to strengthen our digital platform. This is our digital transformation road map. We have kicked off projects for each of the 3 strategies. We have drawn a timeline to be realized towards 2030. First, Chugai itself must become a digital-ready company this year, so we must strengthen the digital platform. We also plan to work on value chains and drug discovery and development starting from Phase I to be able to provide and realize society changing innovative services in the end. As you can see here, we are trying to build a portfolio of IT investments and investments into strategic areas by each division so that we can share with the top management what kind of investments we are planning to make in which areas in order to optimize our investments. We divide IT and digital costs and investments into transformation, growth and operation to show what kinds of investments we will make in which areas or where to reduce the budget. We are sharing this with the top management. Regarding the structure of digital strategy promotion, Strong backup support, leadership and cooperation from our top management; Kosaka, CEO; Okuda, COO and President and CFO, is extremely important in promoting digital transformation across the entire company. They can send a clear message about our digital strategy through New Year's remarks, interviews by analysts and members of the media and our annual reports. We regard this as very important as Dx can be realized with the top management's commitments. Under them, we have the Digital Strategy Committee attended by Value Chain division heads. They meet once a month to discuss Dx-related topics such as portfolio, case sharing and investments. This is very useful in promoting Dx towards the common role shared by the entire company. As for my organization, Digital and IT Supervisory division, it's unique in a sense that there are 2 departments under me. Digital Strategy Department, shown on the left, which is an organization newly established in October this year; and IT Solution Department, former Information Systems Department, shown on the right. Digital Strategy department was launched in October this year, and it's new and young. We brought in members of each division who are not necessarily IT professionals, but who understand our business very well. And we also have the conventional IT department. So in general, there tend to be some conflicts between the department, which is trying to promote something new, and the other department which is responsible for actual system operation and implementation. They may not always be good at reaching consensus with alignment from the same perspective. But by having such an integrated team, we can share the goals we are aiming to achieve, so I think it's a good target for us. From here on, I'd like to talk about specific examples of our initiatives using digital technologies. As for DxD3, my colleagues, Hiroyuki Tsunoda and Nobuya Ishii, will explain later, so I'd like to talk about digital biomarkers. What are digital biomarkers? ECG app has been approved as a medical device for wearable Apple Watch recently. Now tools and apps are becoming available to enable the monitoring of patients' conditions, their clinical course and physical conditions by using wearable and mobile apps as well as implants. We can leverage these technologies to better capture the patient's outcome so that it will lead to drug discovery and the understanding of their diseases. We are aiming to do this through digital technologies. For example, we are now doing research on treatment of endometriosis. Regarding pain, which cannot be evaluated objectively and measured continuously, we are doing experiments to visualize such pain with digital devices and AI for objective evaluation. Also, our company is providing a therapeutic for hemophilia. Hemophilia is an intractable disease where hemostasis is difficult once bleeding starts in patients. What kinds of their daily activities are causing hemorrhagic events, and what are their current conditions? It can be monitored with wearable devices and ePRO apps. This is one of the things we are aiming at right now. As for value chains, one thing we are working on is manufacturing in factories. We have multiple pharmaceutical factories. By visualizing the operations, data accumulation and forecasting, we are aiming to realize more automated future factories with more advanced robotics. Regarding commercial, sales and marketing, my colleague, Takato Shimauchi, will explain later, but these are the initiatives to provide various insights to MRs in the field in a digital fashion. To strengthen the digital platform, we are working on IT infrastructure. Conventionally, in the on-premise environment, researchers used to create various sharing type-environments to do their analysis, but it's very costly and it takes a lot of time to build the environments. They can work more flexibly and more securely with the use of the state-of-the-art clouding technologies. Our researchers handle enormous amounts of high-security information centering on genomic data. They can process such data with high productivity while ensuring security. We are building this as an infrastructure which can be used not just internally but also externally as well in joint research with academia and various partners. As part of the infrastructure, we are also working on our organizational culture. We need to realize digital transformation throughout the entire company, so we need a structure to build company-wide momentum for Dx so that all of us must be united and work together to realize Dx. We are appointing the so-called digital leaders who can serve as influencers from each division. We are realizing initiatives to visualize the budget and share the portfolio through these digital leaders. Also, we are holding digital summits. In February this year before the COVID-19 pandemic, we organized a summit with hundreds of participants. Just last week, we held the second summit in full online settings. On Friday, the last day of the summit, we invited external members with entries from about 1,000 people. And we also opened our website. We'd like to share what we are doing internally and demonstrate externally that Chugai Pharmaceutical is a digital company with various types of digital talent. We want digital talent outside to become interested in Chugai and hope to attract them for hiring. So we are sending out a lot of information. We are issuing press releases on various alliances with external partners. These are just some of the examples. We announced joint press releases with partners working together with Chugai in a variety of areas. We are actively disseminating information on the first initiatives in the industry as well as initiatives which will lead to the industry standards. Regarding human assets development, we are trying to internally develop people who can serve as data scientists and digital project leads. Of course, we are hiring such talent from outside, but we also develop our own curriculum to enhance the skills of our employees. This is about Digital Innovation Lab. We started in February and about 150 digital ideas were generated from within the company. Of course, it's intent for these ideas to be adopted as a real project, but we are using this to create a momentum internally to say it's fine to fail. Let's give it a try first. Roche, our current company in the group, has digital, AI and IT capabilities and resources, more than Chugai. With fundamental agreement for interactive use of such resources, we are hoping to start joint projects with Roche in the respective value chains and areas in the future. Today, analysts are also joining this meeting. As an achievement after 1.5 years of activities, we are selected this year as 1 of the 35 DX Stock 2020. We are the only company from the pharmaceutical industry. I think this raises the visibility of Dx also within our own company as well, so we are very pleased with this. Going forward of course, we will promote our own Dx internally, but also we'd like to promote Dx more broadly for value creation in the society as well. So we appreciate your continued support for the future. That's all for me. Thank you very much.

Ladies and gentlemen, thank you for joining. I'm Hiroyuki Tsunoda, Head of Discovery Technology Department, Research Division at Chugai. I'd talk about innovation in the drug discovery process using AI and AI-powered drug discovery. As Shisai explained earlier, we are promoting digital transformation for drug discovery and development at an in-house project called DxD3. Out of the 3 pillars, I'd like to talk about AI-powered drug discovery. Today, I'd like to explain AI-powered drug discovery in 3 parts. First, about discovery processes of antibody therapeutics. From before, we have been announcing our proprietary antibody engineering technologies as our strength. In addition to the antibody engineering technologies, our antibody discovery platforms are also our major strengths. Specifically, after target validation, we have lead antibody identification followed by lead antibody optimization and clinical candidate selection in the process. In the lead antibody identification process, we are using rabbit B-cell cloning and designed antibody phage library technologies. We are leveraging AI in these elements. We also use AI and lead antibody optimization. We combine antibody and machine learning, one of the AI technologies. We call this project MALEXA internally. As was explained earlier, MALEXA is used in lead antibody identification and optimization processes. We are doing this by developing process-specific machine learning algorithms. First, let me give you an example of the application to the lead antibody acquisition process. This is the antibody library technology I mentioned before for phage display of antibody fragments at the tip of a molecule. Binding molecules are enriched from the antibody library. This process is repeated a few times. In the enrichment process, the whole sequence is analyzed by using next-generation sequencing. Then we will come to know the characteristics of the binding molecules. We cannot tell with human eyes, so we use machine learning for pattern learning, sequence generation and making proposals. By developing machine learning process, a small number of proposals are made in a very short period of time without the usual antibody evaluation process. Researchers can engage in antibody production and evaluation. You can find the results in the middle of this page. If you compare the binding affinity, it was confirmed that the binding affinity was stronger in the antibody molecular group where MALEXA was applied in comparison to the normal selection. We are using these technologies in multiple projects with successful acquisition of antibodies with strong binding affinity. After lead antibody identification, you go into the process of lead antibody optimization. Here, we also have our own platform. We are using multidimensional antibody optimization system. There are about 70 antigen binding regions. We modify each amino acids to produce about 1,300 antibodies for 1 lead antibody. We have built a platform to enable this antibody production in 1 week or so. Our high throughput system is run to evaluate about 1,300 antibodies in 2 to 3 weeks. After antibody data became available, researchers had to identify the characteristics of each antibody to judge the combination of replacements. If I use a simple diagram, you get multidimensional antibody property data, antibody expression amount, the possibility of polymerization and stability in addition to the binding affinity analyzed within this COSMO system, Comprehensive Substitution for Multidimensional Optimization. Before, researchers had to look at each result one by one to evaluate antibodies with different amino acids replacement. Then the combinations of replacements are limited within the scope of the acquired data. We wanted to acquire antibodies with suitable properties and activation from a large amount of data. We thought that we could apply machine learning here and began our initiatives. We obtained experimental data. We developed a sequence generation algorithm and predictive modeling algorithm so that the computer proposes sequences, which is evaluated by researchers. The results are shown in the bottom. By applying MALEXA-LO, we were able to obtain ideal PH-dependent binding antibodies that have high binding affinities that dissociate with low PH conditions. It was also confirmed that these antibodies have good physical properties. So we believe we have created an excellent antibody optimization system. In the machine learning workflow, I would like to talk about sequence generation technology. I earlier mentioned that there are 1,300 antibody sequences. When these 1,300 sequences are combined randomly, you get 10^89 combinations. Here in the universe, it said that the number of stars in the universe is 10^22, so the number of combination is far more than that. It cannot be evaluated considering the current computational limits. So you need to reduce the number of combinations to about 10^10, which is around the current computational limits. The technology called LSTM was used for that. Let's say, there are words to compose sentences such as "I eat curry" and "I ride a car." A learning model is built so that a sentence like "I eat a car" will not be created. In this application, the words are replaced by amino acids and sentences by antibody sequences. The generation model was created to make the machine learn the antibody sequences so that suitable sequences are listed. Then, it is possible to generate various combinations beyond the 1,300 antibody sequences that were actually measured. The computer can analyze up to 100 million sequences and the optimal sequence can be selected from them. This is the MALEXA technology. Machine learning, MALEXA, is applied to both lead antibody identification and its optimization, as I explained earlier. Minimizing immunogenicity is also an important issue for development of antibody drugs. We are currently working on the application of MALEXA for the reduction of immunogenicity to further improve the efficiency of overall process. In addition, Chugai is now working on a new field of drug discovery of midsized molecules. We have already started to apply machine learning technology to the platform for drug discovery of midsized molecules. So such AI technology is being applied to the modality of the new generation. Next, I'd like to talk about AI technology in image analysis. As you know, AI technology is best fitted for image analysis. AI technology has been used for pathology in drug discovery process since a while ago. A field called digital pathology has already been established. This technology is utilized in our company and pathology researchers and data scientists collaborated to develop deep learning algorithm to achieve overwhelming improvement of research productivity. Specifically, about 500 slides can be processed when you apply AI technology, which is a totally different research world compared with the case where researchers examined each slide one by one. Digital pathology has another strength, that is, it allows quantification or digitization. You can extract future amounts of various forms, such as size, shape, color or position. If you are able to extract such information, you can combine them with gene expression or gene mutation, for example, to collect new information. It is a great advantage that you can conduct an integrated analysis like that. Using deep learning, we made AI to learn our image data and inference models were used for remodeling of the information. Thus, we established the technology to recognize shapes. We combine commercial digital pathology software and in-house development to establish highly productive drug discovery process. Now I would like to discuss about TACTICS, which is an activity to generate ideas for drug discovery. We also talked about TACTICS in the media seminar last year. It is a cross-organizational activity to generate ideas for drug discovery. Specifically, we want to promote unique biological discoveries and their sublimation into inventions leading to highly efficacious products. We have a cycle of discussion of ideas and feasibility studies, gradually making them drug discovery projects. One of the issues here is how to process a huge amount of publications to lead to the generation of ideas. As written on the slide, researchers have a limited amount of time for carefully read papers, although the amount of publications is enormous. Also, it is difficult to obtain information from surrounding areas in different fields. We have jointly developed AI technology for text mining in collaboration with FRONTEO and use it for drug discovery process. In FRONTEO's Concept Encoder, words and sentences are vectorized in 300 to 1,000 dimensions, and it is possible to search the vectorized papers. For example, in case of rheumatoid arthritis, it's keywords such as various cytokines or drugs are displayed on the plot. Then, you can collect information from them. We have established infrastructure where idea database from TACTICS and AI technology are matched and utilized. Specifically, we have created a database of unique information from ideas in addition to publications. On the bottom of the slide, there are connections between various shapes such as circles and triangles. Connecting the red star with a red inverted triangle may be difficult for humans. But AI technology is very good at connecting degrees of consistency or degrees of similarity. Thus, AI technology is applied for utilizing structured database of disease etiology or related molecules. On the right side of the slide are our examples. With regard to Factor VIII, for example, it is activated by the proximity effects with Factor IX and Factor X. Bispecific antibodies are quite good at producing proximity effects. So if you have a database of proximity effects, you may find a new treatment approach using antibodies. We utilize such database of ideas. Let me summarize the innovation in the drug discovery process using AI. We are trying to realize various molecular designs with AI technology. We also use AI for image analysis. With regard to text mining, clustering of papers and network analysis, as well as connecting with idea database are promoted. Robots to support experimental work are being developed. And the new laboratory, which is under construction in Yokohama, will be equipped with an IT infrastructure that actively utilizes AI and various digital technologies to create the next-generation laboratory with high productivity. That's all from me. Thank you.

My name is Ishii, Head of Science and Technology Intelligence Department, Project Lifecycle Management unit. I'd like to discuss utilization of real-world data. This is my agenda today. First, I will discuss goals for utilization of medical data, including real-world data; and then, current status and issues for utilization of medical data; lastly, initiatives led by Chugai. The slide shows the image of ideal drugs in the future. As shown on the left, we believe that the required values that we offer will be higher in the future. In recent years, the environment is changing rapidly so that the increase of the required value level will be remarkable and rapid. So far, efforts were made mainly to enhance the value of the drugs themselves to enhance the value that we offer. But going forward, we believe that we have to make efforts to enhance the value by offering drugs plus alpha. What is important then is to offer drugs as well as digitalized information. So digital and data are important factors. An important part of medical data is real-world data. Let me explain real-world data here. It is different from clinical studies where our objective is set in advance and data are taken. Real-world data are collected in daily clinical practice. Thus, the size of the data is much larger compared with the data from clinical trials. But the data are not collected for specific purposes and thus, the quality of the data may be slightly inferior to that of clinical studies. Examples of real-world data are data from claims, DPC, electronic medical records, medical checkup. The results from the analysis of real-world data is called real-world evidence. Real-world evidence is what actually is used. The slide shows how medical data, including real-world data, are utilized. Data from clinical studies account for a small portion of medical data and the majority is real-world data. By utilizing real-world data, the amount of data becomes enormous and when genome data are combined, data can be analyzed in more detail going forward. One scenario of utilization is, as shown in green, extension of healthy life expectancy, including health care and improvement of nursing care service based on science. In addition, it is expected that data can be utilized for community health care partnership, support for clinical decision-making and the provision of optimal treatment. The blue section is quite relevant for us. Through the utilization of real-world data for NDA and drug discovery and basic research, we can develop and provide new health care services. All these will lead to the rapid and continuous return of the values to the public and the patients from whom we receive the data. Recently, a new aspect of real-world data is getting attention that is real-timeness. The first example is related to the Obama Care. Racial disparity in access to timely cancer treatment was a problem. They examined if it was eliminated with real-world data using electronic medical records. It was actually reported that the racial disparity was eliminated where Obamacare was introduced. The bottom graphs show real-time detection of the changes of medical environment and the COVID-19 pandemic. The left graph shows that the number of registered patients in clinical studies decreased with COVID-19 pandemic. The graph on the right shows the number of cancer patients that visit hospitals. As COVID-19 become more prevalent, the number of patients clearly decreases. By understanding real-time changes of the society, health care policies can be implemented on a timely manner. What effects can be expect by utilizing real-world data? I talked about the patients earlier. As for government, effects of health care policies can be assessed and the effects can be studied in the actual clinical settings. As for health care-related companies, it may lead to the creation of new service opportunities based on scientific evidences. For health care providers, they can understand the realities of the care in the institutions, compare the care in the hospital and other hospitals and make decisions based on the real-world data. In academia, studies with the use of real-time medical data can be conducted. For us, pharmaceutical companies, real-world data can be used for NDA, post-marketing surveillance, improvement of R&D efficiency, evaluation of cost effectiveness of drug price negotiation and marketability survey. Regulatory-grade real-world data whose quality is high enough for NDA is getting attention recently. It is expected that when NDA filing based on regulatory-grade real-world data becomes possible, randomized clinical trials can be smaller in size or even be skipped, which means NDA filing can be done earlier and patients can have access to the drugs earlier. The slide summarizes issues to utilize regulatory-grade real-world data. The first issue is to ensure access to real-world data that can be used for NDA. Specifically, the quality of data and the scale or size of data can be the issue. As for the quality of data, papers on the research using real-world data regarding the risk of drug administration to patients with COVID-19 were withdrawn due to data reliability. So the quality of data is critical. As for the scale of data, data are currently generated at individual institutions or our research groups, and they are not combined. So the scale of data is also an issue. The second issue is the establishment of analysis platform of real-world data. Unless a transparent analysis platform is established, the data cannot be turned into evidences. Further, establishment of social environment to utilize the real-world data for NDA is another issue. Guidelines, for example, will be needed. The slide shows the initiatives for utilization of real-world data in the U.S. In 2009, as one of the measures for recovery after a financial crisis, HITECH Act was enacted. It included actions to promote the use of electronic medical records, which promoted the accumulation of real-world data. And in the 21st Century Cures Act in 2016 provided the use of real-world data in regenerative medicine and digital health. NDA filing using real-world data started. This is one such example. Flatiron Health is a Roche Group company based in New York. It collects electronic medical record data in oncology field. They have collected data from about 2 million patients, and they are continuously updated. The data come from over 280 cancer clinics and 7 academic cancer centers. They process this data and provide them to top 15 companies in oncology area. In addition, they cooperate with FDA to solve issues of real-world data utilization. As one result of such efforts, as shown in the charts below cited from a paper, the left chart shows the result of an actual randomized clinical trial with the blue line indicating the investigational drug arm and the red control arm in survival curves. As time passes after the drug administration, the survival ratio continues to decline, but you can see that the investigational drug arm had a longer survival period. The right chart shows the comparison with the real-world data with the blue line using the result of the clinical trial, while the red line showing a control arm created by gathering real-world data of the patients with background similar to the population in the clinical trial. As you can see, survival curves similar to the ones on the left were drawn. Therefore, this is one of the papers implying the possibility of synthesizing a control arm with real-world data. When you apply this approach to not just one but 11 different clinical trials, the result of the evaluation has been the one on the slide. The horizontal axis shows how much improvement was seen by the investigational drug or hazard ratio in the clinical trials, while the vertical axis showing the result when the real-world data was used as a control arm. If the dots are plotted along the dotted line, that shows that accurate prediction was made. As you can see, except 1 dot, fairly accurate calculation of hazard ratios were made, allowing us to believe that the value of using the real-world data is beginning to be proven. Thus, Flatiron Health has used the real-world data to support regulatory applications for drugs or for conducting the cost-effectiveness assessment to negotiate drug pricing and reimbursement. It was reported that the regulatory application has been actually approved. Moreover, as recently reported, FDA possibly accepted the use of a hybrid external control data in the Phase III trial which was synthesized by combining the randomized control arm and real-world data. It was used in the Phase III registrational trial in recurrent glioblastoma. This could lead to a future trend to possibly make use of the real-world data. The charts below show the comparison between a synthetic control arm and the investigational drug arm in the Phase II trial. The blue lines show the investigational drug arms and the red line, the synthetic control arms. You can see the investigational drug arms tend to show longer survival periods. Now what about the status of the real-world data introduction in Japan? First is the creation of registry data. CIN or Clinical Innovation Network, is encouraging the use of patient registries and information collection for cohort studies in Japan. This is also being reported by the media, that the guidelines for the application of registry data are currently being developed, an effort led by the Ministry of Health, Labor and Welfare. In order to improve the statutory framework for the use of the real-world data, the Next Generation Medical Infrastructure Law was enacted in 2018. But under this rule, anonymization of the data made it impossible to keep track of the data for a longer period of time, thus making it difficult to use the data for registration application. Furthermore, documents for application of real-world data in pharmaceutical development has been prepared by Japan Pharmaceuticals Manufacturers Association. Given this context then, how we are going to make more use of real-world data at Chugai is summarized on this slide. In order to create an environment that enables a timely and appropriate use of medical data, including real-world data to ensure decision-making to achieve patient-centric health care and to enable rapid patient access through approval applications, we're working on the use of the real-world data on top of the randomized clinical trials. More specific approaches taken by our company, are shown on this slide. They are divided into 4 quadrants by dividing cases from -- first, from the perspective of whether the product is at the phase of R&D or post launch. Or secondly, from the perspective of whether the evidence obtained will be used internally or externally. We have been taking actions for each. As one example at the top right, expansion of strategies for marketing authorization applications, we added real-world data as reference data in the application documents for a drug in patients with ROS1-positive lung cancer. In the case of R&D internal decision-making, we have used real-world data in analyzing the instance of infections in patients with lupus nephritis and rheumatoid arthritis and presented the result in an academic conference. Over the left, we have used the real-world data to establish product and therapeutic area strategies. Let me share with you one specific example here. We have a drug called entrectinib, which has been approved as NTRK inhibitor. We were aware that it worked for ROS1 gene mutation as well from clinical trials. However, there was an existing drug already approved which we needed to compare our product to. Therefore, we added a reference document in our submission package, the comparison between the result of a month therapy study of our drug and the real-world data of that preceding drug. The regulatory body's position was that there was a limitation to determining the clinical positioning in the market. But the fact that we did have such discussions in and of itself was a great progress, and we hope to continue those efforts going forward. So far, the real-world data we used was all obtained in the U.S. Going forward, we will need to establish real-world data of similar quality in Japan. All parties involved, patients, hospitals, data companies, pharmaceutical companies, health authority, government and academia, should establish the real-world data ecosystem while sharing the benefit for each. To sum up, the use of medical data, including real-world data, is expected to increase the efficiency and sophistication of medical care and provide rapid and sustainable value to the public and patients. In Western countries, there are already examples of the real-world data being used. In Japan, we also must establish a system that allows us to collect high-quality real-world data available for marketing authorization applications. We, at Chugai, aim to achieve goals such as patient-centric health care based on decision-making through improving the environment to use medical data, including real-world data in a timely and appropriate manner and timely patient access to therapies through regulatory filing. That is all for me. Thank you for your attention.

Good afternoon. I am Takato Shimauchi, Head of Customer Solutions Department, Marketing and Sales division. Thank you very much for attending this meeting today. I will talk about our digital marketing strategy, which is not such a difficult presentation. So please bear with me. Having said that, however, it is still part of our broader sales strategy, which makes it difficult for me to go into specifics from time to time. So I'd like to ask for your kind understanding in this regard. My presentation will consist of 3 parts. The first is the website, PLUS CHUGAI, a new customer touch point. The second is creation of system environment for digital marketing. The third is the pilot program that is currently underway for online workshops. First, PLUS CHUGAI, which is a website for health care professionals. It was launched on April 23, 2019. It's been 1 year and 8 months since and we have had 140,000 unique users visiting the site per month on average, reaching as many as 200,000 at its peak. This puts us among the top companies in the pharmaceutical industry in terms of the number of use. PLUS CHUGAI has a 3-layer structure. The first layer at the bottom provides drug information, including safety, to which many physicians can access. The second layer offers a variety of useful contents other than drug information. For those physicians who want to know much more in depth, we ask them to become members to this website. Now for contents, there is a lot of contents available for access by nonmember physicians. You can find the information unrelated to products such as how to create PowerPoint presentations for academic congresses shared by younger physicians, how to write papers in English and the statistical analysis, as well as video clips of surgeries and procedures. Once you become a member, you can get access to content such as online literature search called Scholaria, recently gaining popularity and said to be easier to search than other existing search systems. It is currently available for lung and breast cancers. Another example is a database tool on adverse reactions where doctors and pharmacists can enter a patient's data to look for adverse reactions likely to happen to them. Those tools are accessible to member health care professionals. Online lecture meetings are yet another example, and I'm sure competitors are also offering such services. But ours is not just about products per se, but we offer medical web forums on medical fee revisions. or a series of lectures to explain about different types of cancer pharmacist certifications. What is characteristic about Chugai's online lecture meetings is that all are webcast live, not as recorded video, with some available on demand. So far, what I talked about is online lecture meetings offered nationwide but there are programs in specific limited areas. The beauty of this approach is that, for instance, if the lecture meeting is offered for doctors in Kyushu only, the contents of the lecture can be specifically tailored to what they are particularly interested in. And yet, it could be accessible to doctors in Hokkaido as well. Another approach is to create a close community in the limited geographic area on PLUS CHUGAI and hold a lecture meeting within the community. Now there is a button in PLUS CHUGAI to allow people to directly contact MRs. If physicians type in a question and his or her email address, the question is directly forwarded to the MR in charge, which makes it possible to have interactive communications instead of unilateral ones. What we offer is not just the internal videos in PLUS CHUGAI, but more than 400 of those recorded ones distributed by external media, such as MedPeer, CareNet and Nikkei Medical, are planned to be provided next year so we can communicate our message to health care professionals. Next, I would like to discuss creation of system environment for digital marketing. We're establishing a digital platform for new marketing processes. It is a system for internal use with Chugai's original customer interface, developed in-house, allowing for flexible response. And there are 4 major characteristics. The first is that the product portfolio is focused on oncology and specialty areas as Chugai's proprietary MR digital fusion model. Secondly, there are a lot of so-called package-to-CRM programs available, but what we aim for is not such a waterfall model, but an agile one. In other words, again, we're going to establish what can flexibly respond to environmental changes. The third is to enable giving feedback to MRs by utilizing and analyzing the data with Chugai's unique approach. The fourth is that what is currently under development is a cross-functional project that can be uniformly and commonly used, not just by marketing and sales personnel, but by those in the field across divisions such as medical affairs, drug safety, clinical development and foundation medicine unit, which takes care of genome analysis. This is a process that I was referring to, a cross-functional project across products and organizations to realize customer-centric marketing that merges the digital and real world. This project, designed for organic communication, is named after just that, 0C, internally by using both initials of the 2 words. Many different divisions are involved, but there are mainly 3 projects that are underway. Data visualization, quality improvement of marketing data and approach to new customers. With our customers, all physicians placed in the center by various divisions providing services, we are aiming to enhance customer satisfaction. This slide has already been shared by Shisai earlier, but it is about integrating various data available, not just in sales and marketing division, but in various parts of the company into a single DMP, have AI analyze the data to see what value they can offer, to whom and how, and either feed them back to MRs or use them to facilitate marketing automation via digital methods alone. This can, of course, be used not only by MRs, but by MSL or medical science liaisons or SEs or safety experts. Then, this is the 2 our field team will be equipped with. There are mainly 2 APIs. One is for analysis, where MRs and managers can analyze a large volume of data by themselves using what is called MotionBoard. The other is for action which incorporates AI and make recommendations to MRs, which helps improve the probability of success for MRs in their interactions with physicians, either in terms of getting into contact, or responding to their specific needs. With regard to the relationship between MRs and digital channels, our intention is to select proper channels depending on each customer's needs. For example, we can provide information via digital channels routinely while MR is regularly calling that particular customer, interact with customers in a manner closely linked towards the information provided digitally. In other cases, the MR can offer introductory explanations, inducing physicians to start going to visit websites voluntary. And at the phase of closings, MRs can come back in. In yet other cases, physicians or patients interested in particular topics can be loosely identified via digital channels and taken over by MRs along the way. Then there is also a digital-only approach, obviously. Again, we will select most appropriate channels depending on the situation. Last but not least, I will share with you our online workshops for team medical approach. Before I talk about it, let me first discuss team medical approach. The MHLW and its Phase III basic plan to promote cancer control programs, which describes cancer prevention, improvement of cancer care and cohabitation with cancer, emphasize team medical approach as part of the measures under improvement of cancer care. It also stresses the importance of human resource development for not just doctors but also nurses and pharmacists. In fact, Chugai has been providing support to hospitals in various ways for more than 10 years to facilitate team medical approach where various medical professionals, including doctors, nurses and pharmacists involved in patient care, exercise each specialist knowledge to increase satisfaction of the patients. As part of such efforts, we offer workshops instead of lecture meetings. As you can see, doctors are divided into groups based on the hospitals they belong to and engage in group activities. MRs are also encouraged to join them for discussions as a member of the medical team. Obviously, these workshops have been all either canceled or postponed under the spread of COVID-19 this year. Instead, We have been organizing a fairly large number of such workshops via Zoom in what we call Zoom Day Workshop on a pilot basis. As you may know, we can gather all the participants in the plenary meeting in Zoom while utilizing the breakdown session function to have group discussions or carry out role play sessions in smaller groups before they get back together for presentations at the end. We have many menus available for those workshops to use. Here's an interesting fact. This is a comparison of the same workshops carried out in person last year and remotely this year. And the questionnaire survey was done asking the same set of questions. It turned out that in many of the questions, there are more positive feedback to the online version. Especially high marks were given in the items such as the clear purpose of the meeting, practical workshop and contents usable in the clinical setting. We analyzed why there are such positive feedbacks and learned that in the role-playing exercise, they do not have to care what others may think and that they were able to communicate with doctors in different areas since it was an online workshop. But the biggest benefit seems to have been about time and distance. Nurses with children found it easier to participate. There are also voices saying that under the pandemic, as there are very few opportunities to improve their skills, workshops like these were really appreciated. Now these workshops are targeted at doctors, nurses and pharmacists, but some are still unfamiliar with Zoom and do not even have devices to take part with. And therefore, when we send invitations to them, we also rent an iPad delivered in a traveling bag. And before the actual day of the workshop, there are several opportunities for MRs to check the Internet connections with them. We also use application called LINE WORKS as well and ask nurses and pharmacists with a LINE account to connect with us via LINE WORKS as well in order to make sure we can rescue them through LINE WORKS in case they have lost connection during their actual workshop. But fortunately, this gives us chances to keep communications with nurses and pharmacists even after the workshop. And therefore, we are proactively promoting this approach currently. Of course, there are many other digital approaches that we use in our sales and marketing activities, but that is all from me for today. Thank you for your attention.

Yamaguchi from Citigroup. There are many slides, but I'd like to focus on 2 questions. First, on Page 36, you talked about MALEXA. For deimmunization, MALEXA is under consideration, and you're working on it right now. For the 2 other processes where MALEXA is already applied, since when this has been in operation? And when do you start the application for immunogenicity reduction? You're also planning to apply this to the midsized molecule drug discovery platform. You are now focusing on the development of midsized molecule pipeline. Do you have any challenges in applying MALEXA in multiple stages also for the midsized molecule drug discovery process? Will you implement everything all at once or apply this little by little from earlier processes? Membrane permeability is an issue with midsized molecules. What's your view? This is my first question.

Thank you for your question. First, on the current status of MALEXA program, already as of last year, prototype was established. We are now in a phase to expand this to each project. You also asked a question about immunogenicity reduction. It's now under development, and we cannot clearly say when it will be available. Our basic platform is already completed, and we'd like to apply this technology. There are challenges we need to address, but we think we can develop this not too far away into the future. As for midsized molecule, we are working on it right now. As you know, the midsized molecule drug discovery process consists of a series of different steps. We're expecting a relatively early implementation of the MALEXA technology where it can be applied.

For optimization, how much actual data increase would enable implementation? Data acquisition is also one of the challenges.

We'd like to take this into consideration and build the entire process. That's all from me.

Did you say that MALEXA has been in operation since last year?

Yes.

Understood. So you're using different technologies or methods for the project before MALEXA, correct?

Yes, that's right.

Understood. Next, on Page 57. You talked about real-world data introduction status in Japan. I don't have a clear understanding myself. I think it's on Page 57. When can you get approval based on real-world data in Japan? It's not so clear. It seems U.S. data is being used right now. Around when will you be able to get approval based on real-world data in Japan out of the roads described on this page? What are the bottlenecks right now?

Thank you for your question. To accumulate real-world data, creation of registry data shown at the top of this page is making progress. Capturing data prospectively results in high-quality data. It is reported that MHLW could take the lead in the development of guidelines. So it is expected that it's going to be used sooner or later. On the other hand, to build real-world data from a wide range of sources, such as EHR, electronic health records, there are various laws and regulations related to privacy and personal information. So how to address these issues must be considered. Also, it is necessary to deal with these issues, including the overall framework. So we cannot say when it's going to be possible. Still, we understand that the environment is being put into place to that end.

I think up to the year 2024, is one of the first stages for your digital transformation initiatives. Do you think it will still be difficult in Japan by the timing? Of course, the situation can change as the Japanese government is launching a digital agency. As of now, it can be difficult just in a few years. What do you think?

Even when individual data is being collected, there is a question whether its quality can be guaranteed. If data is going to be collected prospectively, it may take a little more time. But registry data accumulation is already making progress, so people will try to think about how to leverage this data.

Hashiguchi from Daiwa Securities. Thank you for your presentations. I have 2 questions. First, on drug discovery, I'd like to confirm what purpose you are doing this? For example, antibody engineering and midsized molecular technologies have the potential to enable you to make something unprecedented in the world that other pharma companies cannot make, according to my understanding. Using digital technologies will make this possible? Or is it something to raise your betting average or probability of give success in making things other companies may also be able to succeed if they are lucky enough. How should I understand this from which perspective?

Thank you for your question. I understand your question is how much transformation our digital strategy will bring. Important thing is what values can be generated with the data that we obtain in the company. Important factor is how new data, including data that do not exist as real data, can be obtained using AI technology. In terms of molecular design, we are thinking like that.

Second question is when digital becomes more important, considering competitiveness among pharmaceutical companies, if a company has the headquarters or research labs in Japan, could it be more advantageous or disadvantageous? On Page 62, the summary of real-world data, you discussed Western countries and Japan separately, and it seems the situation is different. What data is to be used is very important. Chugai may be doing a good job in Japan, but how much are you concerned about the risk of losing to companies overseas?

Thank you. What we are aiming at in our R&D is to produce global products. So regardless of where the data are taken, we believe we can utilize the global data for R&D. What I mentioned about Japanese data is that the environment here is not ready to utilize real-world data in NDA in Japan. In terms of using the data, overseas data can be used for R&D in the current environment.

Sakai from Credit Suisse. Roche acquired Flatiron Health in 2018 and electronic data were made into real-world data. I think it is already in practical use. How are you involved or not involved in such activities?

I think, simply put, real-world data can be available if claims state become electronic. But it has not been done yet, and there are various electronic medical records with different standards and data are dirty, according to a President of a pharmaceutical company.

In such environment, looking from outside of the industry, I don't really understand how much the ideal can be realized. It may take some time, but how can you shorten the time access until realization considering Roche's activity? Could you explain your strategy a little more specifically? Another question. I believe ultimate short-term goal of digital, including AI, is to reduce the development cost. I don't think it was discussed today. What should I understand about cost efficiency or cost performance? For example, will the R&D cost be halved in 10 years? Will it be realized?

Flatiron is a wholly owned subsidiary of Roche now. They have not started activities in Japan yet. My understanding is that there has not been any discussion on business operations yet. First, data building in Japan is needed now. As for the data of Flatiron, I have cited several points in my presentation, and they were actually used in NDA in the U.S. As you pointed out, real-world data are not collected for specific purposes, so the substance may be an issue. I think that is an important point of Flatiron's business model. They have a process for people to check the data quality taking some time. Thus, they differentiate their data from other real-world data, and this is their advantage. As I mentioned in my presentation briefly, apart from whether it actually happens or not, ideally, by totally replacing the control arm with real-world data, costs can be reduced or development period can be shortened or at least size of the clinical trials can become smaller. This will lead to the reduction of development cost. In addition, if the results of later-stage clinical trials can be predicted based upon the early-stage clinical trials, selective investment can be made, again, leading to the reduction of development cost. In the research part, I understand AI can help improving the efficiency.

This is Oka from SankeI Shimbun. Will drug discovery utilizing real-world data replace the drug discovery using clinical trials in the future? Or will it still be complementary? Also, about the data source, it is registry data, but do you also consider obtaining data directly from medical institutions or data companies?

Let me answer. I understand your question is whether real-world data replace clinical trials. They cannot completely replace clinical trials. Actual patient data are still needed. But it will be possible to reduce the number of patients utilizing real-world data. As for the generation of real-world data, collaboration with medical institutions is important. A very important factor for data is scaling or making the size large. Collaborative research with 1 medical institution cannot achieve this. In addition to the collaboration with medical institutions, scaling of data needs to be carried out.

Another question on Page 15, specific initiative of Chugai endometriosis. You explained this earlier. Could you please discuss a little more specific results or vision? How to make it practical?

On the example of using digital biomarkers in endometriosis, this device had originally been developed to detect pains in a different disease. Endometriosis, pain is an important indicator in clinical studies. But the degree your pain varies from individual to individual, making it difficult to obtain the result in a unified manner, which has been regarded as one of the major obstacles in the clinical development. Therefore, in order to acquire more objective indicators, we're mulling the possibility to use this device to take the patient's vital data on a continuous basis, not just when they visit the hospital, but also at home so that we can obtain the pain indices quantitatively to help enhance the probability of success of the clinical development.

Hashimoto from Nikkei BP. I want to address this question to Ms. Shisai. On Slide 10, you explained how the portfolio looks like in terms of the budget breakdown in percentages. But it is not clear what these percentage numbers mean. So could you elaborate more on, say, what Chugai particularly focus its investment on compared to the peer? Or in which areas you have been increasing the budget over time or plan to increase going forward, if you can.

Thank you for your question. It is fairly difficult to visualize the budget allocation in IT department. For instance, in the expenses, you need to figure out how the digital project implemented by each division should be annotated so that the company-wide digital investment amount can be grasped. Therefore, it is important to strongly request digital leaders in each division to have digital deals in each project annotated in IT and submitted so that those otherwise hidden IT budget at the level of divisions can be picked up and visualized to show how much is being spent. With regard to the comparison with our peers, it's not clear, but what is often said is that 70% to 80% of the IT budget is spent on operation cost. In our case, the portion classified as transformation is the digital budget. On the other hand, growth represents new investments, including the one by the existing back office, which is different from operations such as renewal that comes with growth. The question is how to look at the balance among them as a company? The total amount cannot be expected to steadily be increased over time. That needs to be kept at a certain level. Therefore, the IT department needs to set a clear effort target in terms of how much the operation costs should be optimized or even reduced. For example, you could say about 30% of the total should be allocated to transformation or of the transformation budget, what would be a desirable ratio of investments between optimization of value chain or DxD3. This exercise itself is meaningful, informing consensus in the company on the next goal and the assessment of the current status of investments. And it is important to show the figure like this first in that endeavor.

I have another question, which is for Mr. Ishii and related to real-world data. On Slide 60, you said that when you use real-world data in your submission dossier to PMDA, the response was that there is a limitation to using the data for clinical positioning. Was that because of the nature of the data, such as the fact that it was not Japanese data? Or was it more about attitudes on the part of PMDA? FDA took more positive attitude and accepted the data, while PMDA raised the limitation which gave me the impression that PMDA was more hesitant. So was it because of the attitude of regulatory authority? Can you give us your take on that?

Thank you for your question. My choice of words was not appropriate, which may have misled you. That was not definitely the case at all. I would assume what they meant was that real-world data is too insufficient and lacks track record to be used as scientific base. The fact that the data was foreign origin may have been a factor, but the perception is that these data are still immature in terms of analytical methods, among others. Therefore, by repeatedly using the real-world data to further enhance the accuracy and build a track record, we hope to open up the path to use the data for registration applications.

Then am I correct to understand that in Japan, like in the case of FDA, there is a trend to use the real-world data more proactively going forward?

As I said, given the fact that the work is underway to develop guidelines, we believe that, that is most probably the future trend, but I'm not in a position to say that it is the case.

Namata, Editor in Chief of Mix. I have a question on digital marketing. You explained about customer interfaces. What is extremely important is the data input and the data gathering. According to the presentation material, daily sales report, sales data and other daily activities will be entered as input. Daily sales reports are often used by many companies, but in Chugai, as you used interface, have you put together any internal rules or formats or requested MRs for their cooperation? Another question you showed us 4 different ways to use channels. One of them is digital-only approach, which I think is fairly tricky. How much improvement in productivity are you expecting to see from this approach? Furthermore, when you use different channels for different situations, how much impact would it have on your profit compared to the current approach of using mostly MRs?

Thank you for your questions. Reports from MRs is only a small part of the data that we use. There is a lot of data other than that, which is entered as input. I did not explain in details earlier, but at the bottom of the Slide 75, there is a box with a question mark. This is what we call our secret recipe for success. There, we put in various data where Chugai has its strengths so as to be utilized to generate output. As for the digital-only channel, long-listed products, for instance, may be the one that it is applied to. However, I would like to decline to comment on the productivity and ratios for today. Thank you.