Points of View Success Probability Analysis of Drug Discovery -Focusing on Clinical Trials

Printable PDF

Yosuke Takahashi, Senior Researcher, National Institute of Biomedical Innovation Policy

Introduction

The pharmaceutical industry is said to be a high-risk, high-return industry. It generally takes more than 10 years from the start of research and development to the launch and marketing of a product, and a large amount of R&D investment is required in the process. In addition, the success rate of drug discovery is currently low, as R&D must often be abandoned due to various factors, such as insufficient efficacy, safety concerns, or lack of marketability. As a result, there are countless cases where large investments over a long period of time cannot be recovered at all, and this is considered a high-risk industry unparalleled in other industries.

In terms of lowering risk (≒ improving the probability of success in drug discovery), various efforts are being made against the backdrop of advances in science and technology, and a certain amount of success has been achieved in bringing innovative new drugs to ^market1). In fact, new drugs have been created for various diseases, including lifestyle-related diseases, to satisfy the medical needs of patients, and treatment satisfaction is increasing year by ^year2). As a result, in recent years, there has been a trend toward shifting R&D targets to areas where there are many unmet needs. However, these intractable and rare diseases are generally considered to have a high degree of difficulty in drug discovery, resulting in a lower probability of success, and will continue to be a high risk type industry for the foreseeable future.

In this paper, we analyze and discuss the actual R&D success rate of various new modalities, such as biopharmaceuticals, as they are being put into practical use, and contrast it with the results of previous studies on success probability.

Organizing Information on Previous Studies

There have been several previous studies on the success probability of drug discovery, ^including a comprehensive report by Yagi et al. ³⁾, a former senior researcher at the Pharmaceutical Industry Policy Institute, a report by DiMasi et al. ⁴⁾, and a report by Paul et al. ⁵⁾. In these previous studies, the probability of success (probability of transition to the next phase) in Phase I to Phase II, Phase II to Phase III, Phase III to Filed, and Filed to Approved was calculated based on the results of original questionnaire surveys from different populations. The probability of success of drug discovery from Phase I to Approved (in this case, the probability of success of clinical trials) is calculated by multiplying the product of these probabilities. The data collected covered the periods 2000-2008, 1995-2007, and 2000-2007, respectively, and projects that moved to the next phase or discontinued development during these periods were counted, with no conditions placed on when the project was initiated. The Yagi et al. report is based on the results of a questionnaire survey of Japanese pharmaceutical companies, and distinguishes between the probability of success for projects developed domestically by Japanese companies and the probability of success for projects developed overseas. Since the other two reports were based on surveys of pharmaceutical companies in Europe and the U.S., the results for the probability of success overseas are cited from the Yagi et al. report, and a comparison of the figures from the three reports is shown in Figure 1.

The three reports showed differences in the probability of success in Phase I and Phase III, with the highest probability of 0.74 in Phase I reported by Yagi et al. Although it is a matter of speculation, it is possible that the higher probability of success for Phase I was calculated due to the following reasons: Japanese pharmaceutical companies are more likely to proceed to Phase I through a rigorous and sophisticated screening process at the preclinical stage, and programs that were developed first in Japan are sometimes followed by overseas Phase I programs. On the other hand, the success probability of Phase III may have been higher than that of Phase I. On the other hand, the success probability of Phase III is the lowest reported by Yagi et al. It can be inferred that, at least in the 2000s, the Phase III success probability was higher for foreign firms.

Calculation of success probability of drug discovery taking into account recent data

The data used in the previous studies were limited to the period up to the 2000s and do not necessarily reflect the probability of success in recent years. Although cases of antibody drugs obtaining approval have gradually emerged since the 2000s, it was not until the 2010s that the number of items grew ^{significantly1)}. The practical application of new modalities such as nucleic acid drugs, gene therapy, gene-cell therapy, and antibody-drug conjugates (ADCs) has been in the works since the 2010s. In light of this, the current probability of drug discovery success may be changing with the diversification of modalities. Against this background, we decided to examine the recent status of success probability of drug discovery using EvaluatePharma, a database of drug pipelines.

Two methodologies were used to calculate the probability of success in this study: The first methodology (referred to as Method 1) is the PRTS method, which is the standard method for calculating the probability of success in EvaluatePharma. The first methodology (Method 1) is a method to calculate the success probability from Phase I to Approved by calculating the Phase Progression Probability for each phase using the following formula and taking the product of the ^two6), which is similar to the three previous studies described above. The number of PASS or FAIL benchmarks used in the calculation of Phase Progression Probability is set for the periods shown in Table 1, with the intention of ensuring that the results reflect relatively recent conditions. For the later phases, the period covered is set longer to ensure a sufficient sample size. In this methodology, each program is counted separately, so that even if a single drug candidate compound (product) is being tested for multiple indications, each is counted separately.

The second methodology (Method 2) is a methodology to obtain the percentage of the number of approved products for all products listed in the database whose development success or failure is known (excluding products that are currently under development). In this methodology, products are counted on a product-by-product (≠program-by-program) basis, so that if a product is being developed for multiple indications, even one successful case is counted as APPROVED. The older products, those developed in the 1980s, are incorporated into the calculation of success probability, and the success probability derived from this method can be viewed as the success probability evaluated by averaging over the entire period of time. It should be noted that some of the products listed as discontinued in the database may have been discontinued before entering the clinical stage, in which case the probability of success may have been estimated lower.

Success probability by disease area (based on Method 1)

The probability of success from Phase I to Approved for each disease area is shown in Figure 2a, and the number of programs used to calculate the probability (number of benchmarks) is shown in Appendix 1 at the end of this report. The number of programs used to calculate the probability (number of benchmarks) is shown in Appendix 1 at the end of this report. Since there are many products that are being developed for multiple disease areas in a single product, we did not calculate the probability of success for each disease area in Method 2.

The probability of success in Phase I did not differ significantly by therapeutic area, and was approximately 70%, which is reasonable because the primary endpoints in Phase I are often safety and pharmacokinetic data. On the other hand, the probability of success in Phase II and Phase III differed by disease area, and the probability of success in Phase II and Phase III had a significant impact on the final probability of success (from clinical entry to approval). Cancer, Diabetes, Hepatic & biliary, Neurology, Psychiatry, and Respiratory had the lowest probability of success, while Blood, Hormone, and Infections had the highest probability of success. The probability of success for all disease areas (Total in the figure applies) was 0.67 (Phase I-Phase II), 0.36 (Phase II-Phase III), 0.55 (Phase III-Filed), 0.94 (Filed-Approved), and 0.13 (Phase I- Approved), respectively. Approved), and these values were generally similar to those reported in previous studies (Figure 1).

In the following sections, we will discuss the disease areas where the probability of success is low, Within the Hepatic & Biliary disease category, the majority of products in development are for non-alcoholic steatohepatitis (NASH) or non-alcoholic fatty liver disease (NAFLD), and the probability of success is only about 3% when focusing on these two diseases. The high level of difficulty in drug discovery for these two diseases was the reason for the low probability of success in this therapeutic area. The low probability of success for the other diseases, Cancer, Neurology, Psychiatry, and Respiratory, is due to the following reasons: the construction of non-clinical animal models reflecting actual pathological conditions has not yet been solved, the patient population is heterogeneous with respect to the disease concept, the disease biology is not well understood, and it is difficult to deliver drugs to the target tissues. The reasons for this are thought to include the fact that many diseases have a high degree of difficulty in drug discovery, such as difficulty in drug delivery to the target tissues.

For low-molecular-weight compounds and antibody drugs, for which there are relatively many examples of practical application, Figure 2b shows the probability of success by disease area limited to the relevant modality, and the number of programs used to calculate the probability is shown in Appendix 2 and Appendix 3. The number of programs used to calculate the probability of success is shown in Appendix 2 and Appendix 3. The analysis of the high probability of success of antibody drugs in the Skin domain showed that there were a large number of psoriasis drug programs in this domain, many of which were characterized by success in Phase III trials (data omitted), which increased the overall probability of success. Similarly, when analyzing the Musculoskeletal therapeutic area, there were a large number of programs for rheumatoid arthritis and psoriatic arthritis in antibody drugs, many of which succeeded in clinical trials at rates higher than those in all disease areas (data omitted), increasing the probability of success. The common feature of these drugs is that they are designed to treat diseases by suppressing immune responses, and there may be many diseases for which a drug discovery approach that targets molecules on the surface of cells with antibody drugs is more suitable. In the area of cardiovascular diseases, there are many cases of success in familial hypercholesterolemia and dyslipidemia, and in neurological diseases, there are many cases of success in hypercholesterolemia and dyslipidemia. In the area of Neurology, there were many cases of neuromyelitis optica and migraine, which drove the high probability of success (data omitted), but as in the above, it was difficult to find the intrinsic cause of the high probability of success.

Success probability by modality (based on Method 1)

The probability of success in each Phase by modality and the probability of success from Phase I to Approved (Approved) were calculated by Method 1 and are shown in Figures 3a and 3b and Table 2. The number of programs used to calculate the probabilities (number of benchmarks) is shown in Appendix 4 at the end of this report. The classification method of the modalities was based on the classification criteria in ^{EvaluatePharma8)}.

The probability of success in Phase I was generally around 70% for all modalities, but was slightly higher for gene therapy, at over 80%. A close examination of the benchmarks used to calculate the Phase I success probability for gene therapy revealed that many of the programs were for intractable or rare diseases, and in these programs, there were several cases where the First in Human trial was conducted as a Phase I/II trial for patients (data omitted). In these cases, the analysis was performed as if Phase I had been passed to Phase II (in other words, the success or failure of the Phase I/II trial was counted as a result of the Phase II trial), and this may have increased the apparent probability of success of Phase I.

The probability of success in Phase II was generally lower than the probability of success in other Phases in any modality, making it the most likely to drop in the course of research and development. This result was similar to the results in the previous study shown in Figure 1.

The probability of success in Phase III varied widely among modalities, and was generally 0.5 for low-molecular-weight compounds. The modalities with the highest probability of success were antibody drugs, gene-cell therapy, and Others. The probability of success for antibody drugs is thought to be increasing because they have reached maturity as a modality and drug discovery strategies have been established. As for the high probability of success of gene-cell therapy, although it is difficult to say definitively at this point due to the small sample size (see Appendix 4), it can be inferred that the Phase III study was planned with a high degree of certainty based on the remarkable efficacy that was confirmed in the clinical trials up to Phase II.

The probability of success from the start of Phase I to Approved was approximately 13% for all modalities. The probability of success for small molecule drugs, which accounted for the majority of all modalities, was about 10%, slightly lower than the overall probability. Antibody drugs had a relatively high success probability of about 21%, the highest probability according to the modality classification method in this study. The high success rate of diagnostics and plasma fractions (data omitted), among the various modalities (e.g., natural products, genome editing, etc.), increases the overall success rate of Others.

Success probability by modality (based on Method 2)

Figure 4 shows the results of calculating the success probability by modality using Method 2, which calculates the ratio of the number of approvals to the sum of the number of approvals and discontinuations for all products in the database. Therefore, only the probability of success over the entire development period can be calculated, not the probability of success for each phase.

The success probability for all modalities combined was approximately 12%. For small molecules, which account for the majority of all modalities, the success probability was about 12%. On the other hand, the success probabilities for antibody drugs, nucleic acid drugs, gene cell therapy, and gene cell therapy were all remarkably low, and tended to be lower for modalities that have recently begun to be put into practical use. As in the case of Method 1, the success probability of Others was as high as about 26%, driven by the high success probability of diagnostics and plasma fractionation products included here (data omitted).

Success probability by modality (comparison of Method 1 and Method 2)

Success probabilities by modality calculated for Method 1 and Method 2 are shown in Table 2.Method 1 calculated success probabilities by setting benchmarks for the time period as described above, and thus the success probabilities are considered to reflect relatively recent conditions. On the other hand, Method 2 covers all data in the database, so it can be read as a success probability for the entire period from the old (1980s) to the most recent period.

For small molecules, the success probability is about 10% for all analytical methods, and it is thought that the success probability is always stable at this level. Figure 1 shows the success probabilities in the previous studies. Since more than 90% of the drugs in the previous studies conducted in the 2000s were small molecules, the result that the success probability of small molecules was about 10% in this analysis is the same as the success probability from Phase I to Approved of the previous studies, and is a consistent result. This result was consistent with the results of the previous study, which showed that the probability of success for small molecules was about 10%.

The results of the evaluation of success probability for each drug discovery modality showed that there were several modalities with different success probabilities for Method 1 and Method 2, and that the success probability for Method 2 was significantly low for antibody drugs, nucleic acid drugs, gene therapy, and gene cell therapy. For antibody drugs, the overall probability of success (Method 2) is very low, but in recent years the probability of success (Method 1) has been higher than that for small molecule drugs. The reason for this may be that while there were many failures in the early days of the antibody drug modality, the technology gradually matured by utilizing the knowledge and know-how gained from these ^failures9), and has now reached the harvest stage as a drug discovery modality, leading to the high success rate in recent years. In the same vein, the use of the drug discovery modality has been growing in recent years. From a similar perspective, let us look at the status of so-called new modalities such as nucleic acid medicine, gene therapy, and gene-cell therapy, which have seen an increase in the number of practical applications since the late 2010s. While the overall probability of success (Method 2) is very low, the probability of success (Method 1) has been slightly higher in recent years, and is slightly lower than or equal to or higher than that of small molecule drugs. This indicates that these modalities are just coming out of their infancy, and we expect that they will be applied to a number of innovative drugs in the future.

The recent success rates (Method 1) for antibody and gene-cell therapies far exceed those of small molecule drugs, and these are particularly noteworthy as modalities with a high probability of success. ⁽ BIO), Informa Pharma Intelligence, and QLS Advisors in 2021, which reported on the probability of success by modality calculated using a method similar to Method 1 ^.10) The modality classification methods and analyses are described in the following sections. Although strict comparisons cannot be made due to differences in the modality classification methods and the time period covered by the data used in the analysis, the data here show high success probabilities for CAR-T, siRNA/RNAi, and monoclonal antibodies, and it can be read that the success probabilities are increasing for some new modalities.

Summary and Discussion

In order to understand and grasp the drug discovery risks faced by the pharmaceutical industry, we surveyed the current status of R&D success probability and discussed the differences in success probability by disease area and by drug discovery modality.

Although this paper analyzes the probability of success after Phase I, there are in fact a significant number of cases where R&D is abandoned in the initial process from the start of basic research to Phase I. Taking this into account, there is a substantial number of cases where R&D is abandoned in the initial process from the start of basic research to Phase I. Taking this into account, it can be inferred that the actual probability of success is far lower than the figures presented in this paper. However, it is clear that the medical contributions of new modalities and their impact on society are very large, as exemplified by the significant contributions made by new modalities such as mRNA vaccines during the COVID-19 pandemic. It is clear that the medical contributions of novel modalities and their impact on society are very large, as exemplified by the significant contributions made by novel modalities such as mRNA vaccines during the COVID-19 pandemic, and the importance and necessity of promoting R&D from a long-term perspective.

In order to make drug discovery a viable business and to create an environment in which innovative new drugs can be created one after another, even though the probability of success is extremely low, it is essential to be able to recover the large R&D investment in many projects that had to be discontinued halfway through their development by the profits generated by the successful development of some innovative new drugs. It is essential that the large amount of R&D investment in the many projects that had to be abandoned halfway through development be recovered through profits from some of the successful innovative new drugs, and it is important that the drug price system, social security system, and taxation system be supported to make this possible. It is also important to increase the probability of success, and measures to support the revitalization of basic research, especially in academia, will be necessary.

I would like to close this paper with the hope that the pharmaceutical industry, in close cooperation with government, academia, and other stakeholders, will consider ways to reduce and diversify risks from both academic and regulatory perspectives, thereby contributing to the creation of a better future.

Return to News List

TOP