Topics Econometric Analysis of International Diffusion of New Drugs: Focusing on Japanese and European Approval of New Drugs Approved in the U.S.

Printable PDF

Sadao Nagaoka, Director, Pharmaceutical Industry Policy Institute; Professor, Tokyo Keizai University
Junichi Nishimura, Visiting Fellow, Pharmaceutical Industry Policy Institute; Professor, Gakushuin University
Masao Yoshida, Senior Researcher, Pharmaceutical Industry Policy Institute

1. background and objectives

The international dissemination of new drugs is critical to the promotion of innovation because it increases the diversity of medicines available to patients and also encourages research and development of new drugs by increasing the benefits from new drug creation.

Although there have been previous studies on the international diffusion of new drugs, such as Cockburn, Lanjouw, and Schankerman (2016) ²⁾, there are still many issues that have not been empirically clarified as to the causes of this variation. In addition, the recent resurgence of the drug lag in Japan is a concern, and this is an important topic for analysis of the causes of the drug lag. In the previously published Policy Research Institute News (No. 63 and No. 66), we pointed out that the rate of unapproved new drugs in Japan has been expanding in the latter half of the ^2010s3) and thoroughly analyzed the potential causes of this trend. In this news item, Yoshida (2022) ⁴⁾ provides a new comparative analysis of Japanese and European approvals for items approved in the US.

This paper attempts to complement these analyses by providing a quantitative analysis of the importance of various factors on the international diffusion of new drugs. We analyze the characteristics of new drugs (US FDA breakthrough therapy designation or orphan drug designation) ⁵⁾, the characteristics of new drug developers (whether they are start-ups or not), and the importance of the term of intellectual property protection on the international diffusion of new drugs, and the differences between Japan and Europe in these factors. For this purpose, we conducted a quantitative analysis of new drugs (drugs containing new active ingredients) approved in the U.S. from 2010 to 2021 by connecting the approval information of Japan, the U.S., and European countries (until early 2022) with IQVIA's market launch data.

The structure of this paper is as follows. The next section describes the preceding studies, followed by an explanation of the data construction method used for the analysis in this paper, then the model of market entry and its estimation model, and then the estimation results. In the analysis, the COX proportional hazards model is used for new drugs that were launched in the U.S. first, and a linear probability model is used for new drugs that were launched in the U.S. in the same year as those approved in Japan or Europe.

2. previous studies

The most representative study of the global diffusion of new drugs is Cockburn, Lanjouw, and Schankerman (2016). They analyze the process by which a new drug that is first launched globally subsequently diffuses to different countries, and find that price regulation has a negative impact on whether a new drug is launched in a given country, that the strength of patent protection (e.g., substance patent and length of protection period) has a positive impact, and that even after a new drug is created, the diffusion of the drug to different countries is affected by clinical trial The study also shows that even after a new drug is discovered, incentives for clinical trials are important for its diffusion in each country (based on 672 drugs with new active ingredients launched in 76 countries during the period 1983-2002). Estimates suggest that strong price controls reduce the probability of a new drug launch in a country by 15%, resulting in a launch lag of about 25% (above the expected launch lag). In countries with substance patent regimes, the launch lag is 55% shorter. Furthermore, there is an interdependence between patent systems and price regulation, indicating that the impact of patent protection is offset under strong price ^regulation6).

Gaessler and Wagner (2022) ⁷⁾ analyzed the impact of exclusivity protection periods, focusing on data protection periods. The analysis covers new drug patents for which an opposition to the European Patent Office (EPO) is filed during the development of a new drug, and the impact on development progress and the probability of approval in Japan, the U.S., and Europe is evaluated using a linear probability model by measuring the extent to which the market exclusivity protection period for the new drug in question is reduced as a result of this opposition. The results of the estimation showed that, on average, the loss of one year of market exclusivity protection resulted in a 4.9 percentage point decrease in the probability of approval. In particular, the impact of objections in the early stages of clinical trials was found to be significant, with the probability of approval decreasing the larger the company involved in the development of the drug.

Imai and Narukawa (2022) ⁸⁾ analyzed new drug development delays in Japan for drugs containing new active ingredients that were approved in both the U.S. and Europe between 2010 and 2020, and for which approval in Japan was slower or not yet approved as of the end of 2020, Cox regression analysis was performed. The results indicate that development is delayed in Japan for new drugs for which no similar drugs exist, for new drugs for which the company that obtained approval in the US or Europe does not have a Japanese subsidiary, and for new drugs for which the originator company and the company that obtained approval in Japan do not coincide.

This paper quantitatively analyzes how the characteristics of a new drug (breakthrough or orphan designation), the characteristics of the company developing the new drug (whether it is a startup or not), and the term of intellectual property protection predict the approval rate of the new drug in Japan and Europe, including differences between Japan and Europe, as well as the international co-promotion of the new drug. The new contribution is the analysis of approvals in the same year, which are thought to be the result of international joint clinical trials, which are becoming increasingly important in the international dissemination of new drugs.

3. data construction

The analysis in this paper uses two main types of data. The first is data on actual approvals in Japan and Europe of new drugs approved in the U.S. This is the data used by Yoshida (2022) in this news item, which is constructed by the National Institute of Biomedical Innovation Policy. The sample analyzed is 481 new active ingredient-containing ^drugs9) approved by the U.S. (FDA) during 2010-2021. Approvals in Europe and Japan are known up to the most recent date, with some new drugs approved in 2022. The U.S. market is a highly innovative market for new ^drugs10) and accounts for about 40% of global new drug sales, and according to the Pharmaceutical and Industrial Policy Institute's (PIIPI) ¹¹⁾ fixed-point observation, U.S.-based companies account for a majority of the top-selling products worldwide, suggesting that the U.S. market is of great importance. For this reason, this paper will focus on new drugs approved in the US.

Another data source is IQVIA's Pricing Insights. This data source contains data on market launch dates, prices, intellectual property protection periods, etc. for drugs launched in each country. In this study, the approval data were connected to IQVIA by ingredient names (Molecules) and product names in the US. We visually checked for any differences in notation (i.e., distortion) in the names of ingredients and ^products12). Our Pricing Insights data covers data through early 2019, and the analysis using this data covers new drugs approved in the US through 2018.

4. model of market launch and analytical methods

4.1 Clinical Development Investment Decisions

For a new drug to be approved and launched in a given country, a company must make clinical development investments sufficient for review in that country, and for this to happen, the profits from launching the new drug in that country must exceed the clinical development ^{investments13)}.

Expected profit from the launch of a new drug > Expected cost of clinical development investment (1)

The higher the degree to which the innovativeness of a new drug is evaluated (incentives such as drug prices) and the size of the market (number of patients), the higher the expected profit from the launch of a new drug, and the longer the patent protection period, the longer the period during which profits can be earned from the launch of a new drug. These factors increase the probability that the above equation (1) will be satisfied, making it easier to bring a new drug to market.

Since market and clinical trial cost conditions change over time, even if equation (1) does not hold at a certain point in time (i.e., the probability of failure is high and the expected profit from launching the product is small), it may come to hold at a future point in time. Also, even if equation (1) holds, it may be advantageous for a company to delay conducting a clinical trial if the uncertainty is reduced by waiting, since the clinical trial cost is a sunk ^cost14).

On the other hand, the period of exclusive implementation under patent protection is fixed at the time the new drug is invented and patented, and if the implementation of clinical trials is delayed, the period of possible exclusive implementation under patent protection will be shortened. This effect encourages companies to conduct clinical trials earlier. However, if the clinical development investment is protected by a data protection period, that period is counted from the time of review and approval, so unlike protection by patent rights, there is no effect of encouraging early clinical trials.

Thus, clinical development investment is a risky upfront investment, and if the company holding the new drug faces a risk financing barrier, clinical trials will not be conducted even if equation (1) holds. In addition, startups have a weak sales base in each country and need to collaborate with outside firms in the area of sales promotion, which may delay the implementation of clinical trials and result in smaller expected profits.

In the following, we statistically analyze the extent to which these factors explain the probability of approval of a new drug in Japan and Europe, and the difference between Europe and Japan, focusing on the innovativeness of the new drug, the type of company developing the new drug, and the duration of intellectual property protection. For each of these, the following predictions can be made.

Regarding the difference between Japan and Europe for these effects, we expect that the market is larger in Europe than in Japan, and that the degree to which innovativeness is evaluated is greater in Germany, the UK, etc. than in Japan,

In the following, we will examine whether these expectations are valid.

4.2 Methodology

There are two types of new drugs approved in the U.S.: those approved in the U.S. prior to those approved in Japan or Europe, and those approved in the same year as Japan or ^Europe15). Companies first choose whether or not to obtain approval in the same year, which is analyzed below using a linear probability model. The Cox proportional hazard model is used to analyze the probability of approval in Japan and Europe in the case where the drug is not approved in the same year but is approved in the U.S. first. In both cases, the process by which a new drug approved in the U.S. is approved in Japan and the U.S. is captured as a stochastic event.

The Cox proportional hazards model assumes that the factors that affect the hazard rate (the probability of an approval event occurring per unit of time) act in the same proportion (proportionally) regardless of the elapsed time. While the condition that the factors act proportionally regardless of elapsed time is a rather strong assumption, this model is superior in that it can be estimated by consistently incorporating all available data on the occurrence of events, including censored data where no event has ^occurred16). The Cox proportional hazards model also has the advantage of making no specific distributional assumptions about the probability of an event occurring. Due to data limitations, elapsed time in this estimation is captured as years.

The probability h that a new drug i is approved in a given country j within an elapsed year t is denoted by _xij, where xij is an attribute (or vector if multiple factors are present) per drug unit or country that affects the probability of approval, with the strength of the estimated coefficient β. λ(t) is the time distribution of the approval probability for each elapsed time when _xij are all zero. Omitting the error term, we obtain

h(t│xij) = λ(t)exp( _βxij ) (2)

Taking the logarithm,

_lnh (t│xij)=lnλ(t)+βxij

and it can be seen that a one-unit change in attribute _xij changes the logarithm of the hazard rate by a factor equal to β. In reality, this is not the case in the U.S., where there are international clinical trials.

In reality, in many cases, drugs are approved in the same year in the U.S. and Japan, or in Europe, through international clinical trials. In the hazard model, the probability of an event occurring at elapsed time 0 is zero, making it difficult to include approvals in the same year as the U.S. approval in the analysis. To avoid this limitation, this paper uses the following linear regression model with the probability of approval in Japan or Europe in the same year as the U.S. approval (H( _xij )) as the explained variable (error term omitted).

EH(t│xij) = _δxij+α (3)

In this analysis, it is particularly important to control for the effect of censoring of approval data, so we include unapproved new drugs in Japan and Europe in the sample for analysis and control for the effect of censoring with the FDA approval year dummy.

In both models, _xij is an exogenous variable and is assumed to be independent of the error terms (random factors) in equations (2) and (3), as well as the error terms in the two equations. In practice, however, there may be factors that we cannot ascertain and that are not random. We mitigate the influence of such factors to some extent by introducing control variables such as the ATC classification of the drug, but they are still present, so the following analysis is not an inference of causal inference. It should be noted that each variable also captures the influence of correlated but missing factors. In addition, there are two sets of observations for each new drug, one for Japan and the other for Europe, and the standard error measures are clustered to control for the correlation in the error terms.

5. overview of data

Table 1 below shows the characteristics of the new drugs and the approval rates in Japan and Europe when the drugs are classified as US first, same year as the US, and Japan/EU first (all approved in the US). 280 new drugs were approved in the US first and either not approved or approved later than the US in Japan and Europe. 132 new drugs were approved in Europe or Japan the same year. The probability of a breakthrough designation for these new drugs is similar to the U.S. lead, at 27% and 24%, respectively, although the probability is slightly higher in the case of U.S. approval. The probability of orphan designation is also almost equal for both. In each case, the approval rate in Japan is much lower than that in Europe. In the case of the U.S. lead, 34% in Japan and 52% in Europe have been approved as of early 2022. In addition, when either the U.S. or Japan/Europe approvals are in the same year, almost all (96%) are approved in Europe, but only 67% in Japan. In other words, more often than not, a drug is approved in Europe the same year as it is approved in the U.S., and even when it is not, it is approved earlier in Europe than in Japan. In the case of Europe, the number of new drugs approved in the same year (132 x 0.96 = 127) and the number of new drugs approved later than in the U.S. (280 x 0.52 = 146) are quite close, suggesting the growing importance of approval in the same year through international joint trials. In the case of same-year approvals, the probability that the clinical trial company is an emerging company is much lower: 45% in the case of U.S. first approval and 20% in the case of same-year approval.

The last column in Table 1 is for cases where either Japan or Europe is ahead, which exists in 14% of all cases. The percentage of breakthrough designations is less than half that of the U.S. first or the same year as the U.S..

Cox Proportional Hazards Model Analysis of Prior US Approval of New Drugs

Table 2 shows the results of estimating equation (2) for new drugs approved earlier in the U.S. than in Japan and Europe between 2010 and 2021 (including new drugs not yet approved in Japan and Europe), and for the review status in Japan and Europe up to the latest date (including early 2022). Table 2 also shows the list of factors (explanatory variables) that affect the probabilities. In all models, a U.S. approval year dummy (cohort year dummy) is introduced to control for environmental changes over time common to Japan and Europe, such as changes in clinical trial methods. The estimated coefficients for each variable are not hazard ratios, but rather how much the logarithm of the hazard rate (hazard rate, the probability that approval will occur in each year) will change (approximately what percentage the hazard rate will change ⁾¹⁷⁾. A higher hazard rate results in a shorter lag to approval and a higher probability of approval in a given period.

Model 1 is the simplest estimation model and evaluates the difference in the logarithm of the hazard rate of approval in Japan compared to Europe (the estimated value is negative, so the degree of decrease) separately for the sample of U.S. approval in the first half of the 2010s (2010-2014) and the sample of approval in the second half (2015-2021). According to the estimation results, the probability of approval in Japan is statistically significantly lower than in Europe for both new drugs approved in the first half of the 2010s and new drugs approved in the second half of the 2010s: the hazard rate is about 48% lower for new drugs approved in the first half of the 2010s (the estimated value of the coefficient corresponding to β in equation (2) is -0.477) and for new drugs approved in the second half of the new drugs is also about 76% lower. The length of the period over which the hazard rates are measured differs between the first and second halves, and the magnitude cannot be directly compared; as will be confirmed in Section 7, this probably reflects the fact that, compared to the approval process in Japan, approvals in Europe are concentrated at an earlier stage after approval in the U.S. ¹⁸⁾.

The fact that the probability of approval in Japan is significantly lower than that in Europe does not change when new drug characteristics and characteristics of new drug development companies are introduced in Models 2 and 3 (Model 2), or when ATC large category dummies are controlled (Model 3). ¹⁹⁾ The possibility that the magnitude of the effect of factors such as innovativeness on Japan and Europe may differ is analyzed in the estimation of effects by Japan and Europe below.

Next, we analyze the effect of whether a new drug is designated as a breakthrough or orphan drug in the US. According to Model 2, breakthrough designation is accompanied by an increase of about 56% in the probability of approval in Japan and Europe (the estimated value of the coefficient corresponding to β in Equation (2) is 0.556), but the estimation with the addition of the ATC large category dummy variable (Model 3) loses statistical significance, indicating a large effect of disease area (see (The overall approval rate is higher for diseases with more breakthrough designations). However, when the estimation is restricted to the fields of L (anti-cancer agents and immunomodulators) and J (systemic anti-infectives) (approximately half of all new drugs), which are frequently designated as breakthroughs (Model 4), the breakthrough designation is still significant (the coefficient is 33%, close to 5% significance), suggesting that highly innovative new drugs tend to spread easily in Japan and Europe. This result further suggests that highly innovative new drugs are more likely to spread in Japan and Europe. As shown in Models 5 and 6, the coefficient for the breakthrough designation is also significant and at the 40% level in the sample limited to drugs launched in the U.S. by early 2019, even with the addition of ATC classification and the degree of intellectual property protection (Model 6). Overall, although not strong evidence, consistent with the prediction (1) in Section 4.1, highly innovative new drugs are more likely to be launched earlier in Japan and Europe.

On the other hand, there is no significant relationship between orphan designation and approval rates in Japan and Europe. One of the requirements for orphan designation is that the number of patients must be small, which is a negative factor for clinical development investment. On the other hand, new drugs are innovative in that they satisfy unmet demand for treatments for which there are no existing treatments, and there are incentives in Japan and Europe specifically for orphan drugs, which may be offsetting the negative effect of orphan designation. However, in the L and J areas, the estimated coefficient is -30%, which is statistically significantly lower than the probability of approval.

Next, when we look at the impact when the company developing the clinical trial is a ^startup20), it is robustly large and negative in all models: according to Model 3, which controls for the ATC field, the probability of approval is 37% lower, while the estimates specific to the L and J fields (Model 4) show a 56% lower probability. This result suggests the importance of financial constraints faced by startups and is consistent with prediction (2) in section 4.1 above; large negative coefficients are also observed when NPOs and universities are sponsors, but they are not statistically significant (possibly due to the small sample size and weak power).

Estimates of the effect of the remaining IP protection period (log) in the U.S. at the time of U.S. approval are shown in Model 5 and Model 6, which additionally controls for the ATC large category, and are significantly positive. The logarithm is used to reflect the fact that the decision to invest in clinical development is likely to be a critical issue, especially when the IP residual protection period is short. Information on IP residual protection duration is only available for approved drugs launched in the U.S. by early 2019, and these two models have a small number of observations (about half). Model (5) suggests that a 10% increase in the length of the protection period increases the annual probability of approval by about 6%; controlling for ATC classification is also significant (model (6)), with an effect of about ^5%21).

The above estimates of the remaining IP protection period may be an underestimate. The estimates use the remaining term of IP protection at the time of U.S. approval as a proxy variable for the remaining term available to companies in Japan and Europe in the event of successful clinical trials, which is only valid for the protection of patent rights. The starting point of protection by patent right is the same in Japan, the U.S., and Europe. For example, if the residual term is longer at the time of approval in the U.S., the residual term of protection by patent right will also be longer in Japan and ^Europe22). On the other hand, if the period of data protection (Japan's reexamination system) as a system to protect the profitability of clinical trial investment determines the period of exclusive license, protection by that system is counted from the time of approval in each country, so there is no international linkage of inter-market variation in protection period as in the case of patent right protection (applicable to each type of product in each country) (This is due to the system of common term for each type of drug product in each country). The data we use does not allow us to distinguish between the two because we do not know to what extent data protection is predominant. However, an analysis using data on new drugs actually launched in Japan, the U.S., and Europe shows that the variation in the residual IP protection period in the U.S. significantly explains the variation in the residual IP protection period in Japan and ^Germany23).

Table 3 below shows the extent to which approval rates vary depending on whether a new drug has a breakthrough designation, whether the developing company is a startup, and whether the remaining IP protection period in the U.S. at the time of FDA approval is long (>9 years) in Europe and Japan, respectively. The approval rate is high in both Japan and Europe when the new drug is designated as a breakthrough, and low when the company developing the drug is a startup. It should be noted, however, that the difference is much smaller in Europe than in Japan (12% vs. 26%). The approval rate for new drugs with short residual intellectual property protection is low in both Japan and Europe.

Table 3 shows the possibility that the effect of innovativeness and other characteristics of a new drug on its approval rate may differ between Japan and Europe, as discussed in Section 4.1, Projection 4, and the statistical significance of this effect is examined in Table 4 below. The estimation model is an extension of Table 2, extended to allow for the different effects of the characteristics of the three new drugs in Table 3 in Japan and Europe. As in Table 2, data are pooled for approval events in Japan and Europe, and the time distribution of the hazard underlying the COX proportional hazards model is assumed to be the same in Japan and Europe (this allows for comparison between Japan and Europe).

Model 7 allows for differences between Japan and the U.S. only in the coefficient of breakthrough designation, Model 8 allows for differences between Japan and the U.S. in the impact of startups, and Model 9 allows for differences in the remaining intellectual property protection period. Model (7) is based on the case where a new drug designated as a breakthrough is developed by an emerging company or an NPO and is approved in Europe. According to model (7), the probability of approval of a new drug with breakthrough designation in Japan is 43% lower than in Europe. However, in the absence of breakthrough designation, the approval rate of new drugs in Japan is even lower than that in Europe (the difference in the coefficient between no breakthrough designation x Europe and no breakthrough designation x Japan is 0.651). Therefore, the prediction (4) in 4.1, that the positive impact of innovativeness is greater in Europe, is not valid.

Model (8) shows that being a startup or a non-profit organization has a very large negative impact in Japan compared to Europe. The impact on the probability of approval is negative 82%. At the same time, the negative impact of being a breakthrough designation on Japan's approval probability is much smaller and statistically insignificant. Furthermore, according to model (9), only Europe has a positive and significant impact on the remaining IP protection period, while no significant impact of a longer period on the approval probability is observed in Japan (however, the difference is not significant). It remains established that new firms or NPOs are largely negative and significant only in Japan.

Overall, the prediction (4) in Section 4.1 regarding the difference between Japan and Europe holds for the impact of start-ups and the effect of intellectual property protection. However, the former negative effect is much smaller in Europe, and the latter is significantly positive only in Europe, but the difference between Europe and Japan is not statistically significant. Therefore, the most important difference in the probability of approval between Japan and Europe may be due to the difficulty of investing in clinical development in Japan when the startup is a development company.

Dynamics of Approval of Prior US Approved New Drugs in Japan and Europe 8.

The dynamics of approval differ significantly between Japan and Europe. Figure 1 shows how the probability of approval in Japan and Europe varies with the elapsed year starting from the year of approval in the U.S. (Kaplan-Meier failure estimates) for the new drugs analyzed above that are approved in the U.S. first. Panel A shows new drugs approved in the U.S. in the early 2010s (2010-2014), and Panel B shows new drugs approved in the late 2010s (2015-2021). According to this, approval in Europe is high immediately after prior approval in the U.S., exceeding 50% of the U.S. level in the second year in the first half of the decade and in the third year in the second half of the decade, and flattening out to over 75% in the fifth year in the first half of the decade. In contrast, in Japan, approval is slow to start, increasing at an almost constant rate each year, and it is not until the 10th year in the first half of the 2010s that it exceeds the 50% level in the U.S. Figure 1 also shows confidence intervals, but even in our relatively small sample, the probability of approval in Europe is significantly higher up to the fifth year. Thus, it should be noted that not only the level of approval but also the pattern of its variation by elapsed time differs between Europe and Japan, and the difference in the average approval probability between Europe and Japan varies depending on the length of the elapsed time (the assumption that the hazard rates are proportionally different between Europe and Japan is not valid).

Analysis of the probability of approval in the same year in the U.S. and Japan/EU 9.

Table 5 below tabulates the frequency of new drugs approved by Japan and Europe in the same year as the U.S., respectively. The population is the number of new drugs approved by the U.S. in the same year as the U.S. and Japan or Europe (there were no cases in the same year for Japan, the U.S., and Europe). The population is the sum of cases approved by the U.S. and the U.S. in the same year (there were no cases approved by the U.S., Japan, and Europe in the same year).

In Japan, the probability of approval in the same year is low at about 8.5% in both the first half and the second half of the 2010s. In Europe, the rate was 34% in the first half of the decade and 26% in the second half, which is significantly higher than in Japan, although the rate declined in the second half of the decade compared to the first half. Breakthrough and orphan designations in the U.S. for new drugs approved in the same year increased in the second ^half24), and at the same time, the percentage of startups as developers also increased from 16% to 20%. In the following, we statistically analyze the relationship between these characteristics and the probability of approval in the same year. The estimation model is a linear probability model with equation (3).

The estimation results are shown in Table 6. Models 1 through 4 use the same explanatory variables as in Table 2 to analyze the factors that contribute to the probability of approval in the same year in Japan or Europe as in the U.S. Model 1 is the simplest estimation model and is the basic model controlling only for the FDA approval year dummy. Model 2 adds dummy variables controlling for breakthrough designation, type of company developing the new drug, etc. Model 3 further controls for 15 ATC category dummies. Model 4 is a sample in which estimates are restricted to the L (antineoplastics and immunomodulators) and J (systemic anti-infectives) sectors. Models 5 and 6, on the other hand, estimate Japan and Europe separately.

The estimation results for Model 1 in Table 6 confirm the descriptive statistics in Table 5, with Japan's probability of approval in the same year as the U.S. being 25 percentage points lower in the first half and 17 percentage points lower in the second half compared to Europe's probability of approval in the same year as the U.S. Thus, the difference between Japan and Europe is declining, but as can be seen in Table 5, this is due to a decrease in Europe's same-year approval rate, not to an increase in Japan's same-year approval. Note that unlike the COX proportional hazards model, the earlier and later periods can be directly compared.

Next, according to Model 2, whether a new drug has breakthrough designation in the U.S. or orphan designation has no statistically significant relationship, on average, with the probability of approval in the same year as the U.S. in both Japan and Europe. As Models 5 and 6 show, the same is true for results from separate estimation models for Japan and Europe: in Model 4, which restricts the estimation to the L (antineoplastics and immunomodulators) and J (systemic anti-infectives) sectors, the breakthrough designation has a negative and significant coefficient. Although the results are omitted, the remaining IP protection period is also insignificant, suggesting that institutional barriers may be more important than incentives for approval in the same year. For example, a certain percentage of novel breakthrough therapies are subject to the U.S. Accelerated Approval system (of the 481 FDA-approved new drugs in 2010-2021, 109 have breakthrough designation, 41 of which are expedited approved in the same year as Japan and Europe). The ATC classification is a model that controls for therapeutic areas.

According to Model 3, which controls for territory by ATC classification, the likelihood of approval in the same year is significantly lower by about 14 percentage points when the company developing the clinical trial is an emerging company. This result is consistent with the estimated probability of approval limited to prior approval in the U.S. in Table 2, suggesting that financial constraints on startups also make it difficult to conduct international clinical trials. As Models 5 and 6 show, unlike the results for the U.S. prior, the negative coefficient for start-ups is larger for European approvals (19 percentage points lower in Europe and 9.6 percentage points lower in Japan). The approval in the same year is inherently lower in Japan, and it is possible that the binding constraint of the developer being an emerging company is less significant for the approval in the same year, but further analysis is still needed.

9. conclusions

This paper quantitatively analyzes how the diffusion of new drugs approved first in the U.S. into Japan and Europe is affected by the innovativeness of the new drug, the characteristics of the company developing the new drug, and the duration of intellectual property protection, as well as the significance and factors behind the differences between Japan and Europe. Furthermore, a similar analysis was conducted for approvals in the same year as the U.S. In this study, the year of dissemination is identified by the year of regulatory approval. The main findings are as follows.

First, there are large differences in the probability of approval in Japan and Europe for a new drug that was previously approved in the U.S. Controlling for these differences, the probability of approval in Japan and Europe tends to increase when the innovative nature of the new drug (as evaluated by the FDA's breakthrough designation) is high. Although the probability of approval in Japan is significantly lower than in Europe, there is no significant difference between highly innovative and non-innovative new drugs.

Second, due to international joint clinical trials, etc., the approval of new drugs in the same year as that in the U.S. is high in Europe, but the level in Japan is low, at about one-third of that in Europe in the late 2010s. When a new drug is highly innovative, the probability of its approval in the same year in Japan and Europe does not tend to be high, and may even decrease. This suggests that institutional factors may be more important than incentives for approval in the same year. For example, highly innovative breakthrough therapies are often subject to the Accelerated Approval system in the U.S., which may make it difficult to obtain approval in the same year as in Japan and Europe.

Third, if the company developing the new drug is an emerging company, the probability of approval of the new drug in Japan and Europe is greatly reduced. In the case of prior approval in the U.S., the average probability of approval in Japan and Europe is about 37% lower, and the probability of approval in the same year is also about 14 percentage points lower. The impact is significantly greater for Japanese approvals, and is also a very important factor in the disparity in approval rates between Japan and Europe.

Fourth, the probability of approval increases significantly in Europe when the remaining IP protection period at the time of U.S. approval is long. The probability of approval increases by about 5% when the residual protection period is one year longer. On the other hand, the effect of IP protection period is not significant in Japan. This may be due to limitations in the environment and systems that complement IP protection, such as the size of the market, the level of drug prices, and pharmaceutical regulations. In both Japan and Europe, when the data protection period is important, the estimated results may be an underestimate because the remaining IP protection period at the time of U.S. approval does not predict the exclusive implementation period in each country.

Thus, the approval rate of new drugs in Japan is significantly lower than in Europe, both in terms of diffusion of new drugs from the country of prior approval (the U.S.) and in international approval in the same year. In addition to the basic clinical trial environment in Japan (e.g., low case-accumulation and high costs of CROs and SMOs), this suggests that there are areas for improvement in the incentives for clinical development investment and the regulation of clinical trials, including international clinical trials, in Japan. One of the key issues is to strengthen clinical trial capabilities in Japan through inter-company collaboration when the developing company is an emerging company.

Even if a new drug is approved by the regulatory authorities in each country, if it is not launched in the market, it will not reach patients and the company will not be able to recoup its investment. In this study, we attempted to construct such data, but the coverage of the database we used was not always ^complete25) and international comparability was not clear, so this is a topic for future research.

Return to News List

TOP