PMS have been mandatory in Japan for several decades and are a required part of the drug development process, postapproval. However, preapproval global drug development programs have evolved to be far more comprehensive than in the past and global pharmacovigilance of drug safety has also evolved significantly. Therefore, it is reasonable to ask if mandatory PMS still contribute to the understanding of human drug benefit/risk in Japan.

The strength of this analysis is a large dataset extending over two decades with over 98,000 patients’ data covering a wide range of therapy areas. The 79 Japan PMS identified were conducted for 43 different products. For all products, the reexamination outcome was category 1, “usefulness is confirmed”, which means that based upon the PMS data, together with other data available during the reexamination period (e.g., pharmacovigilance), the products could continue to be marketed and that no critical updates to the product labelling were necessary.

Despite mandatory safety reporting in postmarketing observational studies, the ADR rate was generally lower than in the phase 3 clinical trials, which may be due to the less formal nature of the PMS studies and the relative lack of source data verification compared to that required for the phase 3 interventional studies. One or more new (unlabeled) ADRs were found in most of the postmarketing studies (74%) but not in all studies. By contrast, unlabeled ADRs were found in the postmarketing pharmacovigilance databases for all 30 products that were the subject of the available 39 RERs. PMDA’s discussion of unlabeled ADRs, in the RERs, focused only on those ADRs found in the postmarketing pharmacovigilance database for each product, rather than on unlabeled ADRs that might have been found in an individual study. This may reflect that the RERs are very concise documents in which the PMDA places more emphasis on global pharmacovigilance data with its much larger patient-years exposure, compared to the smaller postmarketing studies. However, none of the unlabeled ADRs in this analysis was considered to have had a meaningful impact on product benefit/risk since no additional risk minimization measures were required for any of the products.

GDUI/SI studies are designed to assess all adverse events in the Japanese population prescribed the drug per the product label, under conditions of routine medical care. The decision to prescribe is independent of the study and those patients studied are a subset of all Japanese patients receiving a prescription.

From the primary analysis subset of 37 PMS with a CSR and a RER and regardless of study type (GDUI or SI), there was no evidence that the Japanese population have meaningful differences in safety or efficacy to the global population studied in phase 3. This indicates that the J-NDA process consistently assures that benefit/risk is acceptable at the time of approval and that the PMS do not yield important additional insights as, generally, efficacy and safety in Japanese patients are similar to the global population. Consequently, despite the current statutory requirement for their conduct and despite updated guidance issued from the PDMA in 2018 focused on improving their scientific focus, efficiency, and methodology [8], the necessity of regulatory-required PMS in Japan should be reconsidered.

Instead, routine pharmacovigilance, which is already conducted for all drug products, could be employed as an alternative approach to the mandatory GDUI or SI. A theoretical advantage of a GDUI or SI study is that AE reporting is obligatory (versus spontaneous reporting in routine pharmacovigilance); however, our finding that routine pharmacovigilance identified greater aggregate numbers of previously unknown ADRs, compared with the GDUI or SI studies, suggests that this potential disadvantage may be outweighed by the far larger numbers of patients subject to routine pharmacovigilance. Kanmuri and Narukawa [9] also concluded that routine pharmacovigilance was more effective; the authors conducted a review of PMS performed for 150 products in Japan and found that safety-related changes to the product label were predominantly driven by routine pharmacovigilance and were only infrequently prompted by a PMS finding.

When there exists a particular scientific uncertainty about the product, on the part of the health authority, a study designed to answer a specific scientific question is an appropriate requirement. In such circumstances, novel methodologies such as utilizing electronic health records data [e.g., the Medical Information Database Network (MID-NET [10]) or Medical Data Vision (MDV [11])] or the global pharmacovigilance database may be a more efficient and robust approach for answering those questions rather than the usual uncontrolled prospective observational methodology of GDUI or SI with limited statistical power. In addition, the question should be asked whether this needs to be addressed in Japan or could potentially be addressed by a study in another region.

This study was a retrospective analysis for which there are several limitations. Importantly, it represents the experience of a single sponsor (Pfizer) for a specified time period (from 2000 to 2020), which may or may not be generalizable. The time limit was placed since PMS design and execution have been consistent over this period. All the PMS identified were either general investigations or special investigations. There were no data from interventional comparative investigations, other postmarketing clinical trials, or postmarketing database studies. There were also significant amounts of missing data, since the CSR and RER were not available for all 79 PMS and hence the primary analysis subset included only 37 PMS (47%), although some limited assessment of the nonincluded studies and RERs was performed. Data were missing because not all RER are available on the PMDA website and because, over time, some products were divested from Pfizer to other companies; CSRs for divested products and some legacy products are not available within Pfizer’s global document management system.

However, the impact of missing data is likely not significant. Given the consistency of findings from the primary analysis subset, the RER-only subset, and the CSR-only subset and since all products had a category 1 outcome, the PMS indicated that, together with other data available during the reexamination period, they did not yield important new insights impacting benefit/risk balance. In this regard, it must be acknowledged that product labels could have been updated with new safety information prior to completion of the reexamination period, e.g., based on routine global pharmacovigilance data, annual safety reports, periodic safety update reports, or, in some cases, interventional clinical trials completed outside of Japan. Therefore, although all products were found to be category 1 by the Ministry of Health, Labour and Welfare, it is possible their reexamination outcome reflected product labels that had already been updated during the reexamination period, based on activities unrelated to the regulatory-required postmarketing studies.