Because of the limitations of trials before marketing, such as small sample size and short duration of medication, it is difficult to find some delayed or rare adverse reactions. But to some extent the lack of clinical drug safety information can be compensated by mining the FEARS database. This study analyzed the ADE reports from FEARS and captured 50 high-frequency signals by ROR mining method. The results showed that some adverse reactions involved in the drug label were essentially included in the signal list ,such as diarrhea, nausea, fatigue, vomiting, loss of appetite, dehydration, etc. and most of them were also ranked high, which improved the reliability of this study.
15 new signals of suspected ADRs not listed in abemaciclib 's label were identified in this study. We searched PubMed and clinicalTrials database for these new adverse reaction signals[20–27] and found that “weight decreased”, “blood creatinine increased”, “hypokalaemia”, “constipation”, “stomatitis”, “pleural effusion”, “pneumothorax”, “hot flush”, “lacrimation increased” were mentioned in the safety profiles of some trials, like MONARCH1, MONARCH2, MONARCH3 and MONARCH E, but in three trials, MONARCH2, MONARCH3 and MONARCH E, the incidences of “pleural effusion” and “hot flush” in the control group were higher than in the experimental group. In addition,A phase 1b trial of necitumumab in combination with abemaciclib for stage IV non-small cell lung cancer mentioned 7% hypokalemia[28]. “Taste disorder” was mentioned in MONARCH3. In our study, there are two new signals under the SOC “renal and urinary disorders”, i.e. “renal disorders” and “renal impairment”. “Renal impairment” was mentioned in MONARCH3, but “renal disorders” were not mentioned in any clinical trial. In trial MONARCH1 it is mentioned that one patient withdrew because of elevated serum creatinine and 0.9% of patients in VERZENIO plus aromatase inhibitors group permanently discontinued due to renal impairment. MONARCH1, MONARCH2, MONARCH3 and MONARCH E reported renal and urinary disorders with different incidences.
No literature was retrieved regarding the PT of “eating disorder”. Eating disorder is one of psychiatric adverse reactions, which might be related to abemaciclib's better lipid solubility[29]. That means the drug molecules can easily enter the central nervous system and cause related effects.
It has been documented[18] that diarrhea, neutropenia, increased ALT and/or AST were more common adverse events of abemaciclib in Japanese. In our study,the adverse event cases reported from Japan with high proportions were diarrhea (3.86%), neutropenia (37.78%), neutropenia (5.79%), increased ALT(18.52%) and increased AST (18.52%). Giving that the reporting source region field in FEARS database does not accurately reflect the race of patients, more accurate race data are needed for correlative argumentation.
Analysis of spontaneous reporting systems is a useful approach to identify possible signals, and the FAERS database is considered one of the largest sources of data. However, our study has some limitations[30]. First, there are missing information and duplicate entries in the FAERS database. We manually removed duplicates and ignored missing information data. Second, a slight increase in ROR values does not imply that there is a risk of AE in clinical practice, i.e., it may be relevant. FDA requires that adverse events be reported without the need to demonstrate a causal relationship between adverse events and the drugs, and reports do not always contain sufficient details for proper assessment of the event. Therefore, these values only provide a safety signal and do not provide an actual risk. Third, the disease may have some complications over time. These suspected complications of disease progression deserve further evaluation, particularly in the context of the relatively small number of cases involved. Finally, FAERS did not perform a separate statistical analysis of patients' underlying disease and could not rule out an impact of underlying disease on the risk of ADEs. However, the FAERS database has a large amount of data and covers a wide population, and post-marketing pharmacovigilance signal mining can partially compensate for the above shortcomings.