Abstract
Background/Aim: Lenalidomide (LND) is an oral anticancer drug used to treat various malignant hematologic diseases, including multiple myeloma. The most common adverse events with LND are myelosuppression, thrombosis and pneumonia. Myelosuppression is reversible, and thrombosis can be treated with prophylactic administration of antithrombotic drugs. Pneumonia is less common and its incidence profile in clinical studies is unclear. This study aimed to evaluate the incidence and onset timing of LND-related lung toxicity and outcome details using the Japanese Adverse Drug Event Report (JADER) database. Patients and Methods: Adverse events with LND reported between April 2004 and March 2021 were selected. Data on lung adverse drug reactions (ADRs) were analyzed, and safety signals were estimated using reported odds ratios (RORs) and 95% confidence intervals (CIs). We also estimated the timing of onset of lung toxic signs. Results: A total of 10,929 ADRs were attributed to LND. Of these, 908 were lung toxicities. The most frequently reported ADRs with significantly high RORs were pneumonia (559 cases, ROR=3.89, 95% CI=3.57-4.24) and bacterial pneumonia (38 cases, ROR=2.02, 95% CI=1.46-2.78). Median onset of pneumonia and bacterial pneumonia were 84 and 74 days, respectively. Prognoses of patients who received LND and had pneumonia and bacterial pneumonia were poor, with 10-20% non-recovery and deaths. Conclusion: The findings indicate that pneumonia and bacterial pneumonia, among all LND-related lung toxicities, may be associated with immunosuppression, suggesting the importance of monitoring respiratory symptoms within the first 3 months of treatment.
Lenalidomide (LND) is an oral anticancer drug for the treatment of multiple myeloma (MM), myelodysplastic syndrome with deletion of the long arm of chromosome 5 (5q-MDS), relapsed or refractory adult T-cell leukemia lymphoma (ATLL), and relapsed or refractory follicular lymphomas (FL). The pharmacological effects of LND, an immunomodulator, on MM are related to its ability to regulate cytokine production, and inhibit angiogenesis and hematopoietic tumor cell growth in the presence of dexamethasone (1). Pharmacological effects of LND on 5q-MDS are associated with growth inhibition and hematopoietic effects on tumor cells. The immunomodulatory effects of LND enhance antibody-dependent cellular cytotoxicity of rituximab, an anti-CD20 monoclonal antibody, on FL.
The clinical efficacy of LND has been demonstrated in some pivotal clinical studies. For instance, LND with dexamethasone can significantly prolong progression-free survival (PFS) than placebo in patients with relapsed or refractory MM (11.1 vs. 4.7 months) (2).
Hematologic adverse events (AEs) with grade ≥3 for this combination therapy have been neutropenia, anemia and thrombocytopenia (41.2%, 13%, and 14.7%, respectively). Non-hematologic AEs included infections (67.8%). These were infectious pneumonia and upper respiratory tract infections. Venous thromboembolism (VTE) was also more common in the LND group (14.7%). In cases with 5q-MDS, a significantly higher number of patients achieved red blood cell transfusion-free status with LND-alone than those with placebo (56.1% vs. 5.6%). Myelosuppression and VTE were the most common AEs (3). In a phase II study of LND in patients with relapsed ATLL, the overall response rate was 42%, and the PFS and overall survival were 3.8 and 20.3 months, respectively. Grade ≥3 AEs were neutropenia (65%), lymphopenia (38%) and thrombocytopenia (23%), all manageable and reversible. Dyspnea and pulmonary edema occurred in 8% and 4% of patients, respectively (4). A phase III AUGMENT study of LND plus rituximab in patients with relapsed or refractory FL showed a significantly longer PFS than those with rituximab-alone (39.4 vs. 14.1 months) (5). Pneumonia, pulmonary embolism, febrile neutropenia and VTE were the most common serious AEs with LND combination (2-3%).
LND is used as the standard treatment for MM, MDS, ATLL and FL. Myelosuppression, including neutropenia and thrombocytopenia, is frequent, but reversible; but there are differences depending on concomitant medications. However, hepatitis B virus reactivation due to immunosuppression, pneumonia, VTE and pulmonary embolism are serious AEs associated with LND. Neutropenia and VTE are serious AEs, but <30 cases of lung toxicity have been reported in randomized controlled studies on relatively large number of patients with FL (n=178) (5) and MM (n=177) (2). It is necessary to enroll at least 3,000 cases to detect at least one adverse drug reaction with a 0.1% frequency of occurrence and at least 95% probability. Previous clinical studies have failed to specify the details of symptoms, timing of onset, and outcomes of LND-induced lung toxicity due to the small number of patients and a low frequency of occurrence. Japanese are prone to lung toxicities with anti-EGFR multi-kinase inhibitors and other drugs (6, 7). Nevertheless, lung toxicities in 36 Japanese patients in the AUGMENT study have not been reported (8). Moreover, the objective and detailed profile of LND-induced lung toxicities in Japanese patients remains to be elucidated. Lung toxicities, such as interstitial pneumonia, have poor prognosis (9). Therefore, lung toxicities should be carefully monitored for the safe use of LND. This requires validation of detailed profiles of LND-induced lung toxicities in a larger patient population.
Data on adverse drug reactions (ADRs) obtained in clinical studies prior to drug approval are from a relatively small patient population with clearly defined backgrounds. In the real-world, the use of an approved drug in a large number of patients with diverse backgrounds reveals previously unknown profiles of ADRs. Pharmacovigilance studies aimed at monitoring drug safety are important for the optimal use of all drugs (10-12). Pharmacovigilance is performed using a spontaneous reporting system (SRS) that reflects the actual clinical practice (13). The Pharmaceuticals and Medical Devices Agency (PMDA), Japan, has established the Japanese Adverse Drug Event Report (JADER) database as an SRS. Analysis of spontaneous adverse event reporting databases is an effective strategy to hypothesize the relationships between unknown or potential ADRs and drugs, and may contribute to the prompt provision of information to healthcare professionals. This study aimed to retrospectively analyze the JADER database to determine the type and temporal profiles of concomitant toxicities in patients treated with LND.
Patients and Methods
Data source. We used data from the public releases of JADER (14-16). This database is available for free; data can be downloaded from the PMDA website (http://www.pmda.go.jp) and include ADR cases. We analyzed ADR reports recorded between April 2004 and March 2021. Data structure of JADER comprises four datasets: patient demographic information (DEMO), drug information (DRUG), ADRs (REAC) and medical history (HIST). ADRs in the JADER were coded according to the terminology recommended by the Medical Dictionary for Regulatory Activities/Japanese version 24.1 (www.pmrj.jp/jmo/php/indexj.php).
First, we removed duplicates from the DRUG and REAC tables (17, 18). Further, we used the identification number for each ADR case to merge corresponding case data from the DRUG, REAC and DEMO tables. The medication contributions to the ADRs were classified as “suspected medicine”, “concomitant medicine”, and “interaction.” We only extracted cases that were classified as “suspected medicines.”
To investigate the association between LND and lung toxicity, we analyzed JADER, that contains spontaneous ADR reports submitted to the PMDA. Each lung-related ADR, coded according to the terminology recommended by the Medical Dictionary for Regulatory Activities, was referred to as lung toxicity in this study.
Statistical analyses. Data on lung toxicity with more than five reported cases were extracted, and the safety signals of ADRs were estimated using the reporting odds ratio (ROR). ROR is frequently used in the spontaneous reporting database as an indicator of the safety signals of ADRs. We used the analysis data table and constructed 2×2 tables based on two classifications: the presence or absence of “lung toxicity” and presence or absence of LND use. The ROR was calculated as a ratio of the reported rate of ADRs attributable to LND and the reported rate of the same ADRs attributable to other drugs in the database. The signal of ADRs was considered positive if the lower limit of the 95% confidence interval (95% CI) of the ROR was >1 (12).
The time to onset of ADRs was calculated and the number of cases was counted for reports in which the dates of onset of ADRs, and start and end of administration were described in year/month/day or year/month (9). The onset time was calculated as: (onset date of adverse event) − (administration start date) +0.5 (13). If the period of non-administration was >1 year, the first administration date of the most recent continuous administration period was used. The time to onset of ADRs for analysis was limited to 2 years (730 days). The Weibull distribution is represented by a scale parameter α and shape parameter β. The scale parameter α represents the scale of the distribution function. It is the quantile where 63.2% of the ADRs occur (14). A large value of the scale indicates a wide distribution, while small value indicates narrow distribution. The shape parameter β represents the change in hazard over time in the absence of a reference population. Depending on the shape parameter β value, the upper limit of β value of 95% confidence interval (CI) <1 indicates that the hazard initially increases and then decreases (early failure type); a β value containing 1 or almost 1 and 95% CI of 1 indicates that the hazard remains constant throughout the exposure period (random failure type); and the lower limit of β value of 95% CI >1 indicates that the hazard increases over time (wear-out failure type).
Results
Incidence of lung toxicity with LND. We combined three tables–DRUG (3,875,874 reports), REAC (1,096,193 reports), and DEMO (693,295 patients)–by the ID numbers. We removed duplicate data from the DRUG and REAC tables (10, 11). The causes of ADRs were distributed into three categories: “suspected drugs”, “concomitant drugs”, and “interactions.” Of these, data included in the “suspected drugs” category was extracted and used as the “data table” (1,772,494 reports).
We analyzed this data table and obtained 10,929 reports on ADRs caused by LND. Of these, 908 lung toxicities were associated with LND (Figure 1). Patient characteristics are shown in Table I. Approximately 61.2% of the patients were men. According to the age distribution, lung toxicity was more common in patients in their 70s (38.7%), followed by those in their 60s (28.9%).
Construction of the data analysis table.
Characteristics of the patients exhibiting lenalidomide-related lung toxicities.
Among the types of lung toxicities caused by LND, pneumonia, interstitial lung disease, bacterial pneumonia, and Pneumocystis jirovecii pneumonia were the most common lung toxicities caused by LND, with 559, 126, 38, and 23, respectively (Table II). The RORs with a lower limit of 95% CI of >1 for pneumonia, bacterial pneumonia, pneumococcal pneumonia, influenzal pneumonia, upper respiratory tract inflammation, pulmonary artery thrombosis, cryptococcal pneumonia, and upper respiratory tract infection were 3.89 (95% CI=3.57-4.24, p<0.001), 2.02 (1.46-2.78, p<0.001), 4.17 (2.50-6.96, p<0.001), 14.0 (8.25-23.7, p<0.001), 4.00 (2.06-7.75, p<0.001), 2.15 (1.07-4.32, p=0.037), 2.62 (1.24-5.54, p=0.020) and 4.22 (1.74-10.3, p=0.008), respectively. Signals were detected for 8 (pneumonia, bacterial pneumonia, pneumococcal pneumonia, influenzal pneumonia, upper respiratory tract inflammation, pulmonary artery thrombosis, cryptococcal pneumonia, and upper respiratory tract infection) out of the 21 lung toxicity cases (Table II).
Number of reports and reported odds ratios (RORs) of lenalidomide-related lung toxicities.
Time to onset of lung toxicity with LND. A histogram of the time to onset of the nine lung toxicity signals showed that they occurred between 13 and 93 days after LND administration (Figure 2). The median onset (quartiles, 25-75%) of pneumonia, bacterial pneumonia, pneumococcal pneumonia, influenzal pneumonia, upper respiratory tract inflammation, pulmonary artery thrombosis, cryptococcal pneumonia and upper respiratory tract infection due to LND were 84 (26-223), 74 (18-287), 93 (30-237), 60 (33-252), 74 (39-154), 33 (18-45), 71 (14-344) and 13 (8-38) days, respectively. The Weibull distribution of the histogram of the time to onset showed that the range of 95% CIs for the shape parameter β of pneumonia, bacterial pneumonia, influenzal pneumonia, and cryptococcal pneumonia were <1; for pneumococcal pneumonia β»1 and for other ADRs β >1 (Table III).
Histogram of lung toxicity for: 1) pneumonia, 2) bacterial pneumonia, 3) pneumococcal pneumonia, 4) influenzal pneumonia, 5) upper respiratory tract inflammation, 6) pulmonary artery thrombosis, 7) cryptococcal pneumonia and 8) upper respiratory tract infections.
The medians and Weibull parameters of lung toxicities.
Outcomes after the occurrence of ADRs. The percentage of outcomes (recovery, remission, not recovered, with sequelae, death, and unclear) after the onset of eight ADRs is shown in Figure 3. Of the eight items for which signals were detected, fatal outcomes were observed in five ADRs.
Percentage of eight adverse drug reactions (ADRs) associated with lenalidomide by outcome.
Discussion
The reported frequency of lung toxicity with LND is not high–only 22 and 5 cases of pulmonary toxicity in 177 and 178 patients with MM and FL, respectively (1, 5). In comparison, the present study analyzed cumulative 559 cases of pneumonia for >17 years since 2004. The present study is large and objective because it accumulated data on 908 cases, including other lung toxicity profiles, and thoroughly examined them. Pneumonia was the most frequently reported lung toxicity due to LND. RORs were significantly high for bacterial pneumonia, pneumococcal pneumonia, influenzal pneumonia, upper respiratory tract infection, upper respiratory tract infection and cryptococcal pneumonia. In contrast, the ROR for interstitial pneumonia with LND was not high and showed a different lung toxicity profile than existing anti-EGFR and ALK-targeted oral anticancer drugs (gefitinib and alectinib). This may be influenced by the immunomodulatory effects of LND, and concomitant use of steroids and rituximab. For example, routine respiratory tract infections caused by Pneumocystis jirovecii occur in healthy individuals, but these rarely lead to pneumonia (19). However, immunomodulatory conditions caused by steroids, infliximab, or LND can lead to opportunistic infections, such as severe pneumonia (pneumocystis pneumonia; PCP) caused by Pneumocystis jirovecii (20). For instance, the amount of steroid used in combination with LND for treatment of MM is approximately 40 mg/day of dexamethasone. Antimicrobial therapy is recommended for prophylaxis of PCP when administering steroids with a prednisolone equivalent of ≥20 mg per day for >1 month (21). Prophylactic administration of trimethoprim–sulfamethoxazole (TMP/SMX) is effective in preventing pneumonia during steroid administration (22, 23). Rituximab targets CD20 on the surface of B cells, leading to liquid immunodeficiency; suppression of antibody production (hypogammaglobulinemia) persists until B cell recovery (24). In addition, neutropenia occurs in approximately 27% of patients treated with rituximab after approximately 90 days (25). Furthermore, rituximab causes cellular immune depression by decreasing levels of CD4-positive T cells (26). Therefore, the combination of LND and rituximab should be considered as a pharmacologic precaution against infectious pneumonia.
In the study (2) of LND for MM, prophylactic antimicrobial agents were not administered; but, the results suggest that compliance with prophylactic antimicrobial use during LND administration may be necessary in the future.
All LND-induced lung toxicities, except upper respiratory tract infection and pulmonary artery thrombus, occurred between 60 and 93 days after LND administration. When the histogram of time to onset of lung toxicity was evaluated by the Weibull distribution, the 95% CIs for the shape parameter β for pneumonia and bacterial pneumonia were <1. Drugs with β <1, as per the Weibull distribution, may cause symptoms early in the relatively prolonged PFS in patients in clinical studies with LND.
The current study also analyzed each pulmonary toxicity outcome that was not thoroughly reviewed in the previous clinical studies. Fatal outcomes were observed in 5 out of 8 cases with lung toxicities. Moreover, 10-20% of infective pneumonia cases had a fatal or unrecovered outcome.
Recent studies have raised concerns that lung injury due to coronavirus disease 2019 (COVID-19) leads to poor prognosis in patients with malignant hematologic diseases (27,28). Furthermore, 23% of patients with MM have no immune responses after two doses of COVID-19 mRNA vaccine (29). In contrast, inhibition of proinflammatory cytokine production by LND may be helpful against COVID-19-related infections and their complications (30). The incidence profile of pneumonia in the current study failed to distinguish between concomitant administration of steroids or rituximab and prevention of pneumonia with TMP/SMX. However, given the relatively long PFS in clinical studies with LND, alerting patients to lung toxicity early in treatment, around 60-90 days into therapy, is important information for prevention strategies. For patients receiving LND, frequent monitoring of respiratory symptoms and immunosuppression-associated pneumonia prophylaxis may be important, especially for COVID-19.
The present study has some limitations. Firstly, unlike clinical and observational studies, the JADER database is based on patients’ self-reports. Therefore, it fails to track all patients who received treatment. Furthermore, toxicity expression results may be over-reported. Secondly, the calculated date of onset data failed to reflect all reported data because data without the date of treatment initiation were excluded. Thirdly, the SRS, such as JADER, have missing data, lack of a denominator, and the presence of confounding factors (e.g., concomitant medications, comorbidities, and severity of ADR). Moreover, risk factors affecting pulmonary toxicities (e.g., smoking and underlying lung diseases) have not been evaluated in the database. However, we showed that observation of respiratory symptoms within 3 months of administration is important in managing pulmonary toxicities of LND, including pneumonia and bacterial pneumonia.
Conclusion
The present study highlighted the importance of accumulating data and analyzing the clinically infrequent toxicities with LND. We showed that LND-induced lung toxicity may occur within 3 months of therapeutic administration. The study findings should allow physicians and pharmacists to closely monitor lung toxicities during LND treatment. These data are important to alert physicians about patients receiving LND, especially with regards to lung toxicities in the ongoing COVID-19 pandemic.
Acknowledgements
We are grateful to Professor Yoshihiro Uesawa at the Department of Medical Molecular Informatics, Meiji Pharmaceutical University. The authors are grateful to Tadashi Hirooka (TAIHO PHARMA Corporation) for his lecture on Hirooka methods using JADER.
Footnotes
Authors’ Contributions
Junya Sato, Tsuchiya Eren, Saeko Murata and Tadashi Shimizu: Data curation; Writing – original draft; Writing – review and editing. Mayako Uchida: Conceptualization; Supervision; Writing – review and editing.
Conflicts of Interest
All Authors declare no conflicts of interest.
- Received September 25, 2022.
- Revision received September 29, 2022.
- Accepted October 6, 2022.
- Copyright © 2022 International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved.









