eMethods. Search strategy for Medline (using PubMed)
eTable 1. Subjects without Nivolumab (n=15,338)
eTable 2. Subjects with Nivolumab (n=5,178)
Customize your JAMA Network experience by selecting one or more topics from the list below.
Oshima Y, Tanimoto T, Yuji K, Tojo A. EGFR–TKI-Associated Interstitial Pneumonitis in Nivolumab-Treated Patients With Non–Small Cell Lung Cancer. JAMA Oncol. 2018;4(8):1112–1115. doi:10.1001/jamaoncol.2017.4526
Does treatment with nivolumab affect the incidence of epidermal growth factor receptor-tyrosine kinase inhibitor (EGFR–TKI)-induced interstitial pneumonitis (IP) in non–small cell lung cancer (NSCLC) patients?
In this observational study including 20 516 patients with non–small cell lung cancer, the adjusted odds ratio for EGFR–TKI-associated IP in cases with and without nivolumab was 5.09 and 1.22, respectively. The difference was statistically significant.
Treatment with nivolumab was associated with increased EGFR–TKI-induced interstitial pneumonitis.
Nivolumab and epidermal growth factor receptor–tyrosine kinase inhibitors (EGFR-TKIs) are now the standard-of-care therapies in non–small cell lung cancer (NSCLC). Although EGFR-TKIs are well understood and have well-defined safety profiles, our experience with immune checkpoint inhibitors is still growing, particularly regarding the use of combinations of different classes of antitumor agents, including both the concomitant and sequential use of such agents.
To determine whether nivolumab increases EGFR–TKI-associated interstitial pneumonitis (IP).
Design, Setting, and Participants
A database study of 20 516 participants with NSCLC in the US Food and Drug Administration Adverse Event Reporting System (FAERS) database, performed between April 2015 and March 2017.
Main Outcomes and Measures
We compared the incidence of EGFR–TKI-associated IP in patients receiving and not receiving nivolumab treatment.
The mean (SD) age of participants treated with EGFR-TKI, with and without nivolumab, was 64.4 (15.5) and 68.9 (11.8) years, respectively, and the proportion of men was 40.0% and 53.8%, respectively. Of the 20 516 participants with NSCLC, 985 cases (4.80%; 95% CI, 4.51-5.10) developed IP. Of 5777 patients treated with EGFR-TKI, 265 developed IP (4.59%; 95% CI, 4.06-5.16). Of 70 patients treated with both EGFR-TKI and nivolumab, 18 developed IP (25.7%; 95% CI, 16.0-37.6). The adjusted odds ratio for an interaction between EGFR-TKI and nivolumab was 4.31 (95% CI, 2.37-7.86; P < .001), suggesting the existence of an interaction. When we further stratified the patients by treatment with and without nivolumab, the odds ratio of EGFR–TKI-associated IP in cases with and without nivolumab treatment was 5.09 (95% CI, 2.87-9.03) and 1.22 (95% CI, 1.00-1.47), respectively.
Conclusions and Relevance
We found a higher proportion of reports of IP for nivolumab in combination with EGFR-TKI vs treatment with either drug alone. Owing to the limitations of this study, the results warrant further confirmation. However, careful consideration should be given to the possibility of an increased risk of IP when EGFR-TKI is administered in combination with nivolumab, including concomitant and sequential use, and careful monitoring for IP is recommended.
Immune checkpoint inhibition therapy is being increasingly applied to treat various solid tumors, including non–small cell lung cancer (NSCLC), using programmed cell death-1 (PD-1) pathway inhibitors, such as monoclonal antibodies that target either PD-1 or PD-1 ligand (PD-L1). Although this can cause various adverse drug reactions, there is limited information concerning the interaction between immune checkpoint inhibition therapy and other antitumor agents, including in the context of molecularly targeted therapy. Recently, in a case series in Japan,1 7 patients pretreated with a PD-1 pathway inhibitor, nivolumab, developed severe interstitial pneumonitis (IP) after epidermal growth factor receptor (EGFR)-tyrosine kinase inhibitor (TKI) treatment; 3 of these patients died.
During EGFR-TKI treatment, IP can be fatal unless it is diagnosed sufficiently early and appropriate treatment is initiated. Although a combination of treatment with nivolumab and subsequent administration of EGFR-TKI was a suspected risk factor for IP onset in the previously mentioned case series, to our knowledge no comparative study has assessed the link. The US Food and Drug Administration (FDA) runs a program assessing the safety of approved products. Adverse reactions can be reported through the MedWatch program by health care professionals, consumers, pharmaceutical firms or the general public. The submitted adverse events are stored in the FDA Adverse Event Reporting System (FAERS) database,2,3 which is open to the public and scientists can download it from the FDA’s home page. Herein, we analyzed the large postmarketing FAERS database and performed a comparative study to investigate whether there was an association between IP and the combination of nivolumab and EGFR-TKI.
Institutional review board approval was not required because FAERS is an unlinkable anonymized database open to the public. We extracted data for patients with NSCLC registered between April 1, 2015, and March 31, 2017, from FAERS. The EGFR-TKIs in the study included afatinib, erlotinib, gefitinib, and osimertinib. Treatment with or without nivolumab was also investigated. Subsequently, we identified patients with IP (eMethods in Supplement 1). Adjusted odds ratios were analyzed by multivariable logistic regression. The variables considered in the models were age, sex, and the use of EGFR-TKI or nivolumab. Data manipulation and statistical analyses were performed using Microsoft Access (2010, Microsoft) and R software (version 3.3.3, R Foundation),4 respectively. The patient data set is included in Supplement 2.
We identified 20 516 NSCLC cases. Among them, 5777 and 5178 cases were treated with EGFR-TKI and nivolumab, respectively (Table 1 and Table 2). Seventy patients were treated with both EGFR-TKI and nivolumab, of which 18 patients (25.7%; 95% CI, 16.0-37.6) developed IP, implying an increased risk of IP with the combination treatment (Table 2).
To confirm this observation, we conducted multivariable logistic regression analysis of the interaction between nivolumab and EGFR-TKI. The adjusted odds ratios for IP with EGFR-TKI treatment and nivolumab treatment were 1.21 (95% CI, 1.00-1.47) and 1.79 (95% CI, 1.50-2.13), respectively (Table 2). The adjusted odds ratio of the interaction effect was 4.31 (95% CI, 2.37-7.86), supporting the existence of the interaction (Table 2).
We conducted further analyses using multivariable logistic regression models stratified by nivolumab treatment for IP. The adjusted odds ratios for IP with EGFR-TKI, with and without nivolumab treatment were 5.09 (95% CI, 2.87-9.03) and 1.22 (95% CI, 1.00-1.47), respectively (eTable 1 and eTable 2 in Supplement 1).
Among the 18 patients with IP treated with both EGFR-TKI and nivolumab, we identified the order of administration in 15 cases; all were treated with EGFR-TKI after nivolumab (eResults in Supplement 1).
Of 3266 Japanese patients with NSCLC, 468 developed IP (14.3%; 95% CI, 13.1-15.6). As reported previously,5 the proportion of patients with IP appeared to be higher in Japan than elsewhere (3.0%; 95% CI, 2.8-3.3). Of 39 Japanese patients with NSCLC treated with both drugs, 16 developed IP (41.0%; 95% CI, 25.6-57.9). Although the number of patients was limited, the combination treatment appeared to result in a higher proportion of IP reports in the Japanese subgroup. Reports from the United States comprised about 40% of all reports; of 20 516 patients with NSCLC, 9046 (approximately 44%) were reported from the United States. However, only 21 patients were treated with EGFR-TKI and nivolumab in combination, of which IP was reported in only 1 (approximately 5%). We could not assess the risk of IP associated with combination treatment in the United States subgroup.
Nivolumab and EGFR-TKIs are now standard-of-care therapies for NSCLC. Although EGFR-TKIs are well understood and have well-defined toxic effects profiles, our experience with immune checkpoint inhibitors is still growing, particularly regarding the use of combinations of different classes of antitumor agents, including both the concomitant and sequential use of such agents. In this study, we noted a higher proportion of IP reports in association with nivolumab and EGFR-TKI combination treatment vs either drug alone. The results of multivariable logistic regression supported the existence of an interaction.
Interstitial pneumonitis is one of the most serious adverse reactions related to EGFR-TKI; the reported incidence is approximately 1.6% to 4.3% in Japanese populations and 0.3% to 1.0% in non-Japanese populations, with a mortality rate of 20% to 50%.5 Furthermore, in phase III trials of nivolumab in patients with NSCLC, approximately 4% of the nivolumab treatment groups developed IP.6,7 Our data suggest that the risk of EGFR–TKI-induced IP may increase markedly when in combination with nivolumab treatment. To our knowledge, no comparative study has assessed the effect of combination treatment with nivolumab and EGFR-TKI on IP. Notably, in a phase Ib clinical trial, TATTON (NCT02143466), the incidence of IP was 2.9% with osimertinib combination and 2.0% with durvalumab monotherapy, compared with 38% with the 2 drugs in combination.8 Durvalumab blocks PD-L1 binding to PD-1 and CD80, and osimertinib is an EGFR-TKI. Thus, PD-1 pathway inhibitors may increase the risk of EGFR–TKI-induced IP. We could not assess the risks of other PD-1 pathway inhibitors owing to the limited number of combination-treatment cases in the FAERS database (Table 3).
Health care professionals, consumers, pharmaceutical firms and anyone who experiences potential adverse reactions can report these as adverse events to the FDA through the MedWatch program. The adverse event reports are stored in the FAERS database. Although pharmaceutical firms are required to submit safety reports on known individual cases, other entities are not. Thus, the information stored in the FAERS database is essentially based on spontaneous reports. In addition, not all serious adverse events in drug-treated patients are captured by MedWatch, resulting in reporting biases. Furthermore, certain important information, such as underlying diseases, previous treatment, or even the start and end dates of use of the suspected drugs, or of adverse event onset, may be missing. Moreover, the reported events are not always confirmed in a standardized manner. Although the FAERS database is useful for monitoring events in real-world clinical practice events quickly, the imperfection of the data that it contains represents a disadvantage. Owing to these limitations, the data only indicate an increased risk with a particular drug, requiring further confirmation. To confirm our findings, a further observational study using real-world data, including a medical record database or medical claims database, is warranted. Such data would capture patients’ background information, which is usually not well represented in the FAERS database. Moreover, drugs including EGFR-TKIs and PD-1 pathway inhibitors have already been used in medical practice, and sufficient data for analyses might have already been compiled.
Although not confirmed, careful consideration should be given to the possibility of an increased risk of IP when EGFR-TKI is administered in combination with nivolumab, including both concomitant and sequential use, and careful monitoring for IP is recommended.
Corresponding Author: Yasuo Oshima, MD, PhD, FACP, The Institute of Medical Science, The University of Tokyo, 4-1-6 Minato-ku, Tokyo-to, 108-8639 Japan (email@example.com).
Accepted for Publication: October 23, 2017.
Published Online: January 11, 2018. doi:10.1001/jamaoncol.2017.4526
Author Contributions: Dr Oshima had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Oshima, Yuji.
Acquisition, analysis, or interpretation of data: Oshima, Tanimono, Tojo.
Drafting of the manuscript: Oshima, Yuji.
Critical revision of the manuscript for important intellectual content: Tanimono, Tojo.
Statistical analysis: Oshima.
Administrative, technical, or material support: Oshima, Tojo.
Study supervision: Yuji, Tojo.
Conflict of Interest Disclosures: Dr Oshima reports personal fees from Novartis Pharma K.K., personal fees from sanofi K.K., outside the submitted work. Dr Tojo reports grants from Pfizer, grants from Chugai Pharmaceutical, grants from Daiichi Sankyo, grants from Sumitomo Dainippon Pharma, personal fees from Taiho Pharmaceutical, personal fees from Novartis, personal fees from Otsuka Pharmaceutical, personal fees from Sysmex, personal fees from Bristol-Myers Squibb, outside the submitted work. No other disclosures are reported.
Create a personal account or sign in to: