Risk factors for pneumonitis in patients treated with anti‐programmed death‐1 therapy: A case‐control study

Abstract Immune checkpoint blockade‐related pneumonitis is a rare but potentially life‐threatening adverse effect, but its risk factors are not completely understood. This case‐control study was conducted to identify pneumonitis risk factors in patients treated with anti‐PD1 monoclonal antibodies (mAbs), including all the patients who developed pneumonitis after anti‐PD‐1 mAbs treatment in the Cancer Center of the Chinese People's Liberation Army from September 2015 to September 2017. Two controls per case were matched according to a propensity‐score matching algorithm to account for confounding effects caused by individual baseline variables. Demographic and clinical information was obtained from medical records. In total, 55 cases and 110 controls were included in the study. No association was observed between smoking status or primary lung cancer and risk of pneumonitis. Significant risk factors for pneumonitis related to anti‐PD‐1 mAbs were prior thoracic radiotherapy, prior lung disease and combination therapy with odds ratios of 3.34 (1.51‐7.39), 2.86 (1.45‐5.64) and 2.73 (1.40‐5.31), respectively. The associations remained significant in the multivariable logistic regression model. The risk of pneumonitis induced by anti‐PD‐1 mAbs is associated with prior thoracic radiotherapy, prior lung disease, and combination therapy. Clinicians should monitor these features in patients receiving anti‐PD‐1 therapy to optimize clinical safety and efficacy.

immune homeostasis and results in immune-related adverse events (irAEs), including dermatologic, hepatic, endocrine, gastrointestinal, and pulmonary complications as well as other less common side effects. 2,3 Among these irAEs, pneumonitis, which is defined as focal or diffuse inflammation of the lung parenchyma, 4 is rare but potentially life threatening. In general, the incidence of pneumonitis ranges from 0% to 10%. 5 Although most events are low grade and are improved/resolved with drug holding/immunosuppression [22], pneumonitis is life threatening in rare cases. In an early-phase study of an anti-PD-1 monoclonal antibody (mAb), pneumonitis accounted for 3 deaths. 6 Recent studies have discussed the clinical features, diagnosis, and management of pneumonitis related to ICBs thoroughly; however, only a few studies have attempted to identify potential risk factors.
In a case report, 3 of 25 patients who received thoracic radiotherapy and anti-PD-1 therapy developed pneumonitis, suggesting a possible association between pneumonitis and prior thoracic radiotherapy. 7 In several meta-analyses of published randomized clinical trials, pneumonitis incidence is relatively increased in smokers; non-small-cell lung cancer (NSCLC) patients; patients with previous lung disease, including chronic obstructive pulmonary disease; and those who received combination therapy. [8][9][10] However, these data were not directly compared but were instead extracted from separate randomized controlled studies, and the variability of these randomized clinical trial characteristics, such as inclusion or exclusion criteria, dosage and the stage of disease, may have confounded the results. Radiomic features can classify and predict patients at baseline who will subsequently develop immunotherapyinduced pneumonitis. 11 However, small sample sizes in casereports and potential confounders in meta-analyses may limit the reliability and validity of data. Using a larger sample size in the present case-control study, we aimed to identify the association between factors weakening pulmonary function and the risk of pneumonitis induced by anti-PD-1 mAbs.

| Study population
Eligible patients included those with advanced tumor who developed pneumonitis after anti-PD-1 mAbs treatment in the Cancer Center of the Chinese People's Liberation Army from September 2015 to September 2017. Pneumonitis was diagnosed by a radiologist according to computed tomography scans. Patients were excluded if there was a clear alternative etiology, such as proven malignant lung infiltration or active lung infection. Pneumonitis grading was performed by the treating investigators in real time using Common Toxicity Criteria for Adverse Events (version 4.0). Controls were defined as advanced tumor patients treated with anti-PD-1 mAbs who did not develop pneumonitis. This study was approved by the Ethics Committee of Chinese People's Liberation Army General Hospital.

| Data collection
For all the patients, the following pretreatment demographic and clinical information was obtained from medical records: age, sex, smoking status, weight, metastatic sites, baseline lactic dehydrogenase, previous anticancer treatments, treatment agents with anti-PD-1 mAbs, the number of anti-PD-1 mAbs treatment cycles, primary tumor type, and prior lung disease. Smoking status was classified as no history of smoking or a positive history. Combination therapy was defined as anti-PD-1 treatment with chemotherapy, targeted therapy or CTLA-4 blockade. Previous anticancer treatment (6 months prior) was classified as chest surgery, thoracic radiotherapy, targeted therapy or chemotherapy. Prior lung disease included pneumothorax, pleural effusion, and pneumonitis before anti-PD-1 mAbs therapy. In patients with pneumonitis, the clinical features of pneumonitis were collected retrospectively from medical records. The severity of the pneumonitis was described using the criteria for interstitial lung diseases. [12][13][14] The median follow-up time for the case group was 239 days with a range from 12 to 610 days. For the control group, the follow-up time was 142.5 days with a range from 34 to 793 days.

| Statistical analysis
Statistical analyses were performed using SPSS statistical software (version 24.0; SPSS, IBM Corporation, USA). Continuous variables were described with the median (range) and were compared by 2-sided t tests or the Mann-Whitney U test. Categorical variables were compared by Chi-square test or Fisher's exact test. Propensity-score matching was used to create case and control groups with similar baseline characteristics. Matching with a ratio of 1:2 was performed using the optimal-matching algorithm in MatchIt R package 3.4.2. 15 Propensity scores were estimated by age, sex, weight, metastasis, number of treatment cycles, and baseline lactic dehydrogenase. Estimates of odds ratios (ORs) and 95% confidence intervals (CIs) of risk factors were obtained using logistic regression. Variables with significance were included in a multiple logistic regression analysis. P = .05 was set as the level of significance. All the reported Pvalues were 2-sided.

| Patient characteristics
Overall, 55 cases and 110 controls were included in this analysis. The characteristics of cases and controls are described in Table 1. No differences were observed in age, sex, weight, metastasis, number of treatment cycles and baseline lactic dehydrogenase between the case and control groups after matching.

| Risk factors for anti-PD-1 mAbsrelated pneumonitis
Immune-mediated pneumonitis is likely to preferentially attack patients with characteristics that worsen pulmonary conditions, including smoking status, prior treatment, combination therapy, primary tumor type, and prior lung disease. Therefore, we performed logistic regression to identify the association between the occurrence of ICB-related pneumonitis and these selected variables. However, no prominent association was observed between the risk of pneumonitis and smoking status or lung cancer. Significant factors associated with anti-PD-1 mAbs-related pneumonitis included prior thoracic radiotherapy, prior lung disease and combination therapy (P < .01). The odds ratios were 3.34 (1.51-7.39),  (Table 3). Next, prior thoracic radiotherapy, prior lung disease, and combination therapy were included in multiple logistic regression analysis. Similar results were observed as prior thoracic radiotherapy, prior lung disease, and combination therapy were significantly associated with the risk of pneumonitis (P < .01, Table 4).

| DISCUSSION
In this case-control study, risk factors, including prior thoracic radiotherapy, prior lung disease, and combination therapy, for pneumonitis induced by anti-PD-1 therapy were identified. No significant association was observed between smoking or lung cancer and the risk of anti-PD-1 therapyrelated pneumonitis. As lungs represent the core organ affected by immunemediated pneumonitis, factors influencing pulmonary functions may be associated with immune-mediated pneumonitis.
Previous meta-analyses have demonstrated that NSCLC patients exhibit an increased incidence of pneumonitis compared with that in melanoma patients. 9,10 However, these data were extracted from separate randomized controlled studies and were not compared directly. Clinical study variables, such as inclusion or exclusion criteria, dosage, and stage, varied in these randomized trials, potentially limiting an accurate interpretation of the results. In addition, a similar incidence among patients with melanoma or NSCLC for monotherapy or combination therapy was revealed in a previous observational study. 16 Consistent with this finding, no significant association between lung cancer and ICB-related pneumonitis was detected in this study. In addition, no relation between smoking status and pneumonitis was revealed. However, the possibility of an association between smoking and pneumonitis cannot be fully excluded because detailed information regarding smoking quantity and duration were not available.
However, we observed an association between pneumonitis and prior thoracic radiotherapy, prior lung disease or combination therapy. Thoracic radiotherapy or prior lung disease were associated with poor pulmonary function. NSCLC patients are typically submitted to thoracic radiotherapy and often exhibit other lung complications. The observed increased incidence of pneumonitis in NSCLC patients from previous studies might be attributed to the poor lung function induced by prior thoracic radiotherapy or NSCLC itself rather than the anti-PD-1 therapy. Pneumonitis incidence might increase if anti-PD-1 mAb was combined with other agents, such as anti-CTLA-4 mAb, 9 also known to carry a risk of pneumonitis. 2 In this study, we confirmed these observations and further clarified that anti-PD-1 treatment combined with chemotherapy, targeted therapy or CTLA-4 blockade was also associated with pneumonitis risk, indicating the additive effects of multiple agents on lung-toxic effects.
However, some limitations of this study should be addressed. First, the number of pneumonitis patients might not be sufficient to provide reliable information concerning the significance of each risk factor. Second, the risk of ICBrelated pneumonitis for each combined therapy (chemotherapy, targeted therapy, CTLA-4 blockade or radiotherapy) might vary due to distinct pharmacological mechanisms. However, the sample size was not adequate to address each combined therapy individually. Third, drug-induced pneumonitis remains a diagnosis of exclusion and requires the consideration of competing diagnoses. Fourth, we did not include patients treated with anti-PD-L1 mAbs who may share similar risk factors for pneumonitis with anti-PD-1 mAbs. Fifth, prior lung disease did not include asthma, chronic obstructive pulmonary disease or interstitial lung disease because this information was not provided. Other limitations common to case-control studies are the risk of observational bias and confounding bias, which was minimized by applying the propensity-score matching protocol.
In conclusion, we identified prior thoracic radiotherapy, prior lung disease and combination therapy as risk factors for the life-threatening pneumonitis observed in patients receiving anti-PD-1 immunotherapy. Although the detailed role and mechanism underlying these risk factors for the development of pneumonitis requires further investigation, patients with these characteristics require intensive care in cases of anti-PD-1 mAbs-related pneumonitis. To the best of our knowledge, this is the first case-control study systematically seeking to identify risk factors for anti-PD-1 mAbs-related pneumonitis development in a Chinese population of multiple types of advanced cancer. This study fills an important gap in the literature and supplements the limited number of published reports on anti-PD-1 pneumonitis. 5

CONFLICT OF INTEREST
The authors have declared no conflict of interests.