Pre-existing interstitial lung disease is associated with onset of nivolumab-induced pneumonitis in patients with solid tumors: a retrospective analysis

Nivolumab is a selective, fully humanized IgG4 monoclonal antibody that blocks binding between programmed death 1 (PD-1) and programmed death ligand-1 (PD-L1)/PD-L2, encouraging the antitumor activity of T cells. Nivolumab has demonstrated antitumor efficacy in clinical trials of various malignant tumors, including melanoma, renal cell cancer, non-small-cell lung cancer (NSCLC), head and neck cancer (HNC), and gastric cancer (GC) [1‐6]. However, immune check point inhibitors (ICI) including nivolumab can induce immune-related adverse events (irAEs) such as skin rash, colitis, endocrine disorders, hepatotoxicity, and pneumonitis.

Among them, pneumonitis is a relatively common but potentially life-threatening irAE. We recently reported that pre-existing interstitial lung diseases (ILD) is a risk factor for anti-PD-1-induced pneumonitis in patients with NSCLC; however, it is not clear if this tendency applies to other types of tumors [7]. Although the incidences of most irAEs are considered to be similar regardless of tumor type, in the case of pneumonitis these frequencies may differ across tumor types. In the previous phase III clinical trials of nivolumab, the incidence of nivolumab-induced pneumonitis was reported as 3.5–5% in NSCLC, 2.1% in squamous cell HNC, and < 1% in GC, and thus it appears to vary among the different cancer types [2‐5]. We speculated that these tumor-specific bias in the incidence of pneumonitis may be related to pre-existing ILD, regardless of the tumor type. Moreover, patients complicated with ILD are routinely excluded from clinical trials, therefore, these questions need to be resolved by real-world data.

The aim of this study was to evaluate the association between ILD on chest computed tomography (CT) and nivolumab-induced pneumonitis among cohorts with different tumor types, including NSCLC, HNC, and GC.

In this study, the overall incidence of all-grade pneumonitis was 11.7%, similar to a previous retrospective Japanese cohort study in patients with NSCLC treated by nivolumab [10, 11]. Although there was no significant difference in the univariate analysis, the NSCLC group tended to have a higher incidence of pneumonitis (14.6%) compared with the HNC/GC group (8.7%), which was related to the high trend of complicating pre-existing ILD in the NSCLC group (P = 0.047).

The present analysis revealed that pre-existing ILD and male sex were significant independent risk factors for nivolumab-induced pneumonitis in patients with solid tumors. Previous evidence has strongly indicated that pre-existing ILD is a risk factor for chemotherapy-induced pneumonitis, and these reports were particularly prevalent in lung cancer [12‐17]. These findings are related to the fact that both lung cancer and ILD are closely associated with cigarette smoking. There is substantial epidemiological evidence that patients with ILD have a high risk of lung cancer [18, 19], whereas it has been reported that 10–20% of patients were complicated with ILD when lung cancer was diagnosed [13, 20]. Underpinning this epidemiological fact is that lung cancer and ILD share multiple common genetic, molecular, and cellular processes, such as epithelial-mesenchymal transition, endoplasmic reticulum stress, transforming growth factor expression, and oxidative stress [21, 22]. However, to our knowledge, there are few reports on the development of non-lung cancer and ILD. In our study, there was no significant difference in smoking rates between the NSCLC and HNC/GC groups, but the comorbidity rate of pre-existing ILD was significantly higher in the NSCLC group than in the HNC/GC group. The high prevalence of ILD in the NSCLC group, despite no difference in smoking history, may supports a correlation between the development of NSCLC and ILD. We previously reported that pre-existing ILD is a risk factor for anti-PD-1-induced pneumonitis [7]. However, it has not been elucidated whether pre-existing ILD is at risk of pneumonitis when using anti-PD-1 agents for other type of tumors. In previous phase III clinical trials, HNC and GC had a lower incidence of nivolumab-induced pneumonitis compared with NSCLC [2‐5]. While a recent meta-analysis indicated that anti-PD-1-related pneumonitis of all grades develops more frequently in NSCLC and renal cell cancer than in melanoma [23, 24]. In our analysis based on pre-existing ILD on chest CT, NSCLC and HNC/GC showed similar trends in the development of pneumonitis. Interestingly, in patients with no fibrosis, the incidence of nivolumab-induced pneumonitis was approximately 5% in both NSCLC and HNC/GC, similar to results reported in previous large clinical trials. Moreover, in patients with pre-existing ILD, the incidence of nivolumab-induced pneumonitis jumped to approximately 30%. Although the biological mechanism of ICI-induced pneumonitis is poorly understood, it can be hypothesized that deregulation of immune effectors and T-cells in the lung interstitium is involved, leading to the increased production of inflammatory cytokines and resulting in an inflammatory response. Patients with ICI-induced pneumonitis were reported to have an increased number of activated T-cells in their bronchoalveolar lavage fluid compared with healthy control subjects, therefore T-cells are thought to participate in the onset of ICI-induced pneumonitis [25]. The activation of various immune cells likely targets the fragile lungs of chronic ILD patients, causing ICI-induced pneumonitis. The results of our study support the hypothesis that the risk of developing pneumonitis depends on the patient’s background lung condition, regardless of the tumor type.

It should be emphasized that increased risk of anti-PD-1-induced pneumonitis is not synonymous with the avoidance of anti-PD-1 treatment. Previous studies have demonstrated that the development of irAEs is associated with clinical benefit of ICIs in melanoma and NSCLC [26‐31]. In a recent retrospective cohort study in patients with NSCLC treated with nivolumab (n = 613), landmark analyses of PFS at 2 months revealed no significant differences in PFS between patients with or without pneumonitis (7.9 vs. 5.9 months; P = 0.872). Moreover, in our previous report, patients with anti-PD-1-induced pneumonitis had relatively favorable survival outcomes [7]. Most cases of pneumonitis associated with nivolumab are relatively mild and classified as grade 1 to 2. It has been reported that nivolumab-induced pneumonitis responds well to steroid therapy and its mortality rate is low compared with gefitinib-induced pneumonitis, which is characterized by rapid progression and a high mortality rate of around 30 to 40% [12]. Therefore, ICI-induced pneumonitis may not lead to poor outcomes, unlike pneumonitis with a cytotoxic chemotherapy or molecularly targeted therapy that has been reported to have a poor prognosis [32]. Therefore, indications for ICI treatment should be judged comprehensively, taking into consideration pre-existing ILD as well as pulmonary function, which affects the severity of pneumonitis. All severe cases of pneumonitis (grade 3 or higher) were observed in patients with pre-existing ILD. Patients with pre-existing ILD were at increased risk of developing pneumonitis, and these patients also had impaired respiratory function, suggesting that pneumonitis was more likely to become severe.

A history of thoracic radiotherapy would also affect the incidence of chemotherapy-related pneumonitis in patients with NSCLC. Tamiya et al. reported that the incidence of nivolumab-induced pneumonitis was significantly higher in a group with thoracic radiation history than in a group without thoracic radiation history [10]. A higher incidence of nivolumab-induced pneumonitis in patients with thoracic radiation history may be because nivolumab can induce radiation recall pneumonitis. Therefore, our analysis excluded patients with a history of thoracic radiation to prevent bias.

Male sex has been reported as a risk factor for drug-induced pneumonitis in several studies following treatment with gefitinib, panitumumab, and pemetrexed [33‐35]. However, we should keep in mind that males have a higher incidence of smoking than females, and multivariate analysis may not completely eliminate this bias. In the present study, only two out of 55 female patients (3.6%) had pre-existing ILD, and these two patients were classified as “inconsistent with UIP.” Therefore, to investigate the relationship between pre-existing ILD and nivolumab-induced pneumonitis particularly in females, further validation of another cohort is needed.

There are several limitations in the present study. First, this was a single-site retrospective study that used chart reviews of a relatively small patient cohort. Second, the number of patients with HNC/GC was smaller, since nivolumab was approved earlier for NSCLC than HNC/GC. NSCLC had a larger sample size and a higher incidence of pneumonitis than HNC and GC, which may have affected the overall results. In addition, the number of cases of HNC and GC is relatively small, and data on the incidence of pneumonitis in HNC/GC should be carefully interpreted. To compare the incidence of pneumonitis for each type of cancer, it is necessary to analyze it in a larger population. Third, we excluded patients who previously received thoracic radiation or other ICIs to prevent biases; however, since the standard treatment differs for each cancer type it is possible that the effects of prior treatment could not be completely excluded. Fourth, the patient cohort was across multiple cancer types, which prevented cancer-specific survival analysis. Fifth, lung function tests were not included in this analysis as some patients with HNC were unable to undertake them because of their post-tracheostomy status so they were not routinely performed in our institution. Sixth, bronchoalveolar lavage fluid was not routinely collected at our hospital and could not be incorporated into the analysis.

Our findings showed that pre-existing ILD is a common risk factor for nivolumab-induced pneumonitis in solid cancer treatment, including NSCLC, HNC, and GC. The difference in the incidence of anti-PD-1-induced pneumonitis across various tumor types seems to have occurred due to the difference in the pre-existing ILD complication rate. To investigate the association between nivolumab-induced pneumonitis and pre-existing ILD across whole tumors, further analyses in a larger multicenter collaborative study are required. Likewise, further investigations are also required for anti-PD-L1-antibodies and chemo-ICI combination therapies to examine the relationship between pre-existing ILD and ICI-induced pneumonitis.