*3.3. Characteristics of Patients with Anti-Cancer Drug-Induced Pneumonia*

The characteristics of the 19 patients with anti-cancer drug-induced pneumonia are listed in Table S1. Although previous radiotherapy did not confer a risk for anti-cancer drug-induced pneumonia (Table 2A), 10 of 19 these patients had received radiotherapy. The development of radiological interstitial abnormalities in the ten patients did not correspond to the exposure to radiation. The HFS did not correlate with the grade of anti-cancer drug-induced pneumonia. Specific regimens were not associated with the risk of anti-cancer drug-induced pneumonia.

### **4. Discussion**

To the best of our knowledge, the present study is the first to provide the following important findings. The quantitative HFS is positively correlated with the risk of anti-cancer drug-induced pneumonia in lung cancer patients. In particular, the HFS was associated with a risk of anti-cancer drug-induced pneumonia in patients with mild pre-existing ILD, even in the absence of honeycombing or traction bronchiectasis.

The current IIP guidelines do not assess the severity of pre-existing ILD, particularly mild ILD. IPF diagnosed using the IIP guidelines was observed in 2%–8% of lung cancer patients [21–23]. In contrast, 28.5% of lung cancer patients that were administered anti-cancer drugs had mild pre-existing ILD confirmed by the HFS in the present study. In two computed tomography lung cancer screening trials that defined the patterns of mild interstitial lung abnormalities, ILD was detected in 15.7%–21.2% of patients with a history of smoking [24,25]. Since the IIP guidelines do not necessarily consider the radiological process of pulmonary fibrosis, mild ILD may have been overlooked. Precise evaluation of the risk of anti-cancer drug-induced pneumonia requires a quantitative method capable of detecting all levels of ILD, including mild ILD.

In the present study, the quantitative HFS was significantly correlated with the risk of anti-cancer drug-induced pneumonia. Little is known of the relationship between anti-cancer drugs and mild pre-existing ILD [1,2]. A nested case-control study investigated the risk factors for pre-existing ILD associated with chemotherapy and gefitinib [1]. Although this study referred to the severity of ILD, including mild grade, the severity was not clearly defined. The adjusted odds ratios indicated that patients with mild and moderate–severe ILD had an approximately five-fold increased risk of developing acute ILD compared to patients without ILD, suggesting no difference between mild and moderate–severe ILD. Another prospective cohort study identified the risk factors for pre-existing ILD associated with erlotinib [2]. This study evaluated only the extent of interstitial abnormalities, but not the fibrous pattern on HRCT. The study population of patients with pre-existing ILD predominantly comprised patients with mild ILD, in which abnormalities involved <5% of the bilateral lower lobes. The multivariate logistic regression analysis yielded an odds ratio of 4.0 (95% CI, 1.3–12.1) between the presence and absence of ILD. Thus, the present study first revealed that an increase of HFS point was associated with a 16% increased risk of anti-cancer drug-induced pneumonia, even in mild ILD without honeycombing and traction bronchiectasis.

Pre-existing mild ILD might induce drug-induced pneumonia because a histological UIP pattern may be involved. The 2018 IPF guidelines categorize mild ILD as an indeterminate for the UIP HRCT pattern (early UIP pattern) [13]. This category includes the subset of patients with reticular patterns in the absence of honeycombing and traction bronchiectasis, predominantly in the subpleural and basal fields. Lung cancer patients with UIP HRCT patterns exhibited exacerbation of ILD more frequently than did those with non-UIP HRCT patterns [3]. In precision medicine, many novel anti-cancer drugs have been developed, some of which are likely to cause drug-induced pneumonia. The early recognition of the subtle fibrosing findings on HRCT may enable physicians to provide better management of lung cancer patients receiving anti-cancer drugs.

The present study had several limitations. First, all cases of pre-existing ILD incidentally included mild cases. Thus, we did not necessarily verify the utility of the HFS for moderate and severe ILD. However, a previous case-control study revealed that the odds ratio of the development of anti-cancer drug-induced pneumonia was high in patients with moderate/severe pre-existing ILD as well as those with mild ILD [1]. Second, the pulmonary function test results could not be evaluated, due to missing values in clinical practice. Another quantitative HRCT scoring system has been replaced by the diffusion capacity of carbon monoxide test for the prognosis with IPF patients [26]. Third, this was a single institutional analysis, and the number of anti-cancer drug-induced pneumonia cases was small. The present findings are therefore potentially subject to selection bias. Further prospective investigations involving large cohorts are required to confirm our findings.

In conclusion, HFS may correlate with the risk of anti-cancer drug-induced pneumonia in lung cancer patients, even in the absence of honeycombing or traction bronchiectasis. This quantitative HFS could be a predictor of anti-cancer drug-induced pneumonia, which could improve the management of lung cancer patients with ILD.

*J. Clin. Med.* **2020**, *9*, 1045

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2077-0383/9/4/1045/s1, Table S1: List of anti-cancer drug-induced pneumonia patients (*n* = 19).

**Author Contributions:** Conceptualization, H.G., H.Y. and H.M.; Data curation, S.S. and K.A.; Formal analysis, H.G., H.I. and K.A.; Investigation, H.G., H.S. and M.F.; Methodology, H.Y., H.T., T.K., K.N., N.S. and H.S.; Supervision, H.Y., K.N., Y.O., H.S. and H.M.; Writing —original draft, H.G.; Writing—review & editing, H.Y., N.S., Y.O. and H.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** We would like to thank Editage (www.editage.jp) for English language editing.

**Conflicts of Interest:** The authors declare no conflicts of interest.
