Interstitial Lung Disease in Japanese Patients With Lung Cancer: A Cohort and Nested Case-Control Study

Rationale: Interstitial lung disease (ILD) occurs in Japanese patients with non-small cell lung cancer (NSCLC) receiving gefitinib. Objectives: To elucidate risk factors for ILD in Japanese patients with NSCLC during treatment with gefitinib or chemotherapy.

Methods: In a prospective epidemiologic cohort, 3,166 Japanese patients with advanced/recurrent NSCLCwere followed for 12weeks on 250 mg gefitinib (n = 1,872 treatment periods) or chemotherapy (n = 2,551). Patients who developed acute ILD (n = 122) and randomly selected control subjects (n = 574) entered a case-control study. Adjusted incidence rate ratios were estimated from casecontrol data by odds ratios (ORs) with 95% confidence intervals (CIs) using logistic regression. Crude (observed) incidence rates and risks were calculated fromcohort data.

Measurements and Main Results: The observed (unadjusted) incidence rate over 12weekswas 2.8 (95%CI, 2.3-3.3) per 1,000 person- weeks, 4.5 (3.5-5.4) for gefitinib versus 1.7 (1.2-2.2) for chemotherapy; the corresponding observed naive cumulative incidence rates at the end of 12-week follow-up were 4.0% (3.0-5.1%) and 2.1% (1.5-2.9%), respectively. Adjusted for imbalances in risk factors between treatments, the overallOR for gefitinib versus chemotherapywas 3.2 (1.9-5.4), elevatedchieflyduringthe first4weeks (3.8 [1.9-7.7]).Other ILD risk factors in both groups included the following: older age, poor World Health Organization performance status, smoking, recent NSCLC diagnosis, reduced normal lung on computed tomography scan, preexisting chronic ILD, concurrent cardiac disease. ILD-related deaths in patients with ILD were 31.6% (gefitinib) versus 27.9% (chemotherapy); adjusted OR, 1.05 (95%CI, 0.3-3.2).

Conclusions: ILD was relatively common in these Japanese patients with NSCLC during therapy with gefitinib or chemotherapy, being higher in the older, smoking patient with preexisting ILD or poor performance status. The risk of developing ILD was higher with gefitinib than chemotherapy,mainly in the first 4 weeks.

Keywords: non-small cell lung cancer; interstitial lung disease; Japanese patients; gefitinib, chemotherapy

Epidermal growth factor receptor (EGFR) tyrosine kinase inhibitors are a well-established therapy for the treatment of non- small cell lung cancer (NSCLC) in many countries. They are generally well tolerated and not typically associated with the cytotoxic side effects commonly seen with chemotherapy.

The EGFR tyrosine kinase inhibitor gefitinib (IRESSA; Astra- Zeneca, London, U.K.) was first approved for the treatment of advanced NSCLC in Japan in July 2002. In clinical trials and in preapproval compassionate clinical use, some reports of interstitial lung disease (ILD)-type events had been observed. As the drugwas made more widely available in Japan after approval, however, an increasing number of spontaneous reports for ILD appeared.

ILD is a disease that affects the parenchyma or alveolar region of the lungs (1). When associated with drug use, it can present precipitously with acute diffuse alveolar damage, which is fatal in some patients (2). Chest imaging shows ground-glass density and patients present with severe breathlessness. There is no specific treatment, but supportive therapy including oxygen, corticosteroids, or assisted ventilation is indicated. Acute exacerbations of ILD have previously been considered relatively rare in many settings, with Japan as a notable exception (3), but recent studies of patients with idiopathic pulmonary fibrosis (IPF) have challenged this and underlined this important risk (4).

ILD, especially IPF, is a known comorbidity in patients with NSCLC and has also been associated with many other lung cancer therapies (5). Rates of acute ILD events up to and exceeding 10% have been reported in patients receiving chemotherapy and radiotherapy (6-11). It is recognized that ILD is more common in Japan than elsewhere (5, 6, 12, 13).

When safety reports of acute ILD-type events in gefitinibtreated patients appeared in Japan, there was limited knowledge about ILD in patients with NSCLC. There was a need to better understand baseline incidence on different treatments, risk factors for developing ILD, and whether gefitinib might be associated with increased risk of ILD, or if patient selection or other aspects were involved. A pharmacoepidemiologic study was designed and conducted by an independent academic team together with scientists from AstraZeneca to define the risk and increase understanding of ILD in Japanese patients with NSCLC. Some of the results of this study have been previously reported in the form of conference abstracts (14, 15).

METHODS

See also the online supplement for further details on methods.

Overall Study Design

A nonrandomized cohort study with a nested case-control study component was conducted between November 12, 2003, and February 22, 2006, in 50 centers across Japan. Patients with advanced or recurring NSCLC who had received at least one chemotherapy regimen were eligible for cohort entry. Patients and their physicians selected the most appropriate treatment (gefitinib 250 mg or chemotherapy) and the patients were followed for up to 12 weeks after treatment initiation. Basic data were collected at the start of follow-up and included sex, age, World Health Organization (WHO) performance status (PS), and tumor histology. If a patient switched to a new treatment, he or she could be re-enrolled for a new treatment period of 12 weeks, provided he or she was still eligible.

Patients who developed acute ILD events during the follow-up were registered to the case-control study nested within the cohort as clinically diagnosed potential cases. For each potential case, four patients who had not yet developed ILD were randomly selected as appropriate control subjects from patients registered to the cohort at that time, and extensive clinical and demographic risk factor data were collected on cases and control subjects (see Figure E1 in the online supplement).

The study followed Good Clinical Practice procedures. An independent external epidemiology advisory board provided advice on design, conduct, and analysis of the study.

Diagnosis of ILD

To ensure an accurate diagnosis of ILD, several study design components were implemented: (1) an information card to all cohort patients, alerting them to the symptoms of ILD; (2) internationally agreed criteria for the diagnosis of ILD and a diagnostic algorithm (see Figure E2) developed from the American Thoracic Society/ European Respiratory Society consensus statement (1); and (3) a blinded diagnostic review of all clinically diagnosed potential ILD cases registered to the study by an independent case review board (CRB) of radiologists and clinicians.

Evaluation of Preexisting Lung Conditions

The CRB also blindly evaluated pretreatment computed tomography (CT) scans for the presence of a number of pulmonary conditions: preexisting (chronic) ILD (mainly IPF), drug-induced lung disease, pulmonary emphysema, radiation pneumonitis, lymphangitis carcinomatosis, and healed tuberculosis, and evaluated the extent of normal lung, as well as the extent of areas adherent to pleura.

Detailed Data Collection

For cases and control subjects, detailed data on NSCLC treatment, demography, cancer histology, clinical stage and the presence of metastases, WHO PS, smoking, previous cancer treatments, past and current medical history, surgical history, and concomitant medication and therapy were collected. Data on serious adverse events (SAEs) and hence all-cause mortality were collected for the gefitinib-treated patients in the cohort only; thus, information on mortality from causes other than ILD in chemotherapy-treated patients is not available from this study.

Statistical Analysis

From cohort data, we estimated observed person-time incidence rates as well as two measures of the observed “risk” of acute ILD to a patient; a naive estimate of observed cumulative incidence (incidence proportion, “frequency”), and risk up to 84 days by the Kaplan-Meier method.

Control subjects for the nested case-control study were sampled using incidence density sampling, and consequently the odds ratio (OR) obtained from the case-control analysis estimates the study incidence rate ratio (and approximately estimates the risk ratio) (16).

For the case-control statistical analysis, it was initially verified that the convenience matching for calendar time implicit in the risk set control sampling could be disregarded. In tabular analyses, we then identified potential confounders and risk factors, using as selection criteria a 10% change in the OR estimate for gefitinib versus chemotherapy treatment when stratifying for each factor separately, and a risk factor crude OR of less than 0.5 or more than 2.0, respectively. We also identified potential interactions between treatment and other risk factors, or between two potential risk factors. Modeling using logistic regression then proceeded in the corresponding four steps. Few previous data were available on risk factors for ILD in patients with NSCLC and so a hypothesis-free stepwise process with loose P value criteria (P

ILD-related mortality among the patients who developed acute ILD on gefitinib or chemotherapy treatment was obtained. Modeling of risk factors for ILD-related mortality followed a similar process to the ILD risk factor modeling. For gefitinib-treated patients, two additional data items were available: total all-cause mortality, which was analyzed by the Kaplan-Meier method, and SAEs, for which frequencies and possible consequences in terms of treatment discontinuation and death were calculated.

RESULTS

Cohort Subjects and Treatments

Cohort participation rates were high. In 10 sampled study centers, 89.6% of eligible patients were enrolled to the cohort. The number of treatment periods and subjects are summarized in Table 1. In total, 4,423 treatment periods in 3,159 subjects were available for analysis. In the cohort, 70.8% of patients had only one treatment period, 21.5% had two periods, and the remaining 7.8% of patients had three or more treatment periods registered (Table 1). Chemotherapy included a wide range of treatments, the most common being taxane monotherapy, followed by taxane+platinum and gemcitabine1vinorelbine combinations.

Cases and Control Subjects

In the overall cohort data of all treatment periods, clinicians reported 155 suspected cases of acute ILD during the follow-up, of which 122 were confirmed by the CRB after blinded review of CT and clinical data-79 of 103 gefitinib-treated (76.7%) and 43 of 52 chemotherapy-treated (82.7%) subjects. A total of 574 eligible control subjects were sampled from the person-time of the cohort. Almost all ILD cases and selected control subjects consented to participate in the nested case-control study, with final participation rates of 98.1 and 92.0%, respectively. Valid data from the CRB review of CT scans were available for 115 cases and 520 control subjects.

Descriptive Data

On data items available for the full cohort (sex, age, WHO PS, and tumor histology), the control subjects were quite representative of the overall cohort (details not shown). Comparisons of the gefitinib- and chemotherapy-treated control groups as representative of the cohort indicated that the former included more women, never- smokers, adenocarcinoma tumors, and poorer PS, as well as less preexisting ILD and pulmonary emphysema on CT scan (Tables 2 and 3). ILD cases, regardless of treatment, were more likely than cohort control subjects to be older, male, smokers, with squamous cell carcinoma histology, and have poor PS (Tables 2 and 3). The frequency of preexisting ILD and pulmonary emphysema was higher in cases, reflected also in a lower extent of normal lung on CT scan.

Cohort Analysis of ILD Occurrence

The observed incidence rate of acute ILD over the entire 12-week follow-up in the overall cohort was 2.8 per 1,000 personweeks-4.5 in the gefitinib-treated and 1.7 in the chemotherapy subcohort (Table 4). The observed incidence in the gefitinibtreated subcohort was highest in the first 4 weeks after starting treatment, greater than in the chemotherapy-treated subcohort. In the following two 4-week periods, the incidence was lower with no clear difference (Table 4, Figure 1A). The naive cumulative incidence of ILD at 84 days (i.e., observed frequency or proportion of the original cohort that developed ILD in the study) for patients in their first study treatment period was 4.0 and 2.1% for gefitinib- and chemotherapy- treated patients, respectively (Table 4), whereas the estimated theoretical 12-week risk of ILD (i.e., taking competing causes of death and loss to follow-up into consideration; Kaplan-Meier method) was 4.5 and 2.4%, respectively (Table 4, Figure 1B). Thus, the observed cohort rates and risks suggested an association of increased ILD occurrence with gefitinib treatment mainly in the first 4 weeks after treatment initiation. All cohort estimates are unadjusted for imbalances between treatments in other risk factors. Detailed comparisons between the treatments therefore used the adjusted case-control OR (as an estimate of the adjusted incidence rate ratio) to achieve comparability.

Case-Control Analysis of ILD Occurrence and Risk Factors

Major results. The OR of developing acute ILD with gefitinib treatment versus chemotherapy, adjusted for the full predictor model of major confounders together with additional identified important risk factors and interactions, was 3.2 (95% confidence interval [CI], 1.9-5.4) (Table 5). Several risk factors aside from treatment also had strong effects, including WHO PS, as well as smoking status and preexisting ILD together with the extent of normal lung on CT scan, which interacted in a complex way in the model (Table 5, Figure 2). Preexisting ILD was confirmed as a strong risk factor, with OR point estimates ranging from 4.8 to 25.3 depending on the extent of remaining normal lung on CT scan, in comparison with patients without preexisting ILD and high extent of normal lung on CT scan (Table 5). The full set of ILD risk factors in both groups from the final model thus included older age (>/=55 yr), WHO PS (>/ =2), smoking, short duration since NSCLC diagnosis (

When the case-control analyses focused on the first 4 weeks after treatment initiation (because the unadjusted cohort analyses above indicated that the bulk of the association with gefitinib appeared to be for this time interval) the estimated OR adjusted for a full predictor model developed on this period’s data was 3.8 (95% CI, 1.9- 7.7). The same model produced an OR for Weeks 5-8 of 1.6 (95% CI, 0.5-4.8), whereas the final 4-week period had too few cases for an adequate estimate. The estimate for Weeks 5-12 combined, using this same model, was 2.5 (95%CI, 1.1-5.8). The important covariates and predictors were the same in this model as in the model for the full 12-week data, with the exception of age, preexisting cardiac disease, and preexisting pulmonary emphysema, which were not included. Due to sparse data beyond 4 weeks, independent models for Weeks 5-8, 9-12, and 5-12 could not be developed.

Confounding and sensitivity analysis. In the overall 12-week basic analysis, moderately strong confounding by other risk factors was found. The crude OR of developing ILD with gefitinib treatment versus chemotherapy was 2.3 (95% CI, 1.5-3.6). When adjusted for some of the most important potential confounders one at a time, the adjusted OR point estimate for the association of treatment with ILD occurrence ranged from 2.1 to 3.1 (see Table E1 for details). The most important confounder was severity of preexisting ILD with strong negative confounding, and the only one that resulted in a lower adjusted OR than 2.3 (positive confounding) was WHO PS.

The propensity score analysis approach identified the following variables as the most important predictors of selecting gefitinib treatment in this study: female sex; nonsmoking status; non- squamous tumor histology; poor PS; preexisting lymphangitis carcinomatosis; no previous gefitinib treatment; and no preexisting ILD, emphysema, or radiation pneumonitis. The estimated OR of developing ILD for gefitinib treatment when stratifying by the propensity score was 3.3 (95% CI, 1.9-5.5), very similar to the primary result, suggesting that the primary regression modeling approach well captured the confounding in the data.

If some misclassification of ILD diagnosis remains despite the design features aimed to minimize it, the adjusted OR point estimate of 3.2 may apart from random variation be subject to systematic bias. A sensitivity analysis to evaluate the possible magnitude of such bias due to misclassification of ILD diagnosis suggested that the true study point estimate for the adjusted OR would be expected to lie between 2.6 and 4.8, assuming diagnostic sensitivity of more than 80% for both gefitinib- and chemotherapy-treated patients, and specificity of more than 99.0% for gefitinib and more than 99.5% for chemotherapy. Lower values for sensitivity/specificity were considered very unlikely for this serious condition in a cancer patient population, in this study setting.

Analysis of ILD Mortality

Mortality due to ILD among gefitinib- or chemotherapy-treated patients. The mortality due to ILD for the patients who developed acute ILD was 31.6% (95% CI, 21.6-43.1) among gefitinib-treated patients and 27.9% (95% CI, 15.3-43.7) among those with other treatments; the OR was 1.05 (95% CI, 0.3-3.2) for gefitinib versus chemotherapy, adjusted for relevant risk factors. Several other factors were strong predictors of a fatal outcome for patients with ILD, including age of 65 years or older, smoking history, preexisting ILD, CT scan evidence of reduced normal lung (=50%), and/or extensive areas adherent to pleura (>/=50%), with ORs ranging from 2.4 to 11.7 (see Table E2).

Overall mortality among gefitinib-treated patients. In the gefitinib-treated cohort in whom such data were available, an analysis of mortality from all causes by the Kaplan-Meier method showed that cumulative mortality at 12 weeks among the patients who did develop ILD was 58.7%, compared with 14.6% (95% CI, 12.8-16.3) among the large majority who did not develop ILD (Figure 3). For the entire gefitinib cohort, including the subjects who developed ILD, the observed cumulative mortality was 16.0% (95% CI, 14.3-17.8), so that the increased mortality in ILD cases impacted the total survival rate at 12 weeks in the overall gefitinib-treated cohort only to a limited extent, reducing survival from 85.4 to 84.0%. SAEs among Gefitinib-treated Patients

SAEs were only collected for gefitinib-treated patients in the cohort, and a total of 198 patient registrations reported SAEs (10.5%), of which 38 (2.0%) reported SAEs resulting in a fatal outcome. Within this group, there were 142 patient registrations with drug-related (as reported by the physicians) SAEs (7.5%), of which 30 (1.6%) resulted in a fatal outcome. The majority of these (25 out of 30) were due to ILD-type events. There were 122 patient registrations where study treatment was discontinued due to the reported SAEs (6.5%). SAEs seen in the gefitinib-treated patients were generally consistent with the known safety profile of gefitinib and/or the patient’s underlying disease and comorbidities.

DISCUSSION

This study provides important information on ILD in an advanced/ recurrent NSCLC setting in Japanese patients in Japan, and it is the largest prospective study of this condition to date. For the first time, the risk of acute ILD events for a large and relatively unselected chemotherapy-treated NSCLC patient cohort in Japan was determined in clinical practice. The study also quantified the greater risk of developing acute ILD associated with gefitinib treatment than with conventional chemotherapy, mainly in the first 4 weeks after treatment initiation. The study confirmed and further defined risk factors for developing ILD with gefitinib or chemotherapy. The factors included older age, poor WHO PS, smoking, short duration since diagnosis of NSCLC, reduced normal lung on CT scan, preexisting ILD, and concurrent cardiac disease. Several of these factors, or related factors, had been reported previously in bivariate or multivariate analyses from other studies (8, 18, 19). These risk factors were the same for patients treated with gefitinib or chemotherapy in the study, and no treatment-specific risk factors were identified. In particular, patients with CT evidence of preexisting ILD (chronic) were at considerably elevated risk of developing acute ILD during treatment, but there were relatively few subjects with preexisting ILD and the data did not indicate a statistically significant difference in treatment-related risk depending on the preexisting ILD status. Of clinical relevance, some of these risk factors were just as strong as, or stronger than, gefitinib treatment, for example having a poor WHO PS (=2) rather than a good PS (OR, 4.0; 95% CI, 1.85-8.75), implying that they can be used to identify patients at particular risk of ILD in clinical practice. The relationship between ILD and pharmacokinetic characteristics of gefitinib, as well as genetic polymorphisms and proteomics determined in study subjects, were also investigated as secondary and exploratory objectives in this study. These analyses are ongoing and results will be submitted for publication in due course.

Over the whole study follow-up, the average incidence rate for acute ILD events in patients treated with gefitinib was 3.2-fold higher relative to that seen with other chemotherapy treatments, adjusted for imbalances in other risk factors between treatments. The increased risk of ILD associated with gefitinib treatment was seen most clearly in the first 4 weeks after treatment initiation. Thus, increased physician awareness of risk factors and careful surveillance of Japanese patients during this period are indicated to manage risk. Such an approach is in line with current recommendations in Japan (20, 21). Beyond 4 weeks after treatment initiation, the risk of ILD associated with gefitinib treatment appears to fall.

ILD risk factors were found to be the same for both types of NSCLC therapy. Gefitinib is, however, a molecularly targeted agent. There is a significant body of evidence to indicate that gefitinib is a valid treatment option for some patients with NSCLC. In the IRESSA Survival Evaluation in Lung cancer (ISEL) study, a large phase III, placebo-controlled trial (n = 1,692), gefitinib was associated with some improvement in overall survival versus placebo, although this failed to reach statistical significance in the primary analysis of the overall population (22). Preplanned subgroup analyses from the study showed statistically significant differences in survival in favor of gefitinib in patients of Asian origin and those who had never smoked. Furthermore, tumor biomarker data suggest that patients with a high EGFR gene copy number, or an EGFR mutation, may be more likely to benefit (23, 24).

Therefore, the consideration of those patients more likely to benefit from the drug balanced with the better identification of these risk factors associated with ILD enables the physician to make careful judgment of the most appropriate therapy for the individual patient. Patients with several risk factors will generally be at more risk, and patients with risk factors may be at higher risk if gefitinib is used. This approach is facilitated by the fact that evidence to date suggests that subgroups less at risk of ILD tend to be those that respond well to gefitinib treatment (8).

A fatal outcome is the major concern with ILD as an SAE of drug treatment. In other large studies, fatality rates due to ILD in gefitinib-treated subjects of approximately 30% have been seen (8, 25), and a similar mortality was observed in this study in both gefitinib-treated and chemotherapy-treated ILD cases. The main predictors of a fatal outcome were older age (>/=65 yr), smoking history, and preexisting ILD, as well as CT scan evidence of reduced normal lung (=50%) or extensive areas adherent to pleura (>/ =50%). Because mortality is high among patients with NSCLC and the frequency of ILD in Japanese patients with NSCLC is low in comparison, ILD-related mortality impacted the overall survival at 12 weeks, for the cohort of gefitinib-treated patients, only to a limited extent (85.4 to 84%). Accordingly, there needs to be an appropriate individualized risk-benefit evaluation for patients also considering other treatments, many of which have their own problems with treatmentrelated mortality due to SAEs other than ILD.

Some methodologic issues may have influenced the study results and deserve comment. This kind of observational pharmacoepidemiologic study is generally considered sensitive to confounding by indication. Most often, it is assumed that more “sick” or “susceptible” patients will receive a new treatment, leading to possibly more adverse effects in this group, even in the absence of a true relationship to treatment. Attempts to adjust for confounding using collected data would then push the adjusted estimate of effect closer to the null, but if sufficiently precise information on strong confounders cannot be collected, it may be impossible to remove all of the confounding. In conducting this study, the suspected adverse effect of ILD was recognized, and in the clinical setting, recommendations were in place to proceed with caution when treating some patients with suspected elevated baseline risk of ILD. This kind of selection would tend to produce the type of data pattern that was in fact observed in this study, a pattern of negative confounding that produces a more elevated OR when adjustment for confounders is performed. Thus, the results are well in line with what might be expected.

Misdiagnosis of ILD (outcome misclassification) is another concern, but it is expected that the stringent design features have minimized this problem in the present study (see online supplement for details). The diagnostic CRB review is a key feature, but it was still CT based, and biopsies-generally considered the gold standard for ILD diagnosis-were in most cases not taken. Overall, a sensitivity analysis suggested that, under reasonable assumptions about possible misclassification of ILD, the main result would remain similar and the conclusions from the study would not be greatly changed.

Random error is another consideration. However, although random error may be responsible for some bias in the point estimate, the confidence interval is reasonably narrow. The results are also consistent with other recent data. For example, as of January 2006, the estimated reporting rate of ILD-type events in Japan from the AstraZeneca Global Drug Safety Database of patients receiving gefitinib treatment was approximately 3.1% (26); from a retrospective study by the West Japan Thoracic Oncology Group (WJTOG), which studied 1,719 patients receiving gefitinib of whom 69 developed ILD, the frequency was 3.5% (95% CI, 2.8-4.5%) (8); from a postmarketing surveillance (PMS) study conducted by AstraZeneca KK Japan, which included 3,322 gefitinib-treated patients, it was 5.8% (25); whereas from the present study, the cumulative incidence at 12 weeks was 4.0% (95% CI, 3.0-5.1%).

These estimates are quite similar, even recognizing that the populations and selection of patients differ between these samples, and duration of follow-up, although similar, varies. In the present study, for the first time, an estimate of cumulative incidence of ILD after 12 weeks of treatment was obtained also from a chemotherapy-treated patient group; this frequency was 2.1% (95% CI, 1.5-2.9%), providing an estimate of this problem unrelated to gefitinib in patients with NSCLC in Japan. The prognosis for gefitinib-treated patients who were diagnosed with ILD was also quite consistent with other studies. In the PMS study, ILD-related death among patients diagnosed with ILD was 38.6% (25); in theWJTOGstudy it was 44.3% (8); in the AstraZeneca Global Drug Safety Database as of January 2006, the proportion of ILD-type events with a fatal outcome in patients receiving treatment with gefitinib in Japan was 37.3% (AstraZeneca, data on file); and in the present study it was 31.6%. This proportion was quite similar to the chemotherapy-treated group, 27.9% (adjusted OR, 1.05; 95% CI, 0.4- 3.2).

The factors associated with risk of acute ILD observed in this Japanese NSCLC population are largely different or even complementary to factors that predict better response to gefitinib. This would seem to support a hypothesis that the mechanism by which ILD occurs is distinct from the successful cancer response mechanism, offering a potential path toward selecting patients with optimal risk-benefit balance for gefitinib treatment.

Interestingly, the issue of ILD in patients with NSCLC, after gefitinib or other treatments, appears to be a problem largely limited to Japan. From the AstraZeneca Global Drug Safety Database, the reporting rate of ILD-type events in patients receiving treatment with gefitinib was only 0.23% in the rest of the world excluding Japan, based on more than 215,000 patients worldwide estimated to have been exposed to gefitinib (26). Even for neighboring countries, the pattern differs from Japan: the rate for East Asian countries, including Korea and Taiwan but excluding Japan, was 0.17% (26). The proportion of ILDtype events with a fatal outcome was similar, however: 37% in Japan and 31% in the rest of the world. The reasons for this difference in incidence of ILD between Japan and other countries remain unclear, but may relate to both constitutional and environmental factors specific to Japan or Japanese patients. For other drug treatments, too, a higher incidence of ILD has been noted in Japan than elsewhere (12, 13).

Within the study, some exploratory analyses are still ongoing related to genetic and proteomic predictors for ILD in patients with NSCLC, to search for biomarkers for early recognition of ILD and hopefully individualized risk assessment. This may help to shed light on why ILD appears to be a particular issue for Japanese patients and the possible underlying mechanisms.

The EGFR is expressed on a number of constituent cells of the lungs including epithelium, smooth muscle cells, fibroblasts, and endothelium (27). There have been a number of animal studies using bleomycin- and vanadium pentoxide-induced lung injury with EGFR- tyrosine kinase inhibitors to determine the role of EGFR in lung fibrosis. Gefitinib and AG1478 have been used in such studies of mice and, when administered in a range of therapeutic doses, show clear attenuation of both bleomycin- (28) and vanadium pentoxide- induced (29) lung fibrosis, although one study (30) has shown augmentation of bleomycininduced fibrosis (when using a subtoxic dose of gefitinib). The similarity of study design and choice of animal strain in the bleomycin studies make it difficult to explain the discrepant results other than by the excessive dosing. This leaves uncertainty as to the underlying mechanism of lung fibrosis in patients with NSCLC receiving gefitinib.

In summary, the study appears to be of adequate validity to avoid serious systematic biases, random error does not seem to be the most likely explanation for the results, and the observed increased risk of ILD with gefitinib treatment relative to chemotherapy treatment in Japanese patients is consistent with previous studies. Although preexisting ILD was confirmed as an important determinant of developing acute ILD symptoms after treatment with gefitinib or chemotherapy, the results also suggested that risk of ILD may be generally affected by a variety of other factors that decrease the amount of normally functioning lung tissue or affect the capability of tissue repair and recovery. The study thus identified several risk factors apart from treatment, which included preexisting ILD, which were not treatment specific, and which were partly similar to risk factors for idiopathic or rheumatic pulmonary lung fibrosis. These findings taken together suggest that there may be a common etiology that gives some patients a greater susceptibility both to idiopathic or rheumatic pulmonary fibrosis and to acute drug- induced lung injury after various treatments.

Conflict of Interest Statement: S.K. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. H.K. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. Y.N. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. M.F. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. K.N. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. Y.I. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. M.T. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. S.Y. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. K.N. does not have a financial relationship with a commercial entity that has an interest in the subject of thismanuscript. M.S. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. The Japan Thoracic Radiology Group does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. H.J. has been an AstraZeneca employee since 2001. Y.I. has been an AstraZeneca employee since 2000. A.A. is a full-time employee of AstraZeneca. C.W. has been a full-time employee at AstraZeneca since 2001 until present and owns shares in the company. T.H. is a full-time R&D scientist at AstraZeneca, UK, and received stock options. F.N. is a full-time employee of AstraZeneca and owns shares in the company.

Acknowledgment: The authors thank the study monitors, nurses, data managers, other support staff, and the patients participating in the study; the external epidemiology advisory board (Kenneth J. Rothman, Jonathan M. Samet, Toshiro Takezaki, Kotaro Ozasa, Masahiko Ando) for their advice and scientific review of study design, conduct, and analysis; Professor Nestor Muller for his expert input into radiologic aspects of ILD diagnosis; all case review board members (including Japan Thoracic Radiology Group members [with *]) individually (Moritaka Suga, Takeshi Johkoh*, Masashi Takahashi*, Yoshiharu Ohno*, Sonoko Nagai, Yoshio Taguchi, Yoshikazu Inoue, Takashi Yana, Masahiko Kusumoto*, Hiroaki Arakawa*, Akinobu Yoshimura, Makoto Nishio, Yuichiro Ohe, Kunihiko Yoshimura, Hiroki Takahashi, Yukihiko Sugiyama, Masahito Ebina, and Fumikazu Sakai*) for their valuable work in blindly reviewing ILD diagnoses, as well as prestudy CT scans for preexisting comorbidities; and all hospitals and clinical investigators who contributed to the data collection in the study (see below). The authors thank Sarah Charlesworth, from Complete Medical Communications, who provided editing assistance funded by AstraZeneca.

Hospitals and principal investigators contributing to the study: National Hospital Organization Hokkaido Cancer Centre (Hiroshi Isobe), Hokkaido University Hospital (Koichi Yamazaki), National Hospital Organization Dohoku National Hospital (Yuka Fujita), Tohoku University Hospital (Akira Inoue), Sendai Kousei Hospital (Shunichi Sugawara), National Cancer Centre Hospital East (Yutaka Nishiwaki), Nippon Medical School Chiba Hokusoh Hospital (Yasushi Ono), Tokyo Medical University Hospital (Masahiro Tsuboi), Nippon Medical School Hospital (Tetsuya Okano), Toho University Omori Medical Centre (Nobuyuki Hamanaka), Toranomon Hospital (Kunihiko Yoshimura), National Hospital Organization Tokyo Hospital (Atsuhisa Tamura), Juntendo University Hospital (Kazuhisa Takahashi), Kyorin University Hospital (Tomoyuki Goya), Tokai University Hospital (Kenji Eguchi), Kitasato University School of Medicine (Noriyuki Masuda), Kanagawa Cardiovascular and Respiratory Centre (Takashi Ogura), Niigata Cancer Centre Hospital (Akira Yokoyama), National Nishi-Niigata Central Hospital (Hiromi Miyao), Toyama University Hospital (Muneharu Maruyama), Kanazawa University Hospital (Kazuo Kasahara), Aichi Hospital, Aichi Cancer Centre (Hiroshi Saito), National Hospital Organization NagoyaMedical Centre (Hideo Saka), Fujita Health University Hospital (Hiroki Sakakibara), Nagoya Ekisaikai Hospital (Masashi Yamamoto), Shiga University of Medical Science Hospital (Noriaki Tezuka), Kyoto Katsura Hospital (Takeshi Hanawa), National Hospital Organization Kyoto Medical Centre (Yoshiyuki Sasaki), Rinku General Medical Centre Municipal Izumisano Hospital (Hisao Uejima), Kinki University, School of Medicine (Kazuhiko Nakagawa), National Hospital Organization Kinki-chuo Chest Medical Centre (Masaaki Kawahara), Osaka City General Hospital (Koji Takeda), Osaka City General Hospital (Hirohito Tada), Osaka City University Hospital (Shinzoh Kudoh), Osaka Prefectural Medical Centre for Respiratory and Allergic Diseases (Kaoru Matsui), Osaka Police Hospital (Kiyoshi Komuta), Toneyama National Hospital (Soichiro Yokota), Kobe City General Hospital (Keisuke Tomii), Hyogo Medical Centre for Adults (Shunichi Negoro), Kobe University Hospital (Yoshihiro Nishimura), Institute of Biomedical Research and Innovation (Nobuyuki Katakami), Tenri Hospital (Yoshio Taguchi), Okayama University Medical and Dental School Hospital (Katsuyuki Kiura), Hiroshima City Hospital (Hidetaka Sumiyoshi), Hiroshima City Hospital (Noritomo Senoo), National Hospital Organization Shikoku Cancer Centre (Tetsu Shinkai), National Hospital Organization Kyushu Cancer Centre (Yukito Ichinose), Fukuoka National Hospital (Akira Motohiro), University of Occupational and Environmental Health (Masamitsu Kido), University of Occupational and Environmental Health (Kenji Sugio), National Hospital Organization Nagasaki Medical Centre (Akitoshi Kinoshita), Kumamoto University Hospital (Mitsuhiro Matsumoto), Kumamoto-Chuo Hospital (Sunao Ushijima), Okinawa National Hospital (Mutsuo Kuba). AT A GLANCE COMMENTARY

Scientific Knowledge on the Subject

Acute interstitial lung disease (ILD) occurs in Japanese patients with non-small cell lung cancer (NSCLC) receiving gefitinib. There is, however, limited knowledge about risk factors for ILD and the incidence of ILD in patients with NSCLC receiving other treatments.

What This Study Adds to the Field

Acute ILD was common in Japanese patients with NSCLC receiving chemotherapy or gefitinib, with higher risk for gefitinib. Age, performance status, smoking, and preexisting chronic ILD were also important risk factors, aiding clinicians in treatment selection.

References

1. American Thoracic Society. American Thoracic Society/European Respiratory Society international multidisciplinary consensus classification of the idiopathic interstitial pneumonias. Am J Respir Crit Care Med 2002;165:277-304.

2. Inoue A, Saijo Y, Maemondo M, Gomi K, Tokue Y, Kimura Y, Ebina M, Kikuchi T, Moriya T, Nukiwa T. Severe acute interstitial pneumonia and gefitinib. Lancet 2003;361:137-139.

3. Kondoh Y, Taniguchi H, Kawabata Y, Yokoi T, Suzuki K, Takagi K. Acute exacerbation in idiopathic pulmonary fibrosis: analysis of clinical and pathologic findings in three cases. Chest 1993;103:1808- 1812.

4. Wells AU, Hogaboam CM. Update in diffuse parenchymal lung disease 2006. Am J Respir Crit Care Med 2007;175:655-660.

5. Raghu G, Nyberg F, Morgan G. The epidemiology of interstitial lung disease and its association with lung cancer. Br J Cancer 2004;91:S3-S10.

6. Kudoh S, Takeda K, Nakagawa K, Takada M, Katakami N, Matsui K, Shinkai T, Sawa T, Goto I, Semba H, et al. Phase III study of docetaxel compared with vinorelbine in elderly patients with advanced nonsmall-cell lung cancer: results of the West Japan Thoracic Oncology Group Trial (WJTOG 9904). J Clin Oncol 2006;24:3657-3663.

7. Abid SH, Malhotra V, Perry MC. Radiation-induced and chemotherapyinduced pulmonary injury. Curr Opin Oncol 2001;13:242- 248.

8. Ando M, Okamoto I, Yamamoto N, Takeda K, Tamura K, Seto T, Ariyoshi Y, Fukuoka M. Predictive factors for interstitial lung disease, antitumor response, and survival in non-small-cell lung cancer patients treated with gefitinib. J Clin Oncol 2006;24:2549- 2556.

9. Danson S, Blackhall F, Hulse P, Ranson M. Interstitial lung disease in lung cancer: separating disease progression from treatment effects. Drug Saf 2005;28:103-113.

10. Rossi SE, Erasmus JJ, McAdams HP, Sporn TA, Goodman PC. Pulmonary drug toxicity: radiologic and pathologic manifestations. Radiographics 2000;20:1245-1259.

11. SandlerAB,Nemunaitis J,DenhamC, vonPawel J,CormierY,Gatzemeier U,Mattson K,Manegold C, PalmerMC,GregorA, et al.Phase III trial of gemcitabine plus cisplatin versus cisplatin alone in patients with locally advanced or metastatic non-small- cell lung cancer. J ClinOncol 2000;18: 122-130.

12. Azuma A, Kudoh S. High prevalence of drug-induced pneumonia in Japan. Japan Medical Association Journal 2007;50:405-411.

13. Koo L, Clark J, Quesenberry CP, Higenbottam T, Nyberg F, Wolf M, SteinbergM, Forsythe B. National differences in reporting “pneumonia” and “pneumonia interstitial”: an analysis of the WHO drug monitoring database on 15 drugs in nine countries for seven pulmonary conditions. Pharmacoepidemiol Drug Saf 2005;14:775-787.

14. Nyberg F, Hada S, Rothman KJ; Iressa CCS Collaborator Group. IRESSA and interstitial lung disease (ILD) in Japan: lessons from a large nested case-control study to evaluate a safety issue [abstract]. Pharmacoepidemiol Drug Saf 2006;15:S287.

15. Kudoh S, Kato H, Nishiwaki Y, Fukuoka M, Nakata K, Suga M, Jiang H, Itoh Y, Higenbottam T, Nyberg F; Japan Thoracic Radiology Group. A cohort and nested case-control study to quantify the risk of interstitial lung disease (ILD) in Japanese patients with NSCLC treated with gefitinib or chemotherapy [abstract]. Am J Respir Crit Care Med 2007;175:A148

16. Pearce N. What does the odds ratio estimate in a case- control study? Int J Epidemiol 1993;22:1189-1192.

17. Brookhart MA, Schneeweiss S, Rothman KJ, Glynn RJ, Avorn J, Sturmer T. Variable selection for propensity score models. Am J Epidemiol 2006;163:1149-1156.

18. Hotta K, Kiura K, Tabata M, Harita S, Gemba K, Yonei T, Bessho A, Maeda T, Moritaka T, Shibayama T, et al. Interstitial lung disease in Japanese patients with non-small cell lung cancer receiving gefitinib: an analysis of risk factors and treatment outcomes in Okayama Lung Cancer Study Group. Cancer J 2005;11:417- 424.

19. Takano T, Ohe Y, Kusumoto M, Tateishi U, Yamamoto S, Nokihara H, Yamamoto N, Sekine I, Kunitoh H, Tamura T, et al. Risk factors for interstitial lung disease and predictive factors for tumor response in patients with advanced non-small cell lung cancer treated with gefitinib. Lung Cancer 2004;45:93-104.

20. AstraZeneca KK Japan. Japanese patient information for IRESSA, version 17 [in Japanese]. Osaka, Japan: AstraZeneca KK Japan; October 2006.

21. Japan Lung Cancer Society Committee on Preparation of Guideline for Use of Gefitinib. Guideline for use of gefitinib. Chiba, Japan: The Society; 2005. In Japanese.

22. Chang A, Parikh P, Thongprasert S, Tan E-H, Perng R-P, Ganzon D, Yang C-H, Tsao C-J, Watkins C, Botwood N, et al. Gefitinib (IRESSA) in patients of Asian origin with refractory advanced nonsmall cell lung cancer: subset analysis from the ISEL study. J Thorac Oncol 2006;1:847-855.

23. Hirsch FR, Varella-Garcia M, Bunn Jr PA, Franklin WA, Dziadziuszko R, Thatcher N, Chang A, Parikh P, Rodrigues Pereira J, Ciuleanu T, et al. Molecular predictors of outcome with gefitinib in a phase III placebo-controlled study in advanced non-small-cell lung cancer. J Clin Oncol 2006;24:5034-5042.

24. Pao W, Miller VA. Epidermal growth factor receptor mutations, smallmolecule kinase inhibitors, and non-small-cell lung cancer: current knowledge and future directions. J Clin Oncol 2005;23:2556- 2568.

25. Yoshida S. The results of gefitinib prospective investigation [in Japanese]. Medicine and Drug Journal 2005;41:772-789.

26. Armour A. Gefitinib in advanced non-small cell lung cancer: clinical experience in patients ofAsian origin. Asia Pac JClinOncol 2007;3:66-78.

27. Modi S, Seidman AD. An update on epidermal growth factor receptor inhibitors. Curr Oncol Rep 2002;4:47-55.

28. Ishii Y, Fujimoto S, Fukuda T. Gefitinib prevents bleomycin- induced lung fibrosis in mice. Am J Respir Crit Care Med 2006;174:550-556.

29. Rice AB, Moomaw CR, Morgan DL, Bonner JC. Specific inhibitors of platelet-derived growth factor or epidermal growth factor receptor tyrosine kinase reduce pulmonary fibrosis in rats. Am J Pathol 1999; 155:213-221.

30. Suzuki H, Aoshiba K, Yokohori N, Nagai A. Epidermal growth factor receptor tyrosine kinase inhibition augments a murine model of pulmonary fibrosis. Cancer Res 2003;63:5054-5059.

Shoji Kudoh1, Harubumi Kato2, Yutaka Nishiwaki3, Masahiro Fukuoka4, Kouichiro Nakata5, Yukito Ichinose6, Masahiro Tsuboi2, Soichiro Yokota7, Kazuhiko Nakagawa4, Moritaka Suga8, Japan Thoracic Radiology Group9*, Haiyi Jiang10, Yohji Itoh10, Alison Armour11, Claire Watkins11, Tim Higenbottam12,13, and Fredrik Nyberg14,15

1Nippon Medical School, Tokyo, Japan; 2Tokyo Medical University Hospital, Tokyo, Japan; 3National Cancer Center Hospital East, Chiba, Japan; 4Kinki University School of Medicine, Osaka, Japan; 5Nakata Clinic, Tokyo, Japan; 6National Kyushu Cancer Center, Fukuoka, Japan; 7Toneyama National Hospital, Osaka, Japan; 8Saiseikai Kumamoto Hospital, Kumamoto, Japan; 9Japan Thoracic Radiology Group, Shiga, Japan; 10AstraZeneca KK, Osaka, Japan; 11AstraZeneca, Macclesfield, Cheshire, United Kingdom; 12AstraZeneca R&D Charnwood, Loughborough, United Kingdom; 13Sheffield University, Sheffield, United Kingdom; 14Epidemiology, AstraZeneca R&D Molndal, Molndal, Sweden; and 15Institute of Environmental Medicine, Karolinska Institute, Stockholm, Sweden

(Received in original form October 11, 2007; accepted in final form March 12, 2008)

Supported by AstraZeneca.

*A list of Japan Thoracic Radiology Group members may be found in the ACKNOWLEDGMENT.

Correspondence and requests for reprints should be addressed to Fredrik Nyberg, M.P.H., M.D., Ph.D., Epidemiology, AstraZeneca R&D Molndal, SE-413 83 Molndal, Sweden. E-mail: [email protected]

This article has an online supplement, which is accessible from this issue’s table of contents at www.atsjournals.org

Am J Respir Crit Care Med Vol 177. pp 1348-1357, 2008

Originally Published in Press as DOI: 10.1164/rccm.200710-1501OC on March 12, 2008

Internet address: www.atsjournals.org

Copyright American Thoracic Society Jun 15, 2008

Originally published by Kudoh, Shoji Kato, Harubumi; Nishiwaki, Yutaka; Fukuoka, Masahiro; Nakata, Kouichiro; Ichinose, Yukito; Tsuboi, Masahiro; Yokota, Soichiro; Nakagawa, Kazuhiko; Suga, Moritaka; Jiang, Haiyi; Itoh, Yohji; Armour, Alison; Watkins, Claire; Higenbottam, Tim; Nyberg, Fredrik.

(c) 2008 American Journal of Respiratory and Critical Care Medicine. Provided by ProQuest Information and Learning. All rights Reserved.