Causal Relationship Between Indium Compound Inhalation and Effects on the Lungs

Journal of J Occup Health 2009; 51: 513Ð521 Occupational Health

Causal Relationship between Indium Compound Inhalation and Effects on the Lungs

Makiko NAKANO1, Kazuyuki OMAE1, Akiyo TANAKA2, Miyuki HIRATA2, Takehiro MICHIKAWA1, Yuriko KIKUCHI1, Noriyuki YOSHIOKA1, Yuji NISHIWAKI1 and Tatsuya CHONAN3

1Department of Preventive Medicine and Public Health, School of Medicine, Keio University, 2Department of Hygiene, Graduate School of Medical Sciences, Kyushu University and 3Department of Medicine, Nikko Memorial Hospital, Japan

Abstract: Causal Relationship between Indium SP-D and SP-A, respectively. Conclusion: Dose- Compound Inhalation and Effects on the Lungs: dependent lung effects due to indium exposure were Makiko NAKANO, et al. Department of Preventive shown, and a decrease of indium exposure reduced Medicine and Public Health, School of Medicine, the lung effects. An In-S value of 3 ng/ml may be a Keio University—Background: Recent case reports cut-off value which could be used to prevent early and epidemiological studies suggest that inhalation of effects on the lungs. indium dust induces lung damage. Objectives: To (J Occup Health 2009; 51: 513–521) elucidate the dose-dependent effects of indium on the lungs and to prove a causal relationship more clearly. Key words: Cross-sectional study, Indium, Interstitial Methods: A baseline observation was conducted on pneumonitis, KL-6, HRCT, SP-D 465 workers currently exposed to indium, 127 workers formerly exposed to indium and 169 workers without Due to the rapid expansion of flat panel displays and indium exposure in 12 factories and 1 research solar cells, indium demand has increased every year, and laboratory from 2003 to 2006. Indium in serum (In-S) in 2007, Japan consumed over 90% of the indium in the was determined as an exposure parameter, and its world1). effects on the lungs were examined. Results: The Until the early 1990’s, there was little information on the means of In-S in the current, former and no exposure toxic effects of indium on humans or animals. In the mid workers were 8.35, 9.63 and 0.56 ng/ml, respectively. 1990’s, intra-tracheal instillation studies of particles of The current and former exposure workers had compound semiconductors such as indium phosphide, significantly higher levels of KL-6, and showed significant dose-dependent increases in KL-6, SP-D, indium arsenide and indium trichloride revealed that metal and SP-A. Current exposure workers with In-S of 3 compounds containing indium had a strong potential to 2Ð6) 7) ng/ml or above demonstrated a significant increase of induce severe lung damage . In 2003, Homma et al. KL-6 in both GM and prevalence exceeding the disclosed a case of interstitial pneumonia in a 27-yr-old male reference value. Approximately a quarter of the former worker engaged in the wet surface polishing process of exposure workers had interstitial changes as seen on indium-tin oxide (ITO) target plates, high-density plates chest HRCT. In-S of exposed workers who had been made of very hard ceramics composed of indium oxide and working before improvements of the working tin oxide, which are used to laminate transparent electro- environment (Group Bef) and those who started conductive thin-film on flat displays and solar cells. The working after improvements (Group Aft) were 12.29 and worker started this job in 1994, and at the beginning of 1998 0.81 ng/ml, respectively. Adjusted odds ratios indicated was admitted to a hospital with the complaint of increasing 87%, 71% and 44% reductions among Group Aft workers who exceeded the reference values of KL-6, dry cough and breathlessness. After a detailed medical examination, he was diagnosed with interstitial pneumonia, possibly caused by the inhalation of ITO particles. Up to Received May 25, 2009; Accepted Aug 26, 2009 the end of 2008, seven indium-induced cases of lung injury Published online in J-STAGE Oct 16, 2009 7Ð11) Correspondence to: M. Nakano, Department of Preventive were disclosed . After the first case was found, two Medicine and Public Health, School of Medicine, Keio University, epidemiological studies were performed in two different 12, 13) 35 Shinano-machi, Shinjuku, Tokyo 160-8582, Japan (e-mail: populations of indium-exposed workers . Both studies [email protected]) suggested strong relationships between indium in serum 514 J Occup Health, Vol. 51, 2009

Table 1. Characteristics of study population and concentrations of indium in serum

No exposure Former exposure Current exposure

n 169 127 465 Male (%) 81.9 91.1** 91.6** Age (yr): Mean ± SD 40.9 ± 10.89 39.1 ± 9.61 35.9 ± 11.30** Smokers (%) 116 (67.8) 76 (60.8) 278 (72.8) In-S (ng/ml): Mean (range ) 0.56 (n.d.Ð3.0) 9.63 (n.dÐ126.8)** 8.35 (n.d.Ð116.9)** Exposure duration (mo): Mean (range) Ð 60.4 (1Ð252) 55.1 (0.25Ð455) Duration after the end of indium exposure (mo): Mean (range) Ð 58.3 (2Ð201)

**: p<0.01 by Dunnett’s test (reference = no exposure workers) or χ2 test.

(In-S) and effects on the lungs. exposure workers. The mean length of indium exposure After the first cross-sectional study12), we continued to was 55.1 mo (range: 0.25Ð455) in the current exposure make efforts to contact companies manufacturing indium workers and 60.4 mo (1Ð252) in the former exposure products in order to inform the results of the health risk workers. The mean duration from the end of indium of indium compounds to their indium-exposed workers exposure in the former exposure workers was 58.3 mo and to ask the workers to join our epidemiological survey. (range 2Ð201). There were no significant differences in By the end of 2006, 13 factories and 1 research laboratory smoking histories among the three groups. Thirty-seven had agreed to join our epidemiological study. workers (14 current exposure, 10 former exposure, and 13 no exposure workers) had a history of asthma. Subjects and Methods Study subjects Exposure assessment Among the 13 factories that participated, 1 factory Most of the current and former exposure workers were studied by Chonan et al.13) and 4 factories studied by exposed to multiple indium compound dusts such as ITO, Hamaguchi et al.12) were included. Because the results indium trioxide, indium hydroxide, indium chloride, of a pre-analysis of the remaining factories and the indium nitrate, and indium metal. Because information participating laboratory were compatible with the results about the differences in the health effects of each indium of the two studies, we combined the data and analyzed compound is very limited, and the indium concentrations all of them together. We informed all study candidates at the work sites could not be measured, we did not stratify about the health risks of indium exposure and the purpose the current and former workers by the species of indium of the epidemiological study, and obtained their informed compounds. We selected In-S as a quantitative biological consent before they participated. The study was approved marker of indium exposure14). In-S was determined by by the Ethical Committee, School of Medicine, Keio inductively coupled plasma mass spectrometry (ICP-MS) University. A total of 889 subjects gave their informed at the Center of Advanced Instrumental Analysis, Kyushu consent. We excluded one factory (n=93) from the University and the Japan Industrial Safety and Health analysis because In-S was determined by the factory’s Association. The detailed procedure has been reported laboratory using a different detection limit. We obtained elsewhere12, 13). If the level of In-S was below the detection no data from 35 workers regarding their indium exposure limit, 0.05 ng/ml was used for the statistical analysis. or In-S. In total, we exclueded 128 workers, and analyzed As shown in Table 1, the means of In-S in the current 761 workers at 5 ITO manufacturing factories, 4 indium exposure, former exposure and no exposure workers were recycling factories, 3 indium oxide manufacturing 8.35, 9.63 and 0.56 ng/ml, and the maximal factories, and 1 research laboratory consisting of 465 concentrations were 116.9, 126.8, and 3.0 ng/ml, current indium exposure workers, 127 former indium respectively. In-S in the current exposure and former exposure workers, and 169 no indium exposure workers. exposure workers was significantly higher than in the no The no exposure workers worked at a factory that did exposure workers. not process indium or were office workers in indium- To assess the dose-effect and dose-response processing factories. The participation rate was 85.6% relationships, the current and former exposure workers (761/889). were classified into 6 groups by In-S referring to Chonan Table 1 shows the characteristics of the study subjects. et al.13) and Hamaguchi et al.12); namely, workers with an The majority of the subjects were male. The current In-S of 0.9 or below were assigned to Group 1, 1.0Ð2.9 exposure workers were significantly younger than the no to Group 2, 3.0Ð4.9 to Group 3, 5.0Ð9.9 to Group 4, 10.0Ð Makiko NAKANO, et al.: Indium-Induced Respiratory Effects 515

19.9 to Group 5, 20.0 or above to Group 6. (COPD)20). Namely, a score of 0 to 5 was given to each of the 6 lung fields when the area of interstitial or Medical examinations emphysematous changes occupied 0%, 1Ð9%, 10Ð24%, Respiratory symptoms and smoking habit were 25Ð49%, 50Ð74%, or more than 75% of each lung field assessed using the Japanese version of the ATS-DLD assessed20). To estimate the prevalence of interstitial or questionnaire15, 16) with some supplementary questions emphysematous changes, a score of 2 or above in at least regarding past history of respiratory diseases and history one of the 6 lung fields was analyzed, because the quality of exposure to indium and other hazardous substances of the HRCT photographs was not sufficiently uniform such as zinc, cadmium, lead, copper, gallium, selenium, due to differences in the CT equipment. etc. Spirometry was performed using electronic spirometers Statistical analysis (HI-701 or HI-801; Chest MI, Tokyo, Japan). Age- and For continuous data, the data distribution was height-adjusted predicted values of forced vital capacity examined, and an appropriate transformation was (FVC) and forced expiratory volume in one second performed to approximate a normal distribution before

(FEV1.0) were determined by the regression formula the analysis. The mean values of the variables were recommended by the Japanese Respiratory Society compared using the non-paired Student’s t-test, Welch’s (JRS)17), and the rates (%) of observed/predicted values t-test, or analysis of variance followed by Dunnett’s test,

(%FVC, %FEV1.0) were calculated. Statistical analysis if necessary, for adjusting multiple comparisons. was performed only for males because sex differences in Wilcoxon’s rank sum test or the Mann-Whitney U test pulmonary function are not negligible, and the number was applied when an appropriate transformation, such as of female subjects was small. interstitial change scores or emphysematous change Serum KL-6 (Krebs von den Lungen-6) was measured scores, could not be applied. For categorical data, in all study subjects18). Lung surfactant protein D (SP- proportions were compared by the χ2 test or Fisher’s exact D), lung surfactant protein A (SP-A), white blood cell method followed by Bonferroni’s method, if necessary, count (WBC) and C-reactive protein (CRP) were for adjusting multiple comparisons. Testing for trend was measured for almost all subjects. For 44 subjects in 2 performed using a multiple regression model for factories, the first and fifth high GM of In-S among 12 continuous data and a logistic regression model for binary factories, serum biochemical tests for liver function (AST, data. Age-, exposure duration-, and smoking-adjusted ALT, γ-GTP), immunological function (IgE, IgG, IgM, odds ratios were calculated according to exposure IgA), and collagen disease (ANA, RF, ACE) were category by applying the logistic regression model. To performed because some cases with indium-exposed assess statistical significance, two-tailed p<0.05 was used lungs have shown mild liver dysfunction, and throughout the analyses. All statistical analyses were hypersensitive pneumonitis and collagen diseases are well performed using SPSS version 15.0 J (SPSS Inc., known to cause interstitial change in the lungs. All tests Chicago, IL, USA). were done at the Special Reference Laboratory, Tokyo, Japan, which is a nationwide clinical laboratory. Urinary Results 8-hydroxy-2’-deoxyguanosine(8-OH-dG) was measured Table 2 shows the outcomes of the medical by high performance liquid chromatography (HPLC) to examinations according to exposure status. The assess oxidative stress on DNA in consideration of geometric mean of KL-6 was significantly higher in the potential carcinogenicity19). current exposure and former exposure workers than in High-resolution computed tomography (HRCT) of the the no exposure workers. The highest KL-6 levels of the lungs without a contrast agent was performed at the three current exposure and former exposure workers were 6,950 levels of the lungs recommended by the JRS20), namely, U/ml and 4,150 U/ml, respectively, which were 14 times the superior border of the aortic arch (upper lung field), and 8 times higher than the clinically applied reference the carina (middle lung field), and 1Ð3 cm above the right value of KL-6 (<500 U/ml). Approximately a quarter of diaphragm (lower lung field). HRCT was performed the current exposure and one-fifth of the former exposure mainly using a multi-slice CT scanner (Asteion 4, Toshiba workers exceeded the reference value, and the prevalences Medical Systems Co., Tochigi) at 120 kV, 200 mA, and a of KL-6 in the current exposure and former exposure slice thickness of 1 mm or a multi-slice CT scanner at workers were significantly higher than that in the no hospitals near the factories. The scores of interstitial exposure workers. The prevalence of SP-D was also changes and emphysematous changes of the upper, significantly higher in the current exposure workers than middle, and lower lung fields of the right and left sides in the no exposure workers. SP-A was not significantly were assessed by two pulmonologists according to different among the three groups. Spirometry did not modified JRS guidelines for the diagnosis and show any exposure-related changes in the current and management of chronic obstructive pulmonary disease former exposure workers, even after adjusting for 516 J Occup Health, Vol. 51, 2009

Table 2. The mean and prevalence of respiratory effect indices by exposure status

No exposure Former exposure Current exposure Mean or Mean or Mean or n Prevalence Range n Prevalence Range n Prevalence Range

Mean and range KL-6 (U/ml)$ 142 226 92Ð523 125 295** 95Ð4,150 424 337** 108Ð6,950 SP-D (ng/ml)$ 142 49.1 17.2Ð197 96 51.1 17.2Ð415 342 54.8 17.2Ð311 SP-A (ng/ml)$ 142 33.1 10.8Ð109 93 35.3 12.6Ð86.4 305 33.0 10Ð117 CRP (mg/dl)$ 114 0.05 0.01Ð3.81 46 0.05 0.01Ð1.27 119 0.04 0.01Ð1.99 WBC (× 102/mm3)2166.0 44Ð96 53 67.8 33Ð114 168 65.1 29Ð122 %FVC (%)& 109 98.5 71.8Ð138.4 78 98.4 72.3Ð121.7 225 99.4 66.6Ð137.4 & FEV1.0 / FVC (%) 109 80.6 65.7Ð95.6 108 81.4 49.1Ð95.5 338 82.5 59.9Ð99.6 & %FEV1.0 (%) 109 91.7 63.8Ð132.4 78 92.0 69.5Ð116.9 225 93.5 62.5Ð123.8 Prevalence (%) KL-6 (>500) 142 0.7 125 16.8** 424 24.3** SP-D (>110) 142 9.2 96 11.5 342 16.4* SP-A (>43.2) 142 22.5 93 33.3 305 26.9 %FVC (<80) 109 5.5 78 3.8 225 4.0

FEV1.0 / FVC (<70) 109 10.1 108 6.5 338 3.6

%FEV1.0 (<80) 109 15.6 78 15.4 225 12.0 Interstitial change 114 9.6 44 27.3* 99 5.1 on HRCT Emphysematous 114 4.4 44 9.1 99 5.1 changes on HRCT Cough or/and sputum 142 14.8 122 25.4* 376 28.7**

*, **: p<0.05, 0.01 by Dunnett’s test (reference = no exposure workers) or χ2 test. $: Geometric mean. &: Only male data was analyzed because of potential sex differences in spirometry. smoking status. The former exposure workers showed a in the current (Table 3) and former (Table 4) exposure significantly higher prevalence rate of interstitial changes workers. In the current exposure workers, the geometric than the no exposure workers, and approximately a mean, prevalence and AOR of KL-6 were significantly quarter of the former exposure workers showed evident higher in groups with an In-S of 3.0 ng/ml or above than interstitial changes. On the other hand, the in no exposure workers, and the dose-dependent trend emphysematous scores were not different among the was statistically significant when a multiple regression current exposure, former exposure, and no exposure model or a multiple logistic regression model was applied. workers. The prevalence rates of cough and/or sputum The mean and prevalence of SP-D were also higher in in the current exposure and former exposure workers were groups with an In-S of 5.0 ng/ml or above than in no significantly higher than in the no exposure workers. The exposure workers, and the dose-dependent trend was also results for the current exposure workers were very similar statistically significant. SP-A also increased dose- when workers with a history of asthma were excluded dependently, and the geometric mean, prevalence and from the analysis, and when smoking status was stratified AOR were significantly higher in workers with an In-S by current/ex- smokers and non-smokers. The current, of 5.0 ng/ml or above. Pulmonary function tests did not former and no exposure workers showed prevalence rates show a clear dose-dependent decrease, but the AOR of of cough and/or sputum of 26.5%, 20.8%, and 10.4% in %FVC (5.51, 95%CI 1.53Ð19.86) and %FEV1.0 (3.71, nonsmokers, and 29.6%, 28.4%, and 17.0% in smokers, 95%CI 1.32Ð10.48) increased in groups with an In-S of respectively. 20 or above. Smoking status did not affect pulmonary

Figures 1aÐ1d show scatter diagrams of log10(In-S) vs. function outcomes. HRCT findings did not reveal dose- KL-6 and SP-D for current and former exposure workers. dependent trends, but the AOR of emphysematous Sharp dose-dependent increases in KL-6 and SP-D were changes increased in Group 6 (AOR 4.42, 95%CI 0.95Ð seen in both the current and former exposure workers. 20.60). Symptoms of cough and/or sputum did not show Tables 3 and 4 show the means, prevalence of abnormal a dose-dependent trend. findings, and adjusted odds ratios (AOR) by In-S level In the former exposure workers, sharp dose-effect Makiko NAKANO, et al.: Indium-Induced Respiratory Effects 517

Fig. 1aÐd. Dose-effect relationships between KL-6 and SP-D, and In-S for workers with current and former indium exposure. 1a and 1c are satter diagrams of log10(In-S) vs. KL-6 and SP-D for workers with current indium exposure. 1b and 1d are scatter diagrams of log10(In-S) vs. KL-6 and SP-D for workers with former indium exposure. KL-6, SP-D: cut-off values are 500 U/ml and 110 ng/ml, respectively.

relationships of the biomarkers of interstitial changes were reduce indium inhalation. We asked them to report when observed that were similar to those in the current exposure they had finished implementing the first improvement in workers, and the dose-dependent trend was also working conditions, and we were able to obtain this statistically significant, although the former exposure information from 12 factories. We categorized the current workers had left their indium jobs for a period of 58.3 and former exposure workers in the 12 factories into 2 mo on average. Significant irreversible lung interstitial groups, one consisting of 379 workers who had been changes were seen in Groups 3, 5, and 6, and dose- working before the improvements were implemented dependent trend was statistically significant. (Group Bef), and 109 workers who started working in a For 44 workers, serum biochemical tests for liver job involving indium after the improvements had been function, immunological function, and collagen disease implemented (Group Aft). Average exposure durations and urinary 8-OH-dG were performed. We did not find were 67.1 mo and 17.3 mo in the Bef and Aft groups, any significant differences in these indices among the respectively; mean ages were 37.6 yr and 33.0 yr in the current, former and no exposure workers. Bef and Aft groups, respectively; and the male proportions After we had intensively disseminated health risk were 94.2% and 83.3% in the Bef and Aft groups, information to the factories, they all began improving their respectively. These differences between the groups were working environments and using dust-protective masks to all statistically significant. The smoking rate was not 518 J Occup Health, Vol. 51, 2009

Table 3. Dose-dependent relationships between In-S and respiratory effects for workers with current indium exposure

In-S levels of workers with current indium exposure Unexposed Group 1 Group 2 Group 3 Group 4 Group 5 Group 6 p for In-S workers 0.9 or below 1.0Ð2.9 3.0Ð4.9 5.0Ð9.9 10.0Ð19.9 20.0 or above trend

Mean and standard deviation or range KL-6 n 142 166 68 35 52 50 53 GM 226 220 255 333** 450** 511** 943** <0.001 GSD 1.42 1.47 1.42 1.60 1.73 1.70 1.95 SP-D n 142 158 54 29 37 34 30 GM 49.1 40.3 58.9 54.9 67.9** 78.8** 121.4** <0.001 GSD 1.78 1.67 1.62 1.74 1.73 1.73 1.74 SP-A n 142 136 52 24 30 33 30 GM 33.1 28.1 29.8 37.5 43.6** 35.3 51.3** <0.001 GSD 1.51 1.46 1.48 1.65 1.54 1.51 1.54 Prevalence (%) of abnormal finding and age-, exposure duration- and smoking-adjusted odds ratio KL-6 Prev 0.7 3.6 2.9 17.1# 36.5# 50.0# 84.9# <0.001 AOR 1 2.5 2.6 21.3 49.1 94.3 526 95%CI 0.5Ð18.5 0.3Ð22.1 4.5Ð155 12.8Ð328 24.3Ð634 125Ð3770 SP-D Prev 9.2 4.4 9.3 13.8 24.3# 32.4# 66.7# <0.001 AOR 1 0.4 1.0 2.1 3.7 5.7 26.0 95%CI 0.1Ð1.1 0.3Ð3 0.5Ð6.7 1.4Ð9.9 2.2Ð15.1 9.5Ð77.9 SP-A Prev 22.5 13.2 15.4 41.7 50.0# 33.3 66.7# <0.001 AOR 1 1.0 0.9 4.2 4.5 3.0 10.1 95%CI 0.5Ð2.0 0.3Ð2.1 1.5Ð12 1.8Ð11.5 1.2Ð7.6 3.9Ð28.3 n 114 8 27 10 112023 HRCT-I Prev 9.6 0.0 3.7 0.0 0.0 0.0 17.4 0.946 HRCT-E Prev 4.4 0.0 0.0 0.0 9.1 0.0 17.4# 0.110

GM, GSD, Prev: geometric mean, geometric standard deviation, and prevalence. AOR, 95%CI: Age-, smoking- and exposure duration- adjusted odds ratio applying logistic regression model and 95% confidence interval. HRCT-I, HRCT-E: Interstitial or emphysematous changes in lung HRCT. *,**: p<0.05, 0.01 by Dunnett’s test . #: p<0.05 after adjusting for multiple comparison by the Bonferroni method. Trend p was analyzed by applying a multiple logistic regression model or a multiple regression model adjusting for age, smoking and exposure duration.

different between the Bef and Aft groups (71.2 and 66.1%). of this study may have been improved in terms of Table 5 shows the In-S and medical examination results assessing the causal relationships between indium for both groups. In-S was significantly lower in Group exposure and its effects on the lungs. Aft than in Group Bef. KL-6, SP-D and SP-A were also The geometric means and the prevalence of serum significantly lower, and their prevalence rates were biomarkers of interstitial lung changes (KL-6, SP-D) were significantly lower in Group Aft than in Group Bef. significantly higher in current exposure workers than in Adjusted odds ratios indicated 87%, 71% , and 44% no exposure workers. The markers of inflammation such reductions of Group Aft workers exceeding the reference as the CRP and WBC count did not show any differences values of KL-6, SP-D , and SP-A, respectively. The among the groups of current, former and no exposure pulmonary function and HRCT findings did not show workers. These findings were compatible with the results differences between Groups Bef and Aft. of two epidemiological studies12, 13) and 5 clinical case reports7Ð11). As shown in Table 3 and Figs. 1a and 1c, the Discussion dose-effect and dose-response relationships between We combined 3 populations, i.e. those reported by indium exposure and the biological effect indices of Hamaguchi et al.12), Chonan et al.13) , and newly enrolled interstitial lung changes were evident. KL-6 showed the workers, because all 3 populations showed essentially sharpest dose-effect and dose-response relationships, and the same dose-dependent effects of indium on the significantly increased in Groups 3 to 6. Based on these respiratory system. Consequently, the internal validity relationships, we believe that the concentration of In-S Makiko NAKANO, et al.: Indium-Induced Respiratory Effects 519

Table 4. Dose-dependent relationships between In-S and respiratory effects for workers with former indium exposure

In-S levels of workers with former indium exposure Unexposed Group 1 Group 2 Group 3 Group 4 Group 5 Group 6 p for In-S workers 0.9 or below 1.0Ð2.9 3.0Ð4.9 5.0Ð9.9 10.0Ð19.9 20.0 or above trend

Mean and standard deviation or range KL-6 n 142 50 21 9 16 14 15 GM 226 206 216 318 285 392** 1,083** <0.001 GSD 1.42 1.42 1.23 1.42 1.56 1.83 1.94 SP-D n 142 48 16 4 11 9 8 GM 49.1 36.5 55.9 29.5 73.9 85.7* 146.2** <0.001 GSD 1.78 1.66 1.85 1.42 1.47 2.61 2.17 SP-A n 142 46 15 4 11 9 8 GM 33.1 29.9 32.1 39.1 42.1 44.3 56.5** <0.001 GSD 1.51 1.56 1.37 1.30 1.53 1.61 1.38 Prevalence (%) of abnormal finding and age-, exposure duration- and smoking-adjusted odds ratio KL-6 Prev 0.7 0.0 0.0 11.1 12.5# 28.6# 93.3# <0.001 SP-D Prev 9.2 0.0 12.5 0.0 9.1 44.4# 50.0# <0.001 SP-A Prev 22.5 19.6 13.3 50.0# 54.5# 66.7# 75.0# <0.001 n 114 12 8 3 8 8 5 HRCT-I Prev 9.6 8.3 12.5 66.7# 25.0 50.0# 40.0# <0.001 HRCT-E Prev 4.4 8.3 0.0 33.3# 0.0 25.0# 0.0 0.617

Abbreviations and symbols: see Table 3.

Table 5. Effectiveness of the work environment improvements on indium exposure and respiratory effects

Group Aft# Group Bef# Prevalence (%) AOR 95%CI n Mean Range n Mean Range Group Aft Group Bef

In-S 109 0.81** n.d.Ð11.2 379 12.29 n.d.Ð126.8 KL-6$ 105 216** 95Ð698 379 380 124Ð6950 3.8** 29.6 0.13 0.05Ð0.37 SP-D$ 105 41.7** 17.2Ð184 268 61.7 17.2Ð350 5.7** 20.5 0.29 0.12Ð0.74 SP-A$ 100 28.6** 12.4Ð94.1 233 37.7 12.7Ð117 16.0** 37.3 0.56 0.28Ð1.09 Cough and/or sputum 108 369 26.9 27.4 0.90 0.53Ð1.53

#: See text. $: Geometric mean. **: p<0.01 by Student’s t-test or χ2 test. AOR, 95%CI: Age-, smoking- and exposure duration- adjusted odds ratio of Group Aft for Group Bef and 95% confidence interval applying logistic regression model.

should be kept at less than 3 ng/ml to prevent interstitial without pleural plaque, although the difference was not damage to the lungs from indium exposure. statistically significant22). Thus, exposure to beryllium Other occupational toxicants inducing interstitial lung and asbestos may be a potential cause of high KL-6, but damage may be potential candidates for inducing high no former or current exposure workers had an KL-6. There are only two previous articles evaluating occupational history of asbestos or beryllium exposure. the association between occupational diseases and KL- Hamaguchi et al. disclosed that no difference in KL-6 6. Inoue et al. showed that serum KL-6 in 26 patients was seen between indium no exposure workers with and with chronic pulmonary berylliosis was significantly without an occupational history of exposure to non- higher than in 15 beryllium-sensitized workers with no ferrous metals other than indium12). Therefore, it is berylliosis or 32 control workers21). Abe et al. compared unlikely that exposure to occupational toxicants other than the KL-6 levels of asbestos-exposed workers with or indium was a cause of KL-6 elevation observed in this without pleural plaque, and stated that KL-6 was higher study. in 17 workers with pleural plaque than in 16 workers KL-6 was significantly higher in non-smoking current 520 J Occup Health, Vol. 51, 2009 exposure workers than in smoking current exposure in both current exposure and former exposure workers. workers (data not shown). The former exposure workers Additionally, there was some information bias due to showed a similar trend, but the difference was not differences in where and when the first cross-sectional significant. However, no difference in KL-6 was found observation was conducted at each factory. In-S was between non-smoking and smoking no exposure workers, measured at two different laboratories, and the chest and Kobayashi et al.23) reported that KL-6 was unaffected HRCT was taken at a hospital near each factory. The by smoking status. If smoking is a potential suppresser resolving power of the CT scanner was different at each of KL-6 release from type II alveolar pneumocytes and hospital. Although unified instructions for performing bronchiolar epithelial cells, KL-6 in smoking workers chest CT were given to the hospitals, and two with indium exposure might be underestimated. pulmonologists read all of the HRCT films, some Improvement of the working conditions caused a measurement bias may have been unavoidable. The reduction of In-S and of the effects on the lung serum biochemical indices were determined in one interstitium, as shown in Table 5. This improvement may nationwide clinical laboratory, and pulmonary function be evidence of the specificity of a causal association tests were performed by trained staff at our research unit. between indium exposure and its effects on the lungs. Therefore, measurement bias in the serum biochemistry We did not assess intra-individual changes of exposure and pulmonary function test is likely to have been small. and effect indices in the current and former exposure In 1964, the advisory committee to the US Surgeon workers between the periods before and after the General on “Cigarette Smoking and Lung Cancer” improvement of working conditions because we judged established the “Criteria for causal association in it unreasonable to assess the efficacy of the improvement epidemiology”, which included 5 criteria: “Temporal based on the findings that the In-S concentration was relationship”, “Strength of association”, “Consistency of still higher and the effects on the lungs were prolonged association”, “Coherence of association”, and “Specificity in the former exposed workers, and based on an animal of association”. experiment indicating a long biological half-life for The first criterion, “Temporal relationship”, i.e. indium in the lungs19). “Exposure always precedes the outcome”, must be In addition to the current exposure workers, the former absolutely fulfilled to show a causal association. Indium exposure workers also showed significantly higher In-S, exposure in the general environment is negligible, and KL-6 and interstitial lung changes than the no exposure occupational exposure is the only source of indium workers, and the dose-dependent effects on the lungs were exposure. Thus, the temporal relationship was adequately significant. Approximately a quarter of the former fulfilled. The strength of association is very clear. workers had irreversible interstitial changes in the lungs Exposed workers showed higher levels of indium with a and symptoms of cough and/or sputum. These findings higher prevalence of outcome indices related to interstitial indicate that the indium burden in the lungs may slowly changes of the lungs, such as KL-6 and SP-D. The dose- decrease after indium exposure stops, but that indium effect and dose-response relationships between In-S and accumulated in the lungs actively and continuously the outcome indices were also significant, and generates effects in the lungs. These findings are pronounced. In this study, we combined three sets of consistent with the prolonged pulmonary inflammation baseline data, namely data reported by Hamaguchi et al.12), and slow elimination of indium from the lungs that were those reported by Chonan et al.13), and other newly seen in rats and mice exposed to 0.1 or 0.3 mg/m3 of collected unpublished data, because all three sets of data indium phosphide for 21 or 22 wk and to room air for showed the same consequences. Seven indium lung cases approximately 80 wk19). A long-term follow-up of the have been discovered7Ð11). Therefore, the consistency of former exposure workers is needed to determine the as- association criterion seems to be fulfilled. The data were yet-unknown prognosis of the effects of indium on the obtained only in Japan, because indium demand in lungs. countries other than Japan is small1). It may not be This study had several limitations. It is likely that possible to measure the incidence of indium lung effects workers who were eligible for participation, but who had or to conduct an epidemiological study in the USA or left or retired from indium jobs due to indium-induced European countries. It is well known that the inhalation respiratory effects, could not participate in our of hardly soluble hazardous particles, such as crystalline epidemiological study, because the existence of indium- silica, asbestos, and so on, induces pneumoconiosis. In induced respiratory effects was not known until our first animal experiments, intra-tracheal instillation of indium epidemiological study12). Though this limitation may lead compounds induces pneumonitis, proliferation of fibrous our study to produce results that are underestimated, we tissues, pulmonary fibrosis, and other lung diseases3Ð6). believe that the survival selection bias was small because The results of our study are consistent with these the dose-effect and dose-response relationships between information. Thus, the coherence of association criterion indium exposure and interstitial lung effects were evident seems to be fulfilled. In newly exposed workers who Makiko NAKANO, et al.: Indium-Induced Respiratory Effects 521 started their indium job after changes in working 9) Taguchi O, Chonan T. Three cases of indium lung. conditions were made (Group Aft), the indium exposure Nihon Kokyuki Gakkai Zasshi 2006; 44: 532Ð6 (in level was considerably reduced, and the interstitial effects Japanese). on the lungs were reduced. Therefore, we think that 10) Nakano M, Omae K. A case of indium lung diagnosed specificity of association may have been established. by worker’s health examination. Occup Health J 2007; 30: 25Ð9 (in Japanese). In a strict sense, the causal association between indium 11)Takeuchi H, Morikawa T, Uchikoshi A, et al. Center exposure and effects on the lungs has not been completely type interstitial pneumonia on upprer lungs due to proven, but the results of the present study strongly Indium inhalation. Nihon Kokyuki Gakkai Zasshi 2006; support the causal association between indium exposure 44(suppl): 274 (in Japanese). and effects on the lungs, and we think it has been 12) Hamaguchi T, Omae K, Takebayashi T, et al. Exposure substantially established. to hardly soluble indium compounds in ITO production In conclusion, the concentration of In-S should be kept and recycling plants is a new risk for interstitial lung at less than 3 ng/ml to prevent interstitial lung damage damage. Occup Environ Med 2008; 65: 51Ð5. from indium exposure. Follow-up observations are now 13) Chonan T, Taguchi O, Omae K. Interstitial pulmonary in progress to elucidate the biological kinetics of In-S disorders in indium-processing workers. Eur Respir J and the prognosis of the effects on the lungs. 2007; 29: 317Ð24. 14) Miyaki K, Hosoda K, Hirata M, et al. Biological monitoring of indium by means of graphite furnace Acknowledgments: The authors thank staff members atomic absorption spectrophotometry in workers and participants at all factories for their helpful and cordial exposed to particles of indium compounds. J Occup cooperation. This study was supported in part by Grants- Health 2003; 45: 228Ð30. in-aid for Scientific Research (Project No 15390191, 15) Japan Public Health Association. Research on method 17390179) from the Ministry of Education, Culture, Sports, of mesuring influence on health due to air pollution. Science and Technology of Japan (2003-4, 2005-6). Tokyo: Japan Public Health Association; 1984. p. 3Ð Conflict of interest: None. 69. 16) Ferris BG. Epidemiology Standardization Project References (American Thoracic Society). Am Rev Respir Dis 1978; 1) Minami H. Demand, supply, and price trend etc. of 118: 1Ð120. indium. Metallic Resource Report 2007; 37: 171Ð6 (in 17) Clinical Pulmonary Functions Committee of the Japanese). Japanese Respiratory Society. Standard values of 2) Tanaka A, Hisanaga A, Hirata M, et al. Chronic toxicity spirogram and arterial blood gas partial pressure among of indium arsenide and indium phosphide to the lungs Japanese. Tokyo: The Japanese Repiratory Society; of hamsters. Fukuoka Igaku Zasshi 1996; 87: 108Ð15 April 2001. p.1Ð17. (in Japanese). 18) Ohnishi H, Yokoyama A, Kondo K, et al. Comparative 3) Kabe I, Nakashima H, Nomiyama T, et al. In vitro study of KL-6, surfactant protein-A, surfactant protein- solubility and in vivo toxicity of indium phosphide. J D, and monocyte chemoattractant protein-1 as serum Occup Health 1996; 38: 6Ð12. markers for interstitial lung diseases. Am J Respir Crit 4) Uemura T, Omae K, Takebayashi T, et al. Effects of Care Med 2002; 165: 378Ð81. intratracheally administered indium phosphide on male 19) Toxicology and carcinogenesis studies of indium fischer 344 rats. J Occup Health 1997; 39: 205Ð10. phosphide (CAS No. 22398-80-7) in F344/N rats and

5) Blazka ME, Tepper JS, Dixon D, Winsett DW, B6C3F1 mice (inhalation studies). Natl Toxicol O’Connor RW, Luster MI. Pulmonary response of Program Tech Rep Ser 2001; 499: 7Ð340. Fischer 344 rats to acute nose-only inhalation of indium 20) Guideline for COPD diagnosis and treatment, 2nd ed. trichloride. Environ Res 1994; 67: 68Ð83. Tokyo: The Japanese Respiratory Society; 2004. 6) Blazka ME, Dixon D, Haskins E, Rosenthal GJ. 21) Inoue Y, Barker E, Daniloff E, Kohno N, Hiwada K, Pulmonary toxicity to intratracheally administered Newman LS. Pulmonary epithelial cell injury and indium trichloride in Fischer 344 rats. Fundam Appl alveolar-capillary permeability in berylliosis. Am J Toxicol 1994; 22: 231Ð9. Respir Crit Care Med 1997; 156: 109Ð15. 7) Homma T, Ueno T, Sekizawa K, Tanaka A, Hirata M. 22) Abe T, Oda J, Takizawa Y, Suzuki E, Gejyo F. Interstitial pneumonia developed in a worker dealing Evaluation of serum fibrosis markers in asbestos- with particles containing indium-tin oxide. J Occup exposed workers. Nihon Kokyuki Gakkai Zasshi 2008; Health 2003; 45: 137Ð9. 46: 875Ð9 (in Japanese). 8) Homma S, Miyamoto A, Sakamoto S, Kishi K, Motoi 23) Kobayashi H, Kanoh S, Motoyoshi K. Serum surfactant N, Yoshimura K. Pulmonary fibrosis in an individual protein-A, but not surfactant protein-D or KL-6, can occupationally exposed to inhaled indium-tin oxide. predict preclinical lung damage induced by smoking. Eur Respir J 2005; 25: 200Ð4. Biomarkers 2008; 13: 385Ð92.