Potential risk factors for haematological cancers in semiconductor workers

Abstract

Background

There has recently been increased interest in cancer incidence in electronics workers.

Aims

To determine the cancer incidence ratio in electronics workers and the potential factors affecting the risk for development of cancer.

Methods

Epidemiological study performed in electronics workers who were employed between 1999 and 2008 in South Korea. Cancer incidence ratio was analysed with respect to departments, divisions, job titles, gender, age, hepatitis B and C virus infection and work duration. We compared the incidence of haematological cancer in this cohort with that expected in the general population.

Results

The study population was 56283. Overall, the standardized incidence ratio (SIR) for haematological cancer was 0.85. In particular, the SIR for leukaemia was 0.86 and for non-Hodgkin lymphoma (NHL) was 0.93, which were not statistically significant. The SIR for NHL was significantly increased [SIR 5.23, 95% confidence interval (CI) 1.31–20.95] in female office workers. We also found that the SIR for NHL was significantly increased in female workers who tested positive for hepatitis virus infection (SIR 7.69, 95% CI 1.08–54.60).

Conclusions

The raised SIR for NHL among female workers was due to potential risk factors such as hepatitis virus infection although additional research and an ongoing, long-term, prospective epidemiological cohort study is needed.

Introduction

In 2010, 202053 new cancer cases were diagnosed in South Korea (103014 cases in men and 99039 cases in women). For all cancer sites, the age-standardized incidence rate (ASR) was 304.8 per 100000, for both sexes (333.6/10⁵ men and 297.0/10⁵ women) [1]. Recently, there has been an increasing interest in work-related cancer, and there are a growing number of claims for industrial diseases and accidents [2]. According to the data on cancer cases registered in South Korea in 2010, the ASR for non-Hodgkin lymphoma (NHL) was 6.2 (3940 cases) per 100000 population [1].

Epidemiological studies on electronics workers have been performed in the USA, UK and other countries [3–8]. The study results have been inconclusive; studies in the US Semiconductor Industry Association (SIA) showed that the cancer mortality in the semiconductor workplace including clean rooms was partly increased in a follow-up study among 100081 workers in two American major semiconductor companies from 1968 to 2002, but no statistically significant, consistent evidence was found [3,9]. In a Scottish cohort study of 2126 male and 2262 female semiconductor workers [4], the standardized mortality ratio (SMR) in both male and female workers was not significantly increased [4]. Similarly, a study conducted in the West Midlands UK on a cohort of 1807 subjects, standardized registration ratios for haematological cancers were not a significantly elevated [5]. In 2011, Lee et al. calculated the cancer incidence rate among workers in eight major semiconductor companies in South Korea. Although this study overlapped with the previous research by The Korea Occupational Safety and Health Agency (KOSHA) [8], it was different from the previous study in that the emphasis was on risk factors for haematological cancer in the work place [8]. In this study, there was no significant increase in the incidence of leukaemia in the semiconductor workers, but the incidence of NHL among females was increased with increasing duration of work [8]. However, this study had several limitations including non-homogeneous work environments for different products and processes among six companies and inclusion of a small number of cancer cases.

Most of the previous studies did not assess the association between blood cancers and risk factors such as hepatitis B and C virus infection in electronics workers. Therefore, a large-scale, retrospective epidemiological study was conducted to determine the cancer incidence ratio among electronics workers and the potential factors affecting the risk for development of cancer.

Methods

Study subjects were selected by using human resources data from 1 January 1998 to 31 December 2009. Information collected included data on retired and incumbent workers, sex, social security numbers, workplaces, departments, detailed tasks, job positions, status of employment and the first and last days at work.

The study subjects were selected from Samsung Electronics Co. Ltd. factories located in three areas (K: Kiheung, H: Hwasung and O: Onyang). The main product manufactured in factories located in the K and H areas was semiconductor memory, and testing of the semiconductor and packaging process was performed in the factory located in the O area. Workers employed for <1 month were excluded and documents of workers (1999–2008) were obtained from the Korea Central Cancer Registry (KCCR) (Figure 1). Because the number of cases was small, a 10-year period instead of a 5-year period was used for age categorization.

Before starting the research, the purpose of this research was explained to the participants, and their consent for participation in this study obtained. This study was approved by the Institutional Review Board of the Kangbuk Samsung Hospital. The type of cancer was classified according to the 10th amendment of International Statistical Classification of Diseases and Related Health Problems (ICD-10) [10].

The standardized incidence ratio (SIR) was calculated using the observed cases of cancers based on the cancer incidence in South Korea general population, which was standardized according to sex and age. By using data collected from 1998 to 2009, SIR and 95% confidence interval (CI) for the subjects were calculated to confirm the statistical significance. Statistical analysis of the KCCR was performed by using the STATA statistical package. Based on the 10-year data, the incidence ratio was calculated by adding the number of incident cases of each cancer per year. All of the data were expressed after age adjustment, and the expected ratio of cancer incidence was based on the national SIR of cancer.

Results

Study participants comprised 27509 workers in the K area, 17907 workers in the H area and 12819 workers in the O area, which yielded a total of 58235 workers. After excluding the 1952 workers who were employed for <1 month, documents of 56283 workers were obtained. The study subjects were not censored because a retiree was included in the study. The study population comprised 24820 men (44%) and 31463 women (56%), with the highest proportion of workers aged between 20 and 29 years (51%). Total person-years were 328348 (150658 for males and 177690 for females) and the mean follow-up period was 5.6 years. Manufacturing workers accounted for 84% of the study population. In terms of work duration, workers who were employed for 1–5 years accounted for 41% of the study population (Table 1).

Twenty-five haematological cancer cases were obtained from the KCCR. However, we did not count the number of cancer cases in the sub-groups (Tables 2 and 3). These 25 haematological cancer cases included 13 leukaemia patients, 11 NHL patients and 1 multiple myeloma patient. The SIR for all haematological malignancies was 0.85 (95% CI 0.57–1.25), assuming the national haematological malignancy incidence ratio is 1 (Table 2). The SIR for all haematological malignancies in female office workers was 3.10 (95% CI 1.00–9.62) (Table 2). The SIR for leukaemia was 0.86 among all study subjects, and the SIR for leukaemia was 1.13 in females, although this was not statistically significant (Table 3). Leukaemia was not associated with occupational tasks (Table 3). However, the SIR for NHL was high in female workers who tested positive for hepatitis B and C virus infection (SIR 7.69, 95% CI 1.08–54.60). The SIR for NHL in female office workers was also high at 5.23 (95% CI 1.31–20.92) (Table 3).

Discussion

The SIR for haematological malignancy in semiconductor manufacturing workers was 0.85 (0.74 in males and 1.01 in females), which is low compared with the 2008 national cancer incidence rate (0.78 in males and 1.54 in females) in South Korea [8]. The SIR for NHL was high in female office workers but was not associated with work. Moreover, a high SIR for blood cancers was found in female workers who were positive for hepatitis B and C virus infection [11].

A previous cohort study, which was conducted in the USA among 82845 male and 43991 female workers in three semiconductor and memory-manufacturing areas of IBM, did not find evidence for the association between job and increased cancer mortality [6]. A study on the cancer incidence among IBM workers reported significantly low levels of SIR for total cancer among workers in the memory-manufacturing facility at San Jose (SIR 0.87, n = 1.319) and the semiconductor facility in East Fishkill (SIR 0.88, n = 1.541) [7]. Among the semiconductor workers, the significantly decreased SIRs were 0.45 for stomach cancer, 0.60 for males, 0.73 for females and 0.70 for leukaemia. Incidence and mortality were reassessed twice after the original values were published in 1985. The SMR for total cancer was decreased among females, which was statistically significant (SMR 0.69, n = 82), while the SMR for total cancer among males was close to expectations (SMR 1.12, n = 29) [5]. Results of these studies are very similar to those of our study. On the other hand, in a previous study in South Korea, the incidence ratio of NHL was higher in female semiconductor manufacturing workers [8].

There were limitations to the previous studies and this study with respect to the work status such as exposure to chemicals in the work place, seniority, job titles and lifestyle factors such as smoking and drinking alcohol. These confounders were not adjusted on statistical analysis. Another possible limitation of this study is the healthy worker effect (HWE) [12]. As this research was performed using an occupational cohort, the findings may be due to the HWE.

The strength of our study was the use of available data from a large fixed population of semiconductor workers who were representative of Korean electronics workers during a 10-year period.

A previous cohort study demonstrated the relationship between chronic hepatitis B virus infection and elevated NHL risk in Korean workers (adjusted hazard ratio 1.74) [11]. According to the nationwide data in South Korea, 4.61, 4.60, 3.69, 3.01 and 2.98% of hepatitis B surface antigen (HBsAg) carriers were observed in 1998, 2001, 2005, 2008 and 2010, respectively [13].

In another cohort study in parous Taiwanese women (n = 1782401), the association of hepatitis B virus infection with increased NHL risk was demonstrated [14]. A US cohort study in 3888 chronic HBV patients and 205203 control subjects reported that the adjusted (age, race, sex and so on) hazard ratio for chronic HBV infection was 2.8 (95% CI 1.16–6.75) compared with control subjects [15]. According to the previous studies, immunosuppressive therapy for cancer increased the risk of viral infection. Also, an unknown virus similar to the hepatitis virus could stimulate the development of tumours such as NHL [16].

We conclude that the SIR for NHL among female workers was high because of potential risk factors such as hepatitis virus, and additional research on other confounding factors is needed.

Conflicts of interest

None declared.

Acknowledgements

We gratefully acknowledge the contributions of the cancer record statistical analyst of The Korea Central Cancer Registry (KCCR), National Cancer Center in Korea.

References