- 學位論文
利用貝氏階層年齡、時期、出生世代模型探討台灣肝癌發生率長期趨勢
Bayesian Hierarchical Age-Period-Cohort Model for Secular Trend of Hepatocellular Carcinoma Incidence in Taiwan
若您是本文的作者，可授權文章由華藝線上圖書館中協助推廣。
摘要
背景： 在台灣肝細胞癌一直位居癌症發病率和死亡率的重要原因，病毒性肝炎(B型肝炎及C型肝炎)是造成高肝癌發生率和死亡率的主要原因。探究其中，不同出生世代與不同地區可能造成不同的病毒性肝炎感染率，而醫學在不同時代對於病毒性肝炎的與時俱進(如疫苗接種與B型肝炎及C型肝炎的治療)也可能對於肝癌發生率和死亡率有不等程度的影響。因此，在考慮地區分布的病毒性肝炎的帶原率與異質性，著手進行年齡、時期、出生世代的效應對於肝癌發生率和死亡率的影響。 目的： 本研究旨在研究年齡，時期和世代效應在考慮地區分布的異質性合併不同性別與年齡族群B型肝炎及C型肝炎帶原率的資料，對於肝癌發生率和死亡率的影響。 方法： 本研究以群眾為對象，年齡設定在20-79歲之間的成人，以1980年至2009年的資料進行分析，期間共有162,677人發生肝癌（國際疾病分類=155.0），138,582人因肝癌(（國際疾病分類=155.0）死亡。對年齡，時期和世代效應的趨勢進行描述性分析，肝癌死亡率的時間趨勢進一步拆解為：肝癌發生率和致死率。貝氏階層模型進一步分析在不同地區分布異質性下，不同性別與年齡B型肝炎及C型肝炎帶原率，評估年齡，時期和世代效應對於肝癌發生率的影響。 結果: 從1980年至2009年資料分析，肝癌死亡率的趨勢，由1980年的每十萬人口死亡率為23.0增加至2009年每十萬人口死亡率37.8，而致死率由1980年1.98下降至1996年的0.67而後維持穩定。 肝癌死亡率的年齡，時期和世代效應的流行病學趨勢，在描述性分析結果與肝癌發生率的相似，肝癌發生率自1980年後隨時期上升，1996年後有顯著的增加，2004年後則形成高原型態。不論時期，50歲以上的族群每十萬人口的發生率為107.7遠高於20-49歲族群(每十萬人口的發生率為11.5)。就出生世代，1905年到1935年肝癌發生率是有明顯的上升趨勢；1935之後的出生世代，在45歲以上年齡族群，肝癌發生率可見穩定或下降的趨勢，在44歲以下年齡族群，肝癌發生率沒有明顯隨出生世代改變的趨勢。 由模型中發現在不同區域存在有較低的精密參數( 變異數倒數=12.9)，而判斷不同區域間存有異質性。南部地區肝癌發生率最高，中部與東部地區次之，而北部地區肝癌發生率最低；進一步考慮地區B型肝炎及C型肝炎帶原率，區域間的異質性也降低。在考慮地區B型肝炎及C型肝炎帶原率前，時期效應(精密參數估計值=54.4)相較於世代效應(精密參數估計值=172.8)對於肝癌發生率有較大的影響；而考慮地區B型肝炎及C型肝炎帶原率後，世代效應(精密參數估計值=55.5)相較於時期效應(精密參數估計值175.7)對於肝癌發生率有較大的影響。以貝氏模型校正性別、年齡與出生世代效應，預測不同年齡族群的肝癌發生率:20-39歲的年齡族群，肝癌發生率由1980年至1999年呈獻上升的趨勢，自2000年後逐漸下降；40-69歲的年齡族群，肝癌發生率由1980年至2004年呈獻上升的趨勢，自2005年後開始下降；70歲以上的年齡族群，肝癌發生率由1980年至2009年呈獻上升的趨勢，未有下降。 結論: 研究結果表明， 考慮地區異質性及不同地區B型肝炎及C型肝炎帶原率，世代效應為模型中年齡、時期與出生世代三者間對肝癌發生率趨勢影響最大者。較早的出生世代有較高的肝癌發生率。模型預測不同年齡族群肝癌發生率趨勢肯定了疫苗接種與B型肝炎C型肝炎治療，在20-29歲的族群兩者對於肝癌發生率的有至關重要的效用，而B型肝炎C型肝炎治療在40-69歲的族群，能降低肝癌發生率，但是這些效益對於較早的出生世代或大於70歲的族群未見有效的助益。
並列摘要
Background: As hepatitis B virus (HBV) and hepatitis C virus (HCV) are two main causes responsible for the high incidence and mortality of hepatocellular carcinoma (HCC), different birth cohorts and different areas may imply different infection rates and different periods may have different types of intervention (including vaccination and treatment for HBV and HCV infection), leading to different risk profiles of HCC incidence and mortality, the interplay between age, period and cohort effects in relation to HCC incidence and mortality is therefore worthy of being investigated by taking into account the heterogeneity of area possibly resulting from HBV and HCV infection rates. Purpose: This study aims to study the impacts of age, period and cohort effects on mortality and incidence trend of HCC taking into account the heterogeneity of area with information on age-gender-dependent HCV and HBV infection. Methods: A population-based study of 138,582 deaths and 162,677 incident HCC patients (International Classification of Diseases = 155.0) aged 20–79 years between 1980 and 2009 were set up for doing age-period-cohort analysis on time trend of HCC incidence. Age, period and cohort descriptive analysis on time trends on HCC mortality was first conducted. Mortality trend was further decomposed into the two time trends of incidence and case-fatality. Bayesian Hierarchical Model was developed to assess how HCC incidence trend was affected by the effects of age, period, and cohort at individual level and the covariates of age-gender-dependent HBV and HCV infection at area level. Results: Time trends in HCC mortality rates have increased from 23.0/100,000 in 1980 to 37.8/100,000 in 2009 for the overall group while case-fatality rates have dropped from 1.98 in 1980 but been stable to 0.67 after 1996. The epidemiological profiles of age-period-cohort descriptive analysis on mortality were similar to those on HCC incidence. The HCC incidence had an increasing trend with calendar period from 1980 but a remarkable jump since 1996 onwards and became plateau since 2004. Patients aged 50 years or older revealed more remarkably increasing time trend with calendar year (107.7 patients per 100,000 person-years) than those aged 20-49 years (11.5 patients per 100,000 person-years). There was a remarkable increasing trend with year of birth from 1905 to 1935 and a stabilizing or decreasing trend with year of birth after 1935 birth-year was noted for those aged 45 years or older but there was lacking of time trend with year of birth for those aged 44 years or below. The heterogeneity across area was indicated by a lower precision parameters (the inverse of variance=12.9). The incidence was highest in southern area, followed by central and eastern area, and lowest in northern area. Such heterogeneity was attenuated when the covariates of HBV and HCV infection rates were included in the model. Before considering HBV and HCV infections the period effect (precision estimates=54.4) made more contribution to HCC incidence than the cohort effect (precision estimates=172.8) whereas the cohort effect (precision estimates=55.5) had more influence than the period effect (precision estimates=175.7) when both HBV and HCV infections at area level were included. The application of Bayesian to the predicted HCC incidence trend with adjustment for age, gender, and birth cohort effect yielded the findings that HCC incidence trend for age group 20-39 increased from 1980 until 1999 but started to decline form 2000 onwards, that for subjects aged between 40-69 increased from 1980 until 2004 and started to decline form 2005 onwards, and that for the oldest group aged 70 years or older has increased from 1980 and never shown a decline. Conclusions: By taking the heterogeneity of area with information on HBV and HCV infection, we found the cohort effect is the most influential among age, period, and cohort in the time trend of HCC incidence. The early birth old cohort made contribution to higher incidence of HCC. The predicted age-specific time-trend on HCC confirmed vaccination and HBV and HCV treatments play a crucial role in a decreased HCC incidence in age group 20-29 and treatments for HBV and HCV made contribution to the reduction of HCC incidence for subjects aged between 40 and 69 years but they may not be effective in earlier birth cohort or subjects aged 70 years or older.
參考文獻
1. Beasley RP, Hwang LY, Lin CC, Chien CS. Hepatocellular carcinoma and hepatitis B virus. A prospective study of 22 707 men in Taiwan. Lancet 1981; 2(8256): 1129-33.
2. Beasley RP. Rocks along the road to the control of HBV and HCC. Annals of epidemiology 2009; 19(4): 231-4.
3. Stevens CE, Beasley RP, Tsui J, Lee WC. Vertical transmission of hepatitis B antigen in Taiwan. The New England journal of medicine 1975; 292(15): 771-4.
4. Beasley RP, Hwang LY, Lin CC, et al. Incidence of hepatitis B virus infections in preschool children in Taiwan. The Journal of infectious diseases 1982; 146(2): 198-204.
5. Chang MH, Chen DS, Hsu HC, Hsu HY, Lee CY. Maternal transmission of hepatitis B virus in childhood hepatocellular carcinoma. Cancer 1989; 64(11): 2377-80.