聯合模型在雞尾酒療法療效評估之應用—利用CD4/CD8比值探討台灣愛滋病資料

Joint Modeling Approach for Evaluating the Efficacy of HAART via CD4/CD8 Ratio for AIDS Patients in Taiwan

陳瑞霙(Jui-ying Chen)

指導教授 ： 曾議寬

國立中央大學/理學院/統計研究所/碩士(2008年)

摘要

近十年來，利用雞尾酒療法 (高效抗逆轉錄病毒療法) 治療愛滋病 (後天免疫缺乏症候群)，已經被廣泛提倡且使用。目前也有大量的研究顯示，此種療法對於治療愛滋病確實有正面的效果。而在各種衡量愛滋病情嚴重程度的指標當中，CD4/CD8 比值是其中一種衡量的指標。本篇論文則是希望透過該指標，探討雞尾酒療法對於愛滋病人存活時間的影響，進而瞭解 CD4/CD8 比值和愛滋病人存活時間的關係。由於此類臨床資料通常同時包含長期追蹤資料及存活時間，為了能有效的進行長期追蹤資料的分析，且在不失去存活資訊的情況下，本篇論文使用了 Wulfsohn 和 Tsiatis (1997) 所提出的聯合模型，並且針對此模型做延伸，導入蒙地卡羅的數值方法，使得此聯合模型方更簡便也更有效率。同時，本篇論文也利用了三種圖形對於資料做了初步的描述，如趨勢圖、事件歷史圖 (Dubin, M"uller, and Wang, 2001) 和 3-D 平滑曲面圖。以上三種圖形不僅能同時呈現出長期追蹤資料的資訊和存活資訊，亦有助於瞭解資料的特性。最後，利用以上所述之方法應用於台灣 160 位愛滋病已發病病人的資料，實際分析其療法之成效。

關鍵字

聯合模型 ； Cox 比例風險模型 ； 愛滋病 ； 存活分析 ； 隨機效應 ； 長期追蹤資料 ； 雞尾酒療法 ； CD4/CD8 比值

並列摘要

The famous treatment, HAART (Highly Active Antiretroviral Therapy), has been advocated for the recent decade for AIDS (Acquired Immune Deficiency Syndrome) patients. Also, large amount of researches reveal its efficacy for treating AIDS. The biomarker, CD4/CD8 ratio, is one of the important way of detecting the disease progression. The objective of this paper is to probe the impact of HAART on survival time for AIDS patients and identify the association between CD4/CD8 ratio and survival time for AIDS patients. In order to combine effectively both longitudinal and survival information from this kind of data, we propose an extension of joint model in Wulfsohn and Tsiatis (1997). Instead of Gauss-Hermite quadrature, Monte Carlo integration is employed in the joint model to make it easier to implement and more efficient to derive estimates. Meanwhile, graphic skills such as profile, event history graph (Dubin, M"uller, and Wang, 2001), and 3-D surface accompanied contour plots are proffered not only to exhibit simultaneously the longitudinal and survival information but also to detect the features of data. The aforementioned techniques are applied to a clinical data of AIDS patients in Taiwan.

並列關鍵字

CD4/CD8 ratio ； Cox proportional hazard regression model ； Survival analysis ； Random effects ； AIDS ； Joint model ； Longitudinal data ； HAART

參考文獻

[1] Clementi, M., Forabosco, P., Amadori, A., Zamarchi, R., De Silvestro, G., Di Gianantonio, E., Chieco-Bianchi, L., and Tenconi, R. (1999), “ CD4 and CD8 T lymphocyte inheritance. Evidence for major autosomal recessive genes,” Human Genetics, 105, 337–342.

[2] Cox, D. R. (1972), “Regression models and life tables”. Journal of the Royal Statistical Society, Series B, 34, 187-220.

[3] Dafni, U. (1993), “Evaluating surrogate markers of clinical outcome when measured with error,” Ph.D. dissertation, Harvard School of Public Health.

[5] Dubin, J., M"uller, H. G., Wang, J. L. (2001), “Event history graphs for censored survival data,” Statistics in Medicine, 20, 2951-2964.

[6] Egger, M., Hirschel, B., Francioli, P., Sudre, P., Wirz, M., Flepp, M., Rickenbach, M., Malinverni, R., Vernazza, P., and Battegay, M. (1997), “Impact of new antiretroviral combination therapies in HIV infected patients in Switzerland: prospective multicentre study. Swiss HIV Cohort Study,” British Medical Journal, 315, 1194-1199.