透過您的圖書館登入
IP:44.220.43.170
  • 學位論文

高雄市結核病近期傳播之空間分布特性與預測模型建立:一項結合流行病學與基因資訊的研究

Spatial characteristics and prediction model of recent transmission of tuberculosis in Kaohsiung: an integrated analysis of epidemiological and genomic data

指導教授 : 林先和
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 結核病的近期傳播是導致新發結核病個案產生的一項重要原因。在中高結核病負擔的地區,結核病傳播的機制仍未明朗。過往研究因多缺乏全基因定序的資訊,較難準確定義結核病的近期傳播事件。同時空間相關的因子雖然被證實與傳播相關,然而只有少數研究將其與基因資訊結合進行分析。本研究結合了空間、流行病學與全基因定序的資料,期能達到三項目標:(一)探討高雄市結核病近期傳播的空間分布特性(二)找出與近期傳播相關的危險因子(三)建立與驗證結核病近期傳播之預測模型。 方法 研究對象為高雄市2019年1月至2021年7月間結核病菌株培養陽性的個案。近期傳播的判定是透過檢體的全基因定序資訊,以基因距離小於5個單核苷酸多態性(Single-Nucleotide Polymorphism, SNP)定義結核病個案是否為近期傳播事件。分析方法分為三個部分:(一)空間分析中我們先透過曼特爾檢定(Mantel test)分析兩兩個案之間空間距離與基因相似性的相關性,接著採用了全局莫蘭指數(Global Moran’s I index)評估近期傳播是否有空間上的自相關,並利用局部莫蘭指數(Local Moran’s I index)找出近期傳播的空間熱區,最後再透過空間風險函數建立近期傳播的風險地圖。(二)為找出與傳播相關的危險因子,我們先以單變項分析初步探討各項因子與近期傳播的相關性,接著使用多變量羅吉斯回歸模型校正潛在的干擾因子並估計各危險因子對近期傳播的勝算比(odd ratios)與95%信賴區間,再透過赤池資訊準則(Akaike information criterion, AIC)與調整後R平方(Adjusted Mcfadden’s r squared)評估模型的配適程度。(三)預測模型透過多變量羅吉斯迴歸建立,預測因子包括過去研究所證實或本研究所發現的流行病學與空間危險因子,並根據赤池資訊準則以逐步向後消去法選出最配適的模型。模型的表現透過曲線下面積與校正檢定(Hosmer-Lemeshow test)進行評估,並以拔靴法(bootstrap validation)與時間切割驗證法(time-split validation)進行驗證。 結果 2019年1月至2021年7月間,高雄市共有2161名結核病個案具有完整的空間與基因資訊而納入進行空間分析,其中1323名個案具有完整的流行病學資訊,分別為230(17%)名近期傳播個案與1093(83%)名非近期傳播個案。空間分析結果發現到L2世系的結核病菌株其空間距離與基因相似性有顯著的正相關(Mantel statistic r: 0.11, Significance<0.01),莫蘭指數則顯示高雄市結核病近期傳播有顯著的空間自相關(Moran’s I statistic: 0.063, P <0.01),同時風險地圖顯示高雄市的東北部地區發生近期傳播的風險較高。羅吉斯迴歸分析發現到部分流行病學與空間特性與結核病的近期傳播相關,其分別為:年齡小於25歲以下的人相較年齡大於65以上的人勝算比值為4.79(95%信賴區間,2.17-10.68)、有社會救助的需求相較於沒有的人勝算比值為2.34(95%信賴區間,1.06 -5.14)、痰液吐片陽性相較於陰性的人勝算比值為1.49(95%信賴區間,1.07 -2.05)、500公尺內有其他痰液吐片陽性的結核病個案相較沒有的人勝算比為1.62(95%信賴區間,1.09 -2.43)。此外,回歸模型的配適度則以結合空間與流行病學因子的配適程度最好。預測模型的表現方面,根據所有研究樣本建立的模型其曲線下面積達到0.744,校正檢定結果顯示模型預測值與真實值並無顯著差異(P=0.42)。而時間切割驗證的模型預測曲線下面積為0.626,校正檢定結果則顯示預測值與實際值有顯著差異(P<0.01)。 結論 本研究針對高雄市結核病的近期傳播進行了完整的探討,證實高雄市結核病的近期傳播具有空間上的相關性並且有近期傳播熱區與高風險地區的存在,個案間空間上的相似性是了解結核病傳播動態的重要因素。另一方面,較輕的年齡、社會救助的需求、未婚的狀態等流行病學特性對於結核病的近期傳播具有顯著的影響。此研究探討了空間的資訊對結核病分析的重要性,空間、流行病學與基因資訊的結合將可使研究者們全面性地了解疾病的傳播動態。

並列摘要


Background. Recent transmission (RT) is a major cause of new tuberculosis (TB) cases. In areas with moderate to high TB burden, the mechanisms of recent TB transmission remain unclear. Furthermore, although spatial information has been proven to associate with the transmission of TB, only a few studies have combined it with whole-genome sequencing (WGS) data. Therefore, the aims of this study are (a) to identify spatial patterns of recent TB transmission in Kaohsiung, (b) to explore the relationship between individual risk factors and the occurrence of the recent transmission event, and (c) to develop and validate the prediction model for recent transmission of TB Method. From Jan 2019 to Jul 2021, the Mycobacterium tuberculosis isolates from TB patients in Kaohsiung were genotyped by WGS. For spatial analysis, we first used the Mantel test to explore the correlation between pairwise distance and genetic similarity. Later, Global and Local Moran’s I were applied to assess TB cases’ spatial patterns and identify spatial hotspots of recent TB transmission. The spatial relative risk function was used to estimate the relative risk of recent TB transmission in each spatial location of Kaohsiung. To identify individual risk factors, multivariable logistic regression analysis was applied to estimate the odds ratios (OR) and 95% confidence interval (CIs) for epidemiological and spatial factors associated with the recent transmission of tuberculosis. For the development and validation of prediction models, we built models that included epidemiological and spatial factors observed in previous studies or identified by regression analysis in this study. Backward elimination based on AIC was used to improve prediction efficiency. The prediction model’s performance was evaluated by Area Under the Receiver operating characteristic (AUROC) and Hosmer-Lemeshows test (H-L test). Furthermore, we performed time-split validation and bootstrap validation to conduct internal validation. Results. From Jan 1, 2019, to Jul 31, 2021, 1323 cases of culture-positive tuberculosis were included, with 230 RT cases (17%) and 1093 non-RT cases (83%). The Mantel test showed that there was a significant correlation between the spatial distance and the genetic distance in the strains of Mtb lineage 2 (Mantel statistic r: 0.11, Significance<0.01). The spatial autocorrelation analysis revealed that there was a significant positive spatial autocorrelation of recent transmission in Kaohsiung (Moran I statistic: 0.063, P <0.01). Also, the northeastern region of Kaohsiung was found to have a higher risk of recent tuberculosis transmission than other regions. Results of multivariable logistic regression analysis showed that factors such as age under 25 (vs. age over 65, OR: 4.79, 95%CI: 2.17-10.68), the need for social support (OR: 2.34, 95%CI: 1.06-5.14), positive sputum smear result (vs. negative result, OR: 1.49, 95%CI: 1.07-2.05), and the existence of neighboring TB cases within 500 meters (OR: 1.62, 95%CI: 1.09-2.43) were the most influencing factors on recent TB transmission. The AUROC of the prediction model built on whole analysis data was estimated to be 0.744 (H-L test: p=0.40). On the other hand, the AUROC obtained from time-split validation was 0.626 (H-L test: p<0.01). Conclusion. In our study, we identified that the distribution of recent TB transmission in Kaohsiung is spatially dependent and found that neighboring TB cases may play an important role in the recent TB transmission. Meanwhile, epidemiological characteristics including age, unmarried status, and need for social support have been observed in RT cases, improving our understanding of TB transmission dynamics in Kaohsiung. The analysis combing epidemiological, spatial, and genomic information provided valuable information and could be further implemented in the future studies for infectious diseases.

參考文獻


1. Rieder H L (1995). Opportunity for exposure and risk of infection: the fuel for the tuberculosis pandemic. Infection. 1995 Jan-Feb;23(1):1-3.
2. Marcel A Behr et al., (2018). Revisiting the timetable of tuberculosis. BMJ. 2018; 362: k2738
3. World Health Organization (2021). Global tuberculosis report 2021
4. 台灣結核病防治年報2020,衛生福利部疾病管制署
5. World Health Organization (2015). The end TB strategy.

延伸閱讀