- 學位論文
運用多階段疾病進展模式評估新冠肺炎抗病毒藥物治療之臨床效益
Evaluating the clinical efficacy of antiviral therapies for COVID-19 patients: A multi-state process approach
摘要
研究背景 截至2021年9月底，歷時超過一年的COVID-19全球大流行已造成239,878,884例確診案例以及4,888,000（2.0%）例死亡個案，而罹病者中尚有17,769,474（7.6%）位COVID-19患者尚未康復需醫療介入與治療。雖然在2021年初醫療科學領域已開發出有效的新冠肺炎疫苗，可達到抑制疫情傳播以及降低罹病者件展成為重症個案之風險，在新冠病毒的高傳播力下，全球醫療體系在面對為數眾多的罹病個案，對於具有臨床效益的療法仍有迫切需求。COVID-19的致病機轉以及造成的人體病理反應在臨床醫療與生物科學的攜手努力下已有相當程度的了解，在此一基礎上也有需多針對不同疾病嚴重程度的治療藥物在經過臨床試驗取得臨床效益實證後投入全球COVID-19病患治療中。在COVID-19 複雜的病程進展變化下，如何運用不同的藥物發展有效的病患治療策略仍有待實證評估。有鑒於此，本研究主旨為 (1) 建構COVID-19多階段疾病進展模型，並利用臨床隨機分派試驗的結果，量化此由低風險、中風險、高風險，乃至於死亡以及康復兩主要終點之疾病進展與動態變化過程； (2) 根據 (1) 的 COVID-19 疾病進展動態變化評估各臨床治療對於COVID-19病患在康復出院與死亡兩主要結果的臨床療效； (3) 根據 COVID-19不同的疾病風險分類，發展以疾病風險為導向的治療策略，並以前述之實證方法評估其臨床效益。 材料與方法 本研究依據世界衛生組織提出的COVID-19病患於住院時的氧氣治療需求，將病患分類為低風險、中風險，以及高風險三種疾病狀態以及進展至康復出院以及死亡兩終點，建立COVID-19多階段疾病進展模型。在此多階段疾病進展之基礎上，本研究運用已發表的三篇臨床隨機對照試驗的研究實證資料(分別為Adaptive COVID-19 Treatment Trial-1 (ACTT-1)、ACTT-2，以及Randomised Evaluation of COVID-19 Therapy (RECOVERY))量化COVID-19在不同疾病風險狀態間轉移之力量與速率。對於此轉移速率矩陣之估計則採用貝氏馬可夫鏈蒙特卡羅(Markov Chain Monte Carlo (MCMC) Simulation)方法結合前述三個臨床隨機分派研究之實證資料所建立概似函數以無訊息的事前分佈進行參數估計。運用此疾病自然轉移速率矩陣對前述研究中不同藥物介入比較治療組與標準療法組的疾病狀態轉移機率，評估該療法的臨床治療效益。 結果 本研究納入包含瑞德西韋治療 (ACTT-1)、瑞德西韋併用愛滅炎治療 (ACTT-2) 與類固醇治療 (RECOVERY) 三個臨床隨機分派研究之實證資料進行COVID-19多階段疾病進展模型建後以及治療臨床療性評估。在此三個臨床隨機分派研究之實證資料支持下， COVID-19在低風險、中風險，以及高風險疾病狀態的每日惡化速率分別為0.18（95% CI：0.17-0.20，低風險至中風險）、0.06（95% CI：0.04 -0.07，中風險至高風險) 和 0.054（95% CI：0.051-0.057，高風險至死亡）。各風險狀態對應的每日好轉速率分別為 0.1（95% CI：0.09-0.11，低風險至康復）、0.05（95% CI：0.04-0.07，中風險至低風險）和 0.033（95% CI：0.027-0.039，高風險至中風險）。中風險與高風險狀態的 COVID-19 每日康復速率估計分別為 0.092（95% CI：0.086-0.097）與0.001（95% CI：0.0001-0.002）。運用瑞德西韋、愛滅炎與類固醇藥物進行以風險為導向之治療策略下，前述各風險狀態的每日疾病惡化速率分別為0.06（95% CI：0.05-0.07，低風險至中風險）、0.08（95% CI：0.05-0.10，中風險至高風險）和0.01（95% CI：0.01-0.02，高風險至死亡）。而對應每日好轉速率分別為0.14（95% CI：0.13-0.16，低風險至康復）、0.30（95% CI：0.18-0.43，中風險至低風險）和 0.08（95% CI：0.04-0.12，高風險至中風險）。中風險與高風險病患每日病程轉為康復的速率分別為0.002（95% CI：0.001-0.007）與 0.000002（95% CI：0-0.000008）。 以前述COVID-19的多階段疾病狀態進展模型以及量化之轉移速率為基礎，相較於標準治療，運用瑞德西韋、愛滅炎與類固醇藥物進行以風險為導向之治療策略臨床效益評估結果顯示，對於低風險、中風險，與高疾病風險的COVID-19病患可分別降低死亡風險達91%、80% 和 75%；而康復與出院率分別提高了17%、17% 和 1.4 倍。 結論 抗病毒藥物瑞德西韋對於低風險與中風險風險之COVID-19病患具有較佳的臨床療效，而類固醇因具其抗發炎藥物機轉，對於高風險病患之療效較佳。本研究萃取三項大型臨床隨機分派研究實證資料結合COVID-19五階段疾病進展動態模型，根據COVID-19各階段疾病特性，評估不同藥物或治療組合對於病患的臨床療效。藉由風險為導向之治療策略實證評估結果，本研究不只證明了此治療策略之臨床療效，此動態多階段疾病進展方法也可運用於目前多種COVID-19療法之實證評估，以加速其臨床治療之發展。
並列摘要
Background The spread of COVID-19 worldwide has led to a total of 239,878,884 confirmed cases of whom 2.0% (4,888,000 patients) have been dead and 7.6% (17,769,474 patients) have not been recovered as of 31 September, 2021. Although vaccines that are effective in reducing the risk of infection and progression of COVID-19 have been developed with rapid distribution, an effective strategy for treating infected subject is urgently needed to mitigate such a global threat on global health. With the efforts of the field of medical science and clinical medicine the clinical evolution and the pathological change have been elucidated. Following the identification of these pathological mechanisms, a variety of compounds and the combination of these treatments and therapies have been proposed with their clinical efficacy been proved by using the randomized controlled study design. Given the complexity in the clinical evolution of COVID-19, the optimal strategy in the clinical use of the proposed treatments and therapies remain to be assessed and elucidated. In this study, we thus aimed (1) to develop a multistate process to depict the evolution of COVID-19 through the low-, medium-, and high-risk status to the two outcomes of recovery and death by using the empirical data on randomized control trials; (2) to evaluate the clinical efficacy of treatment and therapies on the two outcomes of recovery and death on the basis of the COVID-19 evolution of (1); (3) to develop and assess the clinical efficacy of treatment strategy tailored by the risk status of COVID-19. Material Methodology A multistate model depicting the evolution of COVID-19 through the three transient states of low-, medium-, and high-risk status and the two absorbing status of recovery and death defined by the oxygenation requirement was first established. The transition kernel consists of the rates of disease progression and regression through the defined states was then derived by using the published data on three randomized controlled studies, namely the Adaptive COVID-19 Treatment Trial (ACTT) one and two, and the Randomised Evaluation of COVID-19 Therapy (RECOVERY) Trial. Regarding the derivation of the estimated results on the transition kernel, the Bayesian Markov Chain Monte Carlo method was applied to update the non-informative priors by using the likelihood established from the empirical data mentioned above. The clinical efficacy for individual treatment and therapy and integrated treatment strategy tailored by the risk states of COVID-19 were then quantified by comparing the transition probabilities between the treatment groups with the natural evolution group constructed by the multistate model for COVID-19 evolution. Result Three randomized control trails assessing the clinical efficacy of remdesivir (ACTT-1), remdesivir combined with barcitinib (ACTT-2), and dexamethasone (RECOVERY) were included in this study. Supported by the information abstracted from three randomized trials, the daily progression rate for low-, medium-, and high-risk COVID-19 state were estimated as 0.18 (95% CI: 0.17-0.20), 0.06 (95% CI: 0.04-0.07), and 0.054 (95% CI: 0.051-0.057), respectively. The corresponding daily regression rates were estimated as 0.1 (95% CI: 0.09-0.11), 0.05 (95% CI: 0.04-0.07), and 0.033 (95% CI: 0.027-0.039), respectively. The daily recovery rates for medium- and high-risk COVID-19 state were estimated as 0.092 (95% CI: 0.086-0.097) and 0.001 (95% CI: 0.0001-0.002), respectively. Regarding the estimated results for risk-tailored regiment by using remdesivir, barcitinib, and dexamethasone, the estimated results on the daily progression rates for low-, medium-, and high-risk state were 0.06 (95% CI: 0.05-0.07), 0.08 (95% CI: 0.05-0.10), and 0.01 (95% CI: 0.01-0.02), respectively. The corresponding daily regression rates were estimated as 0.14 (95% CI: 0.13-0.16), 0.30 (95% CI: 0.18-0.43), and 0.08 (95% CI: 0.04-0.12). The daily recovery for the medium- and high-risk states were estimated as 0.002 (95% CI: 0.001-0.007) and 0.000002 (95% CI: 0- 0.000008). On the basis of this five-state model for COVID-19 progression, the clinical efficacy on risk-tailored treatment by using the regimen of remdesivir, barcitinib, and dexamethasone was assessed. Compared with standard care, the risk-tailored regimen reduced the risk of COVID-19 death for low-, medium, and high-risk state by 91%, 80%, and 75%, respectively. COVID-19 patient receiving the risk-tailored regiment also have a higher chance of recovery and discharge by 17%, 17%, and 1.4-fold for those at low-, medium-, and high-risk state, respectively. Conclusion While the anti-viral agent of remdesivir treatment is more effective for COVID-19 patients at low- and medium-risk, the treatment with dexamethasone is more effective for high-risk patient due to its potent anti-inflammatory mechanism. By using the five-state COVID-19 progression model in conjunction with the empirical data of three randomized controlled trials, the clinical efficacy of the risk-tailored regiment taking into account the mechanism of each compound and the process of COVID-19 evolution was assessed. The high effectiveness for COVID-19 patients at low-, medium-, and high-risk demonstrated in this study not only shows the risk-tailored approach as a promising treatment modality but also strengthen the evidence for its clinical use.