![Contact us: CSMUH Chinese Medicine Clinical Trial Center http://www.csh.org.tw/into/herb James Cheng-Chung Wei, MD, PhD ( 魏正宗 ) [email_address]](https://image.slidesharecdn.com/2hr-110609182258-phpapp01/85/2-hr-52-320.jpg)
臨床試驗的研究設計 2 hr
- 1. 臨床試驗之研究設計簡介 Protocol design 魏正宗 James Cheng-Chung Wei, MD,PhD. 中山醫學大學附設醫院 過敏免疫風濕科主任 、 中藥臨床試驗中心主任
- 2. 臨床試驗計畫書 Protocol 研究背景 具體研究目的 試驗設計流程圖 随机化方案 病人入選和排除標準 控制組與盲法 結局指標 樣本數計算 資料收集方法 統計分析
- 3. Science Science is built up with facts, as a house is with stones. But a collection of facts is no more a science than a heap of stones is a house. Measurable, reproducible and comparable . Jules Henri Poincare : La Science et L’Hypothese ( 1908)
- 4. ?
- 5. 新藥開發流程 Scale-up 臨床前研究 臨床試驗 新藥審核 12-24 months 1-4 month 9-12 months 12-24 months 6-18 months Phase II Phase IV Pharmacology and Pharmacokinetics Animal Safety Testing Submit IND Submit NDA NDA Approval Formulation Chemistry and Physical Characterization Phase III 新藥監視 Botanicals with historical documentation of safe human use
- 7. Starting from Ask a good (exciting) question!
- 8. Ask a good (exciting) question Clinical relevant Unmet medical needs Clear Ask your question in one sentence Don’t be too ambitious Innovative Unanswered questions Review & critical appraisal of literatures Answerable Practical methodology
- 9. How to ask a goog question? ~From good daily patients care Delicate care of individual patient Detail records Collect patients Setup database Thinking…
- 10. Ask a good question 問題之結構 (PICOT) P: Participant I: Intervention (E:Exposure) C: Comparison O: Outcome T: Time
- 11. Before study design Literatures search Background, Introduction What’s your hypothesis? Study Aim
- 12. Clinical Trials Design Case report, case series Early exploratory study (Pilot study) Phase II Randomized controlled trial (Proof-of-concept study) Dose range study is necessary DBPC trial is preferred Phase III Randomized controlled trial (Confirmatory study)
- 13. Cases series Drug A in the treatment of obesity Sample size: 30 Trial subjects: obesity Intervention: Drug “A” Design: open study, one arm Duration: 30 days Endpoint: body weight Result: 100 kg to 95 kg. Conclusion: Drug A can reduce body weight?
- 14. Uncontrolled studies One arm study 缺點 : 沒有對照就沒有鑑別 自然病程 ( Natural course) 霍桑效應（ Hawthorne effect ） 安慰劑效應（ placebo effect ） 向均數回歸（ regression to the mean ） Needs Known disease course Objective and well accepted endpoints Clear documentation
- 15. ?
- 16. 臨床試驗設計的 Gold standard ~ 隨機對照試驗 （ randomized controlled trials ）
- 17. Structure of Randomized Control Trials Parallel, two arms design Randomized Endpoints Trial subjects Treatment Control
- 18. Trial subjects selection Inclusive criteria Homogenecity Exclusive criteria Efficacy Safety
- 19. How to find a good intervention for clinical trial? Literatures review Modern research Expert’s opinions Pilot clinical trials
- 20. Intervention (treatment) Clear dosing SOP Dosing ranging study Washout period Run-in period Allowed medication Prohibit medication Rescue medication
- 21. Methods of control Placebo control 目的在於克服對照組病人由於心理因素所造成的偏倚 (bias) 常用無藥理作用的乳糖、澱粉等 ， 安慰劑要求在外觀、顔色、形狀與試驗藥物完全一致 要特別注意醫療道德上的可行性。 Standard control (active control) 研究藥物的療效時，可以公認的常規有效療法作爲對照。
- 22. “Bad” controls Self control (before-and-after study) 如用藥前後的自身對比觀察； 對照與實驗在同一物件身上進行，例如身體對稱部位的比較觀察 Historical control 已先前研究之結果為對照 只適用於”不隨時空環境不同而改變結果”之疾病 Blank control 在對照組不加任何處理的 “ 空白 ” 條件下進行觀察。
- 23. Controlled trial: Drug A and B in the treatment of obesity Sample size: 30 in each arm Trial subjects: obesity Intervention: Drug “A” or “B” Duration: 14 days Endpoint: body weight Result: in arm A: 100 kg to 50kg, in arm B: 80 kg to 40 kg. Conclusion: Drug A is better than B in reducing body weight?
- 24. Double blind +/- double dummy Randomization Endpoints Trial subjects “ A” "Placebo” or “B”
- 25. 盲法（ Blind Method ） 受試者或研究人員，可能導致對效果判斷的偏倚，可通過盲法來控制． 可行性及倫理考量 Single blind 只有研究者知道，而受試者不知道。 Double blind 研究者（包括資料分析者）和受試者雙方都不知道。
- 26. 随机化 Randomization 每一個研究物件或觀察單位都有完全均等的機會被抽取或分配到某一組 不受研究者或研究物件主觀意願所左右。 隨機 才能有”可比較性” ( Baseline comparability)
- 27. Randomization 簡單随机化 ：可通過抛擲硬幣、抽籤、摸球、 亂數表 來完成。 區組随机化 ：根據受試者進入研究的時間先後順序，將其分成內含相等例數 （ eg. 4, 6, 8 ） 的若干區組，而後，區組內的受試者被隨機分配至不同組別。 分層随机化 ： 依研究需要的某些變項（ eg. BMI, gender ）劃分為互斥的若干層，然後確認各層在母群體中的比例，接著按此比例分別自每一層利用隨機抽取樣本 。
- 28. Block Randomization 受試者用藥隨機指派將使用 computer-generated 「區塊隨機指派」：將所有樣本數（ 60 位）分成 5 個區塊，按 2 ： 1 之方式每 6 區塊內分別進行隨機指派，區塊內的受試者個數為處理數（ arms ）的倍數。如此可以調整兩組受試者個數相等或隨意調整分組受試者個數的比例
- 29. Endpoints Scientific: well accepted, clear and operable outcome measurements Survival> QOL>Symptomatic>Laboratory Surrogate endpoints TCM endpoints if operable Primary and secondary endpoints Safety and efficacy
- 30. 主要試驗指標 (primary endpoint) 主要試驗指標通常為療效指標。一般而言，一個經過嚴謹設計的臨床試驗其所用的主要指標往往被用於藥品上市後仿單上所宣稱的適應症 (indication) 。 主要療效指標需與宣稱的療效有直接的關聯，但在某些特殊情形下，直接療效指標評估有實質上的困難時，可改採替代性指標 (surrogate endpoint) ，唯須有充足的文獻或長期的流行病學追蹤研究証實其與直接療效指標的關聯性。
- 31. Dose ranging study anti-dsDNA Ab ≧ 10 IU/ml Meet all criteria Randomization (~ 588 patients) 20% 40% 40% ~ 196 patients ~ 392 patients Both Pre-screening and Screening consent forms obtain prior to study activities .
- 32. ATTAIN study abatacept in inadequate responders to TNF-targeted therapy Day 1 12 months 24 months Randomization Placebo + DMARD (n=133) Fixed-dose abatacept ~10 mg/kg + DMARD (n=258) Randomization=2:1 Anti-TNF inadequate responders with active disease (n=393)* Co-primary endpoints: ACR 20 and Physical Function Fixed-dose abatacept ~10 mg/kg ? + DMARD (n=317) ? Double-blind Open-label phase 6 months Open label phase ongoing *DMARDs continued, anti?TNF washout period (28–60 days); ? Based on patient weight range; ? Patients randomized to the placebo group in the first 6 months switched to abatacept during OL period; LTE=long-term extension Genovese M, et al. N Engl J Med 2005;353:1114–1123
- 33. Active AS MNY criteria BASDAI>6 ESR>28 CRP>1 NSAID failure Randomisation Questionnaire X ray: KUB, C, L spine lat view Bath indices ESR, CRP Stock serum Mo 2, 5, 11 Bath indices ESR, CRP Stock serum Mo 23 X ray: KUB, C, L spine lat view Bath indices ESR, CRP Stock serum Biologics + NSAID + Sulfasalazine Biologics - NSAID - Sulfasalazine Factorial design 2 5 11 Biologics + NSAID - Sulfasalazine Biologics - NSAID + Sulfasalazine Endpoints mSASSS ASAS20 ASQoL SF-36 BMD MRI DNA Biomarkers BS-alkP Osteocalcin CTx Wnt, dkk-1 BMP, MMP 23
- 34. Crossover Clinical Trial Drug B Drug A W A S H O U T Phase 1 Period Eligible Patients /subjects Drug A Drug B Informed consent Drug B Drug A Phase 2
- 35. 平行設計與交叉設計 平行設計 (parallel design) ：每位受試者僅能接受一種治療，也就是說各治療組間彼此互為獨立。由於統計模式的應用上無需做許多的假設即可成立，許多傳統方法皆能適用，因而被較廣泛地使用。 交叉設計 (crossover design) ：交叉設計則每位受試者可以在不同的時段 (period) 接受不同的治療，惟兩個治療時段間必須有沖洗期 (wash-out period) ，以避免繼續效應 (carryover effect) 。
- 37. Double blind double dummy design Non-inferiority vs superiority design Abatacept SC (weekly) Placebo IV (monthly) +IV loading dose Abatacept IV (monthly) Placebo SC (weekly) Abatacept SC (weekly) 15 Day 169 Commercial Availability 720 subjects 720 subjects Long-Term Extension ongoing Short Term 6 months 113 1 43 29 57 85 Days 141 Monthly Phone Visits + Quarterly Office Visits All subjects
- 38. Adaptive designs Stopping trial early due to safety, futility, or efficacy Two stage design Dose finding at stage 1 Interim analysis Sample size re-estimation Adaptive randomization …
- 39. Study flow chart
- 40. Statistics & Sample size estimation 預期的療效 Δ 第一類誤差 (α) 第二類誤差 (β) 變異數 (variance) 預期的中途退出率 (dropout rate)
- 41. Sample size estimation web http://stat.ubc.ca/~rollin/stats/ssize/ Comparing a Mean to a Known Value Comparing Means for Two Independent Samples Comparing a Proportion to a Known Value Comparing Proportions for Two Independent Samples Unmatched Case Control Studies http://www.csh.org.tw/into/herb
- 42. Comparing Proportions for Two Independent Samples
- 43. Comparing Means for Two Independent Samples http://stat.ubc.ca/~rollin/stats/ssize/
- 44. TCM in Patients with Hyperuricemia. A DBPC clinical trial 本研究之第一誤差設定為 0.05 ，第二誤差設定為 0.2 ，即本研究之檢力為 0.8 主要療效指標為尿酸 (UA) ，治療組 (μ1) 預期由 9mg/dl 降至 7mg/dl ，安慰劑組 (μ2) 服藥後降至 8.5mg/dl ，標準差 (SD) 設定為 2 ，則 (μ1-μ2)/SD= 56 預估退出率 20% 的樣本數，計算出所需的總樣本數為 56 x 10/8=70 人。 第一年底進行期中分析 (interim analysis) ， 再評估第二年之樣本數 。
- 45. Ultracet in Ankylosing Spondylitis. A DBPC clinical trial. It was assumed that the BASDAI scores of treatment group ( μ1 ) were reduced to 2 from 6, and the BASDAI scores of control group ( μ0 ) were reduced to 3.5 from 6. A 2-side test was used, with type Ⅰ error ( ? ) = 0.05, statistical power (1- ? ) = 0.8 and standard deviation ( SD ) = 1.5. The expected dropout rate was 20%. Finally, the study required only 19 subjects in each group to achieve 95% power. The target sample size for the study was 38.
- 46. Data Management Process CRF CRF 資料接收簽收單 資料確認表格 CRA CDM 臨床統計師 CRF 資料建檔確認表格 資料確認表格簽收單 Data lock Data entry Data entry: 合併校正 Data validation Data analysis
- 47. 個案報告表〈 Case Report Form 〉 任何觀察與發現均正確且完整的記錄在個案報告表上，紀錄者並應簽署。 在個案報告表上的任何更正或原始資料的備份過程，不得消除原始登錄。 更正錯誤記載時，應在原始登錄上畫線刪除，並在其旁填入新的資料，同時由更正者簽署及註明日期，必要時應記錄更正原因。 電腦資料的處理應由經授權的人員加以輸入及更正，任何更正及刪除均應加以記錄。
- 48. 品質保證及品質管制 品質保證及品質管制 QC,QA 監測 Monitor 稽核 Audit 查核 Inspection
- 49. Research team & Resources Principal investigator (PI), Co-PI, Sub-PI Study nurse / Clinical research coordinator (CRC) Statistician / Epidemiologist
- 50. Chung Shan Medical University Hospital Chinese Medicine Clinical Trial Center GCRC (General Clinical Research Center) SMO (site-management organization) model granted by the DOH, Taiwan SMO (site-management organization) GCRC (General Clinical Research Center) CMCTC Chinese Medicine Clinical Trial Center Administrative 2 assistants Clinical affairs 1 CRA/HN 4 CRC Data management 1 Statistic PhD 2 Statistic MS 2 MD 1 Pharmacist 20 Consultants
- 51. 中山醫學大學附設醫院 中藥臨床試驗中心 研究重點 整合醫學與另類醫學 中草藥臨床研究 保健食品及營養補充品 過敏 , 免疫及風濕病之治療 關節炎、痛風、僵直性脊椎炎 過敏、乾癬 免疫調節 臨床試驗服務平台 中草藥暨機能性食品臨床研究中心 生物科技之產學合作窗口
- 52. Contact us: CSMUH Chinese Medicine Clinical Trial Center http://www.csh.org.tw/into/herb James Cheng-Chung Wei, MD, PhD ( 魏正宗 ) [email_address]
Editor's Notes
- #38: This slide shows a schematic overview over the study design. There will be a 6-month short-term period followed by the open-label long term extension which continues until abatacept SC is commercially available in the country or until BMS elects to terminate the trial. In short-term, 50% of the subjects (720) will get randomized to the Abatacept SC arm, while the other half of the subjects will get randomized to the abatacept IV arm. Therefore, all patients receive active drug irrespective of the study arm, and the majority of subjects will receive the new route of administration, Abatacept SC. Patients, who continue in the long-term extension, will administer weekly abatacept injections only, which is then provided as open-label study drug.