臺灣博碩士論文加值系統

中文摘要
癌症是全球人類死亡的主要原因。許多研究致力於尋找疾病的治療方法，而免疫療法為目前新的治療方法之一。在免疫療法中，最受歡迎的免疫治療藥物是免疫檢查點抑製劑，像是anti-PD-1和anti-CTLA-4。雖然已經證實了這些藥物的功效，但仍有一些患者對於治療的效果沒有很明顯、甚至沒有反應的情況發生。因此，如何辨別那些對免疫療法藥物具有效的病患將是一個重要問題。文獻顯示突變負擔（非同義點突變的總數）可能在免疫療法療效上為有用的預測生物標誌。計算腫瘤的突變負擔需要全外顯子定序才能達到，然而因為成本及花費的時間等因素，全外顯子定序目前無法成為常規臨床檢測。因此，本研究以肺腺癌為例，想要建構由一小群基因所組成的突變負擔估計模型以預測癌症免疫療法的療效。
根據從癌症基因體圖譜資料庫下載的體細胞突變數據，我們發展了一個建構突變負擔估計模型的方法。首先基於突變頻率、CDS長度以及基因突變狀態與突變負擔的關聯性等篩選條件來選擇候選基因。隨後，使用最小平方參數估測法和貝葉斯訊息準則進行模型的建構。我們所建構的肺腺癌突變負擔估計模型是由24個基因所組成，將其應用於兩筆獨立的驗證資料來檢驗估計模型的表現，發現預測和實際突變負擔之間的R2為0.7626。由於其中一筆驗證資料中有包含免疫療法療效反應的資訊，所以預測的突變負擔也被用於將病人分類為持久的臨床益處或不具有持久的效益，這樣的分類有85%的靈敏度，93%的特異性以及89 %的準確性。基於建構的突變負擔估計模型，我們可以設計客製化的基因模組，將模型中選擇的基因放入模組中，藉此基因模組來取代全外顯子定序。如此一來，評估突變負擔所需的成本和時間可以顯著降低，癌症免疫治療功效的預測也更可能成為常規臨床檢測。
關鍵字：突變負擔，免疫療法，數學模型，癌症基因體，次世代定序

Abstract
Cancer is the leading cause of human death worldwide. Many researches are dedicated to finding the therapeutics of the disease, and immunotherapy is one of the new therapeutic approaches. The most popular drugs of immunotherapy is the immune checkpoints inhibitors, such as anti-PD-1 and anti-CTLA-4. Although the efficacy of these drugs have been demonstrated, there are still some patients who do not respond to them. Therefore, how to identify the patients potentially respond to the drugs would be an essential question. Literature evidences have shown that mutation burden (the total number of nonsynonymous point mutations) might be a useful predictive biomarker for treatment responses. However, whole-exome sequencing is needed to be performed to unravel the mutation burden of the tumor, which is not cost and time-effective for standard clinical test. Therefore, focused on lung adenocarcinoma, the objective of this study is to construct a mutation burden estimation model of a small set of genes for predicting efficacy of cancer immunotherapy.
With the somatic mutation data downloaded from The Cancer Genome Atlas (TCGA) database, a computational framework was developed to construct the mutation burden estimation model. Candidate genes were selected based on mutation frequency, CDS length, and the association between mutation status with mutation burden. Subsequently, least squares parameter estimation and Bayesian information criterion were used for model construction. The constructed estimation model consisted of 24 genes, which was applied to two independent data to test the performance. R2 between predicted and actual mutation burden is 0.7626. Since there are treatment response information for immunotherapy in the second independent data, the predicted mutation burden were also employed to classify the samples as durable clinical benefit (DCB) or no durable benefit (NDB) with 85% sensitivity, 93% specificity, and 89% accuracy. Based on the constructed estimation model, we can design a customized panel of targeted sequencing of these selected genes instead of whole-exome sequencing. Consequently, the cost and time needed for assessing mutation burden would be significantly decreased and the efficacy prediction of cancer immunotherapy would be more feasible for standard clinical examination.

Keywords: mutation burden, immunotherapy, mathematical model, cancer genomics, next generation sequencing

Contents
中文摘要......i
Abstract......ii
Contents......iv
List of Figures......v
List of Tables......vi
Chapter 1 Introduction......1
Chapter 2 Materials and Methods......4
2.1 Overview of the method......4
2.2 Data......6
2.3 Mutation matrix construction......6
2.4 Selection of candidate genes......8
2.4.1 Mutation frequency......8
2.4.2 Coding DNA sequence (CDS) length......9
2.4.3 Wilcoxon rank sum test comparing the mutation burden of patients in mutated and wild type groups......9
2.5 Construction of mutation burden estimation model......10
2.5.1 Mutation burden estimation model......10
2.5.2 Least squares parameter estimation......11
2.5.3 Model selection......11
2.6 Performance evaluation and validation......12
Chapter 3 Results......14
Chapter 4 Discussion and Conclusions......25
Reference......30

List of Figures
Figure 1 | Mutation burden and the corresponding clinical responses of patients.......3
Figure 2 | Flowchart of the computational framework for mutation burden estimation model construction.......5
Figure 3 | Schematic diagram of the mutation matrix.......8
Figure 4 | Criteria for candidate gene selection.......10
Figure 5 | Venn diagram of three criteria for candidate gene selection.......16
Figure 6 | Performance evaluation of the constructed model for training data from TCGA.......18
Figure 7 | Performance evaluation of the constructed model for independent validation data from Imielinski et al. and Rizvi et al.......18
Figure 8 | The survival analysis comparing PFS in patients with high predicted mutation burden with those with low predicted mutation burden.......20
Figure 9 | ROC curve for the correlation of predicted mutation burden with DCB.......21
Figure 10 | Performance evaluation of the constructed model for independent validation data from Rizvi et al.......21
Figure 11 | The empirical distribution of R2 for 10,000 random models.......23
Figure 12 | The empirical distribution of prediction accuracy of cancer immunotherapy for 10,000 random models.......23
Figure 13 | Performance evaluation of the constructed model for melanoma data from Snyder et al. (n=64) and Van Allen et al. (n=110).......26
Figure 14 | Performance evaluation of the constructed model for melanoma data.......27

List of Tables
Table 1 | Types of Variant_Classification in TCGA LUAD somatic mutation data.......7
Table 2 | Selected candidate genes and related information.......14
Table 3 | The genes and their corresponding parameters in the constructed mutation burden estimation model.......17
Table 4 | The comparison of genes in our constructed model with other cancer gene panels.......28

Reference
1. Ott PA, Hodi FS, Robert C: CTLA-4 and PD-1/PD-L1 blockade: new immunotherapeutic modalities with durable clinical benefit in melanoma patients. In.: AACR; 2013.
2. Topalian SL, Hodi FS, Brahmer JR, Gettinger SN, Smith DC, McDermott DF, Powderly JD, Carvajal RD, Sosman JA, Atkins MB: Safety, activity, and immune correlates of anti–PD-1 antibody in cancer. New England Journal of Medicine 2012, 366(26):2443-2454.
3. Mahoney KM, Freeman GJ, McDermott DF: The next immune-checkpoint inhibitors: PD-1/PD-L1 blockade in melanoma. Clinical therapeutics 2015, 37(4):764-782.
4. Robert C, Schachter J, Long GV, Arance A, Grob JJ, Mortier L, Daud A, Carlino MS, McNeil C, Lotem M: Pembrolizumab versus ipilimumab in advanced melanoma. New England Journal of Medicine 2015, 372(26):2521-2532.
5. Brahmer J, Reckamp KL, Baas P, Crinò L, Eberhardt WE, Poddubskaya E, Antonia S, Pluzanski A, Vokes EE, Holgado E: Nivolumab versus docetaxel in advanced squamous-cell non–small-cell lung cancer. New England Journal of Medicine 2015, 373(2):123-135.
6. Rosenberg JE, Hoffman-Censits J, Powles T, Van Der Heijden MS, Balar AV, Necchi A, Dawson N, O'Donnell PH, Balmanoukian A, Loriot Y: Atezolizumab in patients with locally advanced and metastatic urothelial carcinoma who have progressed following treatment with platinum-based chemotherapy: a single-arm, multicentre, phase 2 trial. The Lancet 2016, 387(10031):1909-1920.
7. Seiwert TY, Haddad RI, Gupta S, Mehra R, Tahara M, Berger R, Lee S-H, Burtness B, Le DT, Heath K: Antitumor activity and safety of pembrolizumab in patients (pts) with advanced squamous cell carcinoma of the head and neck (SCCHN): preliminary results from KEYNOTE-012 expansion cohort. In.: American Society of Clinical Oncology; 2015.
8. Weinstock M, McDermott D: Targeting PD-1/PD-L1 in the treatment of metastatic renal cell carcinoma. Therapeutic advances in urology 2015, 7(6):365-377.
9. Garon E, Gandhi L, Rizvi N, Hui R, Balmanoukian A, Patnaik A, Eder J, Blumenshein G, Aggarwal C, Soria J: LBA43 - Antitumor activity of pembrolizumab (Pembro; MK-3475) and correlation with programmed death ligand 1 (PD-L1) expression in a pooled analysis of patients (PTS) with advanced non-small cell lung carcinoma (NSCLC). Annals of Oncology 2014, 25(suppl 4):mdu438. 451.
10. Rizvi NA, Hellmann MD, Snyder A, Kvistborg P, Makarov V, Havel JJ, Lee W, Yuan J, Wong P, Ho TS: Mutational landscape determines sensitivity to PD-1 blockade in non–small cell lung cancer. Science 2015, 348(6230):124-128.
11. Campesato LF, Barroso-Sousa R, Jimenez L, Correa BR, Sabbaga J, Hoff PM, Reis LF, Galante PAF, Camargo AA: Comprehensive cancer-gene panels can be used to estimate mutational load and predict clinical benefit to PD-1 blockade in clinical practice. Oncotarget 2015, 6(33):34221.
12. Johnson DB, Frampton GM, Rioth MJ, Yusko E, Xu Y, Guo X, Ennis RC, Fabrizio D, Chalmers ZR, Greenbowe J: Targeted next generation sequencing identifies markers of response to PD-1 blockade. Cancer Immunology Research 2016, 4(11):959-967.
13. Roszik J, Haydu LE, Hess KR, Oba J, Joon AY, Siroy AE, Karpinets TV, Stingo FC, Baladandayuthapani V, Tetzlaff MT: Novel algorithmic approach predicts tumor mutation load and correlates with immunotherapy clinical outcomes using a defined gene mutation set. BMC medicine 2016, 14(1):168.
14. The Cancer Genome Atlas Research Network: Comprehensive molecular profiling of lung adenocarcinoma. Nature 2014, 511(7511):543-550.
15. Imielinski M, Berger AH, Hammerman PS, Hernandez B, Pugh TJ, Hodis E, Cho J, Suh J, Capelletti M, Sivachenko A: Mapping the hallmarks of lung adenocarcinoma with massively parallel sequencing. Cell 2012, 150(6):1107-1120.
16. Kinsella RJ, Kähäri A, Haider S, Zamora J, Proctor G, Spudich G, Almeida-King J, Staines D, Derwent P, Kerhornou A: Ensembl BioMarts: a hub for data retrieval across taxonomic space. Database 2011, 2011:bar030.
17. Snyder A, Makarov V, Merghoub T, Yuan J, Zaretsky JM, Desrichard A, Walsh LA, Postow MA, Wong P, Ho TS: Genetic basis for clinical response to CTLA-4 blockade in melanoma. New England Journal of Medicine 2014, 371(23):2189-2199.
18. Van Allen EM, Miao D, Schilling B, Shukla SA, Blank C, Zimmer L, Sucker A, Hillen U, Foppen MHG, Goldinger SM: Genomic correlates of response to CTLA-4 blockade in metastatic melanoma. Science 2015, 350(6257):207-211.