Preview

PDF, English - main document Download (27MB) | Terms of use

Abstract

Multiple myeloma is a haematological disease characterised by the accumulation of malignant plasma cells in the bone marrow. Gene expression data of 839 patients have been obtained using Affymetrix U133 2.0 microarrays in the Multiple Myeloma Research Laboratory at the University Hospital Heidelberg in extended clinical routine. RNA-sequencing was introduced as novel method thought to be superior to DNA-microarray analysis, regarding precision, lower amount of input RNA, and ability to analyse mutated transcripts and splice variants. Therefore, the primary objective of this thesis was to lay the bioinformatic basis for the implementation of RNA-sequencing in applied translational myeloma research and extended clinical routine. This includes i) to establish a practicable analysis pipeline for RNA-sequencing data, ii) to transfer and connect current stratification and classification methods based on microarrays to future RNA-sequencing technology, iii) to discover novel prognostic genes and to develop a novel RNA-sequencing-based risk stratification, iv) to analyse potential and especially actionable therapeutic targets regarding expression, alternative splicing, and mutations, and v) to prospectively test these theoretical target analysis strategies in a consecutive large patient cohort in three clinical relevant examples: BCMA, in development of the T cell bispecific antibody CC-93269 now in clinical testing, CD38, to assess potential mechanism of upfront resistance against anti-CD38 treatment, and cancer testis antigens as vaccination targets, analysing their clinical applicability. Methods. For risk stratification, RNA-sequencing files of 535 multiple myeloma samples were divided into three groups for classification training (∼40%), validation (∼20%) and testing (∼40%). All samples were aligned with STAR and resulting read counts were normalised using edgeR. Each microarray classification was transferred by i) translating the microarray probe sets to Ensembl gene identifiers, ii) creating RNA-sequencing classification as similar as possible to microarray classification, iii) calculating new cutoffs and iv) stratifying the patients. The classifications were validated by examining the proportions of the groups and comparing the survival performance to the microarray classifications of 534 multiple myeloma patients, respectively. Additionally, running log-rank algorithms were used to determine prognostic genes and create a novel RNA-sequencing-based classification, the HDHRS. For potential target analysis, the expression per target, e.g. BCMA or CD38, was determined, definitions for "absent" and "present" expression were established and alternative splice variants were assessed. A pipeline was developed and tested on 142 asymptomatic, 69 relapsed and 767 symptomatic myeloma patients of the CoMMpass cohort. Results. RNA-sequencing can be performed in 90% (in contrast to 80% by microarray) of all patients and the quality of the RNA-sequencing files is sufficient in 97% of all cases. An RNA-sequencing pipeline was successfully implemented. Proliferation-based risk assessment (GPI), risk stratifications (RS, UAMS70, EMC92 and IFM15) as well as molecular classifications (MC, TC) are as predictive on RNA-sequencing as on microarray and are significant prognostic. The novel HDHRS includes 53 discovered prognostic genes and significantly delineates three groups. All stratifications and classifications were validated on the independent test group. On the CoMMpass cohort the stratifications are significant, with laboratory and bioinformatical variations to consider. Nineteen potential targets, five theoretical targets and the mutated target BRAF were assessed. The actionable targets BCMA, CD38, and CD74 are expressed in all, the cancer testis antigens MAGEA3 (33%), RHAMM (88%) and NYESO1/2 (12%) in subfractions of myeloma patients, and 2% show expression of mutated (V600E) BRAF. To assess a potential mechanism of upfront resistance, splice variants of BCMA and CD38 have been analysed. For both targets only one main transcript is expressed. Discussion. RNA-sequencing has been implemented successfully in applied translational myeloma research and extended clinical routine application: All intended risk stratifications and classifications could be successfully transferred and a novel RNA-seq-based risk score (HDHRS) be developed and validated. The target assessment was exemplified regarding patient cohorts for assessment of BCMA, CD38, and cancer testis antigen as vaccination targets. For BCMA, the proportion of patients and plasma cell precursor stages expressing the target has been assessed, to determine potential clinical side effects of the developed compound CC-93269. This exemplifies the usefulness of RNA-sequencing in assessing potential targets for compound development. Likewise, RNA-sequencing is useful to assess expression prior to therapy, to exclude treatments for patients lacking expression of the respective targets. BCMA and CD38 splice variant analyses showed that alternative splicing does not confer up-front resistance of anti-BCMA and anti-CD38 treatment. Both analysis strategies can and will be directly applied to future compound developments. Hence, this thesis is the proof of principle, that all steps necessary for targeted and risk adapted treatment strategy in myeloma can be performed using RNA-sequencing. The results of this thesis, including the developed pipeline, are the fundamental step for the implementation of RNA-sequencing for risk adapted and targeted treatment strategies in clinical routine.