Investigation of treatment-guiding gene signatures in cancer using RNA-sequencing data

Abstract

Cancer is the leading cause of death in the world. Multiple different treatment options are available, target therapy and immunotherapy being two. One potential target is Fibroblast Growth Factor Receptors (FGFRs) and there are both available treatments and new treatments being investigated for this. However, only around 25% of patients benefit from the immunotherapy treatment Immune Checkpoint Inhibitors (ICIs) and only around 20% of advanced urothelial patients have FGFR3 alterations that can be targeted. This means that methods for predicting which patients can benefit from these treatments are needed, for example gene signatures. After a literature study to find potential gene signature for this, a collection of datasets with samples from bladder and lung cancer patients was analysed in Qlucore Omics Explorer (QOE) and RStudio. Scores were created in R and Kaplan-Meier plots were generated in QOE to see if there was a difference in overall survival and progression-free survival between samples with “High” and “Low” scores. All tested gene signatures for ICI response gave better results than the currently used genes PD-1 (programmed cell death protein-1) and PD-L1 (programmed death-ligand 1), and Expanded immune signature, TIS and IFNγ were the most promising. QOE was also used to find potential genes for creating a gene signature to predict if FGFR3 alterations were present. A combination of the gene expression of FGFR3 and a six-gene signature gave the best result in correctly determining if FGFR3 alterations were present.

Popular Abstract

Cancer is the leading cause of death worldwide and caused almost 10 million deaths in 2022 alone. Many different treatment options are available, such as surgery, radiation therapy, chemotherapy, laser therapy, target therapy and immunotherapy. However, all of them come with their own side effects. In order to reduce negative effects to the patient and the treatment cost while still have a treatment that is as effective as possible, ways of predicting if a treatment is going to be beneficial to the specific patient are needed. This report focuses on using gene expression data obtained by RNA sequencing to predict if the patient will have a beneficial response to a certain type of immunotherapy and a certain type of target therapy.
Immunotherapy treatment stimulates the patients’ own immune system in order for it to recognize tumour cells as something that needs to be killed. This report focuses on immune checkpoint inhibitors (ICIs) that affects the programmed cell death protein-1 (PD-1) and programmed death-ligand 1 (PD-L1). These treatments have a larger frequency of durable response than other treatments, but it is still only around 25%. The methods used today for predicting response to this treatment have proven to be not very accurate.
The oncogenic effect of Fibroblast Growth Factor Receptors (FGFRs) has been reported and is pointed out as a potential therapeutic target. FGFRs help control how different cells grow, survive and move. Alterations to the genes that code for these show strong correlations to the formation of different types of cancer. In bladder cancer this is particularly prominent, since 20% of patients with advanced urothelial carcinoma have alterations to the gene FGFR3, one of the four genes coding FGFRs.
Scores were created using the gene expression data and were compared to the gene expression of two single genes currently used in standard of care in their capabilities of predicting response to ICI treatment. All scores were found in literature and selected based on their positive results in the previous studies. The scores that were created to reflect specific immune cell markers gave slightly improved results compared to the ones currently used, while three of the scores investigated gave better results, most likely reflecting a wider immune response. They showed significantly longer survival for patients with high scores compared to the ones with low scores in multiple datasets. These three, Expanded immune signature, TIS and IFNγ, are quite similar but do show promise in being used in practice in some capacity to predict if the treatment would be beneficial for the patient in question.
Different scores were created to determine if it is possible to say if there is an alteration to the FGFR3 gene based on the gene expression data. Cut-off values for the scores were determined to discriminate between the two groups No alterations and Alterations. The final method for this used both the gene expression of the FGFR3 gene and a score created using the 6 genes found most important for distinguish the groups. This method reduced the amount of false negative results to 1.3% and the amount of false positive results to 1.5%.

• author: Gralén, Kajsa
• supervisor: Caroline Brorsson; Anna Sandström Gerdtsson
• organization: Department of Immunotechnology
• course: KIMM05 20251
• year: 2025
• type: H2 - Master's Degree (Two Years)
• subject: Technology and Engineering
• keywords: Gene signatures, Immunotherapy, Cancer
• language: English
• id: 9205330
• date added to LUP: 2025-06-25 13:14:28
• date last changed: 2025-06-25 13:20:42