Abstract 1704P
Background
Recently, antibody-drug conjugates or chimeric antigen receptor T-cell therapies which specifically bind to cancer specific membrane proteins have shown promising clinical outcomes in various solid cancers. However, a standardized methodology to discriminate cancer specific membrane targets has not yet been fully optimized. This study demonstrates a novel methodology to discover cancer-specific membrane targets by utilizing a multimodal approach including genomic, transcriptomic, and pathologic image data.
Methods
We identified genes that are frequently amplified and distinctly expressed only in matched tumor samples, based on copy number alteration and transcriptomic data gathered from The Cancer Genome Atlas (TCGA). Next, we utilized the Human Protein Atlas (HPA) database to select genes that encode proteins localizing to the plasma membrane. Finally, we applied Lunit SCOPE IO to assess TIL density in samples with target gene amplification.
Results
We found 500 genes which were all amplified in ≥5.0% of all samples analyzed. Most of the frequently amplified genes were located in 8q22-24 (N=251), 3q26-29 (N=228), and 11q13 (N=14). After excluding genes that were not registered in the HPA (N=211), 36 out of 289 (12.5%) genes that encode proteins localized to the plasma membrane were filtered in. Among those, 25 genes had significantly higher expression in cancer samples with target gene amplification compared to cancer samples without amplification and compared to normal adjacent tissue (false discovery rate < 1%). We noted two novel targets; ATP13A5 and KCNK9, located in 3q29 and 8q24.3, respectively, and frequently amplified in lung squamous cell carcinoma (29.8% and 4.3%), esophageal adenocarcinoma (16.5% and 11.5%), and ovarian cancer (16.4% and 26.9%, respectively). Interestingly, tumors with ATP13A5 amplification showed higher stromal TIL density compared to those without amplification (mean fold change = 1.53, P = 8.69 x 10-14).
Conclusions
Integrating copy number variation, transcriptomic data, subcellular protein location information, and pathologic data unveils 25 novel cancer-specific membrane targets which can be considered for further development.
Clinical trial identification
Editorial acknowledgement
Legal entity responsible for the study
The authors.
Funding
Has not received any funding.
Disclosure
C. Ock: Financial Interests, Personal, Full or part-time Employment: Lunit Inc.; Financial Interests, Personal, Invited Speaker: Ybiologics; Financial Interests, Personal, Stocks/Shares: Lunit Inc., Ybiologics. All other authors have declared no conflicts of interest.