3546 Background: PIWIL1 is a cancer testis antigen overexpressed in solid tumors, particularly CRC, and is associated with advanced stage and poor prognosis. It is potentially a new therapeutic target as PIWIL1 is not expressed in adult somatic tissue. A novel T cell receptor/PIWIL1 bispecific antibody is in development, restricted to individuals with human leukocyte antigen (HLA) A-02. We sought to characterize the molecular landscape of PIWIL1 in CRC. Methods: 26,581 CRC tumors were tested by next-generation sequencing (NGS) on DNA (592-gene or whole exome [WES]) and RNA (whole transcriptome [WTS]) at Caris Life Sciences (Phoenix, AZ). dMMR/MSI-H was tested by immunohistochemistry (IHC) and NGS, HLA genotypes by WES and PD-L1 by IHC (SP142, 2+, 5%). Tumor mutational burden (TMB) high was defined as ≥ 10 mt/Mb. RNA expression was used to estimate the tumor microenvironment using QuantiSEQ and RNA signatures (interferon gamma [IFNγ]; T-cell inflamed [TIS]) predictive of immune checkpoint blockade (IO) response. The top (H) and bottom (L) quartiles of PIWIL1 expression were compared using Chi-square/Fisher-Exact, and significance was determined as p adjusted for multiple comparisons (Q<0.05). Real-world overall survival (rwOS) was obtained from insurance claims and calculated from tissue collection to last contact. Results: Patients with PIWIL1-H were older (median age 65 v 61, Q<0.01) and more likely to be female (50.8% v 41.3%, Q<0.01), compared to PIWIL1-L. PIWIL1 median expression was higher in right- v left-sided primary tumors (2.0 v 1.3 transcripts per million [TPM], Q<0.05). No significant difference in PIWIL1 expression was observed among HLA genotypes. dMMR/MSI-H, TMB-H and PD-L1+ were higher in PIWIL1-H v PIWIL1-L (13.9% v 6.5%, 17.6% v 10.4%, 6.4% v 3.4%, all Q<0.05). IO-related gene expression ( LAG3, IDO1, CTLA4 and others) were positively associated with PIWIL1 expression (fold change [FC]: 1.4-2.7, Q<0.05). High IFNγ and TIS were associated with PIWIL1-H (Q<0.001). PIWIL1 expression showed a positive association with infiltration of neutrophils, monocytes, CD4+T, dendritic and NK cells (FC: 1.3-2.9, Q<0.05) while Tregs and M1 macrophages had a negative association (0.8-0.9, Q<0.05). In pMMR/MSS CRC, KRAS, BRAF and TP53 mutations and loss of heterozygosity were higher in PIWIL1-H v PIWIL1-L (81% v 69%, 53% v 41%, 16% v 3%, 14% v 10%, all Q<0.05) but not for PIK3CAmutations (14% v 17%). For rwOS, PIWIL1-H had longer median OS than PIWIL1-L, 31.2 v 28.8 months (HR 0.95; 95% CI 0.90-0.99, p=0.043). There was no significant rwOS difference observed in IO-treated tumors. Conclusions: PIWIL1-H CRC is associated with higher rates of dMMR/MSI-H, TMB-H and PD-L1+ as well as IO-related gene expression and signatures that are predictive of response to IO. These data suggest the PIWIL1-H subpopulation could potentially derive substantial benefit from PIWIL1-targeted immunotherapy which should be evaluated in clinical trials.
Read full abstract