129P - Integrating large data to unveil vulnerabilities for patients with hot tumors resistant to checkpoint inhibition

Background

Most studies aimed to enhance immune checkpoint blockade (ICB) efficacy by converting an immune-inert cold tumor into an immune-inflamed hot one, and almost all previous biomarkers were explained by a hot feature. However, ICB resistance largely occurs even when tumors are highly hot. We and others have sporadically revealed some ICB resistance mechanisms specifically occurring in hot tumors. This study integrates big data to precisely classify hot tumors and identify potential targets.

Methods

We included 7099 in-house and external samples with available transcriptomic data from 73 ICB randomized trials or cohorts across 18 tumor types. Hot level was evaluated by classical mRNA markers. We calculated the meta-effect size of differentially expressed genes (DEGs) between responders and non-responders/long and short survival groups across studies.

Results

We used an iterative algorithm to reveal two populations: 1) those whose ICB efficacy are lacking due to having a cold tumor (Hot-dependent), and 2) those whose tumors are hot enough that hot level no longer influence outcomes (Hot-independent). Across all cancer types, over half (51.4%) of tumors are Hot-independent. Kidney cancer and nasopharyngeal carcinoma are mostly Hot-independent, while urothelial and lung cancer are mostly Hot-dependent (Hot-independent percentage: 82.4%, 81.2%, 23.1%, 31.3%, respectively). The significant DEGs are distinct between the Hot-independent and Hot-dependent groups. We developed a de novo signature (HotR score) to identify patients with hot tumors who still cannot benefit from ICB. Hot tumors with a high HotR score had similar ICB efficacy compared with cold tumors (OS: HR=1.05, P=0.74; PFS: HR=0.90, P=0.38). Hot tumors showed better outcomes of ICB over control treatment, but this advantage lost or even reversed in hot tumors when scoring high by HotR, especially in renal cell carcinoma (OS: HR=1.13, P=0.48; PFS: HR=1.14, P=0.5). Importantly, our signature can prioritize personalized targets to combine with ICB for hot tumors.

Conclusions

Our findings help distinguish patients with pseudo-hot ICB-refractory tumor and select targets to maximize their ICB efficacy.

Clinical trial identification

Editorial acknowledgement

Legal entity responsible for the study

A. Li.

Funding

This study was supported by the National Natural Science Foundation of China (81972898, 82172713, and 81972556), the Guangdong Basic and Applied Basic Research Foundation (2023B1515020008), and the Fundamental Research Funds for the Central Universities, Sun Yat-sen University (22ykqb15).

Disclosure

All authors have declared no conflicts of interest.