Skip to main content
Premium Trial:

Request an Annual Quote

Long Noncoding RNAs Predictive of Immunotherapy Response, Study Finds

NEW YORK – In a new study published in JAMA Network Open, researchers have analyzed genomic, transcriptomic, and immune profiles of patients who participated in a Phase II clinical trial and associated their immune functional states with distinct classes of long noncoding RNA (lncRNA). They found that these classes appeared to be predictive of the patients' immunotherapy outcomes.  

The study, led by Herui Yao and Erwei Song from Sun Yat-sen Memorial Hospital in Guangzhou, China, analyzed genomic profiles and long lncRNA data of 3,370 patients, including 348 bladder cancer patients from the Phase II IMvigor210 trial, 71 melanoma patients from The Cancer Genome Atlas (TCGA), and 2,951 pan-cancer TCGA patients.

Previous studies have shown that lncRNAs interact with the immune microenvironment. In particular, tumor-specific cytotoxic T lymphocytes (CTLs) or tumor cells with high nuclear factor (NF)-kappa-B-interacting lncRNA was associated with poor prognosis. Similarly, high hypoxia-inducible factor 1-alpha-stabilizing lncRNA in tumor-associated macrophages also corresponded with poor prognosis.

The researchers wanted to explore in this study whether lncRNAs could serve as an effective predictive biomarker for cancer immunotherapy. All IMvigor210 participants had their tumors profiled with the FoundationOne CDx sequencing test, and whole-exome sequencing was performed in the TCGA melanoma cohort. Transcriptome RNA sequencing was performed on all tumor tissues.

Approximately 400 patients received immunotherapy, including 348  bladder cancer patients who received treatment with the PD-L1 inhibitor atezolizumab (Roche's Tecentriq) as part of the IMvigor210 trial and 71 melanoma patients from the TCGA who received either anti-PD-1, anti-CTLA4, cytokine tumor vaccine, or other immunotherapy treatments.

A pan-cancer multicohort from TCGA that was not treated with immunotherapy was also included in the analysis. This included patients with lung adenocarcinoma, lung squamous cell carcinoma, breast cancer, bladder cancer, and melanoma.

Two distinct lncRNA-based classes were identified in the IMvigor210 trial participants. These two classes were associated with different overall survivals. Patients who had longer survival tended to have lower expression of immune cells, immune checkpoints, and human leukocyte antigens compared to patients who had shorter survivals.

"We hypothesized that immune cells of patients with less survival benefit were in a nonfunctional state and thus referred to the novel lncRNA profile with favorable survival as the immune-functional class and the profile with poor survival as the immune-nonfunctional class," the authors wrote.

In order to describe the essential lncRNAs that altered the functional immune state, the researchers performed LASSO and random forest analysis on the 49 lncRNAs from the two distinct lncRNA-based classes. Further, the LASSO algorithm that was used to identify 29 of the 49 lncRNAs was used to construct lncRNA scores. Patients with lower scores had greater overall survival, overall response, and duration of complete response. The four most important lncRNAs were AC002116-2, AP000251-1, TMEM147-AS1, and NKILA.

For example, Kaplan-Meier analysis found that in bladder cancer patients who received immunotherapy, low NKILA expression was associated with a longer overall survival. NKILA expression was lower in the immune-functional class than in the immune nonfunctional class.

The authors proposed that targeting NKILA in immune cells could be a promising way to reverse the immune nonfunctional state and enhance effects of immunotherapies.

Next, the researchers combined the lncRNA signature with cytotoxic T lymphocyte (CTL) infiltration to categorize bladder cancer and IMvigro210 patients into four different classes, based on overall response rate and overall survival.

Patients with functional immune response and high CTL infiltration were classified as the immune-active class. Patients with functional immune response and low CTL infiltration were classified as the immune-exclusion class. Patients with nonfunctional immune response and high CTL infiltration were classified as the immune-dysfunctional class. Patients with nonfunctional immune response and low CTL infiltration were classified as the immune-desert class.

The immune-active class had higher expression of immune molecules compared to the immune-exclusion class, and the immune-dysfunctional class had higher expression of immune molecules than the immune-desert class. The immune-active class had the most favorable overall survival and overall response rate, while patients in the immune-desert class had the worst clinical outcomes. Patients in the immune-dysfunctional class had a worse overall survival than patients in the immune-exclusion class.

"The four novel distinct classes identified in this study indicated that immune molecular classification of aspects of both immune exclusion and immune dysfunction could be informative for understanding patterns of immune escape (i.e. cancers could escape immunologic destruction) and for selection of candidates for cancer immunotherapy," the authors wrote. "Specific differences were also found in the signaling pathways that characterize each of the four classes, which can inform novel combination strategies, such as immunotherapy combined with targeted therapy, tumor vaccines, or cellular immunotherapy."

For example, immune-active patients could derive additional benefit from immunotherapy plus p53 inhibitor because the p53 signaling pathway is highly enriched in this class. Additionally, the immune-desert class could be treated with a combination of PI3K-Akt inhibitors and checkpoint inhibitors because the PI3K-Akt signaling pathway is highly enriched in this class.

The authors concluded by offering that lncRNA scores should be added to previously developed multi-omics biomarkers for predicting overall survival benefit from cancer immunotherapy, such as tumor mutational burden, PD-L1 expression, and CTL infiltration.