A transcriptional signature of LKB1 functional loss defines a large, therapeutically addressable patient population across human cancers

Technical Summary: A Transcriptional Signature of LKB1 Functional Loss

Problem Statement
The tumor suppressor LKB1 (encoded by STK11) is a critical regulator of metabolism, cell polarity, and immune evasion. Its loss drives aggressive tumor phenotypes, particularly in lung adenocarcinoma (LUAD), where it confers resistance to PD-1 blockade. However, current clinical identification of LKB1-deficient patients relies almost exclusively on somatic mutations and homozygous deletions. This genomic approach fails to capture tumors where LKB1 is functionally silenced via non-genomic mechanisms, such as promoter hypermethylation, microRNA-mediated downregulation, or post-translational protein loss. Consequently, the true prevalence of LKB1-deficient tumors is likely underestimated, and emerging therapies aimed at reactivating the LKB1 pathway in structurally wild-type (WT) tumors lack a robust method for patient selection.

Methodology
The authors developed a mechanism-agnostic transcriptional signature to identify functional LKB1 loss across human cancers.

• Data Sources: The study utilized 32 TCGA PanCancer Atlas studies (10,443 tumors) for discovery, alongside four independent non-TCGA LUAD cohorts (OncoSG, CPTAC, CAS, SU2C-MARK), cancer cell line data (CCLE/DepMap), and an isogenic add-back dataset (GSE51266).
• Signature Derivation: In LUAD, the authors defined "positives" as tumors with deleterious STK11 mutations or homozygous deletions (n=55) and "negatives" as confidently intact WT tumors with high STK11 expression (n=177). Using elastic-net penalized logistic regression with nested cross-validation, they derived a 30-gene signature (23 upregulated, 7 downregulated in loss).
• Validation and Benchmarking: The signature was validated across independent cohorts and benchmarked against the previously published Kaufman 16-gene classifier.
• Functional Verification: To confirm the signature reflects kinase activity rather than just mutation status, the authors applied it to isogenic cell lines (A549 and H2122) transduced with wild-type LKB1, kinase-dead LKB1 (K78I), or empty vector.
• Pan-Cancer Analysis: The signature was applied to structurally WT tumors across 12 cancer types to quantify functional loss prevalence. Correlations with co-mutations (KRAS, NRF2 pathway) and immune signatures were analyzed.

Key Results

• Signature Performance: The 30-gene signature achieved an out-of-fold AUROC of 0.986 in the LUAD discovery set and 0.926 in an independent LUSC cohort. It generalized to four independent non-TCGA LUAD cohorts (AUROC 0.92–0.98) and outperformed the Kaufman 16-gene classifier in five of six tested cohorts.
• Causal Link to Kinase Activity: In the add-back experiment, re-expression of wild-type LKB1 significantly reversed the signature score in mutant cell lines. Crucially, the kinase-dead K78I mutant failed to reverse the signature, demonstrating that the signature specifically reads out LKB1 kinase function.
• Expanded Prevalence: Applying the signature to structurally WT tumors revealed that functional loss affects 7.2% of all tumors (11.9% of WT tumors), expanding the total LKB1-loss prevalence 3.7-fold over genomic loss alone (from 2.7% to 9.9%). This "hidden" population is most prominent in esophageal (11.0%), colorectal (10.7%), endometrial (9.1%), and melanoma (8.9%) cancers, where STK11 mutation rates are typically low.
• Genomic and Immunological Context:
  • Co-mutation Axis: In LUAD, signature positivity in WT tumors rose monotonically across a KRAS/NRF2 co-mutation gradient, reaching 64% in KRAS/NRF2 co-mutant tumors. Functional loss was strongly enriched in BRAF-mutant colorectal (25%) and HER2-amplified breast (21%) cancers.
  • Immune Phenotype: Signature-high tumors exhibited an "immune-cold" phenotype, correlating negatively with antigen presentation, T-cell-inflamed gene expression, and IFN-γ programs. This association held true even within genomically WT tumors, suggesting functional loss drives immune evasion independently of mutation status.
• Independence from Common Drivers: A multivariable model showed that common driver alterations explained very little of the functional loss phenotype (pseudo-R² = 0.06), confirming that functional silencing is largely independent of the standard genomic landscape.

Significance and Claims
The paper establishes a rigorous, mechanism-agnostic transcriptional signature that defines a broad, therapeutically addressable population of patients with LKB1 functional loss who would be missed by genomic sequencing alone. The authors claim this work:

1. Redefines the Target Population: It quantifies a functionally deficient population roughly 3.7 times larger than the mutation-defined population, concentrated in specific cancer types (esophageal, colorectal, endometrial, etc.).
2. Validates a Biomarker for Pathway Restoration: By demonstrating that the signature responds to wild-type LKB1 kinase activity but not kinase-dead mutants, the authors provide a necessary biomarker for selecting patients for therapies aimed at reactivating the LKB1 pathway in structurally WT tumors.
3. Identifies Clinical Subgroups: The enrichment of functional loss in BRAF-mutant colorectal and HER2-amplified breast cancers offers practical, readily testable subgroups for screening.
4. Clarifies Immune Evasion: The study links functional LKB1 loss to an immune-cold microenvironment in lung adenocarcinoma, independent of STK11 mutation status, reinforcing the biological basis for resistance to checkpoint blockade in this broader population.

The authors conclude that this signature provides a quantitative basis for precision therapeutics strategies dependent on restoring LKB1 function, moving beyond the limitations of mutation-based ascertainment.