Lung adenocarcinoma has high mortality and limited prognostic tools. Lung adenocarcinoma (LUAD) is the most prevalent subtype of non-small cell lung cancer, accounting for approximately 40% of all lung cancer cases worldwide. Despite advances in treatment, LUAD remains a major cause of cancer-related mortality due to late-stage diagnosis, high metastatic potential, and the absence of reliable prognostic biomarkers. Current indicators including TNM staging and histopathological classification fail to capture the molecular heterogeneity of LUAD comprehensively, reducing their accuracy for predicting outcomes and guiding targeted therapies.
Long non-coding RNAs have emerged as critical regulators of cancer biology. LncRNAs are RNA transcripts longer than 200 nucleotides that do not encode proteins but influence gene expression, chromatin remodeling, and cellular signaling pathways. Studies across cancer types have demonstrated that lncRNAs contribute to hallmarks of cancer including uncontrolled proliferation, evasion of apoptosis, and metastasis. In LUAD specifically, lncRNAs such as MALAT1 and HOTAIR have been implicated in promoting epithelial-to-mesenchymal transition and modulating oncogenic signaling pathways.
LncRNAs act through diverse molecular mechanisms. Oncogenic lncRNAs frequently function as competing endogenous RNAs (ceRNAs), sponging tumor-suppressive microRNAs to enhance the expression of oncogenic target genes. Others interact with chromatin-modifying complexes to reshape the epigenetic landscape; for example, HOTAIR recruits the PRC2 complex to silence tumor suppressor genes. Some lncRNAs regulate transcriptional and post-transcriptional processes, and large-scale integrative analyses have shown subsets of lncRNAs are closely associated with tumor immune infiltration and patient prognosis.
This study integrates multiple clinical dimensions to identify robust prognostic lncRNAs. Prior studies often identified lncRNAs based on differential expression alone, without evaluating whether expression patterns consistently track with disease severity and patient outcomes across multiple clinical dimensions. This study addresses that gap by combining stage-wise differential expression, survival association analysis, and monotonic trend analyses across tumor stage, nodal involvement, and tumor size using TCGA-LUAD data from 488 tumors and 58 normal lung tissue samples.
TCGA-LUAD data from the TANRIC platform. Pre-processed lncRNA expression data (reads per kilobase per million, RPKM) were obtained from the TANRIC platform, which quantifies lncRNA expression from TCGA RNA-seq BAM files. The dataset included 488 LUAD tumors and 58 normal lung tissue samples, with expression data for over 12,000 annotated lncRNAs. An expression filter retained only lncRNAs with RPKM greater than 1 in at least 20% of all samples (110 of 546 total samples), reducing the candidate set from over 12,000 to 959 lncRNAs before downstream analysis.
Stage-wise differential expression using Welch's t-test with FDR correction. Differential expression between LUAD tumors and normal lung tissues was assessed independently for each tumor stage (I, II, III, IV). Given unequal sample sizes between tumor and normal groups, a two-sided Welch's t-test was used. Nominal p-values were adjusted using the Benjamini-Hochberg false discovery rate procedure. LncRNAs with FDR less than 0.05 and absolute log2 fold-change of at least 1 (corresponding to at least 2-fold change) were considered significantly dysregulated.
Intersection strategy to identify consistently dysregulated lncRNAs. Rather than focusing on stage-specific signals, the study adopted an intersection-based strategy requiring consistent dysregulation across all four tumor stages. This yielded 68 lncRNAs consistently upregulated across all LUAD stages relative to normal lung, and 67 consistently downregulated. This conservative approach ensured that retained candidates represent robust tumor-associated signals rather than isolated or stage-specific findings.
Survival analysis and trend testing across clinical parameters. Kaplan-Meier survival analyses were performed for all 68 consistently upregulated lncRNAs using TCGA clinical data through the ENCORI/starBase platform, with patients stratified by median expression and survival differences assessed by log-rank test. For progression analysis, Spearman rank correlation was applied to evaluate monotonic expression trends across ordered tumor stages, nodal status (Normal, N0-N3), and tumor size categories (Normal, T1-T4), with FDR correction applied to trend p-values.
Stage-wise differential expression identified 68 consistently upregulated lncRNAs. Among the 959 filtered lncRNAs, 168 were upregulated in stage I, 155 in stage II, 143 in stage III, and 96 in stage IV relative to normal lung tissue. Intersection across all four stages yielded 68 lncRNAs that were consistently upregulated and 67 that were consistently downregulated. The declining number of upregulated lncRNAs at later stages reflects increasing transcriptomic complexity and the smaller sample sizes available at stage IV (n=17 tumors).
Five lncRNAs met stringent survival criteria. Kaplan-Meier survival analysis across all 68 consistently upregulated lncRNAs identified five that satisfied predefined selection criteria: hazard ratio of at least 1.5 and log-rank p-value less than 0.05. These were AC245595.1, FAM83A-AS1, CYTOR, MIR4435-2HG, and AP001453.2. Higher expression of each of these five lncRNAs was significantly associated with poorer overall survival in LUAD patients. No manual curation or additional biological assumptions were applied at the selection stage.
Four downregulated lncRNAs showed favorable survival associations. Among the 67 consistently downregulated lncRNAs, Kaplan-Meier analysis identified four whose higher expression was significantly associated with improved overall survival (hazard ratio of at most 0.65, log-rank p less than 0.05), suggesting potential tumor-suppressive or protective roles. These candidates represent a separate class of lncRNA biomarkers whose loss during LUAD progression may contribute to disease aggressiveness.
Subcellular localization suggests functional roles for top candidates. Cytoplasmic-Nuclear Relative Concentration Index (CN-RCI) data from the lncATLAS database was analyzed for FAM83A-AS1, CYTOR, and MIR4435-2HG across multiple human cell lines. CYTOR and MIR4435-2HG showed predominantly positive CN-RCI values, indicating cytoplasmic enrichment, which is consistent with post-transcriptional regulatory roles such as ceRNA activity and RNA-protein interactions. Localization data were not available for AC245595.1 and AP001453.2.
59 lncRNAs show progressive expression trends across LUAD stages. Formal rank-based trend testing applied to the 68 consistently upregulated lncRNAs identified 59 with statistically significant monotonic trends across ordered stages I through IV (FDR-adjusted q less than 0.05). A representative subset of 12 lncRNAs with the strongest positive trends (trend coefficient greater than 0.8) demonstrated clear progressive increases in expression from stage I through stage IV, establishing that a substantial proportion of the consistently upregulated candidates also track quantitatively with disease progression.
FAM83A-AS1, CYTOR, and MIR4435-2HG show significant monotonic trends with nodal involvement. Rank-based trend analysis across ordered nodal stages (Normal, N0, N1, N2, N3) identified statistically significant positive monotonic trends for FAM83A-AS1, CYTOR, and MIR4435-2HG. This means that expression of these three lncRNAs increases progressively with greater lymph-node involvement, linking their expression to one of the most critical determinants of LUAD prognosis and staging. AC245595.1 and AP001453.2 did not show significant monotonic nodal trends.
Expression also increases with primary tumor size. Additional rank-based trend testing across tumor size categories (Normal, T1, T2, T3, T4) confirmed significant positive correlations between expression rank and tumor size for FAM83A-AS1, CYTOR, and MIR4435-2HG. This parallel behavior across both nodal status and tumor size, two independent measures of disease burden, substantially strengthens the evidence that these three lncRNAs are embedded in molecular programs associated with tumor growth and metastatic spread.
No sex-based expression differences detected. Expression levels of all five selected lncRNAs were compared between male and female LUAD patients using two-sided Welch's t-tests. No statistically significant sex-specific differences were observed for any of the analyzed lncRNAs. This indicates that the prognostic relevance of FAM83A-AS1, CYTOR, and MIR4435-2HG applies across both sexes, supporting their potential utility as sex-agnostic biomarkers for LUAD risk stratification.
CYTOR is a well-established oncogenic lncRNA in LUAD. CYTOR has been repeatedly reported as an oncogenic lncRNA in LUAD and other cancer types, with elevated expression associated with poor prognosis and enhanced tumor progression in multiple independent studies. Published research has connected CYTOR with enhanced migratory and invasive capacities in lung and epithelial cancers, with cytoplasmic mechanisms involving RNA-protein interactions proposed as the primary functional pathway. Its cytoplasmic enrichment observed in lncATLAS data is consistent with these post-transcriptional regulatory roles.
MIR4435-2HG has been implicated in lung cancer progression across multiple studies. MIR4435-2HG has been reported to promote lung cancer progression and has been repeatedly associated with unfavorable outcomes across cancers in multiple independent studies. Its cytoplasmic localization pattern, similar to CYTOR, supports a role in post-transcriptional regulation, potentially through ceRNA mechanisms involving microRNA sponging. The consistent appearance of MIR4435-2HG in published prognostic signatures provides independent validation of the current pipeline findings.
FAM83A-AS1 is less characterized but consistently identified as a poor-prognosis marker. FAM83A-AS1 has been reported as an oncogenic lncRNA in LUAD with its higher expression associated with worse survival and functional roles in tumor progression. Although mechanistic evidence for FAM83A-AS1 remains more limited than for CYTOR and MIR4435-2HG, its consistent association with advanced clinical parameters across independent analyses suggests possible roles in pathways regulating tumor growth or dissemination. Its progressive association with nodal involvement and tumor size in this study adds to the body of evidence implicating it in disease aggressiveness.
Coherence across multiple independent readouts is the study's core strength. The value of this study is not any single association but the coherence of patterns across multiple independent analytical dimensions: tumor-normal differential expression across all four stages, adverse survival association by Kaplan-Meier analysis, progressive stage trends by ordinal trend testing, and monotonic increases with nodal status and tumor size by rank correlation. This convergence reduces the likelihood that the observed associations are driven by statistical artifacts or cohort-specific effects, a key requirement for biomarker credibility.
A prognostic framework applicable across all LUAD patients regardless of sex or subgroup. The identification of FAM83A-AS1, CYTOR, and MIR4435-2HG as consistently poor-prognosis-associated lncRNAs establishes a basis for lncRNA-based risk stratification in LUAD that applies broadly across the patient population. The absence of sex-specific expression differences means these biomarkers could be applied uniformly to male and female patients, broadening their practical utility compared to markers with demographic-specific limitations.
Potential to supplement TNM staging with molecular information. The current TNM staging system captures anatomical tumor extent but not the molecular heterogeneity that drives variable outcomes among patients at the same stage. The expression levels of FAM83A-AS1, CYTOR, and MIR4435-2HG could complement TNM staging by providing a molecular layer of risk information that reflects the underlying aggressiveness of individual tumors, particularly for patients with early-stage LUAD where prognosis is highly variable and treatment intensity decisions are clinically challenging.
CYTOR and MIR4435-2HG as potential therapeutic targets. Beyond their utility as biomarkers, the oncogenic roles of CYTOR and MIR4435-2HG in LUAD progression raise the possibility that these lncRNAs could serve as therapeutic targets. Prior studies have connected these molecules to proliferation, migration, epithelial-mesenchymal transition, and cellular metabolism. Therapeutic strategies targeting cytoplasmic lncRNAs through antisense oligonucleotides or small molecule interference represent an active area of oncology research that could be informed by these findings.
Functional validation and independent cohort confirmation are the critical next steps. While the present study provides strong transcriptomic and clinical association evidence, the findings are derived from a single institutional dataset (TCGA-LUAD). Functional validation experiments to confirm the causal roles of FAM83A-AS1, CYTOR, and MIR4435-2HG in LUAD progression, and prospective validation of their expression as prognostic biomarkers in independent patient cohorts, are essential before these lncRNAs can be integrated into clinical prognostic tools or treatment decision frameworks.
FAM83A-AS1, CYTOR, and MIR4435-2HG are robust poor-prognosis biomarker candidates. This study identifies three lncRNAs as a core set of consistently overexpressed, prognostically adverse markers in LUAD whose expression patterns remain coherent across tumor-normal comparisons at all four disease stages, Kaplan-Meier survival analysis, lymph-node status trends, and primary tumor size trends. This multi-dimensional consistency distinguishes them from the broader set of lncRNAs identified by differential expression analysis alone.
The study's value lies in its integrative, objective framework. By applying predefined, quantitative selection criteria at each analytical step, including FDR correction for multiple testing, hazard ratio thresholds, and rank-based ordinal trend testing, the study avoided ad hoc prioritization and manual curation biases. This framework provides a methodological template for objectively identifying clinically relevant lncRNAs that can be applied to other cancer types or extended to include additional clinical dimensions such as treatment response or immune infiltration.
67 consistently downregulated lncRNAs including four protective candidates warrant further study. The parallel analysis of consistently downregulated lncRNAs identified four candidates whose higher expression associates with improved overall survival, suggesting tumor-suppressive functions whose loss may contribute to LUAD progression. These candidates represent a complementary set of biomarker targets that merit further investigation alongside the poor-prognosis upregulated set.
Future work should prioritize functional studies and multi-cohort prospective validation. The transcriptomic associations established here provide a solid foundation for future mechanistic research into the functional roles of FAM83A-AS1, CYTOR, and MIR4435-2HG in LUAD biology. Prospective studies in independent cohorts with diverse patient populations, combined with functional in vitro and in vivo experiments, will be necessary to establish the clinical utility of these lncRNAs as prognostic biomarkers and to evaluate their potential as therapeutic targets in lung adenocarcinoma.