S48 Molecular characterisation of lung adenocarcinoma histological patterns

• Published in: Thorax Vol. 78; no. Suppl 4; p. A38
• Main Authors: Nastase, A, Willis-Owen, SAG, Domingo-Sabugo, C, Starren, E, Olanipekun, M, Nicholson, AG, Moffatt, MF, Cookson, WOCM
• Format: Journal Article
• Language: English
• Published: London BMJ Publishing Group Ltd and British Thoracic Society 06.11.2023; BMJ Publishing Group LTD
• Subjects: Lung cancer; Mutation; Tumors; ‘Don’t stop believing’ – Tumour biology: implications for treatment
• ISSN: 0040-6376; 1468-3296
• DOI: 10.1136/thorax-2023-BTSabstracts.54

BackgroundLung adenocarcinomas (LUADs) usually show one predominant morphological pattern (either lepidic, papillary, acinar, cribriform, micropapillary or solid) that predicts patient’s survival. In addition, LUADs may be classified according to their predominant subtype (either mucinous, non-mucinous or mixed). Genomic differences that may influence patterns have been minimally studied. Our aim was to identify genetic alterations with pathogenic relevance related to LUAD morphology.MethodsWe genomically profiled 89 LUAD tumour samples (13 lepidic, 13 papillary, 37 acinar, 7 cribriform, 7 micropapillary and 12 solid, sub-typed as 48 non-mucinous, 20 mucinous and 19 mixed) by Affymetrix microarrays, single nucleotide polymorphism genotyping and whole-exome or targeted capture sequencing.Results TP53 mutations were the most frequent abnormalities overall (41%) but were notably absent in cribriform tumours, where CDKN2A alterations were most frequent. EGFR mutations were more common in never-smokers, were mutually exclusive to KRAS in all subjects and not detected in solid, papillary or cribriform patterns. KRAS mutations in the acinar subtype were the only predictor for patient survival.The solid tumours (mostly of the non-mucinous subtype [83.3%]) had a high tumour mutation burden (TMB). Solid tumours also exhibited high levels of expression for VEGFA and CD274 (PD-L1) transcripts level and the highest CD8 T-cell abundance, shown by CIBERSORT analysis, suggesting potential response to checkpoint inhibitors for this pattern.Network analysis discovered 92 modules of co-expressed genes in LUAD tumours. The greatest number of co-expression modules were found in solid and lepidic patterns (11 and 10 networks respectively). These associations were almost entirely opposing in directionality, indicating strongly contrasting mechanisms underlying these two histotypes. The largest co-expression module ME1 (chromosome organization, containing 1,007 genes, with TPX2 as hub gene) showed the strongest associations: Cor 0.53, P 8.92x10-08 [% solid]/-0.52, P 1.38x10-07 [% lepidic]). The ME1 module was also linked to with γδ T cells (strength: 1.00, direction: 0.95).ConclusionsOur study has highlighted existence of genomic heterogeneity in LUAD patterns. Assessing key molecular changes in combination with standard histology in LUAD might have direct impact in classifying responders to targeted therapies, in particular solid pattern patients to immune checkpoint inhibitors.