A machine learning framework predicts oncogenic driver mutations from SNP profiles in lung adenocarcinoma.

Genetic predisposition to lung adenocarcinoma among never-smoking Chinese with different epidermal growth factor receptor mutation status

Introduction: EGFR mutations holds the major targets for drug in lung adenocarcinoma (LUAD). Despite the tremendous study of EGFR mutant (MT) LUAD, the comprehensive interpretation of the heterogeneous character of LUAD harboring EGFR MT remains a key challenge. Here, we investigated the heterogeneity of EGFR MT LUAD and explored the tumor microenvironment (TME) in EGFR MT LUAD. Method: We performed single-cell RNA sequencing (scRNA-seq) from 135 LUAD patients which consist of normal(n=24), EGFR wild (WT)(n=18), and MT(n=93). Also, we used whole genome sequencing and bulk-RNA sequencing to validate with scRNA-seq results. From 898,648 cells, main cell types were classified. To explore the various characteristics of MT LUAD tumor cells, we used two ways: i) We re-clustered epithelial cells populating the normal, WT, and MT. ii) We re-clustered only MT epithelial cells. In each analysis, we identified the tumor character in the clusters using differential expressed genes analysis, lineage tracing, clinical information, mutation, and trajectory analysis. Also, we extracted each main cell type except epithelial cells, and identified subtypes of main cell types. Finally, we revealed the interaction of cellular components in TME. Results: In the analysis of epithelial cells, we identified characteristics of specific EGFR MT by comparing of EGFR WT and MT tumors in clusters with similar biological features. The cluster represented by alveolar type 2 (AT2) known as initiation of LUAD was populating normal, WT, and MT. In this cluster, MT- and WT-associated pathway shared but differently significant between MT and WT in the pathway analysis. The cluster represented by proliferative is mostly comprised tumor cells and we found significantly increased the expression of MDK, CD24 in the MT of the cluster. In the analysis of only MT epithelial cells, 2 of clusters were stage-specific cluster: i) The cluster annotated as early stage cluster, ii) The cluster annotated as advanced stage cluster. Trajectory showed that there is a pseudotemporal continuum, following the stage from early stage cluster to advanced stage cluster. Also, based on the lineage tracing, 2 of clusters revealed lineage-specific clusters: i) The cluster annotated as AT2 was enriched from early stage cells, ii) The cluster annotated as basal cell known as origin of lung squamous cell carcinoma(LUSC) was enriched from advanced stage cells. Psedotemporal ordering of these cluster revealed AT2 cluster transdifferentiate into basal cell cluster which implied the possibility of LUAD to LUSC transition by drug resistance. In the interaction of MT and WT TME, the number of signaling received epithelial cells from myeloid cells, endothelial cells, and fibroblasts as sender increased compared with the interaction of normal. Conclusion: We shed light on the ecosystem of TME according to clinical and biological feature of tumor in EGFR mutant LUAD. Citation Format: You Won Lee, Eun Ji Lee, Seung Yeon Oh, Kyoung-Ho Pyo, Seong Gu Heo, YoungJoon Park, Su-Jin Choi, Kyumin Lim, Ju-hyeon Lee, Jae Hwan Kim, Jii Bum Lee, Ji Yoon Lee, Sun Min Lim, Chang Gon Kim, Min Hee Hong, Mi Ran Yun, Byoung Chul Cho. Phenotype profiling of tumor microenvironment in EGFR mutant lung adenocarcinoma with multi-omics data. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 5935.

Purpose: Epidermal growth factor receptor (EGFR) mutation is the most common target for precision treatment in metastatic lung adenocarcinoma. We investigated the predictive role of 18F-FDG PET/CT and clinicopathological features for EGFR mutations in lung adenocarcinoma with bone metastasis.Methods: Seventy-five lung adenocarcinoma patients with histologically confirmed bone metastasis were included. They all received EGFR status test and PET/CT before systemic treatment. The differences of maximum standardized uptake value (SUVmax) in primary tumor (pSUVmax), regional lymph node (nSUVmax) and bone metastasis (bmSUVmax) between different EGFR status groups were compared, alongside with common clinicopathological features. Multivariate logistic regression analysis was performed to evaluate predictors of EGFR mutations.Results: EGFR mutations were found in 37 patients (49.3%). EGFR mutations were more common in females, non-smokers, expression of Thyroid Transcription Factor-1 (TTF-1) and NaspinA. Low bmSUVmax was significantly associated with EGFR mutations, while no significant difference was observed in pSUVmax and nSUVmax. Multivariate analysis showed that bmSUVmax ≤7, non-smoking, expression of TTF-1 were predictors of EGFR mutations. The area under the curve (AUC) of receiver operating characteristic (ROC) curve was 0.84 for the combination of the three factors.Conclusion: Low bmSUVmax is more frequently in EGFR mutations, and bmSUVmax is an independent predictor of EGFR mutations. Combining bmSUVmax with other clinicopathological features could forecast the EGFR status in lung adenocarcinoma with unavailable EGFR gene testing.

Epithelial-mesenchymal transition (EMT) was recently discovered related to the efficacy of epidermal growth factor receptor-tyrosine kinase inhibitors (EGFR-TKIs) in NSCLC patients and cell lines. In this study, we aimed to explore the association among the E-cadherin gene (CDH1) genetic variants, TK-domain mutations of EGFR, and clinicopathologic characteristics in patients with lung adenocarcinoma. A total of 280 patients with lung adenocarcinoma were recruited between years 2012 and 2015. All subjects underwent the analysis of CDH1 genetic variants (rs16260 and rs9929218) by real-time polymerase chain reaction (PCR) genotyping. The results showed that CA and CA + AA genotypes of CDH1 single nucleotide polymorphism (SNP) rs16260 were significantly reverse associated with EGFR mutation type (Adjusted odds ratio (AOR) = 0.43, 95% CI = 0.20-0.92 and AOR = 0.46, 95% CI = 0.22-0.96, respectively) in female lung adenocarcinoma patients. Moreover, the significantly reverse associations between CA and CA + AA genotypes of CDH1 rs16260 and EGFR hotspot mutations, namely L858R mutation and exon 19 in-frame deletion, were also demonstrated among female patients. Besides, CA + AA genotype of CDH1 rs16260 was noted significantly reverse associated with the tumor sizes (OR = 0.31, 95% CI = 0.12-0.80; p = 0.012). In conclusion, our results suggested that CDH1 variants are significantly reverse associated with mutation of EGFR tyrosine kinase, especially among the female patients with lung adenocarcinoma. The CDH1 variants might contribute to pathological development in lung adenocarcinoma.

Gastric and bladder metastases are rare in non-small-cell lung cancer (NSCLC) and the differential diagnosis with primary tumors requires an in-depth examination. KRAS gene mutations are the main oncogenic driver of lung adenocarcinoma in Caucasian patients, occurring in 25%-30% of cases, but their prognostic and predictive role is complex and not fully clarified. KRAS mutations in lung adenocarcinoma are considered negatively predictive for epidermal growth factor receptor tyrosine kinase inhibitors therapy as well as for adjuvant chemotherapy. In this report, the authors describe a case of lung adenocarcinoma harboring a codon 13 KRAS mutation detected in all the biopsies performed in unusual metastatic sites during an atypical disease course. Recently, a different prognostic significance of various KRAS mutations in lung adenocarcinoma has been suggested. Further studies on rare biomolecular alterations to identify subgroups of patients with different prognostic/predictive characteristics are needed.

Epidermal growth factor receptor (EGFR) mutations are common in lung adenocarcinomas of never smokers, while KRAS mutations are more frequent among heavy smokers. Different clinicopathological and biological characteristics may, therefore, exist in lung adenocarcinoma according to smoking status. In the present study, a retrospective review was performed using 521 patients with surgically resected lung adenocarcinomas. The clinicopathological factors of age, gender, pathological tumor size, nodal status, lymphatic permeation and blood vessel invasion and the EGFR and KRAS mutation spectra were compared between never and heavy smokers. EGFR mutations were detected in 233 (45%) patients, while KRAS mutations were detected in 56 (11%) patients. EGFR-mutated adenocarcinomas had a higher prevalence of females in the never smokers compared with the heavy smokers (P<0.001). KRAS-mutated adenocarcinomas had a higher prevalence of females (P<0.001) and showed less frequent vascular invasion (P=0.018) in the never smokers compared with the heavy smokers. Minor EGFR mutations, excluding exon 21 L858R and exon 19 deletions, were more common in heavy smokers than never smokers (P=0.055). KRAS G to A transition was more common in never smokers, while KRAS G to T and G to C transversions were more common in heavy smokers (P=0.036). The clinicopathological characteristics and the spectra of the EGFR and KRAS mutations in lung adenocarcinoma were different between the never and heavy smokers. Further large-scale studies are required to evaluate the efficacy of molecular targeting agents with consideration to specific EGFR and KRAS mutations.

Epidermal growth factor receptor (EGFR) mutations in lung adenocarcinoma are involved in the tumorigenesis and regulation of cell metabolism via Akt signaling. F-18 fluorodeoxyglucose-positron emission tomography ([(18)F]FDG PET), a functional imaging modality, can be used to measure tumor cell metabolism. Thus, in this study, we hypothesize that there exist correlations between EGFR mutation status and [(18)F]FDG uptake of advanced lung adenocarcinoma. From May 2004 to April 2008, patients with stage IIIB or IV lung adenocarcinoma who underwent [(18)F]FDG PET and EGFR mutation analysis before receiving any treatment were eligible to participate in this study. The association of EGFR mutation status with patient characteristics and the SUV(MAX) from the [(18)F]FDG PET was evaluated. Multivariate logistic regression analysis was used to analyze predictors of EGFR mutations. Seventy-seven lung adenocarcinoma patients were included in this study. EGFR mutations were identified in 49 (64%) of the patients. The [(18)F]FDG uptake was significantly higher in EGFR-mutant (mean SUV(MAX) = 10.5 +/- 4.7) than wild-type (8.0 +/- 3.3) lung adenocarcinoma patients (P = 0.008). The median SUV(MAX) was 9.5, and patients with an SUV(MAX) >or= 9.5 were more likely to harbor EGFR mutations (P = 0.009). In the multivariate analysis, an SUV(MAX) >or= 9.5 remained a statistically significant predictor of EGFR mutations (P = 0.005). Among Asian patients with advanced lung adenocarcinoma, those with higher SUV(MAX) on the [(18)F]FDG PET are more likely to carry EGFR mutations.

The p16 gene is frequently inactivated in lung adenocarcinoma. In particular, homozygous deletions (HD) have been frequently detected in cell lines; however, their frequency and specificity is not well-established in primary tumors. The purpose of this study was to elucidate the prevalence and the timing for the occurrence of p16 HDs in lung adenocarcinoma progression in vivo. Multiple ligation-dependent probe amplification was used for the detection of p16 HDs in 28 primary small-sized lung adenocarcinomas and 22 metastatic lung adenocarcinomas to the brain. Cancer cells were isolated from primary adenocarcinoma specimens by laser capture microdissection. HDs were confirmed by quantitative real-time genomic PCR analysis. HDs were detected in 8 of 28 (29%) primary tumors, including 2 of 8 (25%) noninvasive bronchioloalveolar carcinomas, and 5 of 22 (26%) brain metastases, respectively. No significant associations were observed between p16 HDs and gender, age, smoking history, stage, and prognosis. HDs were detected with similar frequencies (17-29%) among adenocarcinomas with epidermal growth factor receptor (EGFR) mutations, with KRAS mutations, and without EGFR/KRAS mutations, and with similar frequencies (22-28%) between adenocarcinomas with and without p53 mutations. p16 HDs occur early in the development of lung adenocarcinomas and with similar frequencies among EGFR type, KRAS type, and non-EGFR/KRAS type lung adenocarcinomas. Tobacco carcinogens would not be a major factor inducing p16 HDs in lung adenocarcinoma progression.

Background: Metastatic lung cancer kills over 180,000 people in the United States annually. Family history is one risk factor for lung adenocarcinoma (LUAD), suggesting a putative genetic predisposition to the development of LUAD. Single nucleotide polymorphisms (SNPs) are the most common genetic variations in the human genome and major determinants of variations in disease susceptibility, prognosis, response to medication, and toxicity. Recent studies indicate that SNPs play important roles in the progression of LUAD. We have previously shown that breast cancer metastasis suppressor-1 (BRMS1) plays a critical role in inhibiting the progression of lung cancer. We observed that SNP rs1052566 homo-variant in breast cancer metastasis suppressor-1 (BRMS1) variant 2 (v2) causes an A273V mutation. However, the role of this SNP in the progression of lung cancer remains to be determined. Methods: We utilized next-generation DNA sequencing and Taqman SNP assays to detect the prevalence of rs1052566 in 40 surgically resected LUAD specimens. Next, we generated a retroviral expressional system (pBabe puro) encoding V5-tagged BRMS1v2 wild-type (WT) and A273V mutant and created NSCLC H1299 and A549 cell lines that stably express V5-tagged BRMS1v2 WT, A273V, or vector control. The metastatic capabilities of these stable cells were evaluated by invasion chamber assays in vitro and a tail-vein injection metastasis model in vivo. To explore the mechanism(s) through which BRMS1v2 A273V promotes metastasis, we performed RT profiler PCR assay. Results: We determined that SNP rs1052566 is present in 45% of LUAD patients. Of these, the homo-variant occurs in 7.5% of patients (3 of 40). Importantly, BRMS1v2 A273V significantly promotes metastasis of NSCLC cells in vitro and in vivo. Additionally, in our mouse model, A273V was associated with decreased survival, compared with BRMS1v2 WT or control. On RT profiler PCR array, 7 genes, including L1CAM, were found to be significantly differentially regulated by BRMS1v2 WT and A273V. Confirmatory RT-PCR and immunoblots proved that BRMS1v2 A273V increases both the transcript and protein levels of L1CAM. Furthermore, we found that either c-Jun or c-Fos promotes the transcription of L1CAM. However, BRMS1v2 A273V specifically enhances c-Fos-, but not c-Jun-, induced L1CAM transcription. Conclusion: We observed that BRMS1v2 A273V abolishes the metastasis suppressor function of BRMS1 and promotes metastasis. Up-regulation of c-Fos-induced L1CAM transcription is a putative mechanism of BRMS1v2 A273V-induced metastasis. Our data suggest that SNP rs1052566, especially its homo-variant, is associated with an increased risk of metastasis in LUAD. Note: This abstract was not presented at the meeting. Citation Format: Neel P. Chudgar, Yuan Liu, Joseph Montecalvo, Marty W. Mayo, Prasad S. Adusumilli, David R. Jones. SNP rs1052566 (Ala273Val) in BRMS1 variant2 promotes metastasis in lung adenocarcinoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 4840. doi:10.1158/1538-7445.AM2017-4840

EGFR and KRAS mutations are frequently detected in lung adenocarcinoma (LUAD). Tumor mutational signature (TMS) determination is an approach to identify somatic mutational patterns associated with pathogenic factors. In this study, through the analysis of TMS, the underlying pathogenic factors of LUAD with EGFR and KRAS mutations were traced. This was a retrospective study. TMS of LUAD with KRAS and EGFR mutations from the TCGA, OncoSG, and MSK datasets was determined by two bioinformatics tools, namely the "MutationalPatterns" and "FitMS" packages. Elevated microsatellite alterations at selected tetranucleotide repeats (EMAST) of LUAD clinical specimens was analyzed using capillary electrophoresis. In LUAD with KRAS mutations, TMS analysis indicated that the smoking-related SBS4 signature was enriched. For LUAD with EGFR L858R mutation, the smoking-related SBS4 signature was enriched in the Western population from the TCGA database; however, the smoking-related SBS4 signature was not obvious in Asian LUAD patients. LUAD with EGFR exon19 deletion (19Del) exhibited stronger SBS15 signature, which was related to defective DNA mismatch repair. Capillary electrophoresis analysis showed that an EMAST locus was frequently instable in LUAD with EGFR 19Del. Different from the Western population, Asian LUAD patients with EGFR mutations exhibited the enrichment of SBS1, SBS2, and SBS13 signatures, which were associated with the endogenous mutation process of cytidine deamination. TMS analysis reveals that smoking is associated with LUAD with KRAS mutations. Defective DNA mismatch repair and endogenous cytidine deamination are associated with LUAD with EGFR mutations, especially for the EGFR 19Del. The endogenous mutational process is stronger in Asian LUAD patients than Western LUAD patients.

Standard cytotoxic chemotherapy is effective for some cancers, but for many others, available treatments offer only a limited survival benefit. Lung adenocarcinoma is one such cancer, responsible for approximately half of lung cancer deaths each year. Development of targeted therapies is thought to hold the most promise for successfully treating this disease, but a targeted approach is dependent on understanding the genomic state of the tumor cells. Exon-directed sequencing of large numbers of lung adenocarcinoma tumor samples has provided an initial low-resolution image of the somatic mutation profile of these tumors. Such cancer sequencing studies have confirmed the high frequency of TP53 and KRAS mutations in lung adenocarcinoma, have found inactivating mutations in known tumor suppressor genes not previously associated with lung adenocarcinoma, and have identified oncogenic mutations of EGFR upon which the first targeted therapy for lung adenocarcinoma patients was based. Additional candidate oncogenes await functional validation. It is anticipated that upcoming whole-exome and whole-genome lung adenocarcinoma sequencing experiments will reveal a more detailed landscape of somatic mutations that can be exploited for therapeutic purposes.

Based on multiscale computed tomography (CT) radiomics, a better model was established to differentially diagnose benign lesions and lung adenocarcinoma of Lung-RADS 2022 category 4B solid lung nodules. The retrospective study included 178 patients with solid pulmonary nodules were assigned to the training dataset (n = 124) and the testing dataset (n = 54). Gradient boosting decision tree (GBDT) was used to reduce the dimensionality of data and select the best radiomics features. Four models were developed by logistic regression method, namely the clinical and imaging model (CIM), the plain CT radiomics model (PRM), the enhanced CT radiomics model (ERM), and the combined model (CM). Area under the curve (AUC) evaluates the model performance. Net reclassification improvement (NRI) and the integrated discrimination index (IDI) were calculated to compare the performance of different models to determine the best model. In the training dataset, the AUC of CIM, PRM, ERM, and CM were 0.795, 0.791, 0.828, and 0.888. The continuous NRI and IDI of CM was better than that of CIM, PRM, ERM (P < 0.001), CM is optimal. In the testing dataset, the AUC of CIM, PRM, ERM and CM were 0.810, 0.689, 0.864 and 0.881. The continuous NRI of CM was better than that of CIM, PRM, ERM (P < 0.050). The IDI of CM was better than that of CIM and PRM (P < 0.050). The AUC values of the Mayo Clinic (Mayo) model, Veterans Administration (VA) model, Peking University People’s Hospital (PKUPH) model and United Imaging Artificial Intelligence (UI AI) model were 0.419, 0.410, 0.676 and 0.675, CM is still optimal. Radiomics can be used as a non-invasive tool to distinguish between benign lesions and lung adenocarcinoma of Lung-RADS 2022 category 4B solid lung nodules, and CM is the best predictive model.

Activating KRAS mutations in lung adenocarcinoma are characterized with treatment resistance and poor prognosis. As a small molecule inhibitor of vascular endothelial growth factor receptor-2 (VEGFR-2) tyrosine kinase, apatinib has been proven successful in advanced gastric cancer and breast cancer. In this study, we show the result of apatinib as salvage treatment in lung adenocarcinoma patients with KRAS mutation. Four advanced lung adenocarcinoma patients with KRAS mutation were orally administered apatinib (250 mg/d) after second-line treatment. One patient showed progressive disease, while 3 patients showed stable disease response to apatinib, with a median progression-free survival (PFS) of 3.8 months (1.5–5.5 months). The main toxicities were hoarseness and hemoptysis, which were manageable. Therefore, apatinib might be an optional choice for advanced lung adenocarcinoma patients with KRAS mutation in post second-line treatment.

Simple SummaryTargeted therapy against epidermal growth factor (EGFR) mutations has become the standard of care for non-small cell lung cancer, and there has not been an efficient genetic test for non-small cell lung cancer patients. The present study aims to find a novel data-driven genetic testing method that can effectively predict the mutation status of EGFR based on a prediction model combining clinical features. The results of this study provide a powerful theoretical basis for the establishment of an effective mutation prediction model. The prediction model can provide a high reference value aiding in EGFR mutation diagnosis and subsequent treatment course.Data from 758 patients with lung adenocarcinoma were retrospectively collected. All patients had undergone computed tomography imaging and EGFR gene testing. Radiomic features were extracted using the medical imaging tool 3D-Slicer and were combined with the clinical features to build a machine learning prediction model. The high-dimensional feature set was screened for optimal feature subsets using principal component analysis (PCA) and the least absolute shrinkage and selection operator (LASSO). Model prediction of EGFR mutation status in the validation group was evaluated using multiple classifiers. We showed that six clinical features and 622 radiomic features were initially collected. Thirty-one radiomic features with non-zero correlation coefficients were obtained by LASSO regression, and 24 features correlated with label values were obtained by PCA. The shared radiomic features determined by these two methods were selected and combined with the clinical features of the respective patient to form a subset of features related to EGFR mutations. The full dataset was partitioned into training and test sets at a ratio of 7:3 using 10-fold cross-validation. The area under the curve (AUC) of the four classifiers with cross-validations was: (1) K-nearest neighbor (AUCmean = 0.83, Acc = 81%); (2) random forest (AUCmean = 0.91, Acc = 83%); (3) LGBM (AUCmean = 0.94, Acc = 88%); and (4) support vector machine (AUCmean = 0.79, Acc = 83%). In summary, the subset of radiographic and clinical features selected by feature engineering effectively predicted the EGFR mutation status of this NSCLC patient cohort.

We investigated optimal peritumoral size and constructed predictive models for epidermal growth factor receptor (EGFR) mutation. A total of 164 patients with lung adenocarcinoma were retrospectively analyzed. Radiomic signatures for the intratumoral region and combinations of intratumoral and peritumoral regions (3, 5, and 7mm) from computed tomography images were extracted using analysis of variance and least absolute shrinkage. The optimal peritumoral region was determined by radiomics score (rad-score). Intratumoral radiomic signatures with clinical features (IRS) were used to construct predictive models for EGFR mutation. Combinations of intratumoral and 3, 5, or 7mm-peritumoral signatures with clinical features (IPRS3, IPRS5, and IPRS7, respectively) were also used to construct predictive models. Support vector machine (SVM), logistic regression (LR), and LightGBM models with five-fold cross-validation were constructed, and the receiver operating characteristics were evaluated. Area under the curve (AUC) of the training and test cohorts values were calculated. Brier scores (BS) and decision curve analysis (DCA) were used to evaluate the predictive models. The AUC values of the SVM, LR, and LightGBM models derived from IRS were 0.783 (95% confidence interval: 0.602-0.956), 0.789 (0.654-0.927), and 0.735 (0.613-0.958) for training, and 0.791 (0.641-0.920), 0.781 (0.538-0.930), and 0.734 (0.538-0.930) for test cohort, respectively. Rad-score confirmed that the 3mm-peritumoral size was optimal (IPRS3), and AUCs values of SVM, LR, and lightGBM models derived from IPRS3 were 0.831 (0.666-0.984), 0.804 (0.622-0.908), and 0.769 (0.628-0.921) for training and 0.765 (0.644-0.921), 0.783 (0.583-0.921), and 0.796 (0.583-0.949) for test cohort, respectively. The BS and DCA of the LR and LightGBM models derived from IPRS3 were better than those from IRS. Accordingly, the combination of intratumoral and 3mm-peritumoral radiomic signatures may be helpful for predicting EGFR mutations.