Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Molecular signatures in IASLC/ATS/ERS classified growth patterns of lung adenocarcinoma

  • Heike Zabeck ,

    Roles Conceptualization, Data curation, Funding acquisition, Investigation, Methodology, Project administration, Resources, Writing – original draft, Writing – review & editing

    heike.zabeck@med.uni-heidelberg.de

    Affiliation Department of Thoracic Surgery, Thoraxklinik, University Hospital Heidelberg, Heidelberg, Germany

  • Hendrik Dienemann,

    Roles Funding acquisition, Supervision

    Affiliation Department of Thoracic Surgery, Thoraxklinik, University Hospital Heidelberg, Heidelberg, Germany

  • Hans Hoffmann,

    Roles Conceptualization, Supervision

    Affiliations Department of Thoracic Surgery, Thoraxklinik, University Hospital Heidelberg, Heidelberg, Germany, Translational Lung Research Centre Heidelberg (TLRC-H), German Centre for Lung Research (DZL), Heidelberg, Germany

  • Joachim Pfannschmidt,

    Roles Conceptualization, Funding acquisition, Supervision

    Current address: Department of Thoracic Surgery, Lung Clinic Heckeshorn at HELIOS Hospital Emil von Behring, Berlin, Germany

    Affiliation Department of Thoracic Surgery, Thoraxklinik, University Hospital Heidelberg, Heidelberg, Germany

  • Arne Warth,

    Roles Conceptualization, Methodology, Supervision, Validation

    Affiliations Translational Lung Research Centre Heidelberg (TLRC-H), German Centre for Lung Research (DZL), Heidelberg, Germany, Institute of Pathology, University Hospital Heidelberg, Heidelberg, Germany

  • Philipp A. Schnabel,

    Roles Conceptualization, Methodology, Supervision

    Current address: Institute of Pathology, Saarland University, Homburg/Saar, Germany

    Affiliation Institute of Pathology, University Hospital Heidelberg, Heidelberg, Germany

  • Thomas Muley,

    Roles Data curation, Methodology

    Affiliations Translational Lung Research Centre Heidelberg (TLRC-H), German Centre for Lung Research (DZL), Heidelberg, Germany, Translational Research Unit (STF), Thoraxklinik, University of Heidelberg, Heidelberg, Germany

  • Michael Meister,

    Roles Methodology

    Affiliations Translational Lung Research Centre Heidelberg (TLRC-H), German Centre for Lung Research (DZL), Heidelberg, Germany, Translational Research Unit (STF), Thoraxklinik, University of Heidelberg, Heidelberg, Germany

  • Holger Sültmann,

    Roles Supervision

    Affiliations Translational Lung Research Centre Heidelberg (TLRC-H), German Centre for Lung Research (DZL), Heidelberg, Germany, Cancer Genome Research (B063), German Cancer Research Center (DKFZ) and German Cancer Consortium (DKTK), Heidelberg, Germany

  • Holger Fröhlich,

    Roles Data curation, Formal analysis, Methodology, Software, Validation, Writing – review & editing

    Affiliation Institute for Computer Science, c/o Bonn-Aachen International Center for IT, Algorithmic Bioinformatics, University of Bonn, Bonn, Germany

  • Ruprecht Kuner,

    Roles Data curation, Formal analysis, Investigation, Methodology, Software, Validation, Visualization, Writing – original draft, Writing – review & editing

    Current address: TRON—Translational Oncology at the University Medical Center of Johannes Gutenberg University, Mainz, Germany

    Affiliations Translational Lung Research Centre Heidelberg (TLRC-H), German Centre for Lung Research (DZL), Heidelberg, Germany, Cancer Genome Research (B063), German Cancer Research Center (DKFZ) and German Cancer Consortium (DKTK), Heidelberg, Germany

  • Felix Lasitschka

    Roles Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Validation, Visualization, Writing – original draft, Writing – review & editing

    Affiliation Institute of Pathology, University Hospital Heidelberg, Heidelberg, Germany

Molecular signatures in IASLC/ATS/ERS classified growth patterns of lung adenocarcinoma

  • Heike Zabeck, 
  • Hendrik Dienemann, 
  • Hans Hoffmann, 
  • Joachim Pfannschmidt, 
  • Arne Warth, 
  • Philipp A. Schnabel, 
  • Thomas Muley, 
  • Michael Meister, 
  • Holger Sültmann, 
  • Holger Fröhlich
PLOS
x

Abstract

Background

The current classification of human lung adenocarcinoma defines five different histological growth patterns within the group of conventional invasive adenocarcinomas. The five growth patterns are characterised by their typical architecture, but also by variable tumor biological behaviour.

Aims

The aim of this study was to identify specific gene signatures of the five adenocarcinoma growth patterns defined by the joint IASLC/ATS/ERS working group.

Methods

Total RNA from microdissected adenocarcinoma tissue samples of ten lepidic, ten acinar, ten solid, nine papillary, and nine micropapillary tumor portions was isolated and prepared for gene expression analysis. Differential expression of genes was determined using the R package “LIMMA”. The overall significance of each signature was assessed via global test. Gene ontology statistics were analysed using GOstat. For immunohistochemical validation, tissue specimens from 20 tumors with solid and 20 tumors with lepidic growth pattern were used.

Results

Microarray analyses between the growth patterns resulted in numerous differentially expressed genes between the solid architecture and other patterns. The comparison of transcriptomic activity in the solid and lepidic patterns revealed 705 up- and 110 downregulated non-redundant genes. The pattern-specific protein expression of Inositol-1,4,5-trisphosphate-kinase-A (ITPKA) and angiogenin by immunohistochemistry confirmed the RNA levels. The strongest differences in protein expression between the two patterns were shown for ITPKA (p = 0.02) and angiogenin (p = 0.113).

Conclusions

In this study growth pattern-specific gene signatures in pulmonary adenocarcinoma were identified and distinct transcriptomic differences between lung adenocarcinoma growth patterns were defined. The study provides valuable new information about pulmonary adenocarcinoma and allows a better assessment of the five adenocarcinoma subgroups.

Introduction

In 2011, a joint IASLC/ATS/ERS working group introduced a new classification of human lung adenocarcinoma [1]. This new classification defined five different histological growth patterns within the group of conventional invasive adenocarcinomas: lepidic (corresponding to the former non-mucinous bronchioloalveolar pattern), acinar, papillary, micropapillary (newly added) and solid predominant adenocarcinoma. Overall tumor classification was suggested to be done according to the predominant pattern with additional documentation of the percentage distribution of all evident architectures [1,2]. The five growth patterns should be characterised by their typical architecture, but also by variable tumor biological behaviour. The predominant pattern has relevant influence on the disease-free and long-term survival of patients. Whereas lepidic predominant adenocarcinoma seems to be the one with the best prognosis, the micropapillary and solid architectures are associated with a particularly poor prognosis [37]. On the molecular level, lung adenocarcinomas have been characterised by diverse transcriptional profiles [810]. Three molecular subtypes (bronchoid/terminal respiratory, magnoid/proximal-proliferative and squamoid/proximal-inflammatory) were defined and successfully verified across several microarray and sequencing datasets [1113]. Such molecular profiles are used to decipher prognostic/predictive biomarkers and therapeutic target sites specific for patient subgroups. However, the present transcriptomic profiles of lung adenocarcinoma do not necessarily reflect the distinct histological architectures. Our study combined tissue microdissection and molecular profiling of the five lung adenocarcinoma growth patterns in order to precisely identify specific genes signatures. Selected biomarkers and drug targeted candidates were validated by immunohistochemistry (IHC) in the lepidic and solid growth patterns.

Materials and methods

Patients

Cryopreserved tumor tissue of 48 patients was selected based on the histopathological findings following surgery for lung adenocarcinoma. All tumor sections were analysed for their growth patterns including lepidic (10), acinar (10), papillary (9), micropapillary (9) and solid (10) architecture. Cryopreserved tumor tissue was provided by lung biobank Heidelberg, a member of the biomaterial bank Heidelberg (BMBH) and of the biobank platform of the German Center for Lung Research (DZL) (Ethical Approval Number: 2070/2001). Tumor sections from an additional 20 patients were used for IHC validation. Paraffin tissue was provided by the tissue bank of the National Center for Tumor Diseases (NCT, Heidelberg, Germany) in accordance with the regulations of the tissue bank and the approval of the ethics committee of Heidelberg University (Ethical Approval Numbers: 206/2005 and 207/2005). All data was fully anonymized before we accessed them. All patients gave their written informed consent to have samples from their medical records used in research.

Laser-capture microdissection

The isolation of tumor areas representing a specific growth pattern was performed by laser capture microdissection (PALM’s LMPC technology, Carl Zeiss Microscopy GmbH, Göttingen, Germany). Fresh frozen tumor samples were cut in 18 μm thick sections and applied to ZEISS Membrane Slides 1.0 PEN (polyethylene naphthalate), baked for 3.5 h at 180° C, irradiated with UV light (312 nm) for 30 min and cooled to -22°C just before use. The tissue sections were stained with 1% cresyl violet (Sigma) for 15 s. Using the RoboLPC method, between 4 and 10 mm2 of cells were cut out [14]. The tissue areas of interest were circumscribed electronically under the microscope, cut automatically by cold laser ablation and catapulted into an opaque ZEISS AdhesiveCap 500 by damage-free laser-induced propulsion [15]. The composition (frequencies) of all 5 growth patterns in each of the 48 tumors has been analyzed (Table 1).

Table 1 shows the percentage distribution of the 5 growth patterns in 48 fresh frozen tumor samples used for microdissection.

Microarray experiments

Total RNA from microdissected adenocarcinoma tissue sections was extracted and quantified using the RNeasy protocol (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. For RNA quality assessment the Agilent 2100 Bioanalyzer System together with the Agilent RNA 6000 Nano Kit was used according to the manufacturer’s instructions (Agilent Technologies, Santa Clara, CA, USA). About 20 ng of total RNA was prepared for microarray hybridisation using the MessageAmpTM Premier RNA Amplification Kit (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer’s instructions. Fragmented biotinylated amplificated RNA was hybridised on Illumina HumanHT-12 v4 Expression Bead Chip including more than 47,000 Probes (Illumina, San Diego, CA, USA). Processing of the Illumina microarray data was performed using the opensource pipeline “Lumi” [16]. More specifically, this pipeline comprises background correction, quantile normalization, model based variance stabilization (PMID 18178591) and detection p-value based present/absent calling. For the following analysis only Entrez ID allocated transcripts with a presence call in each sample of at least one of the studied patient subgroups were considered.

The microarray dataset has been deposited, MIAME compliant, into the NCBI Gene Expression Omnibus database (GSE58772).

Molecular subtype assignment was done as previously described: TCGA data was DESeq normalised and reduced to the 5761 genes, which could be mapped to the data studied here [1113,17]. Datasets were first Blom-transformed and subsequently adjusted by an empirical Bayes approach to allow for an integration of the data studied here and the TCGA data [18,19]. Genes were further reduced to the overlap with the previously reported signature [12]. Based on the resulting 260 genes, an SVM classifier was trained on the TCGA data in order to predict expression subtypes defined in 12 [12]. The prediction performance of the classifier was evaluated via 10-fold cross-validation, yielding an estimated prediction accuracy of the expression subtype of around 90%. The final SVM model was then asked to make predictions for each sample in our dataset. Respective expression subtype predictions (class probabilities) were illustrated by a clustered heatmap.

Differential expression of genes was determined using the R package “LIMMA” [20]. The overall significance of a signature was assessed via a “global test” [21]. The global test is a set based method, which tries to reject the null hypothesis that all genes in set of interest (in our case all signature genes) show no association to a defined clinical outcome or grouping. Gene ontology statistics were analysed using GOstat [22].

Immunohistochemistry (IHC)

IHC was done for the validation of identified differentially expressed genes. Additional tissue sections from 20 tumors that had been used for the microarray experiment and another 20 specimens from other tumors with solid and lepidic growth patterns were stained with the corresponding antibodies for ITPKA (inositol-1,4,5-trisphosphate-3-kinase-A) (polyclonal anti-rabbit ITPKA, 1/100, Atlas, Stockholm, Sweden), PFKP (phosphofructokinase, platelet) (polyclonal anti-rabbit PFKP, 1/50, (Thermo Fisher Scientific, Waltham, MA, USA), ERRFI1 (MIG6, mitogen-inducible gene 6) (polyclonal rabbit anti-ERRFI1, 1/100, Atlas, Stockholm, Sweden) and angiogenin (ANG) (polyclonal rabbit anti-angiogenin, 1/100, Abcam, Cambridge, United Kingdom) using an automated staining protocol on the DAKO autostainer (antigen retrieval with citrate buffer pH 6.0). Positive control tissue sections were chosen according to the manufacturers’ antibodies information, i.e. cerebral cortex for ITPKA, kidney for PFKP, breast cancer for ERRFI1, and liver for angiogenin. Isotype- and concentration-matched control antibodies (Dako, Hamburg, Germany) served as negative controls.

Semi-quantitative evaluation of protein expression was done using the H-Score method according to Pirker et al [23]. The percentage of tumor cells at different staining intensities was determined by visual assessment at 200-fold magnification, with the score calculated using the formula 1 x (% of 1+ cells) + 2 x (% of 2+ cells) + 3 x (% of 3+ cells) [23,24]. Samples were classified as negative (H-Score 0–50), weakly positive (H-Score 51–100), moderately positive (H-Score 101–200) or strongly positive (H-Score 201–300). The average H-Score values for each growth pattern and each antibody staining were calculated and compared.

Results

Molecular profiling of IASLC/ATS/ERS classified growth patterns

In total microdissected tissue sections of 48 specimens were addressed for RNA extraction and microarray experiments. RIN (RNA integrity number) values between 7 and 8 indicate sufficient RNA quality of microdissected tissues (mean RIN 7.5, SD 0.86) for microarray analysis (S1 Fig). Microarray analyses between the five different growth patterns resulted in numerous differentially expressed genes between the solid architecture and other patterns (S1 Table). Only a few solid-independent comparisons (e.g. papillary vs. micropapillary) indicated significant transcriptome differences. In the following, we focused on the gene signature between the solid and lepidic pattern associated with different clinical outcomes. Earlier, we tested all 48 adenocarcinoma transcriptomes according to their similarities to previously reported gene expression subtypes [12]. Supervised classification using 260 informative genes from the Wilkerson signature showed that our 48 samples could be assigned to each of the three classes proximal-inflammatory (PI), proximal-proliferative (PP) and terminal respiratory unit (TRU) with high confidence (S2 Fig). Furthermore, the unsupervised multidimensional scaling plot of the joint expression data from our samples and the TCGA data indicated, that all our samples fell clearly within the distribution of each of the three classes (S3 Fig). Altogether, the solid patterns (90% of specimens) clearly assigned for the PI subtype, acinar patterns (70%) predominantly assigned for the PP and lepidic patterns (70%) for the TRU subtype (S2 Table). However, differentially expressed genes between the microdissected histological patterns differed from previously reported gene signatures in adenocarcinoma subtypes, as highlighted for the solid-lepidic gene signature in the following paragraph.

Differences in gene expression, cellular processes and signalling pathways in the solid and lepidic patterns

Comparison of the transcriptomic activity between the solid and lepidic patterns revealed 705 up- and 110 downregulated non-redundant genes (FDR (false discovery rate) 5%, fold change >1.5 or < 0.66) (S3 Table). A clear separation of the specimens of both patterns could be confirmed by hierarchical clustering (S4 Fig). Furthermore, only 25 of 815 deregulated genes (3%) overlapped with the reported 506 LAD predictor genes classifying the molecular subtypes [12]. Similarly, further reported sets of differentially expressed genes between intrinsic molecular subtypes displayed a poor overlap (S4 Table).

Gene ontology analysis suggested several biological processes, which are linked to overrepresented, upregulated genes in the solid pattern. Cancer-associated processes included cell motility, proliferation, cell cycle and negative regulation of apoptosis (Table 2).

Table 2 shows gene ontologies for 710 significantly upregulated genes in solid compared with lepidic architecture using GOstat p-value ≤ 0.005. No significant gene ontologies were resulted for the 105 downregulated genes.

Gene expression regulation was represented by RNA processing and splicing. Protein linked processes included ontologies like protein transport, protein targeting, ribonucleoprotein complex biogenesis and ubiquitin cycle. Focusing on signal transduction pathways, members of the MAPK [mitogen-activated protein kinases) signalling (MAP2K1, MAPK6, MAPKAPK5, MAP2K1IP1 and MAPK8IP3) and NF-κB (nuclear factor 'kappa-light-chain-enhancer' of activated B-cells) signalling (IKBKG, LITAF, STAT1, BCL3, TFG and TBK1) were upregulated in the solid pattern.

Independent on DEseq and gene ontology analysis, known oncogenes and tumor suppressors in lung adenocarcinoma have been investigated for gene expression variance. Most of the genes were not informative on transcript level, only MET and MAP2K1 expression indicates upregulation in the solid pattern (S5 Table).

For IHC validation, we selected PFKP, ITPKA and ERRFI1 upregulated in the solid pattern and ANG upregulated in the lepidic pattern as putative novel biomarkers in distinct predominant architectures.

Immunohistochemical validation

The pattern-specific protein expression of ITPKA and angiogenin by immunohistochemistry confirmed the RNA levels. The protein ITPKA was more abundant in the solid pattern, showing cytoplasmic staining, and angiogenin was more abundant in the lepidic pattern, showing nuclear staining. The strongest differences in protein expression between the two patterns using the H-Score was shown for ITPKA (p = 0.02) and angiogenin (p = 0.113, not significant). Cytoplasmic and nuclear expression of PFKP was present in both patterns, slightly more in the solid architecture (not significant). The cytoplasmic expression of ERRFI1 remained below the 50 point H-score level for both patterns, and was slightly higher expressed in the solid architecture (Fig 1).

thumbnail
Fig 1. Immunohistochemistry.

Immunohistochemical staining of inositol-1,4,5-trisphosphate-3-kinase-A (ITPKA), phosphofructokinase, platelet (PFKP), mitogen-inducible gene 6 (ERRFI1) and angiogenin (ANG) in pulmonary adenocarcinoma with predominant lepidic (B, E, H, K) and solid growth patterns (C, F, I, L). 200x magnification. Histograms depict mean staining scores with standard deviation of n = 20 adenocarcinomas each (A, D, G, J).

https://doi.org/10.1371/journal.pone.0206132.g001

Discussion

For a long time, the nature of adenocarcinomas was histologically and clinically seen as a monolithic block. Clinical research within the last decade has led to a precise definition of subtypes considering growth patterns and molecular characteristics which are associated with diverse clinical outcomes [1,5,13]. However, histopathological growth patterns and prognostic diversity did not necessarily match with molecular subtypes defined by transcriptional profiles, CpG island methylation or oncogene mutations [13]. So far, biomarkers and molecular targets specific for tumor growth patterns have been sporadic. For example, an association was reported between higher expression of TTF-1 and the lepidic pattern [25]. In the solid pattern, C4.4A was reported as a surrogate marker for a poor outcome [26]. The objective of this project was to broaden our knowledge of the different molecular structures between distinct patterns of lung adenocarcinoma. Furthermore, this may help to establish novel biomarkers and target sites that are valuable for future individualised treatment strategies in human pulmonary adenocarcinoma. To our knowledge, we present the first screening study to identify growth pattern-specific gene signatures. Our approach included laser-captured microdissection to separate specific growth patterns in tissue sections of lung adenocarcinoma. Subsequent microarray analyses depicted transcriptome differences between the solid pattern and other growth patterns. A gene classifier could be adopted to assign all 48 gene expression profiles to previously defined molecular subtypes linked to lung adenocarcinomas. For example, the majority of solid specimens matched with the PI subtype. This association was also described in a comprehensive study including 230 adenocarcinomas [13]. Here, upregulated genes in the solid pattern were found to be more frequently aberrant in the PI subtype. Moreover, we suggest molecular similarities between the acinar pattern and the PP subtype as well as the lepidic pattern and the TRU subtype. Of note, both these tumor subgroups have been independently found to be prognostically favourable [5,12].

Microarray profiles revealed a clear separation between the solid and lepidic growth patterns upon tissue microdissection. Further downstream analyses of the gene signature suggested oncogenic targets, pathways and gene ontologies in both growth patterns. Genes upregulated in the worse prognostic solid architecture were associated with cell proliferation, cell cycle activation, inhibition of apoptosis and cell motility. Our signature did not show a significant overlap with existing lung adenocarcinoma microarray data. The poor accordance with other signatures is likely reasonable since microdissected growth patterns are not readily comparable with unselected tumor tissues.

Four candidate genes were selected for IHC protein analyses according to their expression differences between the solid and the lepidic pattern. The IHC results indicated a higher abundance of ITPKA in the solid pattern of pulmonary adenocarcinoma. Up-regulation of ITPKA in tumor cell lines with low endogenous ITPKA expression increases migration [27]. ITPKA is preferentially expressed in cell lines derived from metastases of small cell lung cancer and squamous lung carcinoma, whereas pulmonary adenocarcinoma shows a high expression frequency of ITPKA in primary tumor cells. The expression of ITPKA in adenocarcinoma might increase the invasive potential of cancer cells. Furthermore, this enzyme is considered a potential target for anti-metastatic therapy and small molecule inhibitors [28]. Our results suggest ITPKA as a potential target for the development of targeted therapies, particular for predominant solid lung adenocarcinoma.

Angiogenin was shown to be up-regulated in the lepidic pattern on the RNA and protein levels. Angiogenin is known as a pro-angiogenic growth factor that is up-regulated in several types of cancer. Nuclear expression of angiogenin has been shown in about two thirds of lung adenocarcinomas, and target inhibition impairs xenograft tumor proliferation and angiogenesis [29]. A p53 interacting function of angiogenin in anti-apoptosis and survival of cancer cells suggests that targeting angiogenin could be an effective therapy for several cancers [30]. Our data indicate that angiogenin might be of importance in the lepidic architecture, nevertheless our findings were not statistically significant. Upregulation of PFKP and ERRFI1 in the solid architecture could not be confirmed on the protein level. Possible explanations are the discordance between RNA and protein, or additional tumor characteristics as confounding factors. For example, it has been shown that ERRFI1/MIG6 expression is associated with EMT and resistance to EGFR inhibitors in lung cancer xenografts [31].

Overall, this study presents distinct transcriptomic differences between lung adenocarcinoma growth patterns, which could be validated for the solid-expressed ITPKA and the lepidic-expressed angiogenin proteins. As a limitation, these gene signatures and putative targets could only indirectly be linked to prognosis via previously defined prognostic adenocarcinoma subgroups. Our screening and validation cohort was designed for pattern-specific expression analysis, but not adequate to include clinical follow-up data into statistical analyses. Further studies in larger cohorts of IASLC/ATS/ERS classified adenocarcinomas are needed to better understand the associations between molecular heterogeneity and clinical outcome in pulmonary adenocarcinoma.

Conclusion

To our knowledge, this study is the first to identify growth pattern-specific gene signatures in pulmonary adenocarcinoma. With tumor profiling at the molecular level the study provides valuable new information about pulmonary adenocarcinoma and allows a better assessment of the five adenocarcinoma subgroups. As this study at hand is only a modest contribution to basic cancer research further validation of individual biomarkers and target sites in distinct histological patterns are strictliy necessary to pave the way toward novel approaches in pathological diagnostics and personalised therapy in the future.

Supporting information

S1 Fig. RIN (RNA integrity number) values of 48 samples.

Illumina chip raw data quartiles and RIN values were combined for ordered samples LC1-LC48. Left axis defines log2 expression value quartiles for each chip, numbering on the right assign RIN value for each RNA sample used for the chip.

https://doi.org/10.1371/journal.pone.0206132.s001

(TIF)

S2 Fig. Heatmap showing SVM predictions for the expression subtype for all 48 specimens in this study.

The color code indicates the class probability for each of the three expression subtypes, thus visualizing the level of confidence of the prediction. Notably, the SVM model was trained on the data by Wilkerson et al. based on the overlap of their reported signature and our chip (260 genes).

https://doi.org/10.1371/journal.pone.0206132.s002

(TIF)

S3 Fig. Multi-dimensional scaling plot of joint expression data from Wilkerson et al. (2012) and this study.

Samples from Wilkerson et al. (2012) are colored according to the expression subtype (PI (red), PP (green), TRU (black)). Samples from this study are labeled with a number, linking to S1 Table (“LC_” is omitted from the text labels in this plot to save space).

https://doi.org/10.1371/journal.pone.0206132.s003

(TIF)

S4 Fig. Comparison of the transcriptomic activity between the solid and lepidic patterns.

Hierarchical clustering (Ward’s method) of differentially expressed genes between the solid and lepidic growth patterns using Pearson correlation distance.

https://doi.org/10.1371/journal.pone.0206132.s004

(TIF)

S1 Table. Differentially expressed genes between all tumor pattern comparisons.

S1 Table shows LIMMA analyses and the number of differentially expressed genes (FDR 5%, fold change >1.5 or < 0.66) between all tumor pattern comparisons.

https://doi.org/10.1371/journal.pone.0206132.s005

(PDF)

S2 Table. Molecular subtype assignment.

S2 Table shows molecular subtype assignment using the reported nearest centroid subtype predictor overlap (Wilkerson PlosOne, 2012)

https://doi.org/10.1371/journal.pone.0206132.s006

(PDF)

S3 Table. Differentially expressed non-redundant genes between solid and lepidic architecture specimens.

LIMMA analysis revealed 815 differentially expressed non-redundant genes (FDR 5%, fold change > 1.5 or < 0.66) between solid and lepidic architecture specimens. The gene list is ordered according to a decreasing fold change.

https://doi.org/10.1371/journal.pone.0206132.s007

(PDF)

S4 Table. Comparison of the 50 highest upregulated genes in solid or lepidic growth pattern with published gene signatures using GeneSigDB.

S4 Table shows the comparison of the 50 highest upregulated genes in solid or lepidic growth pattern with published gene signatures using GeneSigDB. Table view was restricted to studies with a minimum of 20% overlap (10 genes).

https://doi.org/10.1371/journal.pone.0206132.s008

(PDF)

S5 Table. Comparison of known lung adenocarcinoma oncogenes and tumor suppressor genes between different growth patterns.

Known lung adenocarcinoma oncogenes and tumor suppressor genes were shown with MEAN and SD expression values in all growth patterns. T-test between different growth patterns were mostly not significant (p-value > 0.05, not adjusted). Only one group comparison indicated higher expression of MET and MAP2K1 (MEK1) in solid compared to lepidic growth pattern (p-value 0.04).

https://doi.org/10.1371/journal.pone.0206132.s009

(PDF)

Acknowledgments

This study was supported by the “Stiftung für Krebs- und Scharlachforschung Mannheim” and by the German Federal Ministry of Education and Research: German Center for Lung Research (82DZL00404).

The freezing and the management of the excised tumor samples was carried out by Christa Stolp. Jutta Scheuerer prepared the frozen tissue for laser captured microdissection, assisted in RNA extraction and quality assessment. The tissue used for immunohistochemical validation was provided by the tissue bank of the National Center for Tumor Diseases (NCT) Heidelberg. It was used in accordance with the regulations of the NCT tissue bank. Expression profiling was provided by the German Cancer Research Center (DKFZ) Core Unit Genomics and Proteomics Core Facility.

References

  1. 1. Travis WD, Brambilla E, Noguchi M, Nicholson AG, Geisinger KR, Yatabe Y, et al. International association for the study of lung cancer/american thoracic society/european respiratory society international multidisciplinary classification of lung adenocarcinoma. J Thorac Oncol. 2011; 6: 244–285. pmid:21252716
  2. 2. Travis WD, Brambilla E, Noguchi M, Nicholson AG, Geisinger K, Yatabe Y, et al. Diagnosis of lung adenocarcinoma in resected specimens: implications of the 2011 International Association for the Study of Lung Cancer/American Thoracic Society/European Respiratoy Society classification. Arch Pathol Lab Med. 2013; 137: 685–705. pmid:22913371
  3. 3. Yoshizawa A, Motoi N, Riely GJ, Sima CS, Gerald WL, Kris MG, et al. Impact of proposed IASLC/ATS/ERS classification of lung adenocarcinoma: prognostic subgroups and implications for further revision of staging based on analysis of 514 stage I cases. Mod Pathol. 2011; 24: 653–664. pmid:21252858
  4. 4. Russell PA, Wainer Z, Wright GM, Daniels M, Conron M, Williams RA. Does lung adenocarcinoma subtype predict patient survival?: A clinicopathologic study based on the new International Association for the Study of Lung Cancer/American Thoracic Society/European Respiratory Society international multidisciplinary lung adenocarcinoma classification. J Thorac Oncol. 2011; 6: 1496–1504. pmid:21642859
  5. 5. Warth A, Muley T, Meister M, Stenzinger A, Thomas M, Schirmacher P, et al. The novel histologic International Association for the Study of Lung Cancer/American Thoracic Society/European Respiratory Society classification system of lung adenocarcinoma is a stage-independent predictor of survival. J Clin Oncol. 2012; 30: 1438–1446. pmid:22393100
  6. 6. Solis LM, Behrens C, Raso MG, Lin HY, Kadara H, Yuan P, et al. Histologic patterns and molecular characteristics of lung adenocarcinoma associated with clinical outcome. Cancer. 2012; 118: 2889–2899. pmid:22020674
  7. 7. Hung JJ, Jeng WJ, Chou TY, Hsu WH, Wu KJ, Huang BS, et al. Prognostic value of the new International Association for the Study of Lung Cancer/American Thoracic Society/European Respiratory Society lung adenocarcinoma classification on death and recurrence in completely resected stage I lung adenocarcinoma. Ann Surg. 2013; 258: 1079–1086. pmid:23532112
  8. 8. Bhattacharjee A, Richards WG, Staunton J, Li C, Monti S, Vasa P, et al. Classification of human lung carcinomas by mRNA expression profiling reveals distinct adenocarcinoma subclasses. Proc Natl Acad.Sci USA. 2001; 98: 13790–13795. pmid:11707567
  9. 9. Garber ME, Troyanskaya OG, Schluens K, Petersen S, Thaesler Z, Pacyna-Gengelbach M, et al. Diversity of gene expression in adenocarcinoma of the lung. Proc Natl Acad Sci USA. 2001; 98: 13784–13789. pmid:11707590
  10. 10. Beer DG, Kardia SL, Huang CC, Giordano TJ, Levin AM, Misek DE, et al. Gene-expression profiles predict survival of patients with lung adenocarcinoma. Nat Med. 2002; 8: 816–824. pmid:12118244
  11. 11. Hayes DN, Monti S, Parmigiani G, Gilks CB, Naoki K, Bhattacharjee A, et al. Gene expression profiling reveals reproducible human lung adenocarcinoma subtypes in multiple independent patient cohorts. J Clin Oncol. 2006; 24: 5079–5090. pmid:17075127
  12. 12. Wilkerson MD, Yin X, Walter V, Zhao N, Cabanski CR, Hayward MC, et al. Differential pathogenesis of lung adenocarcinoma subtypes involving sequence mutations, copy number, chromosomal instability, and methylation. PLoS One. 2012; 7: e36530. pmid:22590557
  13. 13. Collisson EA, Campbell JD, Brooks AN, Berger AH, Lee W, Chmielecki J, et al. Cancer Genome Atlas Research Network. Comprehensive molecular profiling of lung adenocarcinoma. Nature. 2014; 511: 543–550.
  14. 14. Burgemeister R. New aspects of laser microdissection in research and routine. J Histochem Cytochem. 2005; 53: 409–412. pmid:15750030
  15. 15. Hergenhahn M, Kenzelmann M, Gröne HJ. Laser-controlled microdissection of tissue opens a window of new opportunities. Pathol Res Pract. 2003; 199: 419–423. pmid:12924444
  16. 16. Du P, Kibbe WA, Lin SM. lumi: a pipeline for processing Illumina microarray. Bioinformatics. 2008; 24: 1547–1548. pmid:18467348
  17. 17. Anders S, Huber W. Differential expression analysis for sequence count data. Genome Biol. 2010; 11: R106. pmid:20979621
  18. 18. Zwiener I, Frisch B, Binder H. Transforming RNA-Seq data to improve the performance of prognostic gene signatures. PLoS One. 2014; 9: e85150. pmid:24416353
  19. 19. Walker WL, Liao IH, Gilbert DL, Wong B, Pollard KS, McCulloch CE, et al. Empirical Bayes accommodation of batch-effects in microarray data using identical replicate reference samples: application to RNA expression profiling of blood from Duchenne muscular dystrophy patients. BMC Genomics. 2008; 9: 494. pmid:18937867
  20. 20. Smyth GK. Linear models and empirical bayes methods for assessing differential expression in microarray experiments. Stat Appl Genet Mol Biol. 2004; 3: Article 3.
  21. 21. Goeman JJ, van de Geer SA, de Kort F, van Houwelingen HC. A global test for groups of genes: testing association with a clinical outcome. Bioinformatics. 2004; 20: 93–99. pmid:14693814
  22. 22. Beissbarth T, Speed TP. GOstat: find statistically overrepresented Gene Ontologies within a group of genes. Bioinformatics. 2004; 20: 1464–1465. pmid:14962934
  23. 23. Pirker R, Pereira JR, von Pawel J, Krzakowski M, Ramlau R, Park K, et al. EGFR expression as predictor of survival for first-line chemotherapy plus cetuximab in patients with advanced non-small-cell lung cancer: analysis of data from the phase 3 FLEX study. Lancet Oncol. 2012; 13: 33–42. pmid:22056021
  24. 24. Mazières J, Brugger W, Cappuzzo F, Middel P, Frosch A, Bara I, et al. Evaluation of EGFR protein expression by immunohistochemistry and the magnification rule: re-analysis of the SATURN study. Lung Cancer. 2013; 82: 231–237. pmid:23972450
  25. 25. Kadota K, Nitadori J, Sarkaria IS, Sima CS, Jia X, Yoshizawa A, et al. Thyroid transcription factor-1 expression is an independent predictor of recurrence and correlates with the IASLC/ATS/ERS histologic classification in patients with stage I lung adenocarcinoma. Cancer. 2013; 119: 931–938. pmid:23096929
  26. 26. Jacobsen B, Muley T, Meister M, Dienemann H, Christensen IJ, Santoni-Rugiu E, et al. Ly6/uPAR-related protein C4.4A as a marker of solid growth pattern and poor prognosis in lung adenocarcinoma. J Thorac Oncol. 2013; 8: 152–160. pmid:23287851
  27. 27. Windhorst S, Kalinina T, Schmid K, Blechner C, Kriebitzsch N, Hinsch R, et al. Functional role of inositol-1,4,5-trisphosphate-3-kinase-A for motility of malignant transformed cells. Int J Cancer. 2011; 129: 1300–1309. pmid:21792881
  28. 28. Schröder D, Rehbach C, Seyffarth C, Neuenschwander M, Kries JV, Windhorst S. Identification of a new membrane-permeable inhibitor against inositol-1,4,5-trisphosphate-3-kinase A. Biochem Biophys Res Commun. 2013; 439: 228–234. pmid:23981806
  29. 29. Yuan Y, Wang F, Liu XH, Gong DJ, Cheng HZ, Huang SD. Angiogenin is involved in lung adenocarcinoma cell proliferation and angiogenesis. Lung Cancer. 2009; 66: 28–36. pmid:19423182
  30. 30. Sadagopan S, Veettil MV, Chakraborty S, Sharma-Walia N, Paudel N, Bottero V, et al. Angiogenin functionally interacts with p53 and regulates p53-mediated apoptosis and cell survival. Oncogene. 2012; 31: 4835–4847. pmid:22266868
  31. 31. Izumchenko E, Chang X, Michailidi C, Kagohara L, Ravi R, Paz K, et al. The TGFβ-miR200-MIG6 pathway orchestrates the EMT-associated kinase switch that induces resistance to EGFR inhibitors. Cancer Res. 2014; 74: 3995–4005. pmid:24830724