Comprehensive genomic profiling of Finnish lung adenocarcinoma cohort reveals high clinical actionability and SMARCA4 altered tumors with variable histology and poor prognosis

Introduction Lung adenocarcinoma is the most common type of lung cancer and typically carries a high number of mutations. However, the genetic background of the tumors varies according to patients’ ethnic background and smoking status. Little data is available on the mutational landscape and the frequency of actionable genomic alterations in lung adenocarcinoma in the Finnish population. Materials and methods We evaluated the gene alteration frequencies of 135 stage I–IV lung adenocarcinomas operated at Turku University Hospital between 2004 and 2017 with a large commercial comprehensive genomic profiling panel. Additionally, we correlated the alterations in selected genes with disease outcomes in 115 stage I–III patients with comprehensive follow-up data. The genomic alterations in a sub-cohort of 30 never-smokers were assessed separately. Results Seventy percent of patients in the overall cohort and 77% in the never-smoker sub-cohort harbored an alteration or a genomic signature targetable by FDA and/or EMA approved drug for non-small cell carcinoma, respectively. In multivariable analysis for disease-specific survival, any alteration in SMARCA4 (DSS; HR 3.911, 95%CI 1.561–9.795, P=0.004) exhibited independent prognostic significance along with stage, tumor mutation burden, and predominant histological subtypes. Conclusions Over two thirds of our overall cohort, and especially never-smokers had an actionable genomic alteration or signature. SMARCA4 alterations, detected in 7.4% of the tumors, independently predicted a shortened overall and disease-specific survival regardless of the alteration type. Most SMARCA4 alterations in our cohort were missense mutations associated with differentiated predominant histological subtypes and immunohistochemical SMARCA4/BRG1 and TTF-1 positive status.


Introduction
Lung cancer is the leading cause of cancer-related mortality worldwide [1] . Most lung cancers are non-small cell lung cancers (NSCLC), and adenocarcinoma is the most common subtype with an increasing trend in relative incidence [2] . Adenocarcinomas harbor targetable genomic alterations in ALK, BRAF, EGFR, and ROS1 genes, and novel treatment options such as KRAS Gly12Cys , MET, NTRK, and RET inhibitors, and possibly ERBB2 guided therapies are broadening the clinical repertoire [3][4][5] . In addition to guiding the treatment of individual cancers, genomic alterations may indicate resistance to targeted therapies [6] and immunotherapy [7] , as well as conventional chemotherapy [8] . As the mutational landscape varies by ethnic background, knowledge of the local mutational frequencies will help allocate resources to detect clinically significant alterations in cases where broad genomic profiling is not routinely used. Finns differ genetically from other European populations, and there is little previous data on the mutational landscape of Finnish lung adenocarcinoma patients [9] .
Most of the lung adenocarcinoma patients in the Western population are current or ex-smokers, but the proportion of never-smokers is expected to increase as smoking declines. Adenocarcinomas in never-smokers are genetically different from those detected among smokers as the latter group has a significantly higher number of individual mutations and different mutation types [10][11][12] . Tumors with EGFR mutations are common in never-smokers, occurring in 27.4-66.7% of European [13][14][15][16][17][18] and 60-78% of East Asian [19] patients without a smoking history. Besides EGFR , neversmoking status is associated with oncogenic gene fusions involving ALK1, ROS1 , and RET . The genomic data on European never-smokers is still relatively scarce, and to our knowledge, there is no published genomic data about adenocarcinomas among never-smokers in the Finnish population.
Here we mapped the mutational landscape of lung adenocarcinomas in a single-center patient cohort via a large commercial comprehensive genomic profiling panel, emphasizing currently recognized actionable alterations and alterations in never-smokers.

Materials and methods
Our retrospective cohort consisted of 135 patients with stage I-IV primary invasive lung adenocarcinoma operated in Turku University Hospital in 2004-2017 with curative intent. Two patients had received neoadjuvant chemotherapy, while none had received radiotherapy. Three patients underwent immunotherapy in 2017-2018. We traced the smoking status and other clinicopathological characteristics from the electronic patient records. A never-smoker was defined as a person who self-reportedly had consumed less than 100 cigarettes in their lifetime. We acquired the day and cause of death through Statistics Finland, with the last follow-up day on 31.12.2018. We excluded patients from survival analysis based on the following criteria: incomplete clinical follow-up data, death within 30 postoperative days, macroscopic (R2) residual disease, and immunotherapy during the follow-up period. The collection of clinical patient data was approved by the administration of the Hospital District of Southwest Finland (T150/16), and the use of tissue material was approved by the Scientific Steering Committee of Auria Biobank (AB14-8689 and AB20-9755). We conducted the study in collaboration with Auria Biobank and Roche Oy (Espoo, Finland).

Genomic characterization
FoundationOne (Foundation Medicine, Inc., Cambridge, MA, USA) comprehensive genomic profiling, described and validated by Frampton et al. [20] , was performed on ten 5 μm thick formalin-fixed paraffin-embedded tissue sections on charged and unbaked slides. We omitted tumors with an insufficient number of non-necrotic tumor cells and those limited to a single tumor block per patient. FoundationOne provided the genomic data classified as short variant mutations (single nucleotide variants (SNVs) and indels of 1-40 base pairs), copy number alterations (amplifications and losses), and rearrangements. These were further categorized as either known pathogenic, likely pathogenic, or variants of uncertain significance (VUS). The FoundationOne panel version used in our study covered the exons of 315 genes and selected intronic regions [20] .

Immunohistochemical staining
Immunohistochemical SMARCA4/BRG1 stainings were performed at Fimlab Laboratories (Tampere, Finland) using an in-house validated protocol. In brief, tissue sections of 4 μm thickness were cut from the same paraffin blocks used for sequencing, deparaffinized with xylene, and rehydrated in a series of ethanol. Antigen retrieval was performed with Cell Conditioning 1 (CC1) solution and heat-induced epitope retrieval (HIER). Staining was performed on Ventana BenchMark ULTRA staining instrument using rabbit monoclonal SMARCA4/BRG1 antibody (1:100, clone EPNCIR111A, Abcam, Cambridge, UK), and the sections were counterstained with hematoxylin. Lymphatic tissue on the same slide was used as a positive external control, and intratumoral lymphocytes were used as internal positive controls.

Statistical analyses
The clinicopathological and genomic data were correlated with the χ 2 test and Fischer ś exact test. Overall survival (OS) and disease-specific (DSS) survival were estimated using the Kaplan-Meier method and the log-rank test. The genomic data were adjusted for clinical parameters with multivariable Cox regression analysis. P-values less than 0.05 were considered statistically significant. Statistical analyses were performed with SPSS (IBM, version 28, 2021). The waterfall plots were created with R (R Project, version 3.6.1, 2020) GenVisR package and modified with GIMP.

Patient characteristics
A total of 135 patients with surgically treated primary lung adenocarcinoma and available FoundationOne genomic data were included in the study. All the patients were of Finnish European descent. After applying the exclusion criteria, 115 patients remained in the survival analyses. The never-smoker sub-cohort consisted of 30 patients (70% women). The mean follow-up period was 56.1 months (range 7.0-173.5 months). The clinical characteristics of the cohort are summarized in Table 1 .

Mutational landscape of tumors
The tumors harbored 2682 individual non-synonymous genomic alterations (2366 SNVs and small indels, 281 copy number alterations, and 35 rearrangements), ranging from four to 99 alterations per tumor. The average median exon coverage was 571. Median tumor mutational burden (TMB) was 7.02 (range 0.0-88.6), and 49 patients had TMB of at least ten mutations/MB. Microsatellite instability was present in one tumor, cooccurring with high TMB (35.1 mutations/MB).
Overall, TP53 was the most commonly altered gene, with 54.8% of tumors having TP53 alterations of any kind. One or more alterations in KRAS were detected in 37.8% of the tumors, followed by LRP1B, SPTA1 , and PRKDC (28.1% each), EGFR (23.7%), and STK11 (20.0%). KEAP1 was mutated in 14.1% and NFE2L2 in 5.9% of the tumors. The genes with mutations in at least 10% of patients in the overall cohort are presented in Fig. 1 A. The frequencies of individual genomic alterations of all the genes included in the panel are reported in Supplementary Table S1 and all the alterations in selected genes relevant to lung adenocarcinoma in Supplementary Table S2

Established genomic biomarkers per European Society for Medical Oncology guidelines
At the time of writing this article, the European Society for Medical Oncology (ESMO) guidelines recommended the molecular subtyping of EGFR, ALK, ROS1, BRAF , and NTRK as predictive biomarkers for first-line targeted therapies in metastatic non-small cell carcinoma [21] . Over a third of our patients (31.9%) harbored a pathogenic or likely pathogenic alteration in these genes, mutually exclusively. A total of twenty-nine patients (21.5%) had one or more alterations in EGFR , nine patients (6.7%) in BRAF , three patients (2.2%) in RET , and one patient each (0.7%) in ALK and ROS1. There were no NTRK fusions.
EGFR exon 19 deletions were identified in 13 patients (44.8% of patients with pathological or likely pathological EGFR alterations) and Leu858Arg EGFR mutation in eight patients (27.6%). Two patients had tumors with compound Gly719Ala and Ser768Ile EGFR mutations.
As our patients were treatment naïve at the time of their surgery, no acquired EGFR tyrosine kinase inhibitor (TKI) resistance mutations were present, and neither were primary resistance mutations observed. EGFR -TP53 co-alterations, shown to associate with shorter progression-free survival following 1 st or 2 nd generation EGFR TKI treatment [22] , were seen in 16 tumors. Four patients harbored EGFR exon 20 in-frame insertion, conferring resistance to EGFR TKIs but sensitizing to amivantamab. EGFR was amplified in four tumors, with three having a concurrent activating EGFR mutation and one having an exon 20 insertion, potentially modifying EGFR TKI treatment response.
Five patients had a BRAF Gly469Val mutation, while four patients had other non-Val600 mutations, and one patient had a compound Gly469Val and Ser605Cys mutation. There were no Val600Glu mutations. A previously reported RET fusion was present in three patients, with two having RET / KIF5B fusions and one having a RET / CCDC6 fusion. One patient had a previously unreported RET/PTPLA fusion with a concurrent KRAS Gly12Val mutation. There was one case of ALK/EML4 and ROS1/CD74 fusion each (0.7%).
In conclusion, 24 patients had an EGFR mutation with an indication for EGFR TKIs (exon 19 deletions, Leu858Arg mutation, and Gly719X mutations), four patients had an exon 20 insertion for which amivantamab is recommended, and eight patients had EGFR alterations predicting a decreased response to EGFR TKIs ( EGFR exon 20 insertion, EGFR amplification). Five tumors harbored fusion genes with available targeted treatment ( ALK, ROS1, RET ) ( Table 2 ).

Emerging genomic biomarkers
The genomic biomarkers not yet recommended by ESMO in routine practice but already present in the NCCN, CAP/IASCL/AMP, and ASCO guidelines include KRAS, ERBB2 , and MET [23] . Additionally, FDA has accepted tumor mutation burden (TMB) as a biomarker for immunotherapy with pembrolizumab.
In KRAS , the most common genomic alteration was Gly12Cys (20 patients, 35.1% of patients with any KRAS alterations). Other pathogenic and likely pathogenic KRAS alterations, to which presently no targeted treatment is available, were present in 37 patients (Supplementary Table S2).
The most common ERBB2 alteration was an exon 20 insertion, present in four patients. Two patients had an ERBB2 amplification. One patient had an activating exon 8 mutation Ser310Phe, and one patient had an activating exon 19 Asp769His mutation. In MET , six exon 14 alterations were identified, including three exon 14 skipping splice site mutations, two Tyr1003X mutations, and one Asp1010His mutation. MET amplification, indicating resistance to EGFR TKIs and sensitivity to small molecule MET inhibitors, was present in three tumors.
In conclusion, 26 patients had a variant (20 KRAS Gly12Cys and six MET exon 14 alterations) targetable with drugs already accepted by the U.S. Food and Drug Administration (FDA) and the European Medicines Agency (EMA). Additionally, four ERBB2 exon 20 insertions and one exon 19 SNV, recommended to be treated with trastuzumab-drug conjugates in the NCCN guidelines, were present. Furthermore, three cases of MET amplification were found.

Additional alterations in homologous recombination repair genes, and co-alterations in key genes relevant to cancer
Defects in genes involved in homologous recombination repair (HRR) are being trialed as histology-agnostic biomarkers for PARP inhibitors in various solid malignancies, either alone or in combination with chemotherapy, targeted therapy, or immunotherapy. In non-small cell lung cancer, the prevalence of HRR gene deficiencies has been reported to be approximately 15% [24] . In our cohort, 18 individual tumors (13.3%) had known or likely pathogenic alterations in the following HRR associated genes: ATM in eleven patients, BRCA1/2 in four patients, BRIP1 in one patient, CHEK2 in three patients, RAD50 and RAD54L in two patients each, and one alteration each in Table 2 Actionable and potentially actionable mutations and genomic signatures across the cohort (n = 135). TKI = Tyrosine kinase inhibitor, FDA = U.S. Food and Drug Administration, EMA = European Medicines Agency, MSI-H = microsatellite instability. FANCM, BARD1, BRIP1 , and MRE11A . We were unable to confirm whether these variants were germline or somatic in origin. Co-alterations involving KRAS, TP53, STK11 , and CDKN2A/B were present in 33% of the whole cohort: KRAS -TP53 in 16.3%, KRAS -STK11 in 8%, TP53 -STK11 in 9%, and KRAS -CDKN2A/B loss in 4% of the tumors. KRAS -STK11 -alterations have been associated with a poor response to immune checkpoint inhibitor therapy [7] . MDM2 amplification, suggested being linked with hyperprogression during immunotherapy in a small series of various malignancies [25 , 26] , was present in 7% of the patients. Two patients with a co-occurring EGFR -TP53 -RB1 alteration, predisposing to transformation to small cell carcinoma or squamous cell carcinoma following EGFR TKI treatment [27 , 28] , were observed.

Clinical actionability in the cohort
In the overall cohort, 59 patients (27 never-smokers) had a genomic alteration, and 49 patients a genomic signature (TMB ≥ 10, MSI-H or both, one never-smoker) for which an FDA and/or EMA approved NSCLCindicated targeted treatment is currently available ( Table 2 , March 2022). Of these patients with targetable genomic alterations, 13 (22%) also had a TMB ≥ 10. Thus, overall 95 of our patients (70.4%) had alteration(s) or a genomic signature with an approved targeted drug available. Furthermore, additional alterations with a treatment option in other solid malignancies were detected in nine tumors (four ERBB2 exon 20 insertions, one ERBB2 Ser310Phe SNV, and four BRCA1/2 alterations). All the targetable or potentially targetable alterations were mutually exclusive except for one patient with a KRAS Gly12Cys and a concurrent BRCA2 variant.

The effect of smoking on genomic alterations
The mean age at the time of diagnosis was 71.5 years for smokers, and 64.6 years for never-smokers. Never-smokers had 332 individual genomic alterations (260 short-variants, 63 copy number alterations, and nine rearrangements; 5-19 alterations per patient).
The most frequently altered genes in never-smokers were EGFR (50.0% of patients harboring one or more alterations), TP53 (30.0%), FRS2 and MDM2 (26.7% each), CDKN2A and CDKN2B (7.9% each), BRCA2, ERBB2, PRKDC, PIK3CA, PTEN, SPEN and SPTA (16.7% each), while alterations in other genes were found in four patients (13.3%) or less. The pathogenic ERBB2 variant Ala775_Gly776insYVMA (four patients) and PTEN variant of unknown significance Asp268Glu (three patients) were found exclusively in never-smokers. Other variants detected only in neversmokers presented in only one or two patients. The never-smoking patients harbored all the gene fusions in the cohort. The frequencies of genes mutated in 10% or more of never-smokers are visualized in Fig. 1 B. Fig. 2 introduces the alteration frequencies in clinically important genes and illustrates the differences of genomic alterations between never-smokers and smokers.
A history of cigarette smoking was associated with more alterations of any type in genes including TP53, KRAS, SPTA1, LRP1B , and STK11. Neversmokers, in contrast, harbored more alterations in EGFR, FRS2, MDM2 , and CDKN2B . ( P < 0.05, data not shown). Eleven patients with a smoking history had RUNX1T1 amplification, a finding of unknown clinical significance. Four of our tumors with the RUNX1T1 amplification were predominantly solid in histology, five were predominantly acinar (with two cribriform tumors), one was micropapillary, and one was lepidic. The impact of smoking on the frequency of selected alterations is presented in Supplementary Table  S3.
Significantly, 23 (77%) out of 30 tumors found among never-smokers had an alteration for which an FDA-or EMA-approved NSCLC-indicated targeted treatment is currently available. Additional four never-smokers had an alteration potentially targetable in the near future ( Table 2 ).

Association of genomic alterations with clinicopathological characteristics and survival
The tumors were histologically classified as described in our earlier manuscript [29] . All tumors with any EGFR alterations except one case had predominant histological subtypes associated with favorable survival (lepidic, acinar, papillary). Pathogenic or likely pathogenic EGFR mutations occurred in 27.7% of tumors with favorable predominant histologic subtypes, but only in 2.9% of tumors with histologic subtypes associated with poor survival (solid, micropapillary, cribriform, fetal) ( P = 0.002). In contrast, known and likely pathogenic alterations in KRAS were associated with predominant histological subtypes conferring poor prognosis ( P = 0.035). The majority (72%, 8/11) of the tumors with MDM2 -FRS2 co-amplification were predominantly acinar in histology.
Out of genes relevant to lung adenocarcinoma, any alterations in SMARCA4 were associated with shortened OS (HR 2.732, 95%CI 1.291-5.782, P = 0.009) and DSS (HR 2.901, 95%CI 1.213-6.939, P = 0.017) in univariable analysis. The effect of SMARCA4 alterations on survival is presented in Fig. 3 A-B. Alterations of TP53, KRAS , or STK11 were not associated with differences in OS or DSS, either alone or in combination with each other ( P > 0.05). Similarly, the MDM2-FRS2 co-amplification status was not associated with either OS or DSS (p > 0.05).
SMARCA4 alterations were present in ten individual tumors. Seven of these tumors had a differentiated predominant histological subtype (acinar in six and papillary in one, Fig. 3 Ca-Cg), whereas the three remaining tumors were predominantly solid ( Fig. 3 Ch-Cj). The tumors with SMARCA4 alterations were composed of more than one histological subtype, except for one uniformly solid tumor ( Fig. 3 , patient Ch). Seven tumors harbored SMARCA4 missense variants, two of which had a concurrent truncating SMARCA4 mutation. Two tumors had a truncating SMARCA4 variant without missense variants, and one tumor had a partial SMARCA4 deletion. The clinicopathological characteristics of the individual tumors with SMARCA4 alterations are presented in Supplementary Table S4.
All the tumors with SMARCA4 alterations were further immunohistochemically stained for SMARCA4/BRG1 ( Fig. 3 Ca-Cj). The tumors with acinar or papillary predominant histology demonstrated either diffusely positive BRG1 staining (patients Ca-c, Cf-g) or heterogeneous staining including both positive and negative tumor cells (patients Cd-e). The solid predominant tumors (patients Ch-j), all harboring truncating SMARCA4 variants, had solid (patients Ch-j) and acinar (patient Ci) tumor areas that were BRG1 negative. Adjacent to these negative areas, the tumors exhibited either patchy positive solid areas (patient i) or diffusely positive lepidic growth (patient Cj) suggesting intratumoral heterogeneity in SMARCA4 mutation status.
Immunohistochemical TTF-1 staining was available from nine SMARCA4 altered tumors. One tumor with acinar predominant histology and a missense SMARCA4 variant (patient Cd) and one with solid predominant histology and a combination of truncating and missense SMARCA4 variants (patient Ci) were TTF-1 negative, while the remaining tumors were TTF-1 positive (Supplementary Table S4).

Discussion
In estimating the actionability of genome profiling results with the 315 gene panel, we primarily focused on alterations or genomic signatures for which a drug is currently (March 2022) approved for NSCLC by the FDA and/or EMA as they reflect the true actionability in the routine clinical setting. In the overall cohort, the majority (70%) of patients had such an alteration or signature dominated by EGFR alterations and TMB ≥ 10. The fraction of patients with actionable alterations remained  substantially high (42%) even when excluding TMB. In our cohort, the majority of samples (73%) were from early stage (stages I and II) disease, complicating comparisons to other cohorts with more advanced disease. However, the overall frequencies of actionable alterations were comparable to those reported by Skoulidis and Heymach [30] for metastatic rather than early-stage disease with the exceptions of ALK alterations that were close to those reported for early-stage disease, and EGFR that falls between those reported for early and metastatic disease. Understanding the mutational landscape in the early-stage setting is becoming increasingly important given that several ongoing trials are investigating targeted drugs in the early-stage setting, and osimertinib has already been approved in the adjuvant setting by the FDA.
In the never-smoker sub-cohort, 77% of patients had a targetable alteration, dominated by alterations in EGFR . There were no KRAS Gly12Cys mutations among never-smokers, whereas all ALK, ROS1 , and RET fusions were detected exclusively among never-smokers. If the potentially targetable ERBB2 exon 20 insertions [4 , 5] are included, the fraction of never-smokers with a targetable alteration in the cohort was 90%. This is in line with a recent study reporting 80% of the never-smokers harboring an actionable alteration [17] . These findings suggest that it is highly likely to find tailored targeted therapy for most, if not all never-smokers with newly diagnosed lung adenocarcinoma.
In survival analyses, we previously identified high TMB (defined in our study as equal to or more than 14 mutations/MB of coding DNA) as a stage-and histology-independent favorable prognostic factor with follow-up data until the end of 2016 [29] . The significance of these parameters in this model persisted with updated survival data extending until the end of 2018. Additionally, any alterations in SMARCA4 , a gene participating in chromatin remodeling, conferred an independent poor prognostic effect in multivariable analysis. Although the number of patients with SMARCA4 alterations in our cohort is small, our results support the earlier observations that SMARCA4 alterations are indicators of poor survival in NSCLC [31][32][33][34] .
The SMARCA4 gene codes SMARCA4/BRG1, an ATP-dependent catalytic subunit of the SWI/SFN chromatin remodeling complex. Most SMARCA4 alterations in NSCLC are missense variants [33 , 35] . Truncating mutations, reportedly present in more than one-third of SMARCA4 mutated tumors, are linked to loss of BRG1 expression indicating BRG1 deficiency, whereas missense mutations mostly leave BRG1 expression intact [36] . Both alteration types are associated with poor clinical outcomes [33] , and missense variants seem to affect chromatin remodeling activity [35] in ways other than loss of BRG1 expression.
The majority of the reports evaluating the histopathology of SMARCA4 altered lung adenocarcinomas have concentrated in tumors with loss of immunohistochemical BRG1 expression [37][38][39][40] and, presumably truncating SMARCA4 variants. These tumors are poorly differentiated, have a TTF-1 negative status, and behave aggressively. In contrast, the majority of our SMARCA4 altered tumors exhibited a differentiated predominant subtype, diffuse BRG1 expression, and diffuse positivity for TTF-1. In conclusion, all SMARCA4 variant types were associated with poor prognosis while tumors with SMARCA4 missense variants were mostly associated with differentiated predominant histological subtypes, retained BRG1 expression, and a positive TTF-1 status.
The SMARCA4 alteration prevalence in NSCLC is approximately 10% [33 , 36] . Our patients had a comparable prevalence of 7.4%. In addition to SMARCA4 alterations providing prognostic information, immunotherapy has yielded some promising results in SMARCA4 deficient tumors with loss of BRG1 [33 , 39 , 41 , 42] . In contrast, tumors with SMARCA4 missense variants did not exhibit a survival benefit on immunotherapy [33] . Furthermore, SMARCA4 deficient tumors have demonstrated sensitivity to cisplatinbased regimes [31] and, in preclinical studies, to an ATR inhibitor [43] , a KDM6 inhibitor [44] , and a CDK4/6 inhibitor palbociclib [45] . After more validation, SMARCA4 may be informative in clinical practice as a prognostic and predictive biomarker. More data are required on whether SMARCA4 missense variants are clinically actionable.
The Finnish population is genetically unique among Europeans due to historical population bottlenecks, and thus population databases such as The Exome Aggregation Consortium (ExAC) and the Genome Aggregation Database (GNOMad) divide Europeans into Finns or non-Finns. To our knowledge, no comprehensive genomic profiling data is available on Finnish lung adenocarcinoma patients. While most of the genes relevant to lung cancer exhibited similar mutational frequencies compared to non-Finn European populations, the frequency of certain alterations was higher in our cohort than in referenced cohorts. These included EGFR (30% vs 7.5-19.0%), ERBB2 (11.9% vs 1.0-5.0%), PIK3CA (10.4% vs 3.8-6.4%), and BRAF (8.4% vs 2.9-8.3%) [32 , 46-48] . When comparing the never-smoking sub-cohort with never-smoking non-Finn Europeans, alterations in TP53 (30% vs 5.8-26.7% [12 , 15 , 16] ) and ERBB2 (16.7% vs. 3.0-6.7% [14][15][16][17] ) were more frequent in our study. An alteration of uncertain significance, RUNX1T1 amplification, was present in eleven patients, all with a smoking history. One group has previously reported RUNX1T1 amplification in combined small cell and non-small cell lung cancer [49] . Interestingly, none of our tumors had a neuroendocrine component previously reported for RUNX1T1 amplified tumors [49] . Further research may elucidate the clinical meaning of the amplification.
Remarkably, only one of our patients had an ALK fusion (0.7%), a considerably low number compared to the reported prevalence of 2-7% in other European populations [50] . Furthermore, the pathogenic non-Val600 BRAF mutations seem to be more common in Finnish patients than in comparable non-Finn European populations, a finding supported by a previous report [9] . However, as the number of patients in our cohort is small, these findings require further validation with a larger cohort.

Conclusions
This study broadens the information on genomic alterations in European and Finnish lung adenocarcinoma patients. The most clinically relevant finding was the large number of actionable alterations found among neversmoking patients, supporting using broad genomic profiling, especially if standard tests are negative. SMARCA4 alterations were stage-, histologyand TMB-independent markers of poor prognosis and were associated with differentiated phenotypes and TTF-1 positivity. BRAF non-Val600 mutations and RUNX1T1 amplifications seem to be unusually prevalent and ALK fusions rare in Finnish lung adenocarcinoma patients.
The strengths of the study include the use of the same analysis method on all the tumors and comprehensive follow-up data with up-to-date staging and histological classification. The main limitations are a retrospective cohort and a small number of patients.