Clinicopathologic features of KRAS G12C-mutated non-small cell lung carcinomas:insights from 279 retrospective cases Authors: Martina Bradová, Petr Slavík, Tomáš Vaněček, Petr Martínek, Petr Grossmann, Stanislav Kormunda, Kristýna Behenská, Martin Svatoň, Miloš Pešek, Tomáš Jirásek, Zuzana Špůrková, Petra Hroudová, Hana Mrázková, Barbora Hořavová, Jaromír Roubec, Martin Baník, Petr Mukenšnábl, Michal Michal, Marián Švajdler Published: 2026 ABSTRACT AbstractKRAS G12C-mutated non-small cell lung carcinoma (NSCLC), caused by a glycine-to-cysteine substitution at codon 12, is associated with poor prognosis and is now targetable with specific inhibitors. We retrospectively analyzed 279 KRAS G12C-mutated NSCLC cases (2017–2023) from our registry with available histologic, immunohistochemical, and molecular data. The cohort included 279 patients (125 females, 151 males; mean age 67 years, range 29–91). Most tumors were primary lung carcinomas (n = 229, 82%), while 45 (16%) were metastatic at presentation. Morphologic evaluation was available in 240 tumors: 37% showed solid squamous cell carcinoma (SCC)-like features, 61% rhabdoid/plasmacytoid morphology, and 17% sarcomatoid features. Adenocarcinoma-associated patterns were present in 67 cases, often mixed, and focal solid growth occurred in 77%. TTF1, Napsin A, and CK7 were positive in 86%, 87%, and 98%, respectively, whereas squamous markers were infrequent (p40/p63 7%, CK5/6 8%). PD-L1 expression was detected in 65%. Co-mutations most commonly involved TP53 (n = 27) and STK11 (n = 12); IDH1/2, PIK3CA, and CTNNB1 mutations occurred in four cases each, MET in two cases, and BRAF, FGFR2, FGFR3, and GNAS in one case each. Two gene fusions were identified (LRP12::NRG1, FGFR3::TACC3). Mean survival was 1.89 years, with one- and five-year survival rates of 54% and 25%. KRAS G12C-mutated NSCLC is clinically aggressive and frequently shows solid growth with rhabdoid, plasmacytoid, or SCC-like morphology, which may lead to misclassification and missed genetic testing. Immunohistochemistry and molecular profiling are essential for accurate classification and enabling targeted therapy.Supplementary InformationThe online version contains supplementary material available at 10.1007/s00428-026-04526-x.Keywords: Lung, KRAS G12C, NSCLC, Adenocarcinoma, Molecular genetics, Statistics Introduction Non-small cell lung carcinoma (NSCLC) accounts for approximately 80–85% of lung cancers and is characterized by a high frequency of oncogenic driver mutations that increasingly guide therapeutic decision-making [1–4]. Among these, KRAS mutations represent the most prevalent molecular alterations in NSCLC, while treatment has evolved dramatically with the advent of molecular profiling, enabling the implementation of targeted therapies, including tyrosine kinase inhibitors, as well as immune checkpoint inhibitors (ICIs) [4]. The RAS gene family encodes membrane-bound GTP-binding proteins (GTPases) that regulate cell signaling [5, 6]. It comprises three isoforms (HRAS, NRAS, and KRAS), which are among the most frequently mutated genes in human cancers, particularly lung, colon, and pancreatic [7, 8]. KRAS is the most commonly altered member, with missense mutations typically at codons 12, 13, or 61 [8]. G12 substitutions are the most prevalent in NSCLC and pancreatic ductal adenocarcinoma, but specific variants differ by tumor type: G12C (glycine → cysteine) is most frequent in NSCLC yet rare in other cancers [8–10]. RAS-targeting was long deemed undruggable due to KRAS's exclusive GTP binding and high affinity even at low GTP levels [5, 6]. On May 28, 2021, Food and Drug Administration (FDA) approved sotorasib (AMG-210) [11] followed by adagrasib (MRTX849) on December 12, 2022 [12] as KRAS G12C inhibitors for adults with locally advanced or metastatic NSCLC. In the present study, we retrospectively analyzed clinicopathologic features of 279 KRAS G12C-mutated NSCLCs. Materials and methods Histology and immunohistochemistry A retrospective search of the authors' molecular-genetic registry was conducted. Among 2,348 NSCLC cases analyzed by next-generation sequencing between 2017 and 2023, 279 primary or metastatic tumors harboring KRAS G12C mutation were identified. Of the 2,348 cases, 1,823 were tested using RNA, 492 using both DNA and RNA, and 33 using DNA only. Among the 279 KRAS G12C-mutated cases, 214 were tested using RNA, 60 using both DNA and RNA, and 5 using DNA only. Histopathologic and immunohistochemical features were reviewed. The study complied with the Ethics Committee guidelines of the Faculty Hospital in Pilsen; informed consent was not required. Follow-up data were collected from referring pathologists, clinicians, hospital records, and insurance reports. For conventional microscopy, tissues were formalin-fixed, paraffin-embeded (FFPE), sectioned, and stained with hematoxylin and eosin. For immunohistochemistry, 4-μm-thick sections were cut from paraffin blocks and mounted on positively charged slides (TOMO, Matsunami Glass IND, Osaka, Japan). Sections were processed on a BenchMark ULTRA (Ventana Medical Systems, Tucson, AZ), deparaffinized and subjected to heat-induced epitope retrieval by immersion in a CC1 solution (pH 8.6) at 95 °C and CC2 solution (pH 6.0) at 92 °C. All primary antibodies used in this study are summarized in (Table 1). Visualization was performed using the ultraView Universal DAB Detection Kit (Roche, Tucson, AZ) or ultraView Universal Alkaline Phosphatase Red Detection Kit (Roche, Tucson, AZ). The slides were counterstained with Mayer’s hematoxylin. Appropriate positive and negative controls were employed. Table 1Antibodies Used for Immunohistochemical StudyAntibody SpecificityCloneDilutionAntigen Retrieval/timeSourceSMARCA4EPNCIR111A1:1000CC1/52 minAbcamSMARCB1MRQ-27RTUCC1/52 minVentanaAE1/3AE1/AE3 + PCK26RTUCC1/20 minVentanaOSCARIsoType:IgG2a1:100EnVision High pH/30 minCovanceCAM5.2CAM5.2RTUCC1/36 minVentanaClaudin 43E2C11:200CC1/52 minThermo Scientific – Life TechnologiesMOC-31MOC-311:50CC1/52 minAbcamCK5/6D5/16B41:50EnVision High pH/30 minDakoCK7OV-TL 12/30RTUEnVision High pH/30 minDakop63DAK-p63RTUEnVision Low pH/30 minDakop40BC28RTUCC1/52 minBiocare MedicalTTF1PD-L122C31:50EnVision Low pH/30 minDakoALKOTI1A4RTUEnVision Low pH/30 minDakoROS1D4D6®1:50CC1/64 minCell SignalingRTU, ready to use; CC1, EDTA buffer pH 8.6 at 95 °C; CC2, citrate buffer pH 6.0 at 92 °C; EnVision High pH 9.0 at 97 °C; EnVision Low pH 6.0 at 97 °C; min, minutes Antibodies Used for Immunohistochemical Study RTU, ready to use; CC1, EDTA buffer pH 8.6 at 95 °C; CC2, citrate buffer pH 6.0 at 92 °C; EnVision High pH 9.0 at 97 °C; EnVision Low pH 6.0 at 97 °C; min, minutes Predictive biomarker testing was performed using established protocols. PD-L1 expression was assessed by Tumor Proportion Score (TPS) calculated as the percentage of PD-L1-positive tumor cells among all viable tumor cells. Cut offs were defined negative (< 1%), low (1–49%) and high (≥ 50%). ROS1 expression was evaluated using rabbit monoclonal antibody (D4D6 clone, Cell Signaling Technology, Danvers, MA, USA) and scored semi-quantitatively based on staining intensity and percentage of positive tumor: 0 (negative) – no staining or faint staining in < 10% of tumor cells; 1 + (weak) – faint, barely perceptible staining in ≥ 10% of tumor cells; 2 + (moderate) – distinct cytoplasmic staining in ≥ 10% of tumor cells; 3 + (strong) – intense granular cytoplasmic staining in ≥ 10% of tumor cells. [13, 14]. Tumors which scores 1 +, 2 +, or 3 + were confirmed by FISH or NGS, while negative cases were not tested further. ALK expression was assessed using the OTI1A4 clone (OriGene Technologies, Rockville, MD, USA) [15] interpreted using a binary scoring system: positive when strong granular cytoplasmic staining was present in ≥ 10% of tumor cells, and negative when staining was absent or limited to background levels. Equivocal or discordant cases were confirmed by FISH or NGS. Molecular genetic testing Next generation sequencing (NGS) Depending on the sample size, up to 3 FFPE Sects. (10 µm thick) were sectioned and DNA and RNA was extracted using DNA/RNA FFPE or DNA/RNA FFPE RSC kits (automated on Maxwell RSC 48 Instrument, Promega, Madison, Wisconsin, USA) according to the manufacturer's instructions. DNA and RNA was quantified using the Qubit Broad Range DNA/RNA Assay Kits (ThermoFisher Scientific) and input for each sample’s library preparation was set to 250 ng RNA or 100 ng DNA for fusion or SNV detection respectively. The PreSeq RNA QC Assay using iTaq Universal SYBR Green Supermix (Biorad, Hercules, CA) was performed to assess RNA quality. Samples with the cycle threshold value smaller than 30 continued with target enrichment. The FusionPlex CTL kit containing 195 targets in 36 genes or FusionPlex Lung v2 with 323 gene specific primers targeting 17 genes mutated in non-small cell lung cancer were used for fusion and SNV detection from RNA. VariantPlex CTL kit was used for DNA SNV analysis in 290 regions of interest across 31 genes. Library preparations were performed following Fusion Plex or Variant Plex Protocols for Illumina (ArcherDX Inc., Boulder, CO) and were quantified following the Library Quantification for Illumina Libraries protocol (KAPA, Wilmington, MA) assuming a 300 or 350 bp fragment lengths respectively. Samples were multiplexed and sequenced on a NextSeq 550 or Novaseq 6000 sequencers (Illumina, San Diego, CA) spiked with up to 20% PhiX to ensure library nucleotide diversity. The multiplexing was calculated to obtain a minimum of 3 milion read pairs for FusionPlex CTL and VariantPlex CTL or 2 milion read pairs for FusionPlex Lung V2. The analysis of the sequencing results was performed using the Archer Analysis software (version 5–7; ArcherDX Inc.). Parameters for filtering strong fusions as determined by the Archer Analysis software were set to a minimum of 5 fusion supporting reads with a minimum of 3 unique start sites. For SNV and indel detection the parameters were set to 5% allelic frequency, minimum depth 40x, alternate observations > 5, unique start sites of alternate observation > 3, less than 0,5% population frequency using GnomAD database and non-synonymous variant consequence. All filtered variant ‘s raw reads were visualy inspected by two molecular biologists to exclude potential artefacts and confirmed variants were reported. Detection of ALK and ROS1 gene rearrangements by FISH Prior to performing FISH, hematoxylin and eosin-stained slides were examined to determine the areas for cell counting. Then, a 4-µm-thick formalin-fixed, paraffin-embedded section was placed onto a positively charged slide. The unstained slide was routinely deparaffinized and incubated in the 1 × Target Retrieval Solution Citrate pH 6 (Dako, Glostrup, Denmark) for 40 min at 95 °C, subsequently cooled for 20 min at room temperature in the same solution and washed in deionized water for 5 min. The slide was digested in protease solution with pepsin (0.5 mg/mL) (Sigma Aldrich, St Louis, MO, USA) in 0.01 M HCl at 37 °C from 45 to 60 min according to the sample conditions. The slide was then rinsed in deionized water for 5 min, dehydrated in a series of ethanol solutions (70%, 85%, 96% for 2 min each), and air-dried. For the detection of the break of ALK resp. ROS1 genes were used commercial probes Human ALK Gene Fusion Detection Probe Detection Kit resp. ROS1 (6q22) Gene Fusion Detection Probe Detection Kit (Wuhan HealthCare Biotechnology Co., Ltd., Wuhan City, China). An appropriate amount of both factory premixed probes was applied on the specimens, covered with a glass cover slip, and sealed with rubber cement. The slide was incubated in the ThermoBrite instrument (StatSpin/Iris Sample Processing, Westwood, MA, USA) with codenaturation parameters of 85 °C for 8 min and hybridization parameters of 37 °C for 16 h. The rubber-cemented cover slip was then removed, and the slide was placed in posthybridization wash solution (2 × SSC/0.3% NP-40) at 72 °C for 2 min. The slides were air-dried in the dark, counterstained with DAPI II (Vysis/Abbott Laboratories, Abbott Park, IL, USA), covered with slip, and immediately examined under an Olympus BX51 fluorescence microscope using a × 100 objective and filter sets Triple Band Pass (DAPI/Spectrum Green/Spectrum Orange), Dual Band Pass (FITC/Spectrum Orange), and Single Band Pass (Spectrum Green or Spectrum Orange). A minimum of 50 randomly selected nonoverlapping interphase tumor cell nuclei were evaluated for the presence of a specific positive break pattern where the samples were considered positive when ≥ 15% of nuclei show break signals as previously described by McCoach et al. [16]. Statistical analysis All analyses were performed using SAS software (SAS Institute Inc., Cary, NC, USA). Descriptive statistics were calculated for continuous variables (mean, standard deviation, variance, median, interquartile range, minimum, and maximum) and categorical variables (absolute and relative frequencies). Selected parameters were visualized using box plots, histograms, and pie charts. Groups comparisons were conducted using nonparametric tests (Wilcoxon two-sample test), and correlations were assessed with Spearman’s coefficient. Overall survival analysis was performed using Kaplan–Meier survival curves. The effects of individual factors were tested using the Log-rank test, Gehan–Wilcoxon test, and Cox proportional hazards regression model. Optimal cut-offs and risk groups were identified using Cox regression by maximizing the test statistic. The clinical impact of individual factors was expressed as a Hazard Ratio. Several multivariate models were constructed. Independent predictors were identified using stepwise Cox regression analysis. The multivariate analysis was further developed using a multivariate Cox regression model presented in the form of Classification and Regression Trees. The statistical significance of the regression tree branches identified by the Cox regression model was verified using the Log-rank test. Statistical significance was set at α = 0.05. Results The cohort comprised 151 men and 125 women (sex unknown in 3 cases), aged from 29 to 91 years (mean age 66.9; median 68). Specimens included 37 surgical resections (13%), 58 excisional biopsies (21%), 150 needle biopsies (54%), and 28 cytology samples (10%). In six cases (2%), the type of procedure was unknown. Cytology specimens were insufficient for assessment some histological parameters and most immunohistochemical tests. Histologic classification followed WHO criteria (World Health Organization Classification: Thoracic tumours) [17], and cytology was categorized using WHO Reporting System for Lung Cytopathology [18]. Clinical, histologic, immunohistochemical, molecular findings, staging, and follow-up are summarized in Table 2. Table 2Clinical features, histological parameters, immunohistochemical, molecular genetic findings and staging of the diseaseClinical featuresNo. of cases (279)Sex Male/Female151/125 (in 3 cases sex not known)Mean age (range)66.9 years (29–91)DiagnosisNo. of cases (279)(%)Invasive non-mucinous adenocarcinoma and/or NSCLC most likely adenocarcinoma23384NSCLC NOS218Invasive mucinous adenocarcinoma124Mixed invasive mucinous and non-mucinous adenocarcinoma83Non-keratinizing squamous cell carcinoma10.4Adenosquamous carcinoma10.4Not known31Smoking historyNo. of cases (230)(%)• Smoker15166• Ex-smoker7432• Non-smoker52Localization of the specimenNo. of cases (279)(%)Lung/bronchus22982Locoregional spread/metastasis• Pleura/pleural cavity/chest wall186• Hilar/mediastinal lymph nodes93Distant metastasis• Neck and axillary lymph nodes52• Skin52• Brain/cerebellum52• Abdominal cavity10.4• Bones20.7• Unknown location52Type of growthNo. of cases (240)(%)• Solid18477• Cribriform2510• Lepidic3414• Acinar198• Papillary, micropapillary3715• Other glandular growth patterns10745• Not assessed3916Other histological featuresNo. of cases (158)(%)• Desmoplasia12076• Necrosis4830• Inflammation3321Cellular featuresNo. of cases (240)(%)• SCC-like8937• Rhabdoid/plasmacytoid/epithelioid14661• Sarcomatoid4017• Small cells31• Adenocarcinoma (cribriform or serrated)6728• Cannot be assessed63ImmunohistochemistryNo. of positive cases/No. of tested cases(%)• TTF1210/24486• Napsin A169/19587• CK787/8998• p40/p6311/1627• CK5/67/838• PD-L1• 1–49%51/23322• ≥ 50%100/23343• negative82/23335• ALK1• positive22/20411• negative182/20489• ROS1• positive15/1798• negative164/17992Molecular genetics of KRAS G12C mutated NSCLCNo. of cases (280)(%)Gene mutations• TP53279• STK11124• CTNNB141• IDH1/241• PIK3CA41• MET21• Other (BRAF, FGFR2, FGFR3, GNAS)41,4Gene fusions• LRP12::NRG1; FGFR3::TACC320.7Follow-upNo. of cases (257)(%)• DOD13452• AWD4216• ANED197• DOUR114• LOF5120• NA229StageNo. of cases (231)(%)• IA2210• IB115• IIA31• IIB146• IIIA2511• IIIB2210• IIIC42• IVA7934• IVB5122ANED alive not evidence of disease, AWD alive with disease, DOD died on the disease, DOUR died on unrelated reasons, LOF lost to follow-up, NA not available, NSCLC non-small cell lung carcinoma, SCC squamous cell carcinoma Clinical features, histological parameters, immunohistochemical, molecular genetic findings and staging of the disease ANED alive not evidence of disease, AWD alive with disease, DOD died on the disease, DOUR died on unrelated reasons, LOF lost to follow-up, NA not available, NSCLC non-small cell lung carcinoma, SCC squamous cell carcinoma Final diagnoses included invasive non-mucinous adenocarcinoma or NSCLC-most likely adenocarcinoma in 233 cases (84%), NSCLC not otherwise specified (NOS) in 21 (7.5%), invasive mucinous adenocarcinoma in 12 cases (4.3%), mixed mucinous/non-mucinous adenocarcinomas in 8 (3%), non-keratinizing SCC in 1 (0.4%), and adenosquamous carcinoma in 1 (0.4%). In three cases (1%), only molecular data confirming KRAS G12 mutated (adeno)carcinoma were available; the exact tumor location, biopsy procedure, histology, and immunoprofile were unknown. Histomorphological features Architecture was assessable in 240 cases (86%) (Figs. 1, 2 and 3). Solid growth predominated, present in 184 cases (77%) and accounting for ≥ 50% of tumor volume when observed (Fig. 1A–D). Acantholysis was occasionally present in SCC-like areas (Fig. 2B). Necrosis was common, appearing as comedo-like or extensive tumor necrosis (Fig. 1D). Adenocarcinoma-like patterns were less common, typically involving ≤ 50% of tumor volume. Lepidic (n = 34, 14%), acinar (n = 19, 8%), papillary and micropapillary (n = 37, 15%), cribriform (n = 25, 10%) and other glandular (n = 107, 77%) growth patterns occurred in various combinations, often with solid growth (Fig. 2A–F). Over 70% of cases were of solid and nested morphology with single cell pattern or single cell lines (A). Some cases developed acantholytic areas throughout tumor content (B). Nested morphology and high nuclear and cellular pleomorphism was predominant in the cohort (C), some showing comedo-like necrosis or large necrotic areas (D) Adenocarcinoma morphology was less obvious and if present than it comprised at most 50% of the tumor volume there were different growth patterns including lepidic (A), acinar (B), pseudoglandular with papillary architecture (C), micropapillary pattern (D), cribriform architecture (E), and mucinous adenocarcinoma morphology (F) Tumor cell morphology was epithelioid (rhabdoid) (A) with squamous-like morphology (B). Basaloid and small cell morphology was present in 3 cases (C). Spindle cell to sarcomatoid appearance was recorded in 17% (D). Cases with adenocarcinoma morphology exhibited features reminiscent in some cases of serrated adenocarcinomas of the intestine (E) and, prostatic adenocarcinomas with cribriform architecture (F) Tumor cell morphology was epithelioid to poorly differentiated/rhabdoid in 61% (n = 146) (Fig. 1A-D, 2A, 2B), with SCC-like features in 37% (n = 89) (including two cases with true squamous differentiation diagnosed as non-keratinizing SCC adenosquamous carcinoma) (Fig. 3A), small cell/basaloid in 1% (n = 3) (Fig. 3C), and sarcomatoid/spindle-cell in 17% (n = 40) (Fig. 3D). Hobnail, clear-cell, serrated-like, and intestinal/penicillate features were present in 184 cases (86%) (Fig. 2A–F, 3E, 3F); in half of these cases the pattern was very focal (< 10%), while in the other half it comprised at most 50% of the tumor volume. Most tumors contained multiple cell phenotypes within a single lesion. Immunohistochemical findings Immunohistochemical results adapted to tumor histological or cytological diagnosis are summarized in Table 3. Cases diagnosed as invasive non-mucinous adenocarcinoma or NSCLC most likely adenocarcinoma, were positive for TTF1 in 192 cases (82%) and positive for Napsin A in 158 (68%). CK7 was expressed in 71 cases (30%), and was expressed in all tested TTF1/Napsin A-negative tumors (n = 11) except one triple-negative case. CK5/6 and p40/p63 were rarely positive (2% and 3%), usually alongside TTF1. NSCLC NOS group consisting of 21 cases showed expression of TTF1 in five cases, Napsin A in two cases and CK7 in seven cases, while CK5/6 and p40/p63 were expressed in two and one case respectively. Invasive mucinous adenocarcinoma (12 tested cases) expressed TTF1 in nine cases, Napsin A in six cases, and CK7 in seven. CK5/6 and p40/p63 were negative in all tested cases (2 and 4 cases, respectively). From eight tested mixed invasive mucinous/non-mucinous adenocarcinomas cases, TTF1 was positive in five, Napsin A in three, and CK7 in one; p40/p63 was negative in all three tested cases. Non-keratinizing squamous cell carcinoma (NKSCC) and adenosquamous carcinoma (ASC) (1 case each) were negative for TTF1 and Napsin A. NKSCC was positive for p63/p40 and CK5/6; ASC expressed CK7 and p40/p63 but was CK5/6-negative. Table 3Immunohistochemical findings related to the histological or cytological tumor diagnosisDiagnosisInvasive non-mucinous adenocarcinoma or NSCLC most likely adenocarcinoma (N = 234)NSCLC not otherwise specified (N = 21)Invasive mucinous adenocarcinoma (N = 12)Mixed invasive mucinous and non-mucinous adenocarcinomas (N = 8);Non-keratinizing squamous cell carcinoma (N = 1);Adenosquamous carcinoma (N = 1)AntibodyExpressionTTF1positive19259500negative1992111not done2371200Napsin Apositive15826301negative1662110not done60134400CK7positive7177101negative110000not done162135710CK5/6positive510010negative7322000not done1561810801p40/p63positive720011negative135104300not done9298500PD-L11–49%4333300 ≥ 50%8971201negative7046100not done3272210ALKpositive2011000negative156137601not done5874210ROS1positive1301100negative141136401not done8085310NSCLC non-small cell lung carcinoma Immunohistochemical findings related to the histological or cytological tumor diagnosis NSCLC non-small cell lung carcinoma Predictive biomarkers expression PD-L1 was evaluated in 234 cases; 46 cases were not tested due to insufficient tissue or technical issues. Among the tested cases, PD-L1 was positive in 1–49% of tumor cells in 52 cases (22.2%), in ≥ 50% of tumor cells in 100 cases (42.7%), and negative in 82 cases (35.0%). The mean PD-L1 expression was 39%. ALK was evaluated in 205 cases; 22 cases (10.7%) were positive and 183 (89.3%) negative. ROS1 was tested in 180 cases; 15 (8.3%) were positive and 165 (91.7%) negative. However, molecular genetic testing did not confirm any underlying genetic alterations. Molecular genetic findings In addition to the KRAS G12C mutation detected in all cases, additional mutations were found in 58 cases (20.8%), most commonly TP53 (n = 27) and STK11 (n = 12), followed by IDH1/2 (n = 4), PIK3CA (n = 4), CTNNB1 (n = 4), MET (n = 3), and single cases of BRAF, FGFR2, FGFR3, and GNAS. Two gene fusions were identified: LRP12::NRG1, and FGFR3::TACC3. Follow-up and survival outcomes At diagnosis, stage distribution was: stage IA 22 patients (10%), stage IB 11 (5%), stage IIA 3 (1%), stage IIB 14 (6%), stage IIIA 25 (11%), stage IIIB 22 (10%), stage IIIC 4 (2%), stage IVA 79 (34%), and stage IVB 51 (22%). Follow-up was available in 257 cases (92%). Nineteen patients were alive without evidence of disease (mean 31.7 months; range 1–54; median 32), 42 alive with disease (mean 35.6 months; range 2–102; median 28), 134 died of disease (mean 9.8 months; range 0–72; median 4), 11 died of unrelated causes (mean 14.2 months; range 1–36; median, 12), and 51 were lost to follow-up (mean 7.7 months; range, 1–36; median, 2). Treatment data were available in 247 patients, who received either single or combined modalities, including chemotherapy in 111 cases (45%), radiotherapy in 82 (34%), immunotherapy in 56 (23%), and surgery in 49 (20%). Overall survival (OS) by variables and hazard ratios is shown in Table 4. Average survival was 1.89 years (SD 0.12); 1-year OS was 54%, dropping to 25% at 5 years (Fig. 4). Table 4Statistical parameters in a relationship to 1-, 3- and 5-year overall survival (univariate analysis)VariableSubgroupsNo of patients1-year OS (%)3-year OS5-year OS (%)Mean(yearsS)SECIOS2575435251.8950.1160.92–1.83Age(p 0.0068) > 73 years654522171.2670.1640.50–1.25 < 73 years1925640292.0680.1380.92–3.00Sex(p = 0.0697Male1424929201.6230.1370.71–1.75Female1155942322.1450.1820.75–4.08Stage(p < 0.0001)IA228968NA1.8340.1011.83-NAIB1110090723.1000.2011.75-NAIIA3100100100NANANAIIB138365641.1140.1020.92-NAIIIA255535241.9410.3680.42–3.92IIIB2270001.3930.2260.58–2.00IIIC425NANA0.6480.2230.17-NAIVA795341202.0330.2110.58–3.33IVB512210NA0.6390.1030.17–0.58TherapySurgeryp < 0.0001No1974725211.5050.1160.50–1.25Yes497870463.06160.2293.25-NARTp = 0.0135No1644832221.7130.1430.50–1.33Yes826643312.2850.20021.25–4.08CHTp = 0.0322No1354735291.6440.1540.33–1.75Yes1116236242.1170.1631.17–2.50ITp = 0.0077No1904833231.7460.1370.50–1.33Yes567245392.1390.1711.50-NASmoking historySmokers1505467241.9060.1490.83–2.00Ex-smokers745332321.5680.1530.58–2.42Non-smokers53030NA0.5840.2250.08-NACHT chemotherapy, CI confidence interval, NA not available, IT immune therapy, OS overall survival, RT radiotherapy, SE standard error Statistical parameters in a relationship to 1-, 3- and 5-year overall survival (univariate analysis) Age (p 0.0068) Sex (p = 0.0697 Stage (p < 0.0001) Surgery p < 0.0001 RT p = 0.0135 CHT p = 0.0322 IT p = 0.0077 CHT chemotherapy, CI confidence interval, NA not available, IT immune therapy, OS overall survival, RT radiotherapy, SE standard error Overall survival of KRAS G12C adenocarcinomas by Kaplan–Meier univariate analysis showing poor 5-year OS with only 25% survival rate Age > 73 years predicted worse outcome (HR 1.606, p = 0.0068) (Fig. 5A). Gender was not statistically significant (p = 0.0697), though males had a lower 1-year (49%) and 5-year (20%) OS versus female patients (59% and 32%); males had a 1.353 times greater risk of dying from disease. Higher AJCC stage strongly correlated with poor OS (HR 3.03, p < 0.0001), with critical cut-off between stages IIB and IIIA (HR 5.09, p < 0.0001) (Fig. 5B). Type of therapy significantly influenced OS; absence of surgery (HR 2.987, p < 0.0001), radiotherapy (HR 1.57, p = 0.0135), chemotherapy (HR 1.43, p = 0.0322), and immunotherapy (HR 1.73, p = 0.0077) was each associated with poorer OS. Overall survival of KRAS G12C adenocarcinomas was associated with patient age, with 73 years identified as the cut-off for the worst outcome (p = 0.0068) (A), and with tumor stage, where higher stage correlated with poorer outcomes (p < 0.0001) (B) Analysis of different therapy combinations with at least 10 patients is presented in Fig. 6. As expected, patients who received no treatment (typically diagnosed at an advanced stage) had the poorest 1- and 5-year OS rates (9% and 0%, respectively, HR 5.998, p < 0.0001). Univariate analysis showed that the most statistically significant combination was surgery without radiotherapy, chemotherapy, or immunotherapy with p < 0.0001 and 1-year and 3-year survival rates 91% and 85%, respectively (the 5-year OS could not be calculated due to limited follow-up data) (Supplementary file 1). The HR for poor clinical outcome in this group was 0.184. Smoking had no statistically significant impact on OS; non-smokers (vs smokers and ex-smokers) showed worse OS with HR 0.730 (p = 0.5814, CI 0.730–0.232). Overall survival of KRAS G12C adenocarcinomas by Kaplan–Meier univariate analysis related to different combinations of therapies Multivariate analysis identified two factors significantly associated with OS: the number of therapies and disease stage. Patients who received one or more therapies and had a lower disease stage demonstrated longer OS. According to the Cox proportional hazards model (stepwise regression), the factors with the strongest negative association with OS were as follows (Supplementary file 2): no therapy (HR 7.204, CI 4.557–11.389), chemotherapy only (HR 2.044, CI 1.201–3.477), radiotherapy only (HR 3.192, CI 1.822–5.594), AJCC stage > IIB (HR 5.320, CI 2.820–10.038), and age over 73 years (HR 1.622, CI 1.095–2.402) (Supplementary File 3 – Regression tree). Discussion Kirsten rat sarcoma viral oncogene homolog (KRAS), a principal isoform of the RAS gene, is the most common oncogenic driver in NSCLC [19]. About 80% of all KRAS oncogenic mutations occur at codon 12, with the most comomn variants being p.G12D (41%), p.G12V (28%), and p.G12C (14%) [7]. Among these, the KRAS p.G12C mutation is the predominant variant in NSCLC, accounting for 45–50% of all KRAS mutations and occurring in 11–16% of lung adenocarcinomas [8]. KRAS protein regulates proliferation by alternating between two stages: an inactive guanosine diphosphate (GDP)-bound form and an active guanosine triphosphate (GTP)-bound form. The p.G12C mutation impairs normal cycling activity between these two stages, by interfering with the binding of GTPase-activating proteins, hindering KRAS inactivation and resulting to accumulation of the pro-proliferative form [20]. The identification of tumor-driver gene mutations and/or fusions, along with the development of specific inhibitors, is essential for advanced treatment strategies and clinical outcomes. While EGFR- and TRK-inhibitors showed positive results in tumor-specific therapy [21, 22], RAS was, until recently, considered undruggable, due to two major challenges. First, KRAS has a small pocket on an otherwise smooth surface, making it unable to bind anything other than GTP. Second, GTP is abundant in normal cells, and the KRAS-GTP interaction is exceptionally strong, even at low GTP concentrations, because of its high binding affinity [5, 6]. On May 28, 2021, the Food and Drug Administration (FDA) approved AMG-210 (sotorasib) [11] and on December 12, 2022 MRTX849 (adagrasib) [12] as RAS GTPase family inhibitors for adult patients with KRAS G12C-mutated, locally advanced or metastatic NSCLC. Sotorasib was approved based on CodeBreaK 100 trial (NCT03600883) that involved 124 patients with disease progression after at least one prior systemic therapy [11, 20], The objective response rate (ORR) was 36%, with a median response duration of 10 months. Approval of Adagrasib was based on KRYSTAL-1 clinical trial (NCT03785249), which evaluated efficacy in 112 patients whose disease progressed after platinum-based chemotherapy and immune checkpoint inhibitor therapy [23, 24]. The ORR was 43%, with a median response duration of 8.5 months [12, 23]. In the present study, we retrospectively analyzed and reported baseline clinicopathologic characteristics of KRAS G12C-mutated NSCLCs that were not treated with RAS inhibitors. These tumors frequently exhibited solid growth, with squamous cell carcinoma (SCC)-like or rhabdoid/sarcomatoid phenotypes, while classic adenocarcinoma morphology was observed in only 28%, usually as a minor component. Most tumors expressed adenocarcinoma markers (TTF1, Napsin A), but some lacked these and showed only CK7 positivity; occasional p40/p63 and CK5/6 expression further complicated classification. Importantly, this overlap of morphology and immunoprofile, combined with SCC-like patterns, may lead to diagnostic inaccuracies and missed molecular testing. PD-L1 expression was high across the cohort, with 22.2% of tumors showing positivity in 1–49% of tumor cells and 42.7% showing positivity in ≥ 50% of tumor cells; 35.0% were PD-L1-negative, this indicates that a substantional proportion of KRAS G12C-mutated NSCLCs may be candidates for immunotherapy, in addition to RAS-targeted therapy. Co-mutations occur in a notable subset of KRAS G12C-mutated tumors, including TP53, STK11, IDH1/2, PIK3CA, CTNNB1, MET, BRAF, FGFR2, FGFR3, and GNAS. The presence of these additional genetic alterations expands the spectrum of potential therapeutic targets. For example, lorlatinib is the first-line therapy for ALK rearranged tumors [25, 26], tepotinib may be considered for tumors harboring MET amplifications or mutations [27]. A comprehensive molecular characterization also facilitates access to emerging therapeutic strategies that are currently investigational. These include gene therapy approaches, small molecules aimed at restoring tumor suppressor function (e.g., p53), and novel immunotherapeutic modalities. Several targeted agents under evaluation or approved for other indications may have relevance in this context: anlotinib, an antiangiogenic agent, for TP53-mutated NSCLC [28], ivosidenib (AG-120) for IDH1-mutated tumors, enasidenib (AG-221) for IDH2-mutated tumors [29], or alpelisib, a PI3Kα inhibitor, which has demonstrated efficacy in PIK3CA-mutated metastatic breast cancer and shows promise in other malignancies with PI3K pathway alterations [30, 31]. PD-L1 expression was high throughout the cohort with 52 cases setting in the expression of 1–49% and 100 cases with the expression over or equal to 50% of TPS score. Eighty cases were PD-L1 negative. When calculated across the entire tested cohort (n = 234, including negative cases), the mean PD-L1 expression was 39%. However, when considering only PD-L1-positive cases (n = 152, excluding negative cases), the mean expression was 60.6%. Such findings highlight the possibilities of the potential therapeutic target in a form of ICIs (pembrolizumab, nivolumab, etc.) in a single or combined therapeutic approach [32]. In lung adenocarcinomas, PD-L1 expression is significantly more frequent in poorly differentiated subtypes (solid and micropapillary) compared with well-differentiated ones [33, 34]. Clinically, KRAS G12C-mutated NSCLC represents aggressive malignancy with mean survival of 1.89 months (SD 0.12); with 1-year OS 54%, and a decline to 25% at 5 years. Poor outcome was associated with age > 73 years, AJCC stage > IIB, whereas combined therapy is associated with improved OS. Within our cohort, TP53 co-mutations (n = 27) and STK11 co-mutations (n = 12) were present but did not reach statistical significance as independent prognostic subgroups; notably, the TP53-co-mutated subset had 11 alive with disease at an average follow-up of 29 months and 14 disease-related deaths at 8.7 months, while the STK11-co-mutated subset included 10 disease-related deaths at 10.3 months and 2 survivors without evidence of disease. Although the overall KRAS G12C target remains clinically important, these data do not support the conclusion that KRAS G12C tumors are intrinsically more aggressive than other driver-defined tumors. Head-to-head comparisons in adequately powered cohorts, with standardized mutational panels and uniform treatment exposure, are required to define relative prognosis and to optimize genotype-guided therapy. We identified another FGFR3::TACC3 adenoid carcinoma diagnosed on a core-cut biopsy. The sample was largely necrotic, containing only a few viable tubular and cribriform areas, with no obvious clear cell morphology evident in the biopsy material. The entire tumor could not be assessed, but the patient was staged as cT4 N2 M1b with multiple lung and brain infiltrations; no additional FDG-avid sites were seen on PET/CT, suggesting a primary lung origin. The patient died of disease two months after diagnosis. The second case, featuring an LRP12::NRG1 gene fusion, showed solid growth with epithelial-to-sarcomatoid tumor cells and SCC-like features on core biopsy, staged cT3 N1 M1b, with death from disease at eight months. In both cases, napsin A and TTF-1 positivity supported a lung primary, and a full resection specimen was not available, autopsy was not indicated. Conclusion KRAS G12C–mutated NSCLCs frequently exhibit solid morphology with SCC–like features, which may lead to misclassification and ommitment of proper molecular testing and, consequently, deprive patients of targeted treatment options beyond traditional chemotherapy and radiotherapy. Notably, KRAS G12C–mutated NSCLCs demonstrate a higher prevalence of PD-L1–positivity compared with non–KRAS G12C tumors, suggesting that immune checkpoint inhibitors could represent a more broadely applicable therapeutic option in this subgroup. Comprehensive molecular testing (e.g. NGS) further broadens treatment possibilities for this clinically aggressive tumor type. Supplementary Information Below is the link to the electronic supplementary material. (PNG 110 KB) High resolution image (TIF 4.41 MB) (PNG 193 KB) High resolution image (TIF 8.47 MB) (PNG 66.9 KB) High resolution image (TIF 1.33 MB) Acknowledgements This study was in part supported by study grant SVV 260773 from the Ministry of Education, Czech Republic, the Cooperatio Program, research area SURG, and the project NationalInstitute for Cancer Research – NICR (Programme EXCELES, ID Project No. LX22NPO5102)- Funded by the European Union - Next Generation EU. Authors’ contribution MB & MŠ: conception and design of the work, acquisition, analysis and interpretation of data. The first draft was performed by MB. PS, KB, PM, and MM – acquisition and interpretation of data. TV, PM & PG – performance and interpretation of molecular-genetic analysis. SK – acquisition of statystical analysis, and its interpretaion. MS, MP, TJ, ZŠ, PH, HM, BH, JR, MBaník—providing the case, reading and revising the MS critically for important intellectual content and scientific integrity. All authors have read the manuscript, revising it critically for important intellectual content,scientific integrity, and approved the final manuscript. Funding Open access publishing supported by the institutions participating in the CzechELib Transformative Agreement. This study was in part supported by study grant SVV 260773 from the Ministry of Education, Czech Republic, the Cooperatio Program, research area SURG, and the project National Institute for Cancer Research – NICR (Programme EXCELES, ID Project No. LX22NPO5102)—Funded by the European Union—Next Generation EU. Data availability Data supporting the findings of this study are available within the article. The complete datasets generated during and/or analyzed during the current study are available from the corresponding author upon reasonable request. Declarations Ethics approval The procedures used in this study adhere to the tenets of the Declaration of Helsinki. Informed consent The Ethics Committee waived the requirement for informed consent as according to the Czech Law (Act. no. 373/11, and its amendment Act no. 202/17), it is not necessary to obtain informed consent in fully anonymized studies. Conflicts of interest The authors have disclosed that they have no significant relationships with, or financial interest in, any commercial companies pertaining to this article. Footnotes The preliminary results of the study were presented as a poster at the 114th Annual Meeting of the American and Canadian Academy of Pathology, held from March 22-27, 2025, in Boston, Massachusetts, USA. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials (PNG 110 KB) High resolution image (TIF 4.41 MB) (PNG 193 KB) High resolution image (TIF 8.47 MB) (PNG 66.9 KB) High resolution image (TIF 1.33 MB) Data Availability Statement Data supporting the findings of this study are available within the article. The complete datasets generated during and/or analyzed during the current study are available from the corresponding author upon reasonable request.