KRAS Mutations Are Associated with Shortened Survival in Patients with Epithelioid Malignant Pleural Mesothelioma

Simple Summary Malignant pleural mesothelioma (MPM) is an aggressive tumor with implacable prognosis. It commonly harbors loss-of-function mutations in BAP1, NF2, CDKN2A and TP53, although their pathogenetic role is still uncertain since these genetic alterations, if isolated, do not cause MPM in mice models. The aim of our study was to describe the mutational landscape in a cohort of 29 patients with MPM and, eventually, correlate mutation profile with MPM phenotypic traits and prognosis. We report herein a correlation between the presence of KRAS mutations and prognosis of the epithelioid MPM and, moreover, the number of mutations correlated with shortened survival. Conversely, improved overall survival was associated with the presence of SNP p.V297I of the KDR gene. With the limitation of a relatively small series, these results contribute to the characterization of the mutation profile of MPM and its impact on prognosis. Abstract Malignant pleural mesothelioma (MPM) is an aggressive malignancy of the pleural surface that includes three major histologic subtypes, epitheliod, sarcomatoid and biphasic. Epithelioid mesothelioma is usually associated with better prognosis. The genetic mechanisms driving MPM, the possible target mutations and the correlation with overall survival remain largely unsettled. We performed target exome sequencing in 29 cases of MPM aimed at identifying somatic mutations and, eventually, their correlation with phenotypic traits and prognostic significance. We found that KRAS mutations, occurring in 13.7% of cases, were associated with shortened median survival (7.6 versus 32.6 months in KRAS wild-type; p = 0.005), as it was the occurrence of any ≥3 mutations (7.6 versus 37.6 months; p = 0.049). Conversely, the presence of KDR single nucleotide polymorphism p.V297I (rs2305948) resulted in a favorable variable for survival (NR versus 23.4 months; p = 0.026). With the intrinsic limitations of a small number of cases and patient heterogeneity, results of this study contribute to the characterization of the mutation profile of MPM and the impact of selected somatic mutations, and possibly KDR polymorphism, on prognosis.


Introduction
Malignant pleural mesothelioma (MPM) is a relatively rare and aggressive neoplasm arising from mesothelial cells that line the serous cavities of the body. MPM is the most frequent among malignant mesotheliomas, accounting for 80-95% of cases, followed by peritoneal mesothelioma (5-20%). Other rare cases of mesothelioma originating from pericardium and tunica vaginalis of the testis have also been described [1,2]. When it involves the pleura, symptoms include breathlessness due to pleural effusion, cough, and chest pain associated with malaise and weight loss.
We retrospectively analyzed 29 out of 54 MPM tissue samples, both small pleural thoracoscopic biopsies and tissue from pleurectomy/decortication procedure, for which enough material for genomic evaluation was available. Samples were collected from first-diagnosed patients referred to the University-Hospital of Florence, Careggi, between January 2013 and December 2017. The specimens were stored and then collected from the routine archive at our institute.
Hematoxylin/eosin-stained (H&E) slides were reviewed according to the most recent edition of pleura tumor classification [1] by two pathologists (C.E.C and M.V.) who also visually selected the area with at least 50% of tumor cells for genomic DNA extraction. Clinical-pathological characteristics, including age, sex, exposure to asbestos, survival and histologic subtype, were retrieved from chart records. Detailed information concerning treatment was available for a minority of cases; therefore, it was not included in the analysis. All cases were anonymized with a number code.

Genomic DNA Extraction
DNA was obtained from tissue after manual dissection of selected areas containing at least 50% of neoplastic cells.
After purification and indexing phase with barcodes, the libraries were quantified using Qubit dsDNA HS Assay kit (Invitrogen, Thermo Fisher Scientific, Mississauga, Canada) and diluted in nuclease-free water to a final concentration of 100 nM.
Emulsion PCR was performed on Ion One Touch 2 (Life technology, Grand Island, NE, USA) while Ion Touch ES Ion Torrent Dynabeats MyOne Streptavidin C1 Beads (Invitrogen, Thermo Fisher Scientific, Mississauga, ON, Canada) were added to the pool.

Next-Generation Sequencing and Variant Calling
DNA sequencing was performed using Ion S5 System Ion Torrent S5 TM platform (Invitrogen, Thermo Fisher Scientific, Mississauga, ON, Canada). Tumor samples were tested with a commercial library kit (Myriapod NGS-LT Onco panel 56G, Diatech Pharmacogenetics, Jesi, Italy) targeting over 500 hotspot mutations in 48 genes.
The variant calling was obtained according to Myriapod NGS Data Analysis Software (Diatech Pharmacogenetics, Jesi, Italy), and the reference genome was GRCh37. Only samples sequenced with a quality score (PHRED score) of at least AQ20 were considered, and the coverage was set at a minimum average depth of 100 reads for each amplicon.
Genetic annotation was performed using COSMIC, LOVD, PUBMED and Clinvar for already known genetic variants, while for single-nucleotide polymorphism (SNP), we used dbSNP and 1000 Genome.

Statistical Analysis
Overall survival (OS) was defined as the time from the date of diagnosis to death or last follow-up using Kaplan-Meier method; hazard ratio (HR) was calculated with Cox proportional regression model, and the statistical differences between clinical-pathological variables were obtained using Chi-squared test. A p value < 0.05 was considered statistically significant. Statistical analysis was elaborated using SPSS v.25 (IBM, Armonk, NY, USA).
According to histopathology, all 29 MPM cases were epithelioid subtype. Moreover, the slides were re-evaluated, and based on the pathological grading proposed by the WHO of Thoracic Tumors 5th edition, they were subclassified as grade 1 (58.6% n = 17), grade 2 (31% n = 9) and grade 3 (10.3% n = 3). All the patients were in pT1 pathological stage (sec AJCC, 8th ed.). In detail, in 19 cases (65%), MPM was limited to the parietal pleura, 27.5% (n = 8) involved parietal and visceral pleura, and in 2 patients (6.8%) the parietal and the diaphragmatic pleura. All the clinical-pathological characteristics are summarized in Table 1.

Somatic Mutations Landscape in MM
NGS analysis revealed a total of 36 somatic mutations. At least one somatic variant was discovered in 23/29 samples (79.3%). In 11 patients (37.9%), NGS sequencing demonstrated the presence of a single mutation, while 2 mutations and 3 or more mutations were identified in 5 (18.5%) and 7 patients (25.9%), respectively.
EGFR, KDR, TP53 and KRAS were the most frequently mutated genes, being discovered in about the 62% of the population, corresponding to 18 of 29 patients.
The somatic variant with the highest allelic frequency (40.2%) was p.N371K in the MET gene, followed by the p.G13A mutation in exon 2 of KRAS (VAF =33.5%) and the p.R175H variant in exon 5 of TP53 (VAF = 32.6%).
Kaplan-Meier analysis of survival demonstrated that median overall survival was significantly shorter in patients with three or more mutations (median 7.6; range 6-9.1) (p = 0.004) compared to patients without mutations (median: 37.6; range: 19.2-56.0) ( Figure 1). No significant differences in terms of overall survival were found between cases with more than three mutations when compared with cases carrying one or two alterations.

Identification of Novel EGFR Mutations
We found eight EGFR activating point mutations (22.2%, n = 36) in 5/29 patients (17.2%): six of these were considered rare variants (p.V774M, p.T751I and p.S752Y in exon 19; p.G719S and p.A702D in exon 18; p.E66K in exon 21) that are usually described in non-small cell lung cancer. Moreover, we found two SNVs, p.S752C in exon 19 and p.P694T in exon 18, not previously reported, that were considered as functionally relevant by in silico analysis (ANNOVAR Variant Annotation). Moreover, the somatic variant p.G719S was described in two patients.
Two patients presented a double EGFR mutation: p.A702D associated with p.S752C in one patient and p.T751I plus p.E866K in the second one.
Statistical analysis did not reveal any correlation between EGFR genetic alterations and either clinical characteristics or survival [22].
In this cohort, mutations in TP53 did not show any correlation with overall survival or clinical characteristics.

KRAS Mutations
A KRAS mutation was found in 13.7% of the patients in the cohort. Genetic analysis revealed five point mutations (13.8%, n = 36) located in exon 2 codon 13 (p.G13A and p.S17N) and in exon 3 codons 48 and 57 (p.G48R, p.G48E and p.D57N), at the sites that encode for the DNA binding domain.
Kaplan-Meier analysis revealed a negative impact of KRAS mutations on survival. In detail, the median overall survival was 7.6 months (range: 4.6-10.4) for patients carrying the KRAS genetic alteration, while OS was significantly longer in KRAS wild-type cases: 32.6 ((range: 26.7-38.5) with a hazard ratio (HR) of (95%CI): 5.1 (1.5-17.9) (p = 0.005)) ( Figure 2). Notably, all four KRAS-mutated patients reported a certain previous occupational asbestos exposure. However, we found no concurrent KRAS and TP53 mutations and no correlation between KRAS somatic alteration and clinical features.
Kaplan-Meier analysis of survival demonstrated that median overall su significantly shorter in patients with three or more mutations (median 7.6; rang = 0.004) compared to patients without mutations (median: 37.6; range: 19.2-56 1). No significant differences in terms of overall survival were found between more than three mutations when compared with cases carrying one or two alte

Identification of Novel EGFR Mutations
We found eight EGFR activating point mutations (22.2%, n = 36) in 5/2 (17.2%): six of these were considered rare variants (p.V774M, p.T751I and p.S75 19; p.G719S and p.A702D in exon 18; p.E66K in exon 21) that are usually describ small cell lung cancer. Moreover, we found two SNVs, p.S752C in exon 19 and exon 18, not previously reported, that were considered as functionally relevant analysis (ANNOVAR Variant Annotation). Moreover, the somatic variant p.G described in two patients.
Two patients presented a double EGFR mutation: p.A702D associated wit in one patient and p.T751I plus p.E866K in the second one.
Statistical analysis did not reveal any correlation between EGFR genetic and either clinical characteristics or survival [22].
In this cohort, mutations in TP53 did not show any correlation with overa or clinical characteristics.

KRAS Mutations
A KRAS mutation was found in 13.7% of the patients in the cohort. Genetic analysis revealed five point mutations (13.8%, n = 36) located in exon 2 codon 13 (p.G13A and p.S17N) and in exon 3 codons 48 and 57 (p.G48R, p.G48E and p.D57N), at the sites that encode for the DNA binding domain.
Kaplan-Meier analysis revealed a negative impact of KRAS mutations on survival. In detail, the median overall survival was 7.6 months (range: 4.6-10.4) for patients carrying the KRAS genetic alteration, while OS was significantly longer in KRAS wild-type cases: 32.6 ((range: 26.7-38.5) with a hazard ratio (HR) of (95%CI): 5.1 (1.5-17.9) (p = 0.005)) (Figure 2). Notably, all four KRAS-mutated patients reported a certain previous occupational asbestos exposure. However, we found no concurrent KRAS and TP53 mutations and no correlation between KRAS somatic alteration and clinical features.

Single Nucleotide Polymorphism (SNP)
Next-generation sequence analysis discovered a series of SNPs in the TP53, SMAD4, KDR and STK11 genes. We focused our attention on the single-nucleotide polymorphism p.V297I of the KDR gene found 5 of 29 cases (17.2%).
Kaplan-Meier analysis showed a longer overall survival (p = 0.026) in patients harboring this SNP with a median overall survival not reached (NR) as compared to patients without this particular KDR SNV, who showed a median overall survival of 23.4 months (0.3-48.9, HR (95%CI): 0.32 (0.02-0.95)).
This evidence suggests that SNP p.V297I in the KDR gene might be considered a fa-

Single Nucleotide Polymorphism (SNP)
Next-generation sequence analysis discovered a series of SNPs in the TP53, SMAD4, KDR and STK11 genes. We focused our attention on the single-nucleotide polymorphism p.V297I of the KDR gene found 5 of 29 cases (17.2%).
Kaplan-Meier analysis showed a longer overall survival (p = 0.026) in patients harboring this SNP with a median overall survival not reached (NR) as compared to patients This evidence suggests that SNP p.V297I in the KDR gene might be considered a favorable prognostic factor in epithelioid MPM patients (Figure 3).

Discussion
Pleural mesothelioma is an aggressive tumor with implacable prognosis often ass ciated with asbestos exposure, and to date, there is no real knowledge about the pathog netic mechanisms of this neoplasm. Moreover, because of its rarity, the number of patien tested in most studies is small [27].
The emerging use of next-generation sequencing, however, has highlighted an in creasing number of mutated genes that may have a role in MPM biology.
Mesothelioma usually harbors a low mutation burden (TMB) and low genetic com plexity with an overall rate of <2 non-synonymous mutations per megabase [16,28], al hough cases carrying several gene mutations and/or co-mutations have been reported.
MPM is preferentially associated with alterations in gene involved in the cell cycl Hedgehog and Hippo pathway, and in the DNA repair mechanism. Furthermore, the si naling pathways most frequently affected are the PI3K/AkT/mTOR and MAPK pathway [29,30].
Another important mechanism impaired in MPM is the epigenetic regulation, parti ularly due to the hypermethylation of three genes: TMEM30B, KAZALD1 and MAPK13 In a recent study [29], genomic analysis of more than 1000 pleural mesothelioma sam ples showed that the most frequent molecular alteration was a loss of function (LOH) specific tumor suppressor genes such as CDKN2A, CDKN2B, BAP1 and NF2 (15.20% the total cases each), which is in line with all previous studies.
Notably, additional mutations in targetable genes such as KRAS, EGFR, PDGFRA/ and FGFR3 were detected, although with a frequency of <5% in their population.
The authors of this study, based on the presence, absence or co-occurrence CDKN2A/B and BAP1 mutations, identified four mesothelioma subgroups with differe prognostic impact.
Moreover, a novel molecular MPM subtype has been described. This new enti showed a distinctive clinical phenotype, being frequent in female patients with young age at diagnosis and being recurrent in TP53 and SETDB1 mutations [20].
Gain-of-function mutations in MPM patients are infrequent. Copy number gains an

Discussion
Pleural mesothelioma is an aggressive tumor with implacable prognosis often associated with asbestos exposure, and to date, there is no real knowledge about the pathogenetic mechanisms of this neoplasm. Moreover, because of its rarity, the number of patients tested in most studies is small [27].
The emerging use of next-generation sequencing, however, has highlighted an increasing number of mutated genes that may have a role in MPM biology.
Mesothelioma usually harbors a low mutation burden (TMB) and low genetic complexity with an overall rate of <2 non-synonymous mutations per megabase [16,28], although cases carrying several gene mutations and/or co-mutations have been reported.
MPM is preferentially associated with alterations in gene involved in the cell cycle, Hedgehog and Hippo pathway, and in the DNA repair mechanism. Furthermore, the signaling pathways most frequently affected are the PI3K/AkT/mTOR and MAPK pathways [29,30].
Another important mechanism impaired in MPM is the epigenetic regulation, particularly due to the hypermethylation of three genes: TMEM30B, KAZALD1 and MAPK13.
In a recent study [29], genomic analysis of more than 1000 pleural mesothelioma samples showed that the most frequent molecular alteration was a loss of function (LOH) of specific tumor suppressor genes such as CDKN2A, CDKN2B, BAP1 and NF2 (15.20% of the total cases each), which is in line with all previous studies.
Notably, additional mutations in targetable genes such as KRAS, EGFR, PDGFRA/B and FGFR3 were detected, although with a frequency of <5% in their population.
The authors of this study, based on the presence, absence or co-occurrence of CDKN2A/B and BAP1 mutations, identified four mesothelioma subgroups with different prognostic impact.
Moreover, a novel molecular MPM subtype has been described. This new entity showed a distinctive clinical phenotype, being frequent in female patients with younger age at diagnosis and being recurrent in TP53 and SETDB1 mutations [20].
Gain-of-function mutations in MPM patients are infrequent. Copy number gains and hotspot promoter mutations in the TERT gene have represented the first recurrent oncogenic gain in MPM [31,32]. Moreover, pleura mesothelioma carrying the TERT promoter mutation with TP53 and NF2 mutations has demonstrated a shorter overall survival [33,34].
In a cohort of 29 patients with diagnoses of epithelioid MPM, we described 36 somatic mutations and 5 single-nucleotide polymorphisms. We did not find any mutations usually reported in MPM, such as BAP1, CDKN2A SNVs, etc.
The most pathologically relevant mutations in our series were identified in the EGFR, KRAS, TP53 and KDR (VEGFR2) genes.
The TP53 gene, located in chromosome 17 (p13.1), encodes for a tumor suppressor protein that acts as a transcriptional factor to induce cell cycle arrest, apoptosis, DNA repair, or changes in metabolism.
According to The Cancer Genome Atlas (TCGA), 16% of all MPM show a mutation in the TP53 gene.
We detected six somatic mutations of TP53, all located in the typical hotspot of this onco-suppressor gene and corresponding to the DNA-binding domain. The highest allelic frequency (VAF: 32.6%) was found in the p.R175H variant, which has been demonstrated to confer oncogenic potential for the p53 protein, underlying a metastatic and chemoresistant phenotype. This mutation consists of a substitution of arginine with histidine, causing a reduction of 100 times in the affinity for zinc and consequently leading to a structural remodeling of the protein. Na et al. demonstrated, in BRCA1 deficiency cell lines with concomitant TP53 R175H , that the use of zinc metallochaperones (ZMCs) increased the zinc's concentration and allowed for the correct folding of the mutant TP53 protein and its reactivation [35]. TP53-mutated MPM in our series represented 13.7% of the cases, although in the literature, it is more frequently reported in sarcomatoid and biphasic subtypes. Moreover, many studies have demonstrated its key role in determining an aggressive mesothelioma phenotype [31,36], mostly when associated with other genetic alterations probably related to a genomic instability [20,37].
In our cohort of epithelioid MPM, Kaplan-Meier analysis did not reveal any significant correlation between TP53 mutations and OVS.
Recently, Sementino et al. [36] demonstrated in a mouse model that the combined inactivation of PTEN and TP53 is able to drive the acquisition of an aggressive MPN phenotype through consistent activation of the AKT-mTOR signaling pathway.
One of our patients carried concurrent somatic mutations in PTEN (p.Q245*), AKT (p.P51L) and TP53 (p.R175H), demonstrating an overall survival of 32.2 months, in line with the median OS of epithelioid MPM. However, this represents only a single case, and more studies in human models are needed.
The EGFR (epidermal growth factor receptor) gene encodes for a transmembrane receptor tyrosine kinase belonging to a cell surface receptor super-family (RTK) that can be activated by several different ligands such as EGF (epidermal growth factor), TGFA (transforming growth factor-Alpha) and HBEGF (heparin-binding EGF-like growth factor) [32,38].
Ligand-dependent EGFR activation transduces multiple signaling pathways, including the RAS/MAPK pathway, the PI3K/AKT pathway, and the phospholipase C (PLC)/protein kinase C (PKC) signaling cascade, leading to cell proliferation, angiogenesis and inhibition of apoptosis [39].
It is well known that the EGFR gene is often mutated in numerous carcinoma, such as breast cancer, lung cancer and colon-rectal cancer.
In malignant mesothelioma, overexpression of the EGFR gene, which is detected by immunohistochemistry or FISH, is frequent, accounting for 44-90% of cases, and is probably caused by reactive oxygen species (ROS) present in the tumor microenvironment. EGFR mutations are infrequent, and in more than 75% of cases, they are rare mutations [29,37,[40][41][42][43].
The SNVs occurring in codons 774, 719 and 751 were already described in other neoplasms from different sites. The variant in codon 702 has only been found in esophageal carcinoma [44].
We have also shown two EGFR mutations not yet described in MM, which are considered functionally pathogenetic, in both exon 18 (p.P694T) and exon 19 (p.S752C). Previously, the authors of [45] reported a pathogenetic EGFR mutation in exon 19 (p.E746_A750del), suggesting the potential role of EGFR tyrosine kinase inhibitors (EGFR TKIs) as target therapy in selected cases of MPM.
KRAS single-nucleotide variations in mesothelioma samples have been described for the first time in a study by Mezzapelle et al. [40], in which a correlation with OVS was not found. However, it was demonstrated that mutations in KRAS were not so rare, as reported in previous studies.
Later, other MPM research studies based on comprehensive genomic analyses failed to find KRAS alterations due to the low frequency of copies of mutated alleles that require more sensitive approaches.
Recently, Trassl et al. [46] highlighted that although KRAS mutations are less commonly reported in MPM, KRAS pathway alterations are very frequent. Additionally, a study by Marazioti et al. [47] demonstrated, in a mouse model, that KRAS point mutations in epitheliod MPM, with or without TP53 alteration, are potentially able to drive MPM development and a more aggressive disease with frequent mesothelial pleural effusion. In this study, we reported five different somatic mutations in KRAS that occurred in 13.7% (4/29) of the patients. Notably, all the variants were characterized by a relatively small, below 5%, variant allele frequency (VAF), as usually reported in the literature, except in one case (KRAS:p.G13A) where the VAF was above 30%. Kaplan-Meier analysis revealed a negative impact of KRAS mutations on patient survival, confirming the data of previous studies, most of which were carried out on cell lines or murine models.
Finally, another interest finding was the association between the single-nucleotide polymorphism KDR (VEGFR2) p.V297I (rs2305948) and a longer survival in MPM patients. Five of the patients showed this variant, and the statistical analysis indicated a relevant increase in median overall survival.
In the past few years, it has been shown that in both cases of pleuritis and pleural mesothelioma, an increase in serum level of vascular endothelial growth factor (VEGFR) occurs, particularly in mesothelioma [48,49]. Catalano et al. [50] demonstrated an inverse correlation between the serum level of VEGF and survival, suggesting that the VEGF signaling pathway can contribute to cancer progression.
The SNP p.V297I of KDR is caused by a missense point mutation in codon 297 located in exon 7, leading to the substitution of valine with isoleucine; this alteration produces a lower binding affinity between the receptor VEGFR2 (KDR) and its specific ligand VEGF, resulting in a softened function. In the literature, the role of this SNP is not yet defined, with a series of conflicting hypotheses. A few studies have described this substitution in KDR as an indicator of bad prognosis and lower overall survival [51][52][53], while the paper by Tinhofer in 2016 supports the hypothesis that it is a good prognostic factor in squamous cell carcinoma of the head and neck [54].

Conclusions
In this study, we found two mutations in exons 18 and 19 of the EGFR gene not previously described in the literature, both identified as potentially pathogenetic by different types of prediction software. In addition, we identified three mutations, p.V774M, p.T751I and p.S752Y, in exon 19 of EGFR not previously described in MPM.
Although a limitation of our study was the small number of cases studied and the lack of patient treatment data, we reported a possible correlation between KRAS mutations and shortened overall survival of patients when compared with wild-type cases. Previously, other studies conducted on cell lines [46] or on murine models [54] underlined the importance of the KRAS mutation in the pathogenesis and the progression of MPM. There are also extensive genomic and/or transcriptomic descriptive analyses of tissues from patients with MPM, showing among others the presence of a KRAS mutation. However, to the best our knowledge, the correlation between KRAS alteration and overall survival in primary neoplastic tissue samples is still lacking.
Moreover, another result that needs to be confirmed in large-scale studies is that ≥3 mutations had a negative influence on survival, compared with the patients carrying only one or two somatic variants.
Finally, at least in our cohort of patients, the presence of SNP p.V297I of the KDR gene seems to confer an advantage for overall survival.
Our study has weaknesses. First, it is retrospective in nature, and while eight patients were still alive at the time of the database lock, they all died or were lost by follow-up at the time of writing of this manuscript. Thus, information regarding therapy was available for only a few patients, which was uninformative as to the impact of treatment on survival. Second, the number of cases was relatively small and included the sole epitheliod MPM subtype. Third, pathological grading of MPM as well as the type of specimen procedures (biopsy or pleurectomy) showed no significant correlation with patient survival, but this may be related to the small amount of patients in our study.
Despite these limitations, our findings are in line with previous studies [45][46][47], pointing to an underestimated importance of the RAS pathway in pleura malignant mesothelioma, as suggested by a significantly worse prognosis for patients carrying KRAS mutations. This study adds some new findings in the molecular profile and prognosis of MPM patients, which remain to be fully validated in broader studies and in the different MPM histotypes.