Concomitant TP53 mutation in early-stage resected EGFR-mutated non-small cell lung cancer: a narrative approach in a genetically admixed Brazilian cohort

TP53 mutations are frequent in non-small cell lung cancer (NSCLC) and have been associated with poor outcome. The prognostic and predictive relevance of EGFR/TP53 co-mutations in NSCLC is controversial. We analyzed lung tissue specimens from 70 patients with NSCLC using next-generation sequencing to determine EGFR and TP53 status and the association between these status with baseline patient and tumor characteristics, adjuvant treatments, relapse, and progression-free (PFS) and overall survival (OS) after surgical resection. We found the EGFR mutation in 32.9% of patients (20% classical mutations and 12.9% uncommon mutations). TP53 missense mutations occurred in 25.7% and TP53/EGFR co-mutations occurred in 43.5% of patients. Stage after surgical resection was significantly associated with OS (P=0.028). We identified an association between progression-free survival and poor outcome in patients with distant metastases (P=0.007). We found a marginally significant difference in OS between genders (P=0.057) and between mutant and wild type TP53 (P=0.079). In univariate analysis, distant metastases (P=0.027), pathological stage (IIIA-IIIB vs I-II; P=0.028), and TP53 status (borderline significance between wild type and mutant; P=0.079) influenced OS. In multivariable analysis, a significant model for high risk of death and poor OS (P=0.029) selected patients in stage IIIA-IIIB, with relapse and distant metastases, non-responsive to platin-based chemotherapy and erlotinib, with tumors harboring EGFR uncommon mutations, with TP53 mutant, and with EGFR/TP53 co-mutations. Our study suggested that TP53 mutation tends to confer poor survival and a potentially negative predictive effect associated with a non-response to platinum-based chemotherapy and erlotinib in early-stage resected EGFR-mutated NSCLC.


Introduction
In 2020, lung cancer was reported to be the main cause of cancer-related deaths and the second more prevalent diagnosed malignancy (1). The most frequent type of lung cancer is non-small cell lung cancer (NSCLC), which accounts for 85-90% of lung cancer patients (2). At the first clinical consultation, most patients already present advanced stages of the disease, so the available treatments are chemotherapy, immunotherapy, and target therapy.
Particularly, the outcomes of lung adenocarcinomas have improved due to the increasing characterization of oncogenic drivers and the possibility of efficiently targeting these drivers (3). This advance mainly relates to the development of minor-molecule tyrosine kinase inhibitors (TKI) targeting the activating mutations in the epidermal growth factor receptor (EGFR) gene (4). In patients with advanced disease, EGFR TKIs have shown satisfactory objective response rates and prolonged progression-free survival (PFS) compared with chemotherapy (5).
However, approximately 30% of patients develop primary resistance to TKIs and/or chemotherapy, and the disease eventually relapses months to years after starting TKIs and chemotherapy. Relatively few studies have been conducted on the mechanism of treatment resistance. Uncommon and multiple somatic mutations have been associated with worse outcomes compared with tumors with a single classical mutation (6,7). Access to next generation sequencing (NGS) allows detection of co-mutations in advanced EGFR mutated-NSCLC patients. This detection suggests that these co-mutations might be one of the mechanisms of drug resistance, among which TP53 mutations were the most frequent co-mutations in all types of lung cancer (8). Some clinical studies suggest a negative prognostic effect of TP53 mutations on NSCLC with adjuvant chemotherapy in patients with completely resected TP53mutant NSCLC (9). Unfortunately, to date, there are no approved drugs that specifically target TP53 in NSCLC.
Since lung cancer investigation has focused basically on Caucasian and Asian cohorts (2), we know little about the EGFR/TP53 co-mutations of NSCLC in the Brazilian population. Current research indicates that race plays a role in the genomics of NSCLC in this population. For example, in patients with adenocarcinoma, the estimated prevalence of EGFR mutations in Asians is around 60%, while it is only 20% percent in Caucasians (4). In the limited series of Brazilian patients reported, we observed a considerable variation in mutation rate in the south of Brazil (19%) and 22-30% in the southeast (particularly in the city of São Paulo). This variation in mutation rate suggests an increased prevalence of Amerindian and Asian ancestries (10). However, there is no information about EGFR/P53 co-mutations in Brazilian patients with NSCLC.
Defining of the combined molecular pathogenesis of NSCLC is crucial in Brazilians, as this population has both a higher incidence of NSCLC and an increased mortality from the disease compared with Caucasians and Asians and eventually greater resistance to adjuvant treatment. Thus, we designed the present study to evaluate the prognostic and predictive value of EGFR/TP53 co-mutation detected by NGS in surgically resected NSCLC patients. We aimed to provide a narrative portrait of the effect of the co-mutation on PFS and overall survival (OS) in patients receiving adjuvant treatment.

Patient population and data collection
Because the new EGFR/TP53 co-mutation can also arise from sequencing artifacts, especially artifacts associated with formalin-related DNA damage, as described in Wong et al. (11), we conducted our investigation in freshfrozen specimens from Brazilian patients with lung cancer. Specimen collection took place during surgical resections conducted from August 2003 to August 2010 at the A.C. Camargo Cancer Center, a tertiary referral center for the treatment of lung cancer in São Paulo, Brazil. Our group obtained a total of 70 fresh-frozen specimens from Brazilian patients with lung cancer from different regions of the country. Two experienced lung pathologists reviewed the histologic diagnoses, assessed the accuracy of the histologic diagnoses based on the World Health Organization (WHO) 2021 classification system (12), and stratified them into non-squamous non-small lung cancer (n=46) and squamous non-small cell lung cancer (n=24). EGFR and TP53 statuses were correlated with baseline characteristics, including age, sex, ethnicity, smoking history, pathologic TNM stage (13), histology, type of EGFR mutation, radiotherapy, platinum-based chemotherapy, tyrosine kinase inhibitors (TKI), and relapse free survival (RFS) and overall survival (OS) after primary surgical resection, relapse, and development of distant metastases.
The study was approved in accordance with the ethical standards of the local committee on human experimentation (Research Ethics Committee of University of São Paulo Medical School -CAAE: 79769017.1.0000.5440; opinion number: 2.673.320). The informed consent was waived due to the retrospective study design and the identity of the subjects was omitted and anonymized.

Targeted gene profiling
The DNA of fresh tumor tissue was extracted using the QIAamp DNA Mini Kit (Qiagen, Germany), according to the manufacturer's recommendations, and quantified using the Qubit s 3.0 Fluorometer (Invitrogen, Life Technologies, USA). The EGFR and TP53 genes were targeted using TruSeq Custom Amplicon Panel v1.5 kit (TSCAP, Illumina, USA), followed by massive parallel sequencing on an Illumina MiSeq platform consisting of 150 bp paired-end reads (300 cycles). All tumor specimens had an average sequencing depth of the target region X100Â and coverage of the target region 490% at 30Â. The Molecular Genetics and Bioinformatics Laboratory of the Experimental Research Unit (UNIPEX) at the Medical School of São Paulo State University (UNESP) performed the sequencing data analyses to reduce the effects of PCR amplification and sequencing artifacts. The raw sequencing data were base-called and demultiplexed using MiSeq Reporter v.1.8.1 (Illumina) with default parameters, and FastQC files were generated for downstream data analysis. Filtered reads were aligned to the human genome (hg19, GRCh37) using the Burrows-Wheeler Alignment tool (BWA) v.0.7.10 (http://bio-bwa.sourceforge.net). After alignment, the SAMtools software (https://www.htslib.org/) was applied to convert the alignment files to an indexed binary alignment map format. The single nucleotide variants (SNVs) and short insertions and deletions (INDELs) were named using the GATK UnifiedGenotyper, including HaplotypeCaller with default parameters based on hg19 and annotated with dbSNP version 144 (gatk.broadinstitute.org). Our group used the following cut-off criteria to reduce falsepositive somatic mutations that might originate from germline variants: number of reads with the altered base in the tumor X10, mutations detected at a position of total read depth of X100, frequency of the reads with the altered base in the tumor X5% except for variants also reported in the COSMIC database, minor allele frequency o0.1% in two publicly available databases, namely 1000 Genomes and Exome Aggregation Consortium. We annotated the variants using the VEP software (grch37.ensembl.org) based on the consequences, predicted impacts, and reported allele frequencies in the population. The variants of unknown significance (VUS) were checked on the ClinVar database (http://www.ncbi.nlm.nih.gov/clinvar/).

Statistical analysis
We conducted the Pearson chi-squared test or Fisher's exact test to compare categorical variables and the one-way analysis of variance to compare continuous variables. OS definition -the primary end point -was defined as the time from surgery to death from any cause or to the last follow-up of surviving patients. We set the PFS as the time from surgery to recurrence or death from any cause or to the last follow-up of surviving patients. As our cases were from a genetically mixed population, we evaluated the prognostic and predictive value of TP53/ EGFR co-mutation status in non-squamous NSCLC and squamous NSCLC. Hazard ratios (HRs) and their confidence intervals (CIs) were estimated via a multivariable Cox proportional hazards model including the core variables with Po0.1 in the univariate analysis: gender, pathological TNM stage, relapse, and adjuvant treatment (chemotherapy and radiotherapy), TP53 status, and EGFR/TP53 co-mutation. We set the statistical significance at Po0.05. Survival curves were based on Kaplan-Meier methods and presented with unadjusted HRs from the Cox model and P-values using log rank statistic. The SPSS software version 22.0 (IBM Corporation, USA) was used for the statistical analyses.

Mutation status
In our cohort, the mutation rate of TP53 was 41.5% (n=29/70) and of EGFR was 32.9% (n=23/70), being that the well-established EGFR mutation in exons 18-21 was 20% (n=14/70) and in other exons was 12.9% (n=9/70). From the mutated EGFR cohort, EGFR/TP53 co-mutation was found in 10 patients (n=10/23; 43.5%), as shown in Table 1. The genomic profile of the TP53 gene identified is summarized in Table 2. It is worth noting that, the majority of TP53 mutations was missense and of pathogenic significance. Tables 3 and 4, respectively, show the association between patient characteristics and EGFR and TP53 status. EGFR uncommon mutations were most common in male patients (P=0.005). There were also significantly more non-squamous NSCLC (P=0.035) with locoregional relapse (P=0.028) in the TP53 wild-type group. EGFR/TP53 co-mutation did not differ among the clinicopathological characteristics of the patients. Finally, we examined the importance of the TP53 mutation identified in EGFR-mutant NSCLC patients stratified in EGFR co-existing pathogenic mutation (N=5) and EGFR co-existing VUS (N=5) (Supplementary Table S1). Clinicopathological characteristics of NSCLC patients with EGFR/TP53 co-mutation are shown in Supplementary  Table S2. Overall, dual TP53/EGFR mutations were found in 10 patients (43.5%), 8 males and 2 females, with a median age of 76 years, mostly from European ancestry, and in early disease stage (n=9). Common clinical characteristics of patients whose tumors harbored TP53 mutation in EGFR co-existing pathogenic mutation were older patients with non-squamous NSCLC. These patients developed distant metastases after surgical resection with partial response to systemic chemotherapy and EGFR-TKI and short survival. In contrast, most patients with squamous NSCLC harboring TP53 and coexistent EGFR VUS mutation were younger, without tumor relapse, received no adjuvant chemotherapy, and progressed with long survival.

Survival analysis
Preliminary examination of Kaplan-Meier survival curves demonstrated that patients with pathological stages IA, IB, IIA, and IIB had approximately the same hazard for survival, with a median survival time of 85 months. Thus, we coded overall pathological stage as a single dummy variable with a value of 1 for stages I and II and a value of 2 for stages IIIA and IIIB. The results of the Cox model analysis are reported in Table 5. Among the entire cohort of 70 patients, there were 38 deaths (54.3%). For the overall sample, in the univariate analysis, stage after surgical resection (IIIA vs I-II; P=0.028; Figure 1A), TP53 status (borderline for wild type vs mutant; P=0.079: Figure 1B), and EGFR/TP53 co-mutation status (borderline for wild type vs mutant; P=0.061) influenced OS. We also assessed the effect of relapse and TP53 status on OS in the subset of patients who received adjuvant chemotherapy (n=24; 12 mutants vs 12 wild types) and had metastases (n=19; 16 locoregional vs 3 distant), and we identified a borderline significance (P=0.07 and P=0.06; Figure 2A and B, respectively). It is worth noting that the abrupt drop of survival curves (black line) in both Figure 2A and B refers to three patients in advanced stage with brain metastases, two of which were treated with erlotinib, therefore reflecting the small number of patients. In multivariate analysis, the significant factors for the high risk of death model (P=0.029) were stage IIIA-IIIB, relapse with distant metastases, non-response to chemotherapy, tumors harboring EGFR uncommon mutations, TP53 mutation, and EGFR/TP53 co-mutations. Continued on next column

Discussion
The narrative portrait of our study population showed that the main reason for the failure of surgical resection and adjuvant treatment in prolonging survival of patients with early-stage NSCLC was that routine pathological analysis failed to predict relapse and metastases. In fact, during follow-up, even the patients in early stage developed locoregional and distant relapse with central nervous system metastases. These patients received adjuvant radiotherapy and chemotherapy. Nevertheless, the mortality rate was 54.3%. Lung cancer is a genetic disease that results from a multistep process involving genetic and epigenetic changes, especially activation of growth pathways and inhibition of tumor suppressor pathways. Therefore, the relevant question is whether supplementary genetic information from tumor tissue combined with classic TNM stage classification can help us to improve risk stratification and patient selection for personalized treatment.
It is evident that lung cancer investigation and treatment have entered an era of personalized medicine, which uses biomarkers to stratify patients who are more likely to benefit from a specific drug. However, not all NSCLC are phenotypically equal, and some drugs that are effective against non-squamous NSCLC, including  (14,15). Personalized treatment is more advanced in adenocarcinoma, for which patients are routinely investigated for oncogenic drivers to select therapy. EGFR target therapy was approved as the standard of care for patients with classical EGFR mutations (exon 19 deletions), but 20-30% of these patients developed primary resistance to EGFR target therapy (16). To improve the therapeutic outcome of these patients, it is crucial that we understand the mechanisms underlying resistance. Uncommon EGFR mutations, such as the exon 20 insertion mutation, could result in primary resistance to EGFR target therapy (17). The advent of NGS created the possibility to detect co-mutations in EGFR-mutated NSCLC patients. These co-mutations might be one of the mechanisms of primary drug resistance, among which TP53 mutations were the most frequent co-mutations (3).
In the present study, we evaluated the clinical outcomes of patients with EGFR-driven early NSCLC based on their TP53 mutational status. In our population, EGFR mutation was found in 32.9% of patients. However, mutations in exons 18-21 were 20%, with the most common mutations being short deletions in exon 19 (E746-A750) (18). These findings are consistent with previous works, showing EGFR mutations as the predominant driver mutations in patients with NSCLC (19). In addition, classical mutations such as an exon 19 short deletion and an exon 21 point mutation, L858R, are the most common mutations, accounting for about 85-90% (20,21). TP53 gene, located on the short arm of chromosome 17 (17p13), is involved in many biological processes, including DNA repair, metabolism, cell cycle arrest, apoptosis, and aging (22). In our population, about 40% of patients harbored TP53 mutations. We found TP53 mutations in 21.4% of non-squamous cell carcinomas and 20% of squamous cell carcinomas, contrasting with studies reporting 40 and 51%, respectively (23). In the current study, TP53/EGFR co-mutation was found in 43.5% of patients with early-stage NSCLC. Previous reports indicated that 17-72% of advanced EGFR-mutant lung cancers harbor TP53 mutations (3,22). Co-mutation status did not differ by age, gender, smoking history, histotypes, and pathologic stage, as previously reported by the LACE-Bio group (22).
Currently, the Brazilian population is one of the most genetically diverse populations in the world. Such diversity results from five centuries of admixture between four ethnic groups: Asian, European, African, and Amerindian. Despite the shortage of information about ancestry in our population, the European ancestry had a higher proportion of classical and uncommon EGFR and TP53 mutations than the Asian and African ancestries, whereas EFGR/ TP53 co-mutations occurred in 16.7% of the Asian ancestry and 1.9% of the European ancestry and remained undetected in the African ancestry. Therefore, the rates of somatic mutations in key pathogenic genes involved in NSCLC may have an ancestry-related effect on the mutational spectrum. Because of the historical admixture of the Brazilian population and the patient cohort comprising individuals from different geographic regions, we are aware that the diversity or contribution of the genetic background can only be assessed or even categorized to a limited extent, considering the reduced power of n=10 early-stage patients with co-occurrence of TP53/EGFR mutations. As we pointed out, the TP53 mutation rate in the Brazilian cohort is half of that found in other studies and population cohorts. This query was addressed explanatorily rather than experimentally, as it was not intended to redirect the survey. We did not use molecular tests for ancestry; the patient's ethnicity was based on information from the medical record. However, regarding the TP53 gene, it is essential to relate it to the miscegenation of the Brazilian population. It is known that Li-Fraumeni syndrome (LFS), caused by the p.R337H variant in the TP53 gene, is rare in the world population but highly prevalent in the Brazilian population (23). Somatic variant databases rarely describe this variant. However, the identification of this variant in the genomic profile of tumors should be a predictive finding for LFS in the Brazilian population. In fact, one patient in our series was positive for the LFS variant, but patient was not EGFR-mutated NSCLC.
In patients with EGFR-mutated NSCLC, the coexistence of a TP53 mutation influenced OS when controlling for age, pathologic stage, relapse, brain metastases, and chemotherapy. This finding suggested that co-mutation is a dependent prognostic marker. Our data contrasted with the report from Labbé et al. (6) who found that concomitant TP53 mutation status was dissociated from OS in patients with EGFR-mutant NSCLC at an early stage who underwent primary surgical resection and received adjuvant chemotherapy. These data indicate that co-mutations were not a strong prognostic marker in earlystage patients. The same study also found that objective response rate is not significantly different between TP53mutant and wild type, and there is a non-significant trend towards shorter OS on EGFR with TP53 mutation in advanced NSCLC patients who received target therapy (6). Therefore, further studies on the utility of EGFR/TP53 co-mutation as a prognostic and predictive biomarker for early EGFR-mutated NSCLC patients are needed.
Investigations have been conducted to verify whether the type of gene mutation influences the prognostic and  predictive effect of TP53 mutations. Anchored to mutations subtypes, TP53 mutations showed a remarkable preference for missense mutations over nonsense and frameshift mutations, which are commonly dominant in other tumor suppressor genes such as RB1 and PTEN (24). The study from Labbé et al. (6) showed that NSCLC patients with TP53 missense mutations have significantly shorter PFS when treated with target therapy. In another published study, TP53 non-missense mutations reduced responsiveness to target therapy and worsened the prognosis of EGFR-mutant advanced NSCLC (25).
Although we demonstrated a predictive and prognostic value of EGFR/TP53 co-mutations in a small cohort of NSCLC, future validation using a similar cohort with a large set of patients is needed to corroborate the observed correlations. The present study is mainly descriptive and exploratory, and extension of our findings is essential.
Overall, this study presented a significant model for high risk of death and poor OS for patients with stage III NSCLC with relapse and distant metastases, non-responsive to platinum-based chemotherapy and EGFR TKIs, and harboring EGFR uncommon mutations, TP53 mutations, and EGFR/TP53 co-mutations. Although not currently a therapeutic target, routine inclusion of TP53 mutation testing by NSG may more accurately determine the effects of this tumor suppressor gene both alone and in combination with other driver mutations in lung cancer and whether there is an interaction with treatment.

Conclusion
Our study suggested that TP53 mutation tended to confer poor survival and potential negative predictive effect associated with a non-response to platinum-based chemotherapy and erlotinib in early-stage EGFR-mutated resected NSCLC. However, our observation remains to be validated.

Supplementary Material
Click here to view [pdf].