Prevalence of targeted therapy-related genetic variations in NSCLC and their relationship with clinicopathological characteristics

Background Non-small cell lung cancer (NSCLC) is the most common cancer type in China. Targeted therapies have been used to treat NSCLC for two decades, which is only suitable for a subgroup of patients with specific genetic variations. The aim of this study was to investigate the prevalence of genetic variations leading to sensitivity or resistance to targeted therapies in NSCLC, and their relationship with clinicopathological characteristics of the patients. Methods Tumor samples were collected from 404 patients who were diagnosed to have NSCLC and underwent surgery, transthoracic biopsy, bronchoscopy biopsy, or pleural aspiration in Sichuan Provincial People’s Hospital from January 2019 to March 2020. Commercial amplification-refractory mutation system kits were used to detect targeted therapy-related genetic variations in those tumor samples. The prevalence of genetic variations and their relationship with patient clinicopathological characteristics were analyzed using statistical software, followed by subgroup analysis. Results In all, 50.7% of the NSCLC patients had sensitive genetic variations to anti-EGFR therapies, and 4.9% of those patients had co-existing resistant genetic variations. Fusions in ALK, ROS1, or RET were found in 7.7% of the patients, including 2 patients with co-existing EGFR exon 19 deletion or L858R. EGFR exon 19 deletion and L858R were more common in female patients and adenocarcinoma. Further subgroup analysis confirmed the observation in female patients in adenocarcinoma subgroup, and in adenocarcinoma in male patients. In addition, smokers were more likely to have squamous cell carcinoma and KRAS mutation and less likely to have EGFR L858R, which were also confirmed after standardization of gender except KRAS mutations. Conclusion Nearly half of the NSCLC patients were eligible for anti-EGFR treatments. In NSCLC, female gender and adenocarcinoma may indicate higher chance of EGFR exon 19 deletion or L858R, and smoking history may indicate squamous cell carcinoma and EGFR L858R.


Methods
Tumor samples were collected from 404 patients who were diagnosed to have NSCLC and underwent surgery, transthoracic biopsy, bronchoscopy biopsy, or pleural aspiration in Sichuan Provincial People's Hospital from January 2019 to March 2020. Commercial amplification-refractory mutation system kits were used to detect targeted therapy-related genetic variations in those tumor samples. The prevalence of genetic variations and their relationship with patient clinicopathological characteristics were analyzed using statistical software, followed by subgroup analysis.

Results
In all, 50.7% of the NSCLC patients had sensitive genetic variations to anti-EGFR therapies, and 4.9% of those patients had co-existing resistant genetic variations. Fusions in ALK, ROS1, or RET were found in 7.7% of the patients, including 2 patients with co-existing EGFR exon 19 deletion or L858R. EGFR exon 19 deletion and L858R were more common in female patients and adenocarcinoma. Further subgroup analysis confirmed the observation in female patients in adenocarcinoma subgroup, and in adenocarcinoma in male patients. In addition, smokers were more likely to have squamous cell carcinoma and KRAS clinicopathological characteristics and genetic variations of tumor in NSCLC, which could help further optimize the usage of targeted therapies in patients with different clinicopathological background. The aim of this study was to investigate the prevalence of genetic variations which indicate sensitivity or resistance to targeted therapies in NSCLC, as well as the relationships between those genetic variations and clinicopathological characteristics of patients. Our study results showed significant associations between genetic variations and age (EGFR L858R), gender (EGFR exon 19 deletion, L858R), smoking history (EGFR L858R, exon 20 insertion, and KRAS mutation), histological type (EGFR exon 19 deletion, L858R), and tumor grade (ALK fusion). In addition, further analysis also revealed significant associations between gender and EGFR L858R, KRAS mutation, histological type in non-smokers, and associations between smoking history and histological type in both male and female patients, or EGFR L858R in male patients.

Patients
Tumor samples were retrospectively collected from patients diagnosed with NSCLC who underwent surgery, transthoracic biopsy, bronchoscopy biopsy, or pleural aspiration in Sichuan Provincial People's Hospital from January 2019 to March 2020. The inclusion criteria included: 1) pathologically-proven NSCLC, e.g. lung adenocarcinoma, lung squamous cell carcinoma, or large cell carcinoma; 2) age � 18 years; 3) with sufficient clinicopathological information. The exclusion criteria included: 1) samples with pathologically-proven small cell lung cancer; 2) sample collected in lung not being the primary lesion; 3) samples from NSCLC patients who received chemotherapy, radiotherapy, or targeted therapies; 4) samples from patients with unclear smoking history or clinical staging information. All those tumor samples were formalin-fixed and paraffin-embedded. After sectioning and H&E staining, pathological diagnosis for each sample was confirmed by two independent pathologists, and slides with tumor content < 30% were excluded from further analysis. This study complied with the Declaration of Helsinki and was approved by Institutional Review Board of Sichuan Provincial People's Hospital. Since this was a retrospective non-interventional study, informed consent was waived by the Institutional Review Board, and all the data were collected and analyzed anonymously.

Amplification-refractory mutation system (ARMS)
Three to eight sections from each tumor sample were firstly de-paraffinized in xylene in a 1.5-ml Eppendorf tube. DNA and RNA were extracted using a commercial DNA/RNA coextraction kit (FFPE DNA/RNA Extraction Kit, AmoyDx, China) following the manufacturer's instruction. After measurement of A260/A280, specific gene variations in tumor samples were determined using cFDA-approved commercial kit (Human Ten-Genetic-Variation Co-Detection Kit, AmoyDx, China) which target genetic variations in EGFR, ALK, ROS1, RET, KRAS, NRAS, BRAF, HER2, or PIK3CA on ABI7500 Real Time PCR system (ThermoFisher, USA) following the manufacturer's instructions.

Statistical analysis
Statistical analysis was performed using IBM SPSS 24.0 software (IBM, USA). The genetic variation results and clinicopathological characteristics from each patient were summarized and the prevalence of positivity, negativity, or a specific subgroup was calculated by dividing the number of positive/negative/subgroup cases by the total number of patients. Relationship between different genetic variations and patient clinicopathological characteristics (age, gender, smoking history, alcohol consumption, histological type, and tumor grade) were then analyzed using χ 2 test and Fisher's exact test. Difference was considered statistically significant if P < 0.05. When statistical difference was observed among three or more subgroups (e.g. stage I, II, III, and IV), post hoc analysis was performed between each of the subgroups using χ 2 test and Fisher's exact test. When significant association was observed between different clinicopathological characteristics, relationship between genetic variations and clinicopathological characteristics were further analyzed in subgroups.

Patient characteristics
In all, there were 405 patients enrolled in this study, in which 1 patient was excluded due to treatment history of targeted therapies. Samples from the rest 404 treatment-naïve patients were finally included in our analysis. As shown in Table 1, the patients were almost evenly distributed in males and females in this cohort. Majority of the patients were non-smokers (68.6%), and did not consume alcohol regularly (87.4%). Adenocarcinoma is the dominant subtype of NSCLC (94.3%) in this cohort. The percentage of tumor grades were Grade 1, 30.7%; Grade 2, 13.9%; Grade 3, 17.1%; Grade 4, 38.4%. Pathological diagnosis of NSCLC was confirmed in all samples.
In the 9 patients carrying EGFR exon 20 insertion, 1 patient also had exon 19 deletion.
In other gene variations tested, the overall positive rate for ALK, ROS1, or RET fusion were 7.7% (31/404), with two samples holding both RET fusion and sensitive mutations in EGFR (exon 19 deletion, or L858R). Three patients carried both KRAS mutations and EGFR exon 19 deletion, and one patient had both BRAF mutation and EGFR exon 19 deletion. In addition, three patients carried both PIK3CA mutation and EGFR mutations (2 patients with exon 19 deletions, and 1 patient with both L858R and T790M).

Relationship between genetic variations in tumor and patient clinicopathological characteristics
As shown in Tables 3 and 4, older patients (age > median) were associated with significantly less prevalence of EGFR L858R mutation, compared to younger patients (age � median). In addition, male patients showed significantly lower prevalence of EGFR exon 19 deletion and L858R mutation but significantly higher prevalence of KRAS mutations, compared to female patients. Smoking was associated with significantly lower prevalence of EGFR L858R mutation and KRAS mutation. In the two subtypes of NSCLC, adenocarcinoma was significantly associated with higher prevalence of EGFR exon 19 deletion and L858R mutation, compared to squamous cell carcinoma. Significant association was also observed between tumor grade and ALK fusion, and post hoc analysis showed that ALK fusion was significantly more prevalent in grade II tumors compared to grade IV tumors (S1 Table). No significant difference was found between alcohol consumption and histological type of NSCLC, or gene variations in tumor.

Relationship between histological type, tumor grade, and demographics of patients
As shown in Table 5, male patients were significantly more likely to be smokers and alcohol consumers. Adenocarcinoma was significantly more prevalent in female patient and nonsmokers, compared to male patients and smokers, respectively. Significant association was observed between tumor grade and age, gender, smoking. Furthermore, advanced tumor was significantly more prevalent in older patients, male patients, smokers, and alcohol consumers.
Post hoc analysis showed that stage III and stage IV tumors were significantly more prevalent in older patients. Compared to stage I tumors, male patients and smoking patients were significantly more likely to have stage III and stage IV tumors (S1 Table).

Association between genetic variations, histological type, tumor grade, and gender or smoking history of patients
Since the percentage of smokers was much higher in male patients than in females, we further studied the relationship among histological subtype, tumor grade, genetic variations, and gender and smoking history of patients. As shown in Table 6, in both male and female patients, smokers are significantly more likely to have squamous cell carcinoma than adenocarcinoma. In male patients, tumors from smokers were significantly less likely to carry EGFR L858R mutation. Interestingly, in non-smokers, male patients were more likely to have squamous cell carcinoma than females. In addition, in non-smokers, more male patients were positive in KRAS mutations than female patients. However, in non-smokers, more female patients were positive in EGFR L858R mutation than male patients. Similarly, since histological type was significantly associated with gender of patients, we further investigated the relationship among histological type, gender, and genetic variations. As shown in Table 7, in adenocarcinoma, female patients had significantly higher prevalence of EGFR exon 19 deletion and L858R compared to male patients. On the other hand, in male patients, adenocarcinoma was associated with higher prevalence of EGFR exon 19 deletion and L858R, compared to squamous cell carcinoma.

Discussion
Several targeted therapies have been approved for the treatment of NSCLC harboring certain genetic variations in EGFR, ALK, ROS1, or BRAF [5][6][7][8][9] and it is required to determine the existence of those genetic variations in tumor before targeted therapies are given to patients. Better understanding of the prevalence of those drug targets in NSCLC and their relationships may provide more guidance to clinical practices. Our study investigated common genetic variations in EGFR, fusions in ALK or ROS1, and BRAF V600X mutations in a cohort of NSCLC patients. For EGFR, there are overall 50.7% of NSCLC patients harboring sensitive genetic variations to anti-EGFR therapies (including exon 19 deletion, L858R, L861Q, G719X, and S768I). This is higher than the percentage (32.3%) reported by a previous systemic review by Zhang et al. [19], which could be due to the different ethnicity of the patient cohorts in the two studies (Chinese patients versus mixed ethnicity of patients), or different sensitivity of measurement methods used in the two studies (ARMS versus a mixture of measurement methods). In all, there were 4.9% (10/205) of those patients holding genetic variations resistant to anti-EGFR therapy, including 2 patients with EGFR T790M mutation, 1 patient with EGFR exon 20 insertion, 3 patients with KRAS mutations, 1 patient with BRAF V600X mutation (BRAF V600E was also indicated as resistant mechanism for Osimertinib in a previous study [22]), and 3 patients with PIK3CA mutations. Prevalence of ALK, ROS1, or RET fusions was relatively lower (7.7%), and only 1.2% of the patients had BRAF mutations. HER2 exon 20 insertion was found in 3.5% of the patients, which is higher than the prevalence (2.2%) as reported in previous study by Arcila et al. [17], possibly due to different screening strategy. Similar to the previous study, HER2 exon 20 insertion did not co-exist with other genetic variations in EGFR, ALK, ROS1, RET, or BRAF, indicating a need for targeted treatments for this specific subgroup of patients. Different from the previous study by Arcila et al. [17] which find two cases with concurrent PIK3CA mutations and HER2 exon insertion, we did not find any co-existing PIK3CA mutations in our patient cohort. MET exon 14 skipping mutation was found in 1.5% (6/404) of the patients in our cohort, which is similar to previous reports (1.6~2.6% of NSCLC) [18]. In our cohort, MET exon 14 skipping mutation did not co-exist with other genetic variations. Similarly, previous studies also showed no co-existing MET exon 14 skipping mutation with other driver gene variations, except amplifications of EGFR in 4 cases [18].
In our results, EGFR L858R mutation was more commonly seen in younger patients, which is conflicting with previous reports by Zhang et al. [20], possibly due to different cut-off value for younger and older patients (age of 50 in Zhang et al. versus median age used for cut-off value in our study). EGFR exon 19 deletion and L858R mutation were more commonly observed in female patients and adenocarcinoma. Similar observation has been reported between L858R mutation and gender, smoking, and histological type in a Japanese cohort [23] and a Polish cohort [24]. Further subgroup analysis confirmed the increased prevalence of those genetic variations in female patients in adenocarcinoma subgroup, and also in adenocarcinoma in male patients, which was not performed in previous studies. Similar to previous findings [23,25,26], smokers were more commonly found to have squamous cell carcinoma and KRAS mutation, and less likely to have EGFR L858R mutation in our cohort, and those associations between smoking and histological type, EGFR L858R mutation (but not KRAS mutation) were also observed after standardization of gender. Similarly, squamous cell carcinoma, KRAS mutations, and EGFR L858R mutation were more commonly found in males than females. Interestingly, after further standardization of smoking status, we found that in non-smokers, male patients were more likely to have squamous cell carcinoma and KRAS mutations and less likely to have EGFR L858R mutation than female patients. Previous study by Xue et al. [26] and Ramlau et al. [24] found that KRAS mutation was more common in male patients in a Chinese cohort, and EGFR L858R mutation was more common in female patients in the Polish cohort, but further standardization of smoking status was not performed in the study. Study by Stapelfeld et al. [27] reported conflicting results that KRAS mutations were more frequently found in women than in men, possibly due to the different race of patients (Caucasian) investigated in that study.
In summary, our study investigated the prevalence of targeted therapy-related genetic variations in NSCLC, as well as the relationship between those genetic variations and clinicopathological characteristics of patients. Compared to previous studies, our results showed overall similar prevalence of the investigated genetic variations in NSCLC. In addition, we also observed higher prevalence of EGFR exon 19 deletion and L858R both in female patients and in adenocarcinoma. Different from previous studies, we further looked into subgroups and confirmed those observations in female patients in adenocarcinoma subgroup and in adenocarcinoma in male patients, which indicate that the observations above were not due to the crosslinkage between gender and histological type but independent association between gender/histological type and the genetic variations. Furthermore, we also observed that smokers were more frequently to have squamous cell carcinoma and KRAS mutations, and less frequently to have EGFR L858R. Similarly, different from previous studies, we also performed subgroup analysis and confirmed those observations (except KRAS mutations) after standardization of gender, further confirming the association between histological type, EGFR L858R mutation and smoking. Our findings further confirmed the association between gender, histological type, smoking, and sensitive genetic variations of anti-EGFR therapies, which may help clinicians and clinical investigator in the patient selection of currently-available or potential targeted therapies. Limitations of this study could be the small patient cohort which may lead to potential statistical bias. Investigations using large patient cohort are required to further confirm the findings.