Detection of E2A-PBX1 fusion transcripts in human non-small-cell lung cancer

Background E2A-PBX1 fusion gene caused by t(1;19)(q23;p13), has been well characterized in acute lymphoid leukemia (ALL). There is no report on E2A-PBX1 fusion transcripts in non-small-cell lung cancer (NSCLC). Methods We used polymerase chain reaction (PCR) to detect E2A-PBX1 fusion transcripts in human NSCLC tissue specimens and cell lines. We analyzed correlation of E2A-PBX1 fusion transcripts with clinical outcomes in 76 patients with adenocarcinoma in situ (AIS) and other subgroups. We compared mutation status of k-ras, p53 and EGFR in 22 patients with E2A-PBX1 fusion transcripts. Results We detected E2A-PBX1 transcripts in 23 of 184 (12.5%) NSCLC tissue specimens and 3 of 13 (23.1%) NSCLC cell lines. Presence of E2A-PBX1 fusion transcripts correlated with smoking status in female patients (P = 0.048), AIS histology (P = 0.006) and tumor size (P = 0.026). The overall survival was associated with gender among AIS patients (P = 0.0378) and AIS patients without E2A-PBX1 fusion transcripts (P = 0.0345), but not among AIS patients with E2A-PBX1 fusion transcripts (P = 0.6401). The overall survival was also associated with status of E2A-PBX1 fusion transcripts among AIS stage IA patients (P = 0.0363) and AIS stage IA female patients (P = 0.0174). In addition, among the 22 patients with E2A-PBX1 fusion transcripts, 12 (54.5%) patients including all four non-smokers, showed no common mutations in k-ras, p53 and EGFR. Conclusions E2A-PBX1 fusion gene caused by t(1;19)(q23;p13) may be a common genetic change in AIS and a survival determinant for female AIS patients at early stage.


Background
Lung cancer has been the leading cause of cancerrelated deaths in developed countries [1]. Non-small-cell lung cancer (NSCLC) accounts for around 80% of all lung cancer cases. Somatic events, such as point mutation, genomic rearrangements (e.g. translocation) and changes in copy number, usually cooperatively cause alterations in oncogenes, tumor-suppressor genes, and microRNA genes, and lead to the multi-step carcinogenesis [2,3]. During tumor initiation and/or progression, encoded oncogenic proteins activated by translocations or mutations can alter cell proliferation and/or apoptosis [3], resulting in transformation events. Fusion transcripts can be caused by chromosomal translocations and may occur more frequently in solid tumors than previously understood [2][3][4].

Patients and tissue specimens
A total of 184 patients were chosen in this study. All eligible patients without preoperative chemotherapy or radiation treatment underwent surgical resection of a primary NSCLC and had adequate mediastinal lymph node staging at UCSF between July 1997 and January 2007. Their clinical information of patients was summarized in Table 1. Information on clinical variables and patient follow-up were obtained from a prospectively maintained database including all subjects with banked tissue in the study. Patients consented to tissue specimen collection prospectively, and the study was approved by UCSF Human Research Protection Program Committee on Human Research. Tissue specimens were snap-frozen in liquid nitrogen at the time of the operation and stored in -150°C freezer.

Statistical analysis
The associations between the status of E2A-PBX1 fusion transcripts and clinical values were analyzed with Pearson Chi-square test and student t test for category variables and continuous variables, respectively. Median survival, 95% confidence intervals (CI) was calculated by Kaplan-Meier model and the log-rank test. A Cox regression model was used in AIS patients to assess the effects of E2A-PBX1 fusion transcripts, adjusted for gender, tumor stage, smoking status, race and Eastern Cooperative Oncology Group (ECOG) performance status. All p values reported were from two-sided tests. All analysis was performed by using SPSS 13.0. A p-value ≤ 0.05 was considered as significant.

Detection of E2A-PBX1 fusion transcripts in NSCLC
We performed nested PCR and detected E2A-PBX1 in 23/184 (12.5%) NSCLC patients as well as in positive control (RCH-ACV cell line [23,24]), but not in negative control (CEM cell line [23,24]) or adult normal lung ( Figure 1A). For the 23 patients with E2A-PBX1 fusion transcripts in their tumor tissues, we did not detect the E2A-PBX1 fusion transcripts in their paired adjacent normal tissues (figures not shown). We searched the sequencing results for all the PCR products in NCBI nucleotide/translated nucleotide/protein databases by BLAST (Basic Local Alignment Search Tool). The alignments showed that all the products we obtained were human E2A-PBX1 fusion gene with 100% identities. We also detected and confirmed E2A-PBX1 fusion transcripts in 3/13 (23.1%) NSCLC cell lines ( Figure 1B). Furthermore, we found that all the junction sites in these specimens were the same as that reported by Nourse J, et al. [5] (sequencing examples of the sequence around the junction site in one positive NSCLC tissue sample and cell line were was shown in Figure 1C).  transcripts in female smokers (14/68 (20.6%)) was significantly higher than that in female non-smokers (2/35 (5.7%)) (p = 0.048). The odds ratio for female smoker/nonsmoker was 4.278, and 95% CI was from 0.914 to 20.026, also suggesting that the expression of E2A-PBX1 fusion transcripts correlated with smoking status among female patients with NSCLC. The frequencies of E2A-PBX1 fusion transcripts in adenocarcinomas, squamous cell carcinomas, carcinoids and large cell carcinomas were 22/152 (14.5%), 0/18 (0%), 0/6 (0%), 1/4 (25%), respectively (p = 0.276) ( Table 1). Interestingly, the frequency of E2A-PBX1 fusion transcripts in patients with AIS (17/76 (22.4%)) was significantly higher (p = 0.006) than that in patients with invasive adenocarcinoma (5/76 (6.6%)) ( Table 1). The odds ratio for AIS/invasive adenocarcinoma was 4.092, and 95% CI was from 1.424 to 11.753, suggesting significant correlation between the expression of E2A-PBX1 fusion transcripts and patients with AIS. Moreover, the mean tumor size in patients with E2A-PBX1 fusion transcripts (4.1 ± 2.8cm) was significantly larger than that in patients without E2A-PBX1 fusion transcripts (3.2 ± 1.7cm) (p = 0.026) ( Table 1). There was no significant association between expression of E2A-PBX1 fusion transcripts and age, gender, race, stage, or recurrence status ( Table 1).

E2A-PBX1 fusion transcripts as a potential genetic biomarker in AIS
Mutations in K-ras, p53 and EGFR are commonly found in NSCLC [16][17][18]. We next screened for mutations in  (Table 3 and Table 4). 8 of these 22 (36.4%) patients with E2A-PBX1 fusion transcripts had K-ras mutations; 3 of the 22 (13.6%) patients had p53 mutations; only 1 of the 22 (4.5%) patient had EGFR mutation. K-ras and EGFR mutations in these patients were mutually exclusive to each other, same as previously reporeted [26]. Notably, 12 of the 22 (54.5%) patients had none of these three common mutations in lung cancer (Table 3 and Table 4). This subgroup of patients included all four non-smokers (patient # 4, 8, 13 and 22) and 8 stage I patients with AIS (Table 3). Among all thirteen NSCLC cell lines that we examined, only H1666 that was derived from a female patient with BAC (non-smoker) had no mutations in K-ras, p53 and EGFR (Note: the histology information of H1666 from ATCC still uses BAC). Taken together, our results suggest that the E2A-PBX1 fusion gene may be a genetic biomarker in NSCLC, especially in its subtype AIS.

Discussion
Somatic genetic changes have been believed to play important roles in human tumorigenesis, but the cancer type in which somatic rearrangement occurs is limited to leukemias, lymphomas and soft tissue tumors [2]. Overexpression of Notch3 was found to be associated with chromosome 19 translocation in lung cancer [27]. EML4-ALK fusion gene [28] and ETS fusion genes [29,30] exist in NSCLC and prostate cancer, respectively. It is still unclear whether chromosome aberrations are important in the initiation of epithelial tumorigenesis. AIS (formerly named BAC) is a subset of adenocarcinoma characterized by non-invasive growth along alveolar septae [19,25]. It is more prevalent in women, nonsmokers, and Asians [25]. Despite the lack of stromal, vascular, or pleural invasion, AIS is malignant and surgical resection is currently the mainstay of curative treatment. We previously discussed about a multi-step model of lung cancer development, especially AIS as carcinoma in situ [31]. Genetic changes can sequentially accumulate   [16][17][18]25], few are specific to AIS and "driver mutations" for AIS progression are still unidentified. Here in this study, we reported in NSCLC the expression of E2A-PBX1 fusion transcripts that have been well documented in leukemias [5][6][7][8][9][10][11][12][13][14][15]. This is the first report of detection of the E2A-PBX1 fusion transcripts in solid tumors. More interestingly, we observed that the E2A-PBX1 fusion transcripts were more frequently found in AIS than other subtypes of NSCLC, and the presence of E2A-PBX1 fusion transcripts were significantly associated with decreased overall survival in female and stage IA patients with AIS. These results suggest that the E2A-PBX1 fusion transcripts may play a critical role in AIS progression, especially for females and non-smokers. Supportive evidence also comes from our analysis of mutations in K-ras, p53 and EGFR that are common in NSCLC and considered as "driver mutations" [16][17][18]. Comparison of the mutational status of these genes in patients expressing the E2A-PBX1 fusion transcripts showed that approximately 55% patients examined in our study cohort were wild type in K-ras, p53 and EGFR. Majority of this subgroup were patients with AIS including all four non-smokers. Because E2A-PBX1 onco-protein has been proved to exhibit transformation potentials by transcribing target genes [5][6][7][8][9][10][11][12][13][14][15], we argue that E2A-PBX1 may serve as one "driver mutation" in AIS and play critical roles during initiation and progression of at least a subset of AIS. E2A-PBX1 may represent a new therapeutic target for NSCLC, especially AIS. Further investigation is needed to evaluate the function of E2A-PBX1 fusion protein, as well as its therapeutic and prognostic values and its correlation with treatment resistance in AIS.
In this study, we only examined in NSCLC specimens the conserved E2A-PBX1 fusion transcripts that are well documented in leukemias [5][6][7][8][9][10][11][12][13][14][15]. It is possible that other forms of E2A-PBX1 fusion transcripts also exist in NSCLC. TCGA (The Cancer Genome Atlas) data may be useful to analyze the frequency of E2A-PBX1 fusion transcriptions in NSCLC. Another limitation of this study is relatively small number of AIS specimens analyzed. Analysis of an independent large cohort of AIS is needed to validate our observation.

Conclusions
Our data demonstrated the presence of E2A-PBX1 fusion transcripts caused by t(1;19)(q23;p13) in lung adenocarcinomas, especially AIS. It may be a common genetic change in AIS and a survival determinant for female AIS patients at early stage. These data may be of significant clinical importance, because finding reliable genetic biomarkers for early-stage lung adenocarcinomas including AIS is becoming increasingly apparent for early identification and management of this deadly disease. Table 4 Summary of mutational analysis in NSCLC patients with E2A-PBX1 fusion transcripts Total (%) K-P-E-K + P-E-K-P + E-K + P + E-K-P + E+ K-P-E+ K + P-E+ K + P + E+