This retrospective study was approved by the Institutional Review Board of our institution (IRB No. 4-2012-0709). A total of 158 non-small cell lung cancer (NSCLC) patients who underwent pulmonary resection followed by adjuvant radiotherapy between 1996 and 2008 were identified. Squamous cell carcinoma was confirmed surgically in 71 patients, and clinical specimens were available from 36 patients. Of these, three patients were excluded from the analysis because of a lack of tumor tissue in their paraffin-embedded tissue blocks. Therefore, tumor samples from 33 patients were available for analysis. The patients’ medical records were reviewed retrospectively for evaluation of clinicopathological characteristics and survival outcomes. After radical resection, all patients were determined to have pathologic TNM stage III tumors according to the sixth edition cancer staging guidelines provided by the American Joint Committee on Cancer [9].

Fluorescent in situ hybridization (FISH) was performed for analysis of SOX2 gene amplifications on 4-μm-thick formalin-fixed paraffin embedded (FFPE) tissue sections. Briefly, the sections were deparaffinized in xylene (twice for 10 minutes each), followed by immersion in 100% ethanol (twice for 5 minutes each) before hybridization. Pretreatment was performed according to the Vysis protocol for FFPE tissue specimens. A SOX2-specific DNA probe (green) and a centromere 3-specific probe (red) were used (ZytoLight SPEC SOX2/CEN 3 Dual Color Probe, ZytoVision, Bremerhaven, Germany). The sections were then denatured and hybridized with ThermoBrite (Abbott Molecular, Des Plaines, IL) at 80°C for 5 minutes, followed by an overnight at 37°C. Post-hybridization washes were performed according to the Vysis protocol for FFPE tissue specimens (Abbott Molecular), and DAPI counterstaining was then performed. Semi-quantitative analysis of the SOX2 amplification status was performed by comparing the number of green signals (SOX2 target regions) to the number of red signals in each sample (reference regions). A non-amplified nucleus showed one green target signal for every corresponding red reference signal, with a green/red ratio of 1:1 (Fig. 1A). Cases containing 2-9 more green target signals than red signals in at least 30% of their tumor cells were defined as having low-level SOX2 amplification (Fig. 1B). Cases with an additional ≥ 10 green signals in a cluster-like formation were defined as having high-level SOX2 amplification (Fig. 1C). All tissue slides were analyzed under a 100× oil immersion objective lens using a fluorescence microscope equipped with the appropriate filters. At least 100 nuclei per case were assessed.

Four-micrometer-thick tissue sections were deparaffinized, rehydrated, and washed twice in buffer. The slides were incubated in hydrogen peroxide for 10 minutes to reduce nonspecific background staining due to endogenous peroxidases, and then washed four times in buffer. Primary antibodies against human SOX2 (1:200, R&D Systems, Minneapolis, MN) were then applied, and slides were incubated according to the manufacturer’s recommended protocols. The slides were washed four times in buffer, incubated with primary antibody enhancer for 20 minutes at room temperature, and then washed four times in buffer. Next, horseradish peroxidase polymer was applied to the slides, which were then incubated for 30 minutes at room temperature before washing four times in buffer, followed by incubation with hematoxylin and chromogen, washed four times in deionized water, and counterstained. The staining results were evaluated and each case was scored from 0 to 2, according to the intensity of nuclear staining in the tumor cells. The staining intensity of the normal bronchial epithelium served as the internal control and was given an arbitrary score of 1. Each tumor was compared to the internal control and given a score of 1 (moderate expression) (Fig. 1D) when the intensity was the same as that of the internal control, 2 (strong expression) (Fig. 1E) when stronger, and 0 (weak expression) (Fig. 1F) when weaker.

Disease-free survival (DFS) was estimated from the time of diagnosis to the time of initial tumor relapse (local recurrence or distant) or death from any cause. Overall survival (OS) time was measured from the time of diagnosis to death or last follow-up date. For univariate analysis, Kaplan-Meier survival analyses were used to estimate OS and DFS, and differences in the survival rates were compared using log-rank tests. A Cox proportional hazards model was used for multivariate analysis to evaluate prognostic factors influencing OS and DFS. Multivariate analysis was performed using backwards elimination to stay in the model. Hazard ratios (HR) are given with 95% confidence intervals (95% CI). Correlations between categorical variables were examined using χ² or Fisher exact tests. Continuous variables were compared to categorical variables using Mann-Whitney U tests. Statistical significance was defined as a p-value of < 0.05 for all analyses. SPSS ver. 20.0 (IBM Co., Armonk, NY) was utilized for all statistical analyses.

The median patient age was 66 years (range, 48 to 73years), and 32 patients were male (97%). Most patients had an Eastern Cooperative Oncology Group (ECOG) performance status of 0 or 1 (93.9%) and a positive smoking history (75.8%). Fourteen patients (42.4%) were classified as T3 or T4, and 27 patients (81.8%) as N2 or N3. A positive margin of resection was found in 11 patients (33.3%), lymphovascular invasion and perineural invasion in seven patients (21.2%) and three patients (9.1%), and extranodal extension in four patients (12.1%). The median total dose and fraction size of radiotherapy were 59.4 Gy (range, 50.4 to 69 Gy) and 1.8 Gy (range, 1.8 to 2 Gy), respectively. The patient characteristics are listed in Table 1.

SOX2 amplification was assessed using FISH and protein expression was determined using immunohistochemistry in lung squamous cell carcinoma patients (Fig. 1). High- and low-level amplification were observed in four (12.2%) and 18 patients (54.5%), respectively (Table 1). SOX2 gene amplification group was defined as low or high level amplification. Strong and moderate expressions were observed in eight (24.3%) and 18 patients (54.5%), respectively (Table 1). SOX2 overexpression was defined as a score of 1 or 2, and no overexpression was defined as a score of 0.

Analysis of the correlation between SOX2 gene amplification and protein expression showed significant association of SOX2 overexpression with SOX2 gene amplification (p=0.002) (Table 2). Next, we analyzed the association between several clinicopathological characteristics and SOX2 gene amplification and protein expression. However, no clinicopathological factor showed significant association with SOX2 gene amplification or protein expression (Table 3).

The median follow-up period for the surviving patients was 58 months (range, 5 to 102 months). The 5-year OS and DFS for all patients was 50.7% and 39.9%, respectively. In prognostic analysis using Kaplan-Meier curves, SOX2 overexpression was the only significant prognostic factor for OS (yes vs. no; median, 73 months vs. 8 months; p=0.01) (Table 4). Kaplan-Meier curves also showed that patients with SOX2 overexpression had significantly better OS, as shown in Fig. 2A. Age and total dose showed better OS with a statistically significant trend. However, SOX2 amplification did not show association with OS (Table 4, Fig. 2B). In multivariate analysis, SOX2 overexpression (HR, 0.1; 95% CI, 0.02 to 0.5; p=0.005) and age (HR, 0.33; 95% CI, 0.11 to 0.98; p=0.046) were the independent prognostic factors for OS (Table 4). In univariate analysis, age (< 65 vs. ≥ 65; median, 96 months vs. 12 months; p=0.02) and total radiation dose (> 59.4 Gy vs. ≤ 59.4 Gy; median, 96 months vs. 30 months; p=0.04) were the significant prognostic factors for DFS (Table 4). SOX2 overexpression showed a significant trend toward better DFS (p=0.08) (Table 4, Fig. 2C). SOX2 amplification also did not show association with DFS (Table 4, Fig. 2D). In multivariate analysis, SOX2 overexpression (HR, 0.15; 95% CI, 0.04 to 0.65; p=0.01) and total radiation dose (HR, 0.13; 95% CI, 0.02 to 0.7; p=0.02) were the independent prognostic factors for DFS (Table 4).