Array-comparative genomic hybridization reveals loss of SOCS6 is associated with poor prognosis in primary lung squamous cell carcinoma

Abstract

Background: Primary tumor recurrence commonly occurs after surgical resection of lung squamous cell carcinoma (SCC). Little is known about the genes driving SCC recurrence.

Methods: We used array comparative genomic hybridization (aCGH) to identify genes affected by copy number alterations that may be involved in SCC recurrence. Training and test sets of resected primary lung SCC were assembled. aCGH was used to determine genomic copy number in a training set of 62 primary lung SCCs (28 with recurrence and 34 with no evidence of recurrence) and the altered copy number of candidate genes was confirmed by quantitative PCR (qPCR). An independent test set of 72 primary lung SCCs (20 with recurrence and 52 with no evidence of recurrence) was used for biological validation. mRNA expression of candidate genes was studied using qRT-PCR. Candidate gene promoter methylation was evaluated using methylation microarrays and Sequenom EpiTYPER analysis.

Results: 18q22.3 loss was identified by aCGH as being significantly associated with recurrence (p = 0.038). Seven genes within 18q22.3 had aCGH copy number loss associated with recurrence but only SOCS6 copy number was both technically replicated by qPCR and biologically validated in the test set. SOCS6 copy number loss correlated with reduced mRNA expression in the study samples and in the samples with copy number loss, there was a trend for increased methylation, albeit non-significant. Overall survival was significantly poorer in patients with SOCS6 loss compared to patients without SOCS6 loss in both the training (30 vs. 43 months, p = 0.023) and test set (27 vs. 43 months, p = 0.010).

Conclusion: Reduced copy number and mRNA expression of SOCS6 are associated with disease recurrence in primary lung SCC and may be useful prognostic biomarkers.

Figure 1. Genomic copy number alterations in lung squamous cell carcinomas.

Plots of high-level amplifications (a) and deletions (b) in 62 lung SCCs from GISTIC analysis of aCGH data. X-axis shows the G score (top) and false discovery rate (q value; bottom) with a green line demarcating a false discovery rate of 0.05. Labels on the right denote the peaks of the most significantly altered regions.

Figure 2. aCGH copy number of genes within 18q22.3 demonstrating preferential loss in SCC recurrence.

The Y-axis represents the derived DNA copy number from aCGH log2 normalized data and the X-axis represents the recurrence phenotype. Mann-Whitney U test to was used to assess for any differences in copy number between recurrence phenotypes and p values<0.05 were deemed significant.

Figure 3. Quantitative PCR (qPCR) validation of array CGH identified candidate genes preferentially lost in SCC recurrence.

The Y-axis represents the derived DNA copy number from qPCR normalized to house-keeper genes, b-actin while the X-axis represents the DNA copy number derived from aCGH. Pearson's correlation coefficient was used to assess for any relationship between the copy number derived from the methods and p values<0.05 were deemed significant. The aCGH copy number of onlySOCS6 (a) was validated by qPCR.

Figure 4. Dot plot of qPCR-derived SOCS6 copy number (y-axis) is compared to the recurrence phenotype (x-axis) in the training (n = 62) and test set (n = 72) subjects.

Figure 4A and 4B represent training set and test set subjects respectively. Mann-Whitney U test to was used to assess for any differences in copy number between recurrence phenotypes and p values<0.05 were deemed significant.

Figure 5. Relationship between SOCS6 mRNA expression, copy number and recurrence phenotype.

Figure 5A and 5C are scatter plots of qRT-PCR derived SOCS6 mRNA expression (x-axis) and qPCR-derived SOCS6 copy number (y-axis) in training set (n = 62) (a) and test set (n = 72) (c) respectively. Figure 5B and 5D represent SOCS6 mRNA levels (y-axis) ad recurrence phenotype (x-axis) in the training and test sets respectively. Pearson's correlation coefficient was used to determine any association between SOCS6 qPCR derived copy number and mRNA levels. Mann-Whitney U test to was used to assess for any differences in copy number between recurrence phenotypes and p values<0.05 were deemed significant.

Figure 6. Kaplan-Meier curves of overall survival and SOCS6 qPCR-derived copy number in study subjects with follow-up duration of 5 years after surgical resection in the training set (n = 62) and test set (n = 72).

Figure 6A and 6B represent overall survival in all training set and TNM early stage subjects while 6C and 6D represent all test set and TNM early stage test set subjects. Censored values (+) indicate the last known follow-up time for those subjects still alive after surgical resection.