This article is mentioned in:

Introduction

Cyclooxygenase (COX) is the key enzyme required for the conversion of arachidonic acid to prostaglandins (PGs). Two COX isoforms have been identified and are referred to as constitutive COX (COX-1) and inducible COX (COX-2). COX-1 is constitutively expressed in many tissues and plays important roles in the control of homeostasis (1). On the other hand, COX-2 is an inducible enzyme that is activated in response to extracellular stimuli, such as growth factors and pro-inflammatory cytokines (2). Some investigators have demonstrated that COX-2 is constitutively overexpressed in a variety of epithelial malignancies, such as lung, breast, pancreas, colon, esophagus, and head and neck cancers, and COX-2 overexpression is usually associated with a poor prognosis (3–6).

Regulatory T cells (Tregs) were initially characterized as having a CD4+CD25+ phenotype, and these cells are thought to modulate the antitumor immune response (7). Tregs can suppress the activity of cytotoxic T cells through direct cell-to-cell contact or via the release of cytokines. The most specific Treg cell marker identified to date is a nuclear transcription factor known as Foxp3 (8,9). A high density of tumor-infiltrating Foxp3+ Tregs is reportedly associated with a higher risk of recurrence and a poor overall survival among patients with non-small cell lung cancer (NSCLC) (10). In 2010, we demonstrated that the tumor-infiltrating Foxp3+ Treg count (Foxp3 score) was positively correlated with the intratumoral COX-2 expression level and was associated with a poor recurrence-free survival period, particularly among patients with node-negative NSCLC (11).

Recently, numerous single nucleotide polymorphisms (SNPs) in the COX-2 gene have been identified, and these SNPs may contribute to differential gene expression and enzyme activities (12,13). In NSCLC, Bi et al (14) reported that a certain COX-2 SNP was a potential predictor of survival in patients with locally advanced NSCLC who were treated with chemoradiotherapy or radiotherapy alone. However, whether COX-2 genetic variants influence the function of COX-2 in NSCLC remains unclear. In the present study, we analyzed five types of COX-2 SNPs and evaluated whether the COX-2 SNPs were correlated with the intratumoral expression levels of COX-2, Foxp3+ Tregs and Ki-67 in NSCLC.

Patients and methods

Study population

Blood and tumor samples were obtained from 80 consecutive patients with NCSLC who underwent a complete resection with systematic lymph node dissection at Kawasaki Medical School Hospital between August 2011 and March 2013. None of the patients had received either radiotherapy or chemotherapy prior to surgery. This study was conducted with the approval of the institutional Ethics Committee of Kawasaki Medical School, and informed consent for the use of blood and tumor specimens was obtained from each of the patients. The histological diagnosis of the tumors was based on the criteria of the World Health Organization, and the TNM stage was determined according to the criteria established in 2009.

Genotyping of COX-2 SNPs

Blood samples were collected at the time of pre-operation. Genomic DNA was isolated from whole peripheral blood and was subjected to DNA amplification using a DNA Extractor WB-Rapid kit. The genomic DNA region containing the SNP was amplified using a polymerase chain reaction (PCR) performed using an Ampdirect Plus kit. The PCR primers used for the detection of the COX-2 −1195G/A, −1290A/G, −765G/C, 1759G/A and 8473T/C SNPs were as follows: −1195F, 5′-TCCACTTCTTTTCTGGTGTGTG-3′ and −1195R, 5′-CTGGGCTTATTGGGGCTAA-3′; −1290F, 5′-CCA CTTCTTTTCTGGTGTGTG-3′ and −1290R; 5′-GGGAGATT TTGACAGTTGGAA-3′; −765F, 5′-CCAAAATAATCCACG CATCA-3′ and −765R; 5′-TACCTTCACCCCCTCCTTG-3′; 1759F, 5′-GGGCTGTCCCTTTACTTCATT-3′ and 1759R, 5′-GACTCCTTTCTCCGCAACA-3′; 8473F, 5′-TGTCACAA GATGGCAAAATGC-3′ and 8473R, 5′-GCTTTTACAGGTG ATTCTACCCTATGA-3′, respectively.

DNA sequencing

The polymorphisms were analyzed using the ABI PRISM® 310 Genetic Analyzer (Applied Biosystems, Foster City, CA, USA). The results were analyzed using GeneMapper Software, ver. 4.0 (Applied Biosystems).

Immunohistochemical study

Immunohistochemical analyses were performed using resected paraffin-embedded lung cancer tissues. After microtome sectioning, the slides were processed for COX-2, Foxp3 and Ki-67 staining using an automated immunostainer (NexES; Ventana, Tucson, AZ, USA). The streptavidin-biotin-peroxidase detection technique using diaminobenzidine as a chromogen was applied. The primary antibodies were used according to the manufacturer’s instructions (COX-2, clone CX-294, 1/50 dilution; DakoCytomation; Foxp3, clone 22510, 1/100 dilution; Abcam; Ki-67, MIB-1, 1/100 dilution; DakoCytomation). The expression of each marker protein was examined and evaluated according to a previously reported original protocol. The slides were examined by an investigator who had no knowledge of the corresponding clinicopathological data.

For COX-2, the slides were scored according to the intensity of staining (0–3), and the percentages of cells with scores of 0 (0%), 1 (1–9%), 2 (10–49%), and 3 (50–100%) were determined. The immunohistochemistry (IHC) score (0–9) was defined as the product of the intensity and the percentage of stained cells. COX-2 expression was judged as positive when the IHC score was ≥4 (groups 3 and 4) (Fig. 1A) (15).

Figure 1 Immunohistochemical staining of (A) cyclooxygenase-2 (COX-2), (B) regulatory T cells (Tregs) and (C) Ki-67. Magnification, ×200.

To evaluate Treg immunostaining, 10 high-power field (HPF) digital images of the tumor areas were selected, and the absolute number of Foxp3-positive lymphocytes in these 10 HPF digital images was determined. The number of immunostained Foxp3 cells was then determined by averaging the 10 HPF digital image cell counts, resulting in the Treg score (Fig. 1B) (16).

The labeling index of Ki-67 was measured by determining the percentage of cells with positively stained nuclei. Ki-67 expression was judged as positive when >10% of the cancer cell nuclei showed positive staining (Fig. 1C) (17).

Statistical analysis

All the statistical analyses were performed using the SPSS statistical package (version 17.0; SPSS, Chicago, IL, USA). The Chi-square test and the Fisher’s exact test were used to examine the association between COX-2 SNPs and various clinicopathological parameters and protein expression levels evaluated using IHC. The vascular score was presented as the mean ± SD, and the difference between groups was analyzed using an unpaired Student’s t-test. The significance level was P<0.05. A prognostic evaluation was performed using the disease-free survival (DFS) period. The DFS was defined as the time from surgical resection until lung cancer recurrence or non-lung cancer-related death. To explore the association between DFS and COX-2 SNPs, a Kaplan-Meier survival analysis was performed by stratifying significant predictor variables that had been identified using the COX proportional hazards model. Two-sided P-values of <0.05 were considered to be statistically significant.

Results

Patient characteristics

The patient characteristics are documented in Table I. The mean age of the 80 patients was 69.9 years, and 50 of the patients were male. The histological type was adenocarcinoma in 61 cases, squamous cell carcinoma in 17 cases, and other types in 2 cases. The median follow-up period was 24 months (range, 9–30 months).

Table I Patient characteristics.

Table I

Patient characteristics.

Characteristics	No. of patients	Percentage
Gender
Male	50	62.5
Female	30	37.5
Age, mean ± SD	69.9±9.6
Histology
Adenocarcinoma	61	76.3
Squamous cell carcinoma	17	21.3
Large cell carcinoma	1	1.2
Pleomorphic carcinoma	1	1.2
Pathological stage
IA	41	51.2
IB	19	23.8
IIA+IIB	12	15.0
IIIA+IIIB	8	10.0
Adjuvant chemotherapy
(+)	20	25.0
(−)	60	75.0

Relationship between the expression status of COX-2 and the Foxp3-positive lymphocyte count

An immunohistochemical study showed that COX-2 was positive in 27 cases and negative in 53 cases. In the COX-2-positive group, the mean Treg score was 9.22. Conversely, in the COX-2-negative group, the mean Treg score was 3.47. The Treg score was significantly and positively correlated with the COX-2 expression level (P<0.001).

Associations between genotypes and clinicopathological findings

The associations between the COX-2 genotypes and the clinicopathological findings are shown in Table II. For the −1195G/A polymorphism, the AA genotype was observed in 53 cases and the GA/GG genotype was observed in 27 cases. Pleural invasion was significantly higher in the AA group than that in the GA/GG group (P=0.040). For the 1759G/A polymorphism, the GG genotype was observed in 56 cases and the GA/AA genotype was observed in 24 cases. The GA/AA group contained more patients who were over 70 years of age than the GG group. For the other genotypes, however, no significant correlations were found between the COX-2 genotypes and the clinicopathological findings.

Table II Association between COX-2 genotypes and clinicopathological findings.

Table II

Association between COX-2 genotypes and clinicopathological findings.

	−1195G/A			−1290A/G			−765G/C
Factor	AA	GG+GA	P-value	AA	GG+GA	P-value	GG	CC+GC	P-value
Age (years)			0.943			0.651			0.459
<70	24	12		27	9		28	8
≥70	29	15		31	13		31	13
Gender			0.360			0.698			0.646
Male	35	15		37	13		36	14
Female	18	12		21	9		23	7
Histology			0.378			0.895			0.994
Adenocarcinoma	42	19		44	17		45	16
Squamous cell carcinoma	9	8		12	5		12	5
Others	2	0		2	0		2	0
Pleural invasion			0.040			0.822			0.624
Negative	31	22		38	15		40	13
Positive	22	5		20	7		19	8
Vascular invasion			0.686			0.291			0.779
Negative	30	14		34	10		33	11
Positive	23	13		24	12		26	10
Nodal status			0.116			0.985			0.934
N0	48	21		50	19		51	18
N1/N2	5	6		8	3		8	3
	1759G/A					8473T/C
Factor	GG		AA+GA	P-value		TT	CC+TC		P-value
Age (years)				0.019					0.503
<70	29		6			27	9
≥70	27		18			30	14
Gender				0.614					0.848
Male	34		16			36	14
Female	22		8			21	9
Histology				0.863					0.788
Adenocarcinoma	43		18			45	16
Squamous cell carcinoma	11		6			12	5
Others	2		0			2	0
Pleural invasion				0.642					0.242
Negative	38		15			40	13
Positive	18		9			17	10
Vascular invasion				0.281					0.862
Negative	33		11			31	13
Positive	23		13			26	10
Nodal status				0.620					0.907
N0	49		20			49	20
N1/N2	7		4			8	3

[i] COX-2, cyclooxygenase-2.

Associations between genotypes and COX-2, Treg and Ki-67 expression levels

The associations between the COX-2 genotypes and the expression levels of COX-2, Tregs and Ki-67 are shown in Table III. No significant correlations were found between the COX-2 genotypes and the COX-2 score or the Ki-67 labeling index. For the −1195G/A polymorphism, however, the mean Treg score was 6.6 in the AA group and 3.0 in the GA/GG group. The mean Treg score was significantly higher in the AA group (P=0.003). Other polymorphisms showed no significant associations with the Treg score (Fig. 2).

Figure 2 Association between cyclooxygenase-2 (COX-2) single nucleotide polymorphisms (SNPs) and regulatory T cell (Treg) score. *P=0.003.

Table III Association between COX-2 genotypes and COX-2, Treg and Ki-67 expression.

Table III

Association between COX-2 genotypes and COX-2, Treg and Ki-67 expression.

	−1195G/A			−1290A/G			−765G/C
Factor	AA	GG+GA	P-value	AA	GG+GA	P-value	GG	CC+GC	P-value
COX-2 score	2.9	2.9	0.932	3.3	2.8	0.205	2.8	3.1	0.947
Treg score	6.6	3.0	0.003	6.0	3.8	0.063	5.5	5.2	0.382
Ki-67 labeling index	28.7	30.0	0.792	28.3	31.0	0.571	29.5	28.4	0.832
	1759G/A					8473T/C
Factor	GG		AA+GA	P-value		TT	CC+TC		P-value
COX-2 score	2.7		3.3	0.730		2.8	3.0		0.150
Treg score	5.2		5.8	0.653		5.8	4.5		0.108
Ki-67 labeling index	27.0		33.7	0.198		29.6	28.2		0.786

[i] COX-2, cyclooxygenase-2; Tregs, regulatory T cells.

Associations between COX-2 genotypes and Treg score according to the COX-2 expression level

Next, we evaluated whether the influence of the COX-2 genotype on the Treg score differed according to the COX-2 expression level (Table IV and Fig. 3). In the COX-2-positive expression group, a significant difference in the Treg scores was observed between the genotypes with the −1195G/A and −1290G/A polymorphisms. For the −1195G/A polymorphism, the mean Treg score was 11.2 in the AA group and 5.3 in the GA/GG group (Fig. 3A). The Treg score of the AA group was significantly higher than that of the GA/GG group (P=0.03). For the −1290G/A polymorphism, the mean Treg score was 11.4 in the AA group and 5.6 in the GA/GG group (Fig. 3A). The Treg score for the AA group was significantly higher than that for the GA/GG group (P=0.033). On the other hand, in the COX-2-negative expression group, a significant difference in the Treg scores was only observed for the −1195G/A polymorphism. The mean Treg score was 4.3 in the AA group and 1.8 in the GA/GG group (Fig. 3B). Similar to the COX-2-positive expression group, the Treg score of the AA group was significantly higher than that of the GA/GG group (P=0.011). These results showed that the −1195AA genotype group had a significantly higher Treg score than the GA/GG group, regardless of the intratumoral COX-2 expression level. For the other COX-2 SNPs, significant differences in the Treg scores were not found when the data were examined according to genotype.

Figure 3 Association between cyclooxygenase-2 (COX-2) single nucleotide polymorphisms (SNPs) and regulatory T cell (Treg) score according to COX-2 expression. (A) COX-2-positive group, (B) COX-2-negative group. *P=0.033 in A; *P=0.011 in B.

Table IV Associations between COX-2 genotypes and Treg score in regards to COX-2 expression.

Table IV

Associations between COX-2 genotypes and Treg score in regards to COX-2 expression.

		COX-2-negative group			COX-2-positive group
Genotype	N	n	Treg score	P-value	n	Treg score	P-value
−1195G/A				0.011			0.030
AA	53	35	4.3±5.1		18	11.2±9.0
GG+GA	27	18	1.8±1.6		9	5.3±4.2
−1290A/G				0.211			0.033
AA	58	41	3.8±4.6		17	11.4±9.3
AG+GG	22	12	2.3±3.1		10	5.6±3.6
−765G/C				0.346			0.797
GG	59	41	3.7±4.8		18	9.6±7.1
GC+CC	21	12	2.8±2.2		9	8.6±10.2
1759G/A				0.340			0.576
GG	56	38	3.1±3.9		18	9.8±8.5
GA+AA	24	15	4.5±5.3		9	8.0±7.6
8473T/C				0.521			0.088
TT	57	37	3.2±4.7		20	10.5±8.7
TC+CC	23	16	4.0±3.5		7	5.6±5.1

[i] COX-2, cyclooxygenase-2; Tregs, regulatory T cells.

Association between genotypes and DFS

The DFS period of the −1195AA genotype group was shorter than that of the GA/GG genotype group; however, the difference was not significant (Fig. 4). For the other COX-2 SNPs, no differences in the DFS period were observed when the data were examined according to genotype.

Figure 4 Disease-free survival of the patients with different cyclooxygenase-2 (COX-2) −1195G/A genotypes.

Discussion

In 2010, we demonstrated that the tumor-infiltrating Foxp3+ Tregs count (Treg score) was positively correlated with the intratumoral COX-2 expression level and was associated with a poor recurrence-free survival period, in particular among patients with node-negative NSCLC (11). In the present study, we examined whether COX-2 SNPs are associated with the expression of COX-2, Foxp3+ Treg and Ki-67 in 80 consecutive NSCLC patients who underwent resection. Our results showed that the AA genotype of the −1195G/A SNP in the COX-2 promoter region significantly contributed to the increased tumor-infiltrated Foxp3-positive lymphocyte count and indicated that NSCLC with an AA genotype for the −1195G/A SNP had a shorter DFS, compared with the GA/GG genotype.

A few studies have described different COX-2 SNPs and the associated clinical outcomes for several types of cancer. Li et al (18) reported that COX-2 SNPs were associated with the prognosis of patients with colorectal cancer. Bi et al (14) showed that genetic polymorphisms in COX-2 were associated with survival in patients with locally advanced NSCLC who had undergone chemoradiotherapy or radiotherapy alone They reported that the AA genotype of the −1195G/A SNP in the COX-2 promoter region significantly contributed to an unfavorable overall survival and progression-free survival, compared with the other genotype. Our results were similar to their results, but this study is the first to point out that the COX-2 polymorphism is associated with the Treg score in NSCLC.

The genotype frequencies for COX-2 −1195G/A SNPs in this study were equivalent to those in a previous study (19). Regarding the function of the −1195G/A polymorphism in COX-2, the −1195G to A change reportedly creates a c-MYB binding site in the COX-2 promoter region, thereby increasing the promoter activity (12). Compared with the −1195G-containing counterparts, the −1195AA carriers showed a significantly higher COX-2 expression level (12). In the present study, no significant correlations were found between the COX-2 expression level and the genotype of COX-2. However, the Treg score for the AA genotype of the −1195G/A polymorphism was significantly higher than that for the GA/GG group. Furthermore, the AA genotype group showed a significantly higher Treg score than the GA/GG group, regardless of the intratumoral COX-2 expression. These results suggest that the polymorphism may influence the inducing capacity of Tregs into NSCLC, as well as the prognosis of patients with NSCLC as a result of the infiltration of Tregs. To validate our hypothesis, the quantity or biological activity of prostaglandin E2 (PGE2) may need to be measured in lung tumor tissue, followed by an investigation of the correlation between COX-2 SNPs and intratumoral PGE2, which is converted from arachidonic acid in the presence of COX-2 as a catalytic substance and is considered to stimulate the infiltration of Tregs into tumor tissue (6).

Recently, a clinical trial by Cancer and Leukemia Group B demonstrated that among patients with increased COX-2 expression levels, survival was better among those who received treatment with a COX-2 inhibitor than among those who did not receive this treatment (15). Considering the present results, it may be necessary to investigate the COX-2 −1195 genetic polymorphism status when deciding upon a treatment strategy for NSCLC in the future.

This study has several limitations. First, the sample size may not be sufficiently large. The sample size of this study was smaller than that of a previous study (14) in which the correlation between the outcome of patients with unresectable NSCLC and the COX-2 polymorphism status was investigated. Second, the present study included only cases of resectable, relatively early-stage NSCLC and did not include any advanced NSCLC cases. Thus, our results may not be representative of NSCLC in general. Our results should thus be validated for a range of disease stages in the future.

In conclusion, our results showed significant differences in intratumoral Treg expression among NSCLC patients with different COX-2 −1195G/A genotypes. The tumor-infiltrating Treg count was significantly higher among the −1195AA genotype group, regardless of the COX-2 expression level. These findings suggest that the COX-2 −1195G/A polymorphism is a potential regulator of the infiltration of Tregs into NSCLC and that it may affect patient prognosis through its influence on Treg infiltration in NSCLC.

Acknowledgements

The authors thank Keiko Isoda for the technical assistance.

References