Genetic polymorphism of the interferon-γ gene in cervical carcinogenesis

Abstract

Beyond human papillomavirus (HPV) infection, host genetic factors may contribute to cervical carcinogenesis. This study aims to test the hypothesis that CA-dinucleotide repeat polymorphism in the first intron of the interferon-gamma (IFN-γ) gene is associated with HPV-initiated cervical carcinogenesis. A hospital-based case-control study including patients with low-grade squamous intraepithelial lesions (LSILs; n = 93), high-grade squamous intraepithelial lesions (HSILs; n = 123) and invasive carcinomas (n = 153) of the uterine cervix, as well as 1:1 age-matched controls, was conducted. The IFN-γ genotype was determined by PCR and capillary electrophoresis with internal standards. HPV genotype was determined by consensus PCR and reverse line blot hybridization. Genotypes containing the 12 or 14 allele (12 or 14 CA repeats) were significantly more common in patients with HSILs than in controls (46% vs. 22%; OR = 3.0; 95% CI = 1.7–5.2; p < 0.0001). In contrast, genotypes containing 13 and 18 were significantly more common in controls than in patients with HSILs (76% vs. 53%; OR = 0.3; 95% CI = 0.2–0.6; p = 0.0001) or squamous cell carcinomas (74% vs. 63%; OR = 0.6; 95% CI = 0.4–1.0; p = 0.037). The frequency of the 12 and 14 genotypes increased significantly in accordance with the severity of cervical carcinogenesis (p_{test for trend} = 0.0002), whereas the 13 and 18 genotypes showed the opposite trend (p_{test for trend} = 0.007). Comparing IFN-γ genotype and HPV status, 18-containing genotypes were more frequently found in HPV⁺ LSILs, and 12-containing genotypes were less frequently found in HPV⁺ HSILs. Compared with non-13 genotypes, 13 genotype HSILs were more frequently infected with HPV58 (70% vs. 45%) and less frequently infected with HPV18 (0% vs. 16%; p= 0.007). Genetic polymorphism of the IFN-γ gene is associated with individual susceptibility to cervical carcinogenesis. This polymorphism correlates with HPV infection in a disease- and type-specific manner. © 2004 Wiley-Liss, Inc.

Cervical cancer remains the major cause of cancer morbidity and mortality for women in developing countries. The causal role of high-risk human papillomavirus (HPV) in cervical carcinogenesis is beyond reasonable doubt.1 However, most HPV infections are transient; only type-specific persistence contributes significantly to the development of cervical neoplasia.2 The immune response to HPV infection and the mechanism for its clearance remain elusive. Several lines of evidence suggest that cell-mediated immune responses are important in controlling both HPV infections and HPV-associated neoplasms.3 The genetic variations of the host immune system to raise effective immune responses against HPV-derived antigens may contribute to the various outcome of HPV-induced cervical carcinogenesis.

In patients with high-grade squamous intraepithelial lesions (HSILs) and invasive cervical cancer (CC), peripheral lymphocytes have deficient cell-mediated immune response to HPV E6, E7 and L1 peptides.4, 5 Lymphoproliferative responses to specific HPV E6 and E7 peptides appear to be associated with the clearance of HPV infection and regression of HSIL.6 Reduced local cellular immunity has also been demonstrated in invasive cervical cancer as compared with premalignant lesions.7 The development of cervical cancer may therefore represent a failure in the local barrier of cell-mediated immunity.

Interferon-gamma (IFN-γ), produced mainly by activated lymphocytes and natural killer cells, is one of the dominant cytokines that polarize helper T cells toward the Th 1 phenotype and inhibit the development of Th 2 cells.8, 9 The polarized type 1 cellular immunity plays a decisive role in the host defense against viral infection and tumor development.10, 11 Thus, high level of IFN-γ production is typically associated with effective host defense against viral infection such as HPV. Moreover, decreased intratumor expression of IFN-γ has been reported to be associated with the poor prognosis in patients with cervical cancer.12

On the other hand, in tumor-associated inflammatory cells, increase of IFN-γ production may contributes to tumor growth and progression.13, 14 It is also well known that chronic inflammation, such as persistent HPV infection and chronic inflammatory bowel diseases, can predispose an individual to cancer.2, 15, 16 Therefore, the role of IFN-γ in tumor development is cell-dependent and complex.

Genetic polymorphisms in the IFN-γ gene17, 18, 19 have been shown to be associated with differing amounts of IFN-γ secretion17 and hence may be of biologic importance. The CA-dinucleotide polymorphism in the first intron of the IFN-γ gene has been shown to be associated with some autoimmune diseases.20, 21 The polymorphic responses of the IFN-γ gene in host encountering HPV infection may influence an individual's susceptibility to cervical cancer development.

The present study tests the hypothesis that dinucleotide polymorphisms in the first intron of the IFN-γ gene are associated with susceptibility to HPV-induced cervical carcinogenesis.

Material and methods

Patients

A hospital-based case-control study was conducted. The study included patients with low-grade squamous intraepithelial lesions (LSILs; n = 93), HSILs (n = 123) and invasive squamous cell carcinomas (SCCs; n = 153) of the uterine cervix, diagnosed and treated at Tri-Service General Hospital, Taipei, Taiwan, between 1993 and 2000. Cytology, histology and clinical data for all patients were panel-reviewed by a group of staff members including colposcopists, cytologists and pathologists to reach a final diagnosis. All patients were investigated and managed under a standard protocol for cervical neoplasia at the same hospital. Age-matched (1:1; ± 3 years) controls were invited from healthy women attending the routine Pap screening during the same period. Exclusion criteria included previous pregnancy, chronic or acute viral infections, history of cervical neoplasia, skin or genital warts, immune-compromised conditions, other cancers and operations on the uterine cervix. The study was approved by the institutional review board of the Tri-Service General Hospital. Informed consent was obtained from each patient and control subject.

Clinical specimens and extraction of genomic DNA

Specimens from patients and control subjects were collected under an established protocol for tissue procurement and banking.22 Briefly, tumor tissues from patients with CC were freshly collected from surgical resections or biopsies. They were snap-frozen in liquid nitrogen and stored at −70°C until use. Part of each specimen was examined pathologically for diagnosis. Genomic DNA was extracted by proteinase K digestion and phenol-chloroform extraction as described previously.23 Cervical scrapings of patients with HSIL and white blood cells of control subjects were collected,22 and genomic DNA was extracted by using a Qiagen DNA extraction kit (Qiagen, Hilden, Germany) according to the manufacturer's protocol. Concentration of the genomic DNA was determined by using fluorescent illumination with PicoGreen dye (Molecular Probes, Eugene, OR).

Determination of IFN-γ promoter polymorphism

The differential lengths of CA repeats at the first intron of the IFN-γ gene were analyzed by PCR with fluorescently tagged primers, followed by capillary electrophoresis.17 In brief, 100 ng of DNA was amplified in a mixture containing 1 × standard PCR buffer, 3.5 mM MgCl₂, 200 μM each of deoxynucleotides, 1.25 units of Taq Gold DNA polymerase (Perkin-Elmer Cetus, Norwalk, CT) and 250 μM each of primers (forward, 5′-FAM-GCTGTCATAATAATATTCAGAC-3′; reverse, 5′-CGAGCTTTAAAAGATAGTTCC-3′). Forty cycles of amplification was performed (30 sec at 95°C, 30 sec at 66°C, 30 sec at 72°C), with an initial denaturation at 95°C for 10 min and a final extension at 72°C for 10 min. The specific size of the PCR products was examined by agarose gel electrophoresis with ethidium-bromide staining.

To determine the number of CA repeats, capillary electrophoresis was conducted from the PCR products together with the Genescan-350 TAMRA standard (ABI, Foster City, CA). The mixture was heated to 98°C for 4.5 min and then immediately chilled on ice. Samples (1–2 μL) of the mixture were loaded and separated electrophoretically in an ABI 310 DNA sequencer and the sizes were then analyzed by GeneScan version 2.1 software (ABI). Further confirmation of CA-dinucleotide polymorphisms18 was conducted by direct DNA sequencing.

Immunohistochemical staining of paraffin-embedded tissue and morphometric analysis

Five μm paraffin-embedded sections were cut onto aminopropyltriethoxysilane-coated slides, dewaxed in xylene and soaked through a series of gradient ethanol for dehydration. Tissues were pressure-cooked in 0.5 M EDTA buffer (pH 8.0) for 10 min to facilitate antigen retrieval. Following washes in TBS, endogenous peroxidase activity was quenched by incubation in 3% hydrogen peroxide in TBS for 10 min. Sections were then washed in TBS and blocked with 10% serum (Dako, Carpinteria, CA) in TBS for 1 hr.

Mouse monoclonal antibody CD3 (Dako) was preabsorbed overnight at 4°C with 10% BSA in TBS. Primary antibody was diluted 1:40 in TBS containing 1% BSA. One hundred μl of antibody was added to each section, and the slides were incubated at 4°C overnight in humidified chamber. The slides were then in washed in TBS, followed by biotinylated antimouse secondary antibody (Dako) for 15 min at room temperature. Following washing in TBS, streptavidin horseradish peroxidase (HRP; Dako) was added for 15 min. After 2 additional washes, the presence of HRP was detected with the substrate 3-amino-9-ethylcarbozol AEC (0.5 mg/mL; Sigma, St. Louis, MO). The reaction was stopped by rinsing in water, and slides were lightly counterstained with hematoxylin followed by repeat rinsing, then applying coverslips with glycerin/gelatin mounting medium.

In order to cover the representative areas of each specimen for morphometric analysis and comparison, 10 hypercellular infiltrated areas were randomly selected underneath the surface epithelium parallel to the basement membrane of HSIL.24 Grid intersection point counting of the infiltrated lymphoid cells in the connective tissue was performed at a magnification of 400×. An eyepiece of the light microscope having 2 sets of 11 crossing lines which consisted of 121 intersections was used for counting. Total numbers of positively staining cells in the grids from 10 hypercellular infiltrated areas were counted in every specimen with different genotypes. The difference of infiltrating T lymphocytes in HSILs with different genotypes was tested.

HPV detection and genotyping

The presence of HPV DNA in cervical scrapings (HSIL) and tumor tissue (CC) was detected by L1 consensus PCR followed by reverse line blot.25 Extracted DNA (100 ng) was PCR-amplified with biotin-labeled primers of HPV (PGMY) and β₂-microglobulin. Part of the amplified product was visualized by ethidium-bromide staining after agarose (1.0%) gel electrophoresis. The integrity of extracted DNA and HPV genotype were detected by hybridization with a strip containing probes for 27 HPV types and for the β₂-microglobulin control and visualized with streptavidin and alkaline phosphatase staining. DNA sequencing was used to verify novel HPV types beyond the detection spectrum of the hybridization. HPV genotypes were grouped into group 16, group 18 and group 58 according to the phylogenetic similarity described previously.26

Statistical analysis

The difference in mean ages between patients and controls, as well as morphometric analysis of immunohistochemistry staining, was evaluated by Student's t-test. Associations between IFN-γ genotype, allele frequencies and disease severity, or HPV status were analyzed by using a chi-square test and Fisher's exact test where necessary. Odds ratios and 95% confidence intervals were calculated. The α-level of statistical significance was set at p = 0.05 for all statistical tests.

Results

Allele frequencies of IFN-γ gene in patients and controls

Eight alleles ranging from 113 to 127 bp in length were observed at the first intron of the IFN-γ gene. These alleles include 11 (113 bp) to 18 (127 bp) CA-dinucleotide repeats (11–18; Table I). The allele frequencies in patients with HSILs differed significantly from that of controls (Table II). Of the 123 patients with HSILs, the 12-repeat (12) allele and 14-repeat (14) allele were present in 53 (22%) and 17 (7%) patients, compared with 29 (12%) and 2 (1%) of the healthy controls, respectively (OR = 2.1, 95% CI = 1.3–3.4, p = 0.004 and OR = 9.1, 95% CI = 2.1–40.0, p = 0.003). The 13-repeat (13) allele and 18-repeat (18) allele were present in 74 (30%) and 3 (1%) of the 123 patients with HSILs, compared with 109 (44%) and 13 (5%) of the healthy controls, respectively (OR = 0.5, 95% CI = 0.4–0.8, p = 0.001 and OR = 0.2, 95% CI = 0.1–0.8, p = 0.002). No significant difference was observed in allele frequencies between LSILs or SCCs and their controls. Comparing HSILs and SCCs together with the controls, the 12 and 14 alleles also demonstrated significant risk (OR = 1.4, 95% CI = 1.0–1.9, p = 0.036 and OR = 8.3, 95% CI = 2.5–28.0, p < 0.001, respectively), whereas the protective effect of 13 and 18 alleles were also significant (OR = 0.7, 95% CI = 0.6–0.9, p = 0.012 and OR = 0.4, 95% CI = 0.2–0.8, p = 0.008, respectively).

Table I. Allele Frequency of the IFN-γ Gene CA Repeat in Patients with LSIL, HSIL, or SCC and Age-Matched Controls

IFN-γ (CA)n1	Amplicon size (bp)	Normal	LSIL	Normal	HSIL	Normal	SCC
		n = 186		n = 246		n = 306
1 n represents the repeat number. 2 11 indicates 11-repeat allele. 3 Number in parentheses represents percentage.
112	113	1 (1)3	0 (0)	1 (0)	5 (2)	1 (0)	0 (0)
12	115	32 (17)	24 (13)	29 (12)	53 (22)	50 (16)	52 (17)
13	117	72 (39)	80 (43)	109 (44)	74 (30)	129 (42)	123 (40)
14	119	2 (1)	7 (4)	2 (1)	17 (7)	1 (0)	7 (2)
15	121	65 (35)	57 (31)	85 (35)	84 (34)	103 (34)	104 (34)
16	123	3 (2)	6 (3)	6 (2)	7 (3)	5 (2)	11 (4)
17	125	2 (1)	2 (1)	1 (0)	3 (1)	2 (1)	1 (0)
18	127	9 (5)	10 (5)	13 (5)	3 (1)	15 (5)	8 (3)

Table II. Allele Distribution and Odd Ratios of IFN-γ Gene CA Repeat in LSIL, HSIL, or SCC and Controls

Allele	Normal	LSIL	OR (95% CI)	Normal	HSIL	OR (95% CI)	Normal	SCC	OR (95% CI)	Normal	≥ HSIL2	OR (95% CI)
	n = 186			n = 246			n = 306			n = 552
1 Alleles other than 12 allele. 2 Includes HSIL and SCC.
12	32 (17)	24 (13)	0.7 (0.4–1.3)	29 (12)	53 (22)	2.1 (1.3–3.4)	50 (16)	52 (17)	1.0 (0.7–1.6)	79 (14)	105 (19)	1.4 (1.0–1.9)
^∧121	154 (83)	162 (87)	1	217 (88)	193 (78)	1	256 (84)	254 (83)	1	473 (86)	347 (81)
p-value	NS			0.0042			NS			0.0363
13	72 (39)	80 (43)	1.2 (0.8–1.8)	109 (44)	74 (30)	0.5 (0.4–0.8)	129 (42)	123 (40)	0.9 (0.7–1.3)	238 (43)	197 (36)	0.7 (0.6–0.9)
^∧13	114 (61)	106 (57)	1	137 (56)	172 (70)	1	177 (58)	183 (60)	1	314 (57)	355 (64)
p-value	NS			0.0012			NS			0.0117
14	2 (1)	7 (4)	3.6 (0.7–18)	2 (1)	17 (7)	9.1 (2.1–40.0)	1 (0)	7 (2)	7.1 (0.9–58.2)	3 (1)	24 (4)	8.3 (2.5–28)
^∧14	184 (99)	179 (96)	1	244 (99)	229 (93)	1	305 (100)	299 (98)	1	549 (99)	528 (96)
p-value	NS			0.0034			NS			0.0006
18	9 (5)	10 (5)	1.1 (0.4–2.8)	13 (5)	3 (1)	0.2 (0.1–0.8)	15 (5)	8 (3)	0.5 (0.2–1.2)	28 (5)	11 (2)	0.4 (0.2–0.8)
^∧18	177 (95)	176 (95)	1	233 (95)	243 (99)	1	291 (95)	298 (97)	1	524 (95)	541 (98)
p-value	NS			0.0197			NS			0.0075

Frequencies of IFN-γ genotypes in patients and controls

The distribution of IFN-γ genotypes with various allele combinations is shown in Table III. Seventeen genotypes were documented in which the genotype containing 15 and 13 alleles were more prevalent (38%). Genotypes containing the 12 or 14 allele (12/14) were significantly more common in patients with HSILs than in controls (46% vs. 22%; OR = 3.0; 95% CI = 1.7–5.2; p < 0.001; Table IV). In contrast, genotypes containing the 13 or 18 allele (13/18) were significantly more common in controls than in patients with HSILs (76% vs. 53%; OR = 0.3; 95% CI = 0.2–0.6; p < 0.001) and SCCs (74% vs. 63%; OR = 0.6; 95% CI = 0.4–1.0; p = 0.037). Further analysis revealed that the 12/14 genotypes carried significantly more risk of the development of HSILs/SCCs (OR = 1.8; 95% CI = 1.3–2.7; p < 0.001), whereas the 13/18 genotypes were protective (OR = 0.5; 95% CI = 0.3–0.7; p < 0.001). Meanwhile, the frequency of 12/14 genotypes increased significantly in accordance with the severity of cervical carcinogenesis (p_{test for trend} = 0.0002), whereas the 13/18 genotypes showed the opposite trend (p_{test for trend} = 0.0072; Fig. 1).

Table III. Genotype Combinations in LSIL, HSIL, SCC, and Controls in Taiwanese Population

Genotype		Normal	LSIL	Normal	HSIL	Normal	SCC
Allele 1	Allele 2	n = 93		n = 123		n = 153
11	11	0 (0)	0 (0)	0 (0)	1 (1)	0 (0)	0 (0)
12	11	0 (0)	0 (0)	0 (0)	1 (1)	0 (0)	0 (0)
14	11	0 (0)	0 (0)	0 (0)	2 (2)	0 (0)	0 (0)
15	11	1 (1)	0 (0)	1 (1)	0 (0)	1 (0)	0 (0)
12	12	5 (5)	2 (2)	4 (3)	7 (6)	6 (4)	5 (3)
13	12	9 (10)	7 (8)	10 (8)	14 (11)	20 (13)	18 (12)
14	12	0 (0)	0 (0)	0 (0)	4 (3)	0 (0)	1 (1)
15	12	9 (10)	10 (11)	9 (7)	20 (16)	15 (10)	20 (13)
16	12	1 (1)	1 (1)	1 (1)	0 (0)	0 (0)	1 (1)
17	12	1 (1)	0 (0)	0 (0)	0 (0)	1 (1)	0 (0)
18	12	2 (2)	2 (2)	1 (1)	0 (0)	2 (1)	2 (1)
13	13	10 (11)	20 (22)	21 (17)	11 (9)	24 (16)	32 (21)
14	13	2 (2)	4 (4)	2 (2)	2 (2)	1 (1)	0 (0)
15	13	35 (38)	18 (19)	45 (37)	31 (25)	49 (32)	34 (22)
16	13	1 (1)	3 (3)	2 (2)	4 (3)	3 (2)	4 (3)
17	13	1 (1)	2 (2)	1 (1)	0 (0)	1 (1)	1 (1)
18	13	4 (4)	6 (6)	7 (6)	1 (1)	7 (5)	2 (1)

Table IV. Genotype Distribution and Odd Ratios of IFN-γ Gene CA Repeats in LSIL, HSIL, or SCC and Controls

Genotype	Normal	LSIL	OR (95% CI)	Normal	HSIL	OR (95% CI)	Normal	SCC	OR (95% CI)	Normal	≥ HSIL	OR (95% CI)
	n = 93			n = 123			n = 153			n = 276
1 12/14 represents genotypes containing 12 or 14 allele. 2 ^∧12/14 represents genotypes other than 12/14.
12/141	29 (31)	28 (30)	1.0 (0.5–1.8)	27 (22)	56 (46)	3.0 (1.7–5.2)	45 (29)	53 (35)	1.3 (0.8–2.1)	72 (26)	109 (39)	1.8 (1.3–2.7)
^∧12/142	64 (69)	65 (70)		96 (78)	67 (54)	1	108 (71)	100 (65)	1	204 (74)	167 (61)	1
p-value	NS			< 0.0001			NS			0.0008
13/18	67 (72)	64 (69)	0.9 (0.5–1.6)	94 (76)	65 (53)	0.3 (0.2–0.6)	113 (74)	96 (63)	0.6 (0.4–1.0)	207 (75)	161 (58)	0.5 (0.3–0.7)
^∧13/18	26 (28)	29 (31)	1	29 (24)	58 (47)	1	40 (26)	57 (37)	1	69 (25)	115 (42)	1
p-value	NS			0.0001			0.0367			< 0.0001

Figure 1. The trend of IFN-γ genotype distribution in the full spectrum of cervical carcinogenesis. Genotypes containing the 12 or 14 allele (12/14) show the tendency of increased risk for cervical carcinogenesis (p_{test for trend} = 0.0072) and vice versa for the protective role of genotypes containing the 13 or 18 allele (13/18; p_{test for trend} = 0.0002).

Immunohistochemical staining of infiltrating T lymphocytes in HSILs with different IFN-γ genotypes

The association of inflammatory cells infiltration and IFN-γ genotypes was tested. During an inflammatory response, T lymphocytes are the main source of IFN-γ secretion. The immunohistochemistry staining of CD3⁺ T lymphocytes adjacent to the epithelium of HSILs with different genotypes was quantified (Fig. 2). The mean counting of CD3⁺ infiltrating T lymphocytes in 100 areas of 10 HSILs with genotypes 12/12 or 12/14 is significantly higher than that in 120 areas of 12 HSILs with genotypes 13/13 or 13/18 (54.6 ± 3.9 vs. 48.9 ± 2.8; p < 0.001).

Figure 2. Representative immunohistochemical staining of CD3⁺ T lymphocyte in HSILs with different genotypes (400×). Morphometric measurement by the grid was shown.

Association of IFN-γ genotypes and HPV in different cervical neoplasms

Comparing the HPV infection status and distribution of IFN-γ genotypes (Table V), the 18-containing genotype was more frequently found in HPV-positive LSILs than HPV-negative LSILs (17% vs. 4%; p = 0.050), whereas in HPV-positive patients with HSILs, 12-containing genotype was less frequently found in HPV-positive cases than HPV-negative cases (30% vs. 55%; p = 0.008). There was no significant association between 13- and 14-containing genotype and HPV status in any cervical disease status, and no association between IFN-γ genotypes and HPV status was found in CC. With regard to HPV genotype, HSIL patients with 13-containing genotype are more frequently infected with HPV 58 (70%) and less commonly infected with HPV 18 (0%) than the non-13-containing genotype (45% and 16%, respectively; p = 0.07).

Table V. Association Between IFN-γ Genotype and HPV Status in Patients with HSIL or LSIL

	HPV status		HPV grouping
	Negative	Positive	Group 16	Group 18	Group 58
1 Genotypes other than the 12 allele-containing genotypes.
HSIL
12	22 (49)	23 (51)	11 (50)	2 (9)	9 (41)
^∧121	18 (25)	54 (75)	31 (61)	6 (12)	14 (27)
P value	0.008		NS
13	20 (34)	39 (66)	24 (30)	0 (0)	13 (70)
^∧13	20 (34)	38 (66)	12 (39)	5 (16)	14 (45)
P value	NS		0.0072
LSIL
18	8 (80)	2 (20)	0 (0)	0 (0)	2 (100)
^∧18	38 (46)	45 (54)	14 (31)	15 (33)	16 (36)
P value	0.0502		NS

Discussion

The present study reveals that genotypes with higher IFN-γ expression levels, the 12/14 genotype (homozygous or heterozygous), have increased risk of cervical carcinogenesis, which is against the hypothesis that higher IFN-γ expression may have a protective effect. Recent data have expanded the concept that chronic infection or inflammation is a critical component for carcinogenesis and tumor progression.13, 14, 16, 27 Cytokines in the microenvironment may provide growth signals and the fuel promoting genomic instability, malignant transformation and migration, which may partly explain the increased risk of cervical cancer in patients coinfected by herpes virus 2, Chlamydia, or other pathogens.28, 29, 30 Our previous study showed that a single nucleotide polymorphism (SNP) creating an IFN-γ response element in the Fas(CD95) promoter was associated with increased risk of cervical cancer, indicating differential activation-induced cell death (AICD) of lymphocytes as a susceptibility factor to cervical carcinogenesis.26 Whether the higher-expression allele of IFN-γ gene induces higher or earlier AICD of lymphocytes resulting in increased risk of cervical cancer requires further investigation.

Results of this study also expanded the spectrum of the dinucleotide repeat polymorphism of the IFN-γ gene. According to the literature, the 12 allele was completely linked with an SNP at the 5′ end of the dinucleotide repeats, in which the T allele creates an NF-κB-binding site and results in a higher level of IFN-γ production.18 In contrast, the 13 allele was reported to have a lower IFN-γ expression level.31 So far, the functional significance of alleles other than 12 and 13 remains undetermined. The present study demonstrates that the 14 and 12 alleles indicate risk of cervical carcinogenesis, whereas the 18 and 13 alleles are protective, which suggests that the 14 and 18 alleles may have higher and lower IFN-γ expression, respectively. This speculation is further supported by the immunohistochemical staining in the present study, which demonstrates that HSILs with 12/12 or 12/14 genotype have more infiltrating T lymphocytes adjacent to the basement membrane than those with 13/13 or 13/18 genotype.

In spite of the reported function of some alleles, the true significance of IFN-γ gene polymorphisms remains speculative. Although the presence of the 12 allele was correlated with higher levels of IFN-γ production from peripheral mononuclear cells stimulated with PHA, no such correlation was observed when intracellular IFN-γ levels were measured through flow cytometry.32 A recent study conducted in South Africa demonstrated ethnic differences in the T/A SNP, but there was no influence on the risk of cervical cancer.33 The results reported here imply that polymorphisms in the IFN-γ gene or a gene in close linkage disequilibrium may have an important role in cervical carcinogenesis. It is possible that the dinucleotide repeats in the IFN-γ locus are linked to one or more as yet unidentified polymorphisms in the region of chromosome 12q21. Identification of other polymorphisms, along with haplotype association, will further clarify the role of IFN-γ polymorphisms in cervical carcinogenesis. Checking along the stretch of chromosome 12q, interleukin-22 and interleukin-26 are in the close proximity of the IFN-γ gene. This cluster of cytokine genes may be associated with diseases resulting from immune dysregulation. Indeed, a recent report demonstrated the haplotype distribution of the IFN-γ/interleukin-26 gene on chromosome 12q is associated with a sex-based differential susceptibility to Th 1-type inflammatory disease.34 Further characterization of this locus in association with patients of cervical neoplasm may increase our understanding of the mechanism of HPV-associated carcinogenesis.

An interesting observation of IFN-γ gene polymorphism in relation to the full spectrum of cervical carcinogenesis was made in this study. The frequency of 12/14 genotypes increases dramatically from LSILs to HSILs, but not in SCCs, whereas the trend is the opposite in 13/18 genotypes. While genes involved in early cell transformation and late invasion stages of cancer development may be different, the same molecule may have different roles at different stages of carcinogenesis. It is speculated that the genetic polymorphism of IFN-γ may affect the process of infection-induced mitogenesis and carcinogenesis but not the invasion and metastasis of cancer. This result emphasized the discrete roles of cancer-related molecules in different stages of carcinogenesis.

The association of the status of HPV infection with IFN-γ gene polymorphism is also interesting. In HSILs, the 12-containing genotype is less commonly associated with high-risk HPV infection, which suggests that high IFN-γ expression may suppress the load of HPV and factors other than HPVs that induce IFN-γ expression may cause morphologic features indistinguishable from HSILs. The 13 genotype in HSILs is associated with a higher infection rate of HPV group 58, a prevalent group of HPV subtypes in Southeast Asia. Why the 13 genotype is associated with HPV group 58 in HSILs but is protective against the development of HSILs remains illusive. The clearance of different types of HPV in different IFN-γ genotypes warrants further investigation. Meanwhile, the distributions of microsatellite polymorphisms in the IFN-γ gene are different in different populations (Table VI). Some alleles, such as 15 and 18, are very rare in Caucasians, but are not uncommon in Asian populations. Populations with different IFN-γ genetic backgrounds may confer at least in part the different geographic distribution of HPV genotypes worldwide. Limited by the case number, the present study do not allow a detailed stratification of HPV types. Further large-scale studies and studies in different populations are warranted in order to understand the interaction between HPVs and host genetic polymorphism of the IFN-γ gene.

Table VI. Allele Frequencies of IFN-γ Gene CA Repeats in Different Populations

Allele	United Kingdom17 (n = 328)	Germany20 (n = 734)	Japan35 (n = 436)	Korea36 (n = 622)	Taiwan (n = 552)
12	0.48	0.43	0.09	0.12	0.16
13	0.43	0.45	0.58	0.52	0.42
14	0.04	0.07	0.01	0.01	0
15	0.05	0.03	0.27	0.32	0.34
16	0		0.05	0.01	0.02
17	0		0	0	0.01
18	0		0.01	0.01	0.05

Acknowledgements

The authors are grateful to Hui-Chen Wan and Rui-Len Huang for excellent technical assistance.