High Myopia Is Not Associated with the SNPs in the TGIF, Lumican, TGFB1, and HGF Genes

Abstract

purpose. Four single-nucleotide polymorphisms (SNPs) in the TGIF, lumican, TGFB1, and HGF genes have been declared to be associated with high myopia in Chinese living in southeast China, but none of them has been confirmed by additional studies. This study was conducted to verify the reported positive association results by analysis of subjects from the same region.

methods. DNA was prepared from venous leukocytes of 288 patients with high myopia and 208 control subjects. The four SNPs (rs2229336, rs3759223, rs1982073, and rs3735520) in the four genes were genotyped by restriction fragment length polymorphism (RFLP) analysis. The allele and genotype frequencies of these SNPs from patients and control subjects were compared by χ² test.

results. Polymorphism at rs2229336 was not detected in all 496 subjects. There were no statistically significant differences between patients and control subjects for the other three SNPs: rs3759223, rs1982073, and rs3735520.

conclusions. The study does not support the association of high myopia with alleles of rs2229336 in TGIF, rs3759223 in lumican, rs1982073 in TGFB1, and rs3735520 in HGF. These results provide a view contrary to those in previous reports. Reasonable criteria as well as replication should be the first priority for genetic association studies to avoid excessive expansion of false-positive results, especially for high myopia.

Myopia is a leading visual problem affecting an average of 30% (3%–84%) of people worldwide, with the highest prevalence in East Asians. ^{1 2 3 4} Its extreme form, high myopia, is the fourth most common causes of irreversible blindness due to the accompanying complications, including chorioretinal degeneration, retinal detachment, and glaucoma. ^{5 6}  

Several lines of evidence have demonstrated that genetic factors play an important role in the development of high myopia. ^{1 7 8 9} High myopia may be inherited as an autosomal dominant, autosomal recessive, or X-linked recessive trait. ^{8 10} Linkage studies have mapped at least eight loci (MYP1, MYP2, MYP3, MYP4, MYP5, MYP11, MYP12, and MYP13) responsible for high myopia with Mendelian inheritance, but none of the responsible genes has been identified. ^{5 11 12 13 14 15 16 17} On the other hand, high myopia is considered to be a complex trait, and genetic association studies are often used to search for genetic factors that may influence the susceptibility to high myopia. 

For Mendelian traits, the presence of defective genes usually completely determines the occurrence of disease, which has been proved to be correct for most loci and genes identified so far. However, for complex genetic traits, the occurrence of disease is predisposed by multiple genetic and environmental factors, and it is more complicated and difficult to identify the susceptibility factors. Genetic association studies have been widely used to search for genetic factors that contribute to complex traits, such as diabetes, inflammatory bowel disease, prostate cancer, breast cancer, asthma, coronary heart disease, and atrial fibrillation. ^{18 19 20 21 22 23 24 25 26 27 28 29} The most important progress for genetic association of common eye diseases is the identification of some single-nucleotide polymorphisms (SNPs) in the complementary factor H (CFH) gene that predispose to age-related macular degeneration (AMD). ^{30 31 32} Unfortunately, most genetic risk factors (∼95%) reported for many other complex traits have been false positives. ^{33 34 35 36 37} Therefore, replication is the cornerstone for genetic association. ^{35 38} It is important to be very careful in making conclusion from single association study between SNPs in a gene and the disease susceptibility before necessary replications have confirmed the findings. 

In recent years, several genetic association studies have been reported on high myopia in East Asian populations. Some SNPs, such as rs2229336 in the TGIF gene, rs3759223 in the lumican gene, rs1982073 in the TGFB1 gene, and rs3735520 in the HGF gene, have been reported to be associated with high myopia, ^{39 40 41 42} it is regrettable that none of these positive reports has been replicated by us or other groups, even though a few researchers treat these preliminary results as key evidence. 

We tested these four SNPs for association with high myopia based on analysis of subjects from the same population: Han Chinese living in Southeast China. Our results do not confirm association of the four reported SNPs with high myopia. The reasons of contradictory findings and the concern for such studies in the future are discussed. 

Materials and Methods

Subjects

The procedure for collecting subjects and obtaining informed consent was the same as previously described. ⁴³ This study adhered to the tenets of the Declaration of Helsinki and was approved by the Institutional Review Board of Zhongshan Ophthalmic Center. A total of 288 Chinese subjects with high myopia were recruited who met the following criteria: (1) bilateral refraction of −6.00 D or lower (spherical equivalent); and (2) no other known ocular or systemic diseases. A group of 208 control subjects met the following criteria: (1) bilateral refraction between −0.50 D and +1.00 D; (2) at least 18 years of age; and (3) no family history of high myopia. The refractive error was measured with cycloplegic autorefraction after mydriasis (compound tropicamide) for all eyes. An ophthalmic examination was performed by two of the authors (ophthalmologists QZ and XG). Genomic DNA was prepared from venous blood of patients and control subjects. 

SNP Genotyping

The four SNPs were genotyped by assaying restricted fragment length polymorphism (RFLP). Four pairs of primers (Table 1)were used to amplify DNA fragments encompassing the four SNPs, including rs2229336, rs3759223, rs1982073, and rs3735520 (http://www.ncbi.nlm.nih.gov/snp/, National Center for Biotechnology Information [NCBI], Bethesda, MD), which have been reported to be associated with high myopia. ^{39 40 41 42} The amplicons were digested by the restriction endonucleases AflIII, PsiI, MspA1I, and BglII, respectively, according to the instructions of the manufacturer (New England Biolabs, Dalian, China; Table 2 ). The digested products were analyzed by agarose gel or polyacrylamide gel electrophoresis. 

Statistical Analysis

Allele frequencies of all SNPs detected were assessed for Hardy-Weinberg equilibrium (HWE). Distribution of allele and genotype frequency in patients was compared with that in control subjects by using the χ² test or the Fisher exact test. P < 0.05 was used as the basic statistical significance based on previous reports. A Bonferroni correction was applied for the multiple tests, ⁴⁴ where the probability for significance was corrected from 0.05 to 0.0125 (α/4 = 0.05/4), since four SNPs were involved in the test. Statistical analyses were performed with commercial software (SPSS software ver. 10; SPSS Science, Chicago, IL). 

Results

A total of 496 unrelated subjects were recruited in this study. The cohort consisted of 288 unrelated cases with high myopia and 208 unrelated control subjects. The mean age of individuals with high myopia was 21.76 ± 16.24 years and that of control subjects was 27.32 ± 7.32 years. The four SNPs were genotyped in all subjects (Fig. 1 , Table 3 ). Allele frequencies and genotypes in patients and control subjects were listed in Table 3 . There was no significant deviation from HWE in any of the SNPs. Based on corrected P = 0.0125 (α/4 = 0.05/4), There is no statistically significant difference of allele frequencies for all four SNPs (rs2229336 in TGIF, rs3759223 in Lumican, rs1982073 in TGFB1, and rs3735520 in HGF) between 288 high myopia cases and 208 control subjects. The SNP rs2229336 in TGIF was monomorphic, where only allele T was detected in all 496 subjects (288 patients and 208 control subjects), which is entirely consistent with HapMap information for Asians (http://www.ncbi.nlm.nih.gov/SNP/snp_ref.cgi?rs=2229336/ provided in the public domain by the National Center for Biotechnology Information, Bethesda, MD) but is completely different from Lam’s report. ⁴⁰ Although the frequency of rs3759223 in lumican showed a difference between high myopia and control subjects (allele χ²= 5.094, P = 0.024; genotype χ²= 6.09, P = 0.048); however, the difference was not significant after Bonferroni correction with a corrected significant level of P < 0.0125. 

Discussion

Previously, four SNPs (rs2229336 in TGIF, rs3759223 in Lumican, rs1982073 in TGFB1, and rs3735520 in the HGF) were reported to be associated with high myopia in Chinese living in southeast China. ^{39 40 41 42} However, none of them was supported by subsequent studies. ^{45 46 47 48 49} As most of such subsequent studies were performed on SNPs other than the reported positive SNPs as well as on populations other than Chinese, it is unclear whether the contradictory findings resulted from differences in ethnic origin of the SNPs. In this study analyzing subjects from the same region, the T allele is present at rs2229336 in all 496 subjects, and the distribution of other three SNPs (rs3759223, rs1982073, and rs3735520) shows no significant difference between 288 subjects with high myopia and 208 control subjects. Besides, the odds ratio with the 95% confidence limits of the four SNP does not confirm association with high myopia. Therefore, we did not find evidence of association between high myopia and the four SNPs, which contradicts results in the original reports about their association with high myopia in Chinese. ^{39 40 41 42}  

TGIF (TGIF1, OMIM 602630; Online Mendelian Inheritance in Man; http://www.ncbi.nlm.nih.gov/Omim/ NCBI) was initially analyzed as a candidate gene for myopia because of its location within the MYP2 region ^{15 50} and its presumed role in ocular growth. ⁵¹ Evidence of a possible role of TGIF in myopia came only from a case–control study (71 subjects with high myopia versus 105 control subjects) of Chinese in Hong Kong, ⁴⁰ where only one SNP, rs2229336 (c.657T/G, synonymous), was of statistical significance after adjustment for multiple testing. However, association of this locus was not supported by four subsequent studies ^{45 47 48 49} where the SNP rs2229336 was not analyzed. In this study, only one allele, T at rs2229336, was detected in 496 Cantonese Chinese subjects living in Guangdong, Cantonese Chinese being the majority of Hong Kong residents. Our results and other previous replication studies indicate that the reported association of TGIF with myopia by Lam et al. ⁴⁰ is spurious based on the following points: First, none of the subsequent five studies support the association, especially data from the original families for MYP2 mapping, as well as data from the same region. In addition, allele G at rs2229336 is a variation usually seen in Africans (HapMap-YRI for Sub-Saharan Africans) but is not present in the 992 chromosomes in our study, or in HapMap-HCB (Han Chinese in Beijing), or in HapMap-JPT (Japanese in Tokyo) for Asians (http://www.ncbi.nlm.nih.gov/SNP/snp_ref.cgi?rs=2229336; National Center for Biotechnology Information). The high frequency of allele G in Hong Kong residents is most likely due to genotyping error, admixture, and population stratification. Moreover, the results of Lam et al. ³⁴ belong to the fifth-class association and are of low quality (small number of subjects and poor analysis of data). Finally, P < 0.05 or P < 0.01 for statistical significance is inadequately used for declaring genetic association (a detailed discussion follows). 

Lumican (LUM, OMIM 600616) was analyzed as a candidate gene for myopia because of its location within the MYP3 region, ¹⁴ its role in controlling sclera collagen fibrillogenesis, ⁵² and the ocular phenotype in LUM-FMOD double-null mice. ⁵³ This gene was excluded as a candidate gene for MYP3 based on analysis of the original MYP3 families, ⁵⁴ as well as sequence analysis of 96 unrelated Chinese patients with high myopia (Wang P, unpublished data, 2007). The only evidence for association of Lumican with human high myopia is from a case–control study (120 patients with high myopia and 137 control subjects) of Chinese at Taipei, ⁴² in which one SNP (rs3759223) located 4406 bp upstream from exon 1 was of statistical significance (P = 0.004528) but, of note, a strong deviation from Hardy-Weinberg equilibrium was present at rs3759223 in the case group (χ²= 21.1796). Based on these data, the observed genotype frequency and expected genotype frequency of rs3759223 could be calculated, where the former (C/C 0.51, C/T 0.27, T/T 0.22) is significantly different from the latter (C/C 0.41, C/T 0.46 T/T 0.13). Unfortunately, this kind of deviation is most likely to be a consequence of genotyping errors or admixture, ^{55 56 57 58} rather than a true difference. Our case–control study of this SNP in 288 patients with high myopia and 208 control subjects does not support an association of rs3759223 with high myopia (P > 0.05). 

TGFB1 (OMIM 190180) was analyzed as a candidate gene for myopia because of its differential expression in experimental chicken myopia ⁵⁹ as well as its functional relation with TGIF. ⁶⁰ In the initial study, only one (rs1982073 at codon 10, c.29T>C, p. Leu10Pro) of the 10 cSNPs in TGFB1 was analyzed and reported to be of significant difference between 201 patients with high myopia and 86 control subjects in a Chinese population at Taiwan (P < 0.001, uncorrected). ⁴¹ The reasons that only one SNP (rs1982073) was analyzed in this initial study were not given. At the same time, 10 SNPs (rs1982073 not included) and related haplotypes in TGFB1 were analyzed in 330 Japanese patients with high myopia and 330 control subjects, but none of them was associated with high myopia, ⁴⁶ and a further study on TGFB1 was suggested. Our results suggested that rs1982073 is not associated with high myopia in the Chinese. Therefore, it is highly likely that the initial report is a false-positive result. 

HGF (OMIM 142409) was analyzed as a candidate gene for high myopia because of its functional role observed in experimental myopia in animal models. ^{61 62 63} In a family-based association study, one SNP (rs3735520) and a haplotype of three SNPs in HGF were found be to be associated with high myopia by an analysis of 128 nuclear families with 133 myopic offspring (mean spherical equivalent ≤ −10.0 D). However, our case–control study does not support the association between rs3735520 and high myopia. Although several genes or their encoded proteins have been found to participate in experimental myopia in animal models, none of them has been demonstrated to contribute to human nonsyndromic myopia. 

To date, variations in several genes have been reported to associate with nonsyndromic high myopia, ^{39 40 41 42 64 65} but none of them have been confirmed by replication studies. ^{45 47 48 49 54 66} With regard to the current status of genetic studies of myopia, there are two key problems that have not been adequately addressed regarding the genetic traits of high myopia and the criteria for identifying genetic association. 

First, high myopia can be transmitted as Mendelian traits (autosomal dominant, autosomal recessive, and X-linked) or as complex trait. However, for high myopia in each individual, whether to define it as a Mendelian trait or a complex trait is still uncertain. In fact, this will affect not only the research design but also the final goal in searching for genes responsible for high myopia. Indeed, it is possible to enrich the patients in each group by selecting patients based on their family history, onset of the disease, degree of high myopia, and pathologic changes in the fundus. For example, Mendelian traits are most likely to be identified in individuals with extreme high myopia or high myopia that presents before school age. Conversely, moderate to high myopia occurring after school age is more likely to be influenced by multiple genetic and environmental factors (common disease). By using association study, it should be difficult to identify causative genes for Mendelian high myopia, which have been demonstrated to have high level of genetic heterogeneity. Therefore, careful selection of cases before initiation of an association study can enrich the samples for complex trait and therefore enhance the possibilities of identifying genes predisposing to myopia. 

On the other hand, although genetic association studies have been widely used to search for genetic factors involved in common diseases, many researchers may not have realized that most positive results of such association studies (∼95%) have turned out to be false positive because of misclassification, small sample size, admixture and population stratification, multiple testing, publication bias, and inappropriate criteria for statistical significance. ^{33 35 37 67 68 69} The most striking problem in performing a genetic association study is the improper interpretation of the probability (P). Consequently, many false-positive results have been published and inappropriately regarded as important findings by authors, reviewers, and editors. The commonly used criteria for statistical significance (P < 5 × 10⁻² or 1 × 10⁻²) are widely misused in genetic association studies. It has been estimated that the false positive is still rather high, even when using a more stringent criteria (P < 10⁻⁵) for genetic association. ^{68 69 70 71 72} To achieve a ratio of true positive to false-positive association of 20:1, P 5 × 10⁻⁵ or even lower has been suggested as a criteria for candidate gene association studies ^{34 68 69 70 71 72} and around 10⁻⁷ to 10⁻⁸ for genome-wide association studies. ^{18 73 74 75 76} Unfortunately, this problem has been neglected in association studies of myopia. Therefore, it should not be surprising to see that none of the associations of the four SNPs previously reported to contribute to high myopia were replicated in our study. We suggest that considering all those “positive findings” as false-positive or suggestive association until meaningful results can provide additional support. All consequent studies based on this level of “positive findings” should also be treated very carefully. ⁷⁷  

 Supported by Grant 30725044 from the National Science Fund for Distinguished Young Scholars, Grants 30572006 and 30772390 from the National Natural Science Foundation of China, Grant 20050558073 from the Ministry of Education of China, and Grant 2006Z3-E0062 from the Bureau of Science and Technology of Guangzhou.

 Submitted for publication July 8, 2008; revised October 30 and November 19, 2008; accepted January 26, 2009.

 Disclosure: P. Wang, None; S. Li, None; X. Xiao, None; X. Jia, None; X. Jiao, None; X. Guo, None; Q. Zhang, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

Primers Used to Amplify DNA Fragments Encompassing the Four SNPs

Gene	SNP	Primer Sequence	Product Length (bp)	Annealing Temp. (°C)
TGIF	rs2229336	F-5′-TCTGCCAGTCGGTCGGTGTG-3′	244	56
		R-5′-TTTGCCTGAAGCTCCATCTC-3′
Lumican	rs3759223	F-5′-CTCTGAAACGCACAAAAT-3′	187	56
		R-5′-AGAAAAACTCCACCTATCC-3′
TGFB1	rs1982073	F-5′-GGGCCTCCCCACCACACCAG-3′	203	62
		R-5′-CGATGCGCTTCCGCTTCACC-3′
HGF	rs3735520	F-5′-ACCCTAATAAAGCACAAGA-3′	285	56
		R-5′-CCTGCCTGATAAGTCCCTGAG-3′

Table 2.

 

Enzymes Used for RFLP Analysis

SNP	Genotype	Enzyme	Resulting Fragments (bp)
rs2229336	T/T	AflIII	244
	T/G		244/151/91/2
	G/G		151/91/2
rs3759223	C/C	PsiI	147/40
	C/T		147/119/40/28
	T/T		119/40/28
rs1982073	C/C	MspA1I	94/57/40/12
	C/T		106/94/57/40/12
	T/T		106/57/40
rs3735520	C/C	BglII	285
	C/T		285/211/74
	T/T		211/74

Figure 1.

 

RFLP analysis of SNP (A) rs2229336, (B) rs3759223, (C) rs1982073, and (D) vrs3735520. Lane M: 50-bp DNA marker. The different band patterns resulting from RFLP analysis of the SNPs are shown.

Table 3.

 

Allele Frequency and Genotypes of Four SNPs in High Myopia Patients and Control Subjects

Gene	SNP	Allele Frequency							χ2	P	High Myopia					Normal Control					χ2	P	OR	95% CI
				High Myopia			Normal Control
				a	b	a	b					a/a	a/b	b/b	a/a	a/b	b/b
TGIF	rs2229336	T (100)	G (0)	T (100)	G (0)	0	1	288	0	0	208	0	0	0	1	0	0
Lumican	rs3759223	C (26.74)	T (73.26)	C (33.89)	T (66.11)	5.09	0.024*	21	112	155	27	87	94	6.09	0.048*	0.708	0.495–1.012
TGFB1	rs1982073	C (56.25)	T (43.75)	C (59.13)	T (40.87)	0.82	0.365	92	140	56	75	96	37	0.937	0.626	0.896	0.566–1.420
HGF	rs3735520	C (58.68)	T (41.32)	C (56.01)	T (43.99)	0.71	0.401	96	146	46	64	105	39	0.791	0.673	1.214	0.759–1.942

 a/a: genotype with homozygous allele a; a/b: genotype with heterozygous alleles; b/b: genotype with homozygous allele b; OR: odds ratio; 95% CI: 95% confidence interval.

*  Statistical difference did not exist after Bonferroni correction by comparing each probability with the adjusted significant level (α/4 = 0.05/4 = 0.0125).

The authors thank all patients and normal control subjects for their participation.