Genetic Epidemiological Analysis of the Keratoconus Genetic Model in the Chinese Keratoconus Cohort Study

Abstract

Purpose

Keratoconus (KC) is a corneal disorder characterized by progressive corneal protrusion and thinning. Our previous studies have demonstrated that genetic factors influence KC occurrence. The purpose of this study was to explore the genetic model of KC from the perspective of genetic epidemiology.

Methods

A total of 157 KC families, including 157 KC probands and their 445 first-degree relatives, from the Chinese Keratoconus (CKC) Cohort Study were included in present study. The genetic model of KC was evaluated by three genetic epidemiological analyses, including the Penrose method, simple segregation analysis, and complex segregation analysis, where the complex segregation analysis was conducted using the Statistical Analysis for Genetic Epidemiology package.

Results

The 157 KC families included 181 affected individuals, 384 unaffected individuals, and 37 unknown individuals. There are 24 individuals diagnosed with KC among the 445 first-degree relatives. The relative frequency calculated by the Penrose method was 33.188, which was close to 1/√q. In addition, the segregation ratio calculated by the simple segregation analysis was 0.046, which was less than 1/4. Furthermore, all the hypotheses of Mendelian, nontransmission and environmental model were rejected by complex segregation analysis. These results fully showed that KC is a disease of multifactorial inheritance.

Conclusions

This study identified that KC followed a pattern of multifactorial inheritance, which is helpful to provide initial guidance for prevention and management of the disease and points out a research direction for future research.

Keratoconus (KC) is a corneal disorder characterized by corneal ectasia and protrusion, progressive corneal thinning, and irregular astigmatism.¹ It mostly occurs in adolescence, progresses gradually and tends to steady at the age of 30 to 40 years.^2,3 The prevalence of KC is 0.138% worldwide.⁴ Additionally, the disease usually manifests as progressive vision loss and reaches the level of legal blindness in severe cases.⁴ As one of the main causes for visual disorder, the prevention and treatment of KC is still limited owing to a lack of understanding of its etiology and pathogenesis.^5,6

Although the etiology and pathogenesis of KC are not clear, there remain a large number of studies demonstrating a strong relationship between KC and genetic factors from various study perspectives. Multiple studies, including ours, investigated the prevalence of KC in first-degree relatives of KC patients and found that it ranges from 2.08% to 27.9%.^7–10 In addition, several studies identified a high heritability of corneal parameters in nuclear families with KC and in population-based studies.^11–13 Furthermore, a series of association studies and linkage analyses identified lots of susceptibility genes that are associated with KC.^14–16 The association between KC and genetic factors has been shown by previous researches; however, the genetic model of KC remains ambiguous, including the patterns of autosomal dominant, autosomal recessive, and non-Mendelian modes of inheritance.^17–20 As an interdisciplinary subject of genetics and epidemiology, genetic epidemiology aims to explore the etiology, distribution, and prevention and control measures of genetically related diseases in the population.^21,22 Genetic epidemiological analyses are generally considered to fall into three main components: familial aggregation analysis, heritability analysis, and segregation analysis.^23–25 Segregation analysis is a statistical methodology that estimate the best-fitting genetic model of disease according to the phenotypes of family members without molecular data,^25–27 which has a large theoretical value and research significance for guiding the prevention and management of disease and widely used in many disease, such as cancers,^28,29 systemic lupus erythematosus,³⁰ facial melasma,³¹ and molar–incisor hypomineralization.³²

In our previous studies, we investigated family aggregation and heritability of KC in the Chinese Keratoconus (CKC) Cohort Study and found that the prevalence in first-degree relatives was up to 8.77% and that there is high heritability of corneal parameters in nuclear families.^8,11 Next, exploring the genetic model of KC is our research priority, which is of great significance to understand the specific role of genetic factors in KC, explore the recurrence risk of relatives, and guide the prevention and control of the disease. Therefore, genetic epidemiological analyses, including the Penrose method, simple segregation analysis, and complex segregation analysis, were performed to explore KC genetic model in the CKC Cohort Study in the present study.^33–35

Methods

Study Participants

A total of 157 families, including 157 KC probands and their 445 first-degree relatives (parents and siblings), from the CKC Cohort Study were included in the current study with the aim of exploring the genetic model of KC.³⁶ The KC probands and their first-degree relatives who consented to participate in this study underwent detailed eye examinations including uncorrected visual acuity, IOP, slit lamp examination, and corneal tomography with Pentacam HR (Oculus, Lynnwood, WA, USA). The diagnostic criteria for KC are that the Pentacam corneal topography reveals an asymmetric bowtie pattern with or without skewed axes and a Belin Ambrosio enhanced ectasia total deviation index value of greater than 2.6, and slit lamp examination that reveals the presence of at least one clinical characteristic, such as Fleischer's ring, Vogt's striae, localized stromal thinning, conical protrusion, or anterior stromal scar.^37,38 The exclusion criteria for a KC family included parents with consanguineous marriage, study participant who underwent corneal refractive surgery, and syndromic diseases or corneal dystrophy. This study was conducted in compliance with the Declaration of Helsinki and was subject to the Ethics Committee of Henan Eye Hospital's approval (ethical approval number: HNEECKY-2019 (5)). The purpose and significance of this study were provided to all study participants, and informed consent was obtained from all participants.

Genetic Epidemiological Analyses

The Penrose Method

The Penrose method is an approach to assess genetic model by comparing the expected values for various modes of inheritance and the relative frequency.³³ The relative frequency refers to the prevalence rates in siblings (S) divided by the prevalence rates in general population (q). The criteria for the judgement are as follows: an S/q ratio close to 1/2q indicates an autosomal dominant inheritance; an S/q ratio close to 1/4q indicates an autosomal recessive inheritance; and an S/q ratio close to 1/√q indicates a multifactorial inheritance.

Simple Segregation Analysis

The goal of simple segregation analysis is to identify genetic model by comparing segregation ratio (P) to the expected proportion in particular genetic hypothesis.³⁴ The definition of segregation ratio is the observed proportion of affected siblings and offsprings, and the value of P is equal to (R − N)/(T − N). The variance of segregation ratio is Sp², and the value is equal to (P × q)/(T − N), and the standard error of SE(P) is equal to (Sp²)^1/2. In the formula, N is the total number of pedigrees, T is the total number of all offsprings, R is the total number of affected offsprings, and the value of q is equal to 1 − P. The following is a summary of criterions: a value of P between 1/2 and 3/4 indicates an autosomal dominant inheritance; a value of P between 1/4 and 1/2 indicates an autosomal recessive inheritance; and a value of P of less than 1/4 indicates a multifactorial inheritance.

Complex Segregation Analysis

Complex segregation analysis was carried out using the multivariate logistic regression model of the Statistical Analysis for Genetic Epidemiology package (Version 6.4, Case Western Reserve University, Cleveland, OH, USA).^35,39 A total of seven segregation models were estimated through the maximum likelihood estimation, including unrestricted general model and six other models with specific parameters. The likelihood ratio test statistic, which is calculated as −2(LnL_general − LnL_specific), was used to test a hypothesis about a specific mode of inheritance, and the statistics values were χ² asymptotically distributed. The degrees of freedom were defined as n − k, with n and k equal the number of independent parameters estimated in the general model and the specific models, respectively. The specific model was accepted when the P value was greater than 0.05. When more than one specific model was accepted, the one with the lowest Akaike information criteria, the presentation of −2LnL + 2k, was considered the best model.

Results

Demographic Characteristics of the Study Subjects

There were 602 individuals from 157 KC families, including 157 KC probands and 445 first-degree relatives, included in this study. Among the 157 KC probands, 122 individuals (77.707%) were male and 35 (22.293%) were female. The mean age at diagnosis of KC probands was 17.025 ± 3.943 years, and the male to female ratio was 3.486:1. Among the 445 first-degree relatives, 314 relatives were parents of KC probands and 131 relatives were siblings. The mean age of parents and siblings of KC probands were 44.866 ± 5.488 years and 16.965 ± 6.816 years, respectively.

The Prevalence of KC in the First-degree Relatives

The 445 first-degree relatives included 24 affected individuals, 384 unaffected individuals, and 37 unknown individuals. As shown in Figure 1 and Supplementary Table S1, the prevalence of KC in the first-degree relatives of KC probands is 5.393% (24/445). Additionally, the 24 affected first-degree relatives included 18 parents of KC probands and 6 siblings; the prevalence of KC in the parents was 5.732% (18/314) and the prevalence of KC in the siblings was 4.580% (6/131). These findings support the role of genetic factors in the development of KC.

Figure 1.

The prevalence of KC in first-degree relatives. The prevalence of KC among first-degree relatives was 5.732% in parents and 4.580% in siblings, with an overall prevalence of 5.393%.

The Genetic Model of KC Calculated by the Penrose Method

As mentioned, the prevalence of KC in the siblings (S) is 4.580%, the prevalence of KC in the populations (q) is 0.138%, and the S/q ratio is 33.188. The calculated S/q ratio is 33.188. As shown in Figure 2 and Supplementary Table S2, the S/q ratio is close to 1/√q (26.919), but far from 1/2q (362.319) and 1/4q (181.159), indicating that KC conforms with a pattern of multifactorial inheritance.

Figure 2.

The comparison of relative frequency and the theoretical values for various modes of inheritance. The relative frequency of KC is 33.188, which is close to 1/√q (26.919), but far from 1/2q (362.319) and 1/4q (181.159).

The Genetic Model of KC Calculated by Simple Segregation Analysis

As shown in Table 1, the total number of pedigrees (N) is 157, the total number of all offsprings (T) is 288, and the total number of affected offsprings (R) is 163. Using the formula of P is equal to (R − N)/(T − N), the value of segregation ratio (P) is 0.046, the value of SE(P) is 0.018, and the 95% confidence interval is 0.028 to 0.064. The value of P of less than 1/4 suggests a multifactorial inheritance mode of KC.

Table 1.

The Distribution of Pedigrees by Size and Number of Affected Offspring

The Number of Offsprings in a Pedigree	The Number of Pedigrees	The Number of All Offsprings	The Number of Affected Offspring
1	50	50	50
2	85	170	88
3	20	60	22
4	2	8	3
Total	157 (N)	288 (T)	163 (R)

N, the total number of pedigrees; R, the total number of affected offsprings; T, the total number of all offsprings.

Complex Segregation Analysis

As shown in Table 2, seven genetic models, including three non-Mendelian inheritance (general, nontransmission, or environmental) and four Mendelian inheritance (dominant, recessive, additive, or major gene), were fitted by the SEGREG program of the Statistical Analysis for Genetic Epidemiology. Compared with the general model, all Mendelian, nontransmission and environmental models were rejected through the likelihood ratio test. First, the rejection of the nontransmission (sporadic) and environmental models of inheritance indicated that genetic factors might play important roles in the pathogenesis of KC. In addition, the hypothesis of Mendelian models was also rejected, which revealed that the inheritance of KC followed a pattern of polygenic inheritance rather than single-gene inheritance.

Table 2.

Complex Segregation Analysis Results of KC Using the Statistical Analysis for Genetic Epidemiology SEGREG Program

	Non-Mendelian			Mendelian
Item	General	Nontransmission	Environmental	Dominant	Recessive	Additive	Major Gene
qA	0.982	–	0.858	0.039	0.971	0.036	0.039
τAA	0.109	–	= qA	1	1	1	1
τAB	0.361	–	= qA	0.5	0.5	0.5	0.5
τBB	0	–	= qA	0	0	0	0
βAA	−1.501	−0.588	−0.647	0.340	−0.703	0.748	0.340
βAB	−0.459	=βAA	−0.685	=βAA	=βBB	0.013	0.340
βBB	0.312	=βAA	1.503	−0.729	0.374	−0.722	−0.729
−2lnL	515.006	648.517	610.325	599.895	603.161	640.338	599.895
AIC	529.006	652.517	620.325	607.895	611.161	650.338	609.895
χ 2		133.511	95.319	84.889	88.155	125.332	84.889
df		5	2	3	3	2	2
P-value		<0.001	<0.001	<0.001	<0.001	<0.001	<0.001

−2lnL, −2 times the log likelihood L; AIC, Akaike information criterion; βAA, βAB, βBB, the baseline parameter for a subpopulation with the genotypes AA, AB, and BB; df, degrees of freedom; qA, frequency of allele A; τAA, τAB, τBB, transmission probability of the genotypes AA, AB, and BB.

Discussion

This study explored the genetic model of KC in 157 KC probands and their 445 first-degree relatives using genetic epidemiological analyses. Among the 445 first-degree relatives, the prevalence of KC in the parents is 5.732% (18/314) and the prevalence of KC in the siblings is 4.580% (6/131). Genetic epidemiological analyses, including the Penrose method, simple segregation analysis, and complex segregation analysis, were used for calculating the genetic model and all evidenced that KC is a multifactorial disease.

The etiology and pathogenesis of KC remained unclear until now, especially whether and how it is influenced by genetic factors. In the present study, the prevalence of KC in first-degree relatives was 5.393%, which was far greater than that in the general population.⁴ The high prevalence of first-degree relatives fully proves the importance of genetic factors in the occurrence of KC in present study. Current understandings have formed general consensus on the importance of genetic factors in the occurrence of the KC, but the genetic model of KC has not come to any uniform conclusion.^4,8,15 Cheng et al.¹⁷ conducted Sanger sequencing and cosegregation analysis on the suspected pathogenic variants in five Chinese families with KC, and found that KC was compatible with autosomal dominant inheritance in four families and autosomal recessive inheritance in one family. Kriszt et al.¹⁸ performed complex segregation analysis in 212 members of 60 sporadic KC families and suggested the inheritances of the Keratoconus Percentage Index, Keratoconus Severity Index, and Fourier 6-mm asymmetry indices are compatible with non-Mendelian major gene effects. Wang et al.¹⁹ suggested the inheritances of Keratoconus Percentage Index, a quantitative index, was an autosomal recessive model with major model effect through conducting complex segregation analysis in 95 KC families. Schmitt-Bernard et al.²⁰ evaluated the incidence of KC in family members of monozygotic twins and pointed out a recessive model of inheritance. Altogether, perhaps particular mutations lead to KC in a dominant manner, whereas others are recessive or polygenic, which illustrates that KC is a heterogeneous disease. Exploring the genetic model of KC plays important roles in elucidating the genetic heterogeneity and exploring the etiology, thereby facilitating the goal of precise diagnosis, prevention, and treatment of the disease. However, there is still no final verdict on a genetic model of KC. Previous researches have shown that KC follows a pattern of non-Mendelian inheritance, autosomal dominant inheritance, and autosomal recessive inheritance, but for the most part chiefly focused on the European populations.^17–20 Notably, there were very little research into the genetic model of KC in China. Considering the genetic heterogeneity in KC, it is essential to investigate the genetic model of KC in the Chinese population.

The progressive visual impairment of KC makes it a growing public health problem with serious implications for the economy and society.^40,41 As described elsewhere in this article, most previous studies paid much attention to the genetic model of KC in certain pedigrees that lose sight of the populations of patients with KC.¹⁷ To have a further comprehension of the genetic mechanisms of KC, thereby controlling and preventing the disease, exploring the genetic model of KC from the perspective of population genetics should be of high priority. With the characteristic of regarded populations and pedigrees as study objects at the same time, genetic epidemiology plays important roles in exploring family aggregation, genetic model, and susceptibility genes of genetically related diseases.^22,42 Therefore, the main research methods assessed the genetic model of KC in this study were genetic epidemiological analyses, which included the Penrose method, simple segregation analysis, and complex segregation analysis.

For the most reliable results, we adopted three methods to assess the genetic model of KC in the present study. After comparing the prevalence rates of KC in patients' siblings within the general population, the Penrose method draws the elementary conclusion that KC follows a pattern of multifactorial inheritance.⁴³ Simple segregation analysis identified that KC is a disease of multifactorial inheritance by comparing the consistency of the segregation ratio and theoretical values in particular genetic hypotheses.³³ It usually falls into complete ascertainment, incomplete ascertainment, and offspring correction of segregation analysis according to probability of ascertainment. The present study adopted the method of incomplete ascertainment, the sibling method, to assess the genetic model of KC owing to the difficulty of recruiting relatives of KC patients and diagnose them.⁴⁴ Compared with simple segregation analysis, which has the limitation of only using data from siblings, complex segregation analysis takes advantage of all pedigree data to fit various genetic models for evaluating the genetic pattern of KC, and the results indicated that KC is a multifactorial disease.^35,45 Notably, this method can also be used to evaluate whether the occurrence of disease is influenced by environmental factors. There are some limitations that need to be taken into account, such as the inaccuracy of major gene effect or the equality of estimated parameters and boundary values.³⁵ To sum up, three different methods were used to estimate the genetic model of KC to improve findings accuracy. The consistency of the results makes the conclusion that KC is a multifactorial disease more persuasive.

Detailed studies of the genetic model of KC using genetic epidemiological analyses suggested that KC is a disease with polygenic inheritance, which indicated that the occurrence of KC might be subject to the manifold influence of genetic factors, environmental factors and their interaction.^46–48 These results contribute to guiding the prevention and management of KC according to the genetic characteristics of polygenic diseases and lay the foundation for follow-up studies. The recurrence risk refers to the probability of recurrence of a particular disease that has already occurred in a family, which has undeniably become one of the greatest concerns of physicians and patients in present clinical practice.⁴⁹ Empirically, the recurrence risk of relatives is probably a connection with the extent of the relation to the proband, the number of KC patients in one pedigree, the severity of the KC probands and the gender differences of patients, which can provide initial guidance for physicians to provide genetic counseling programs and early diagnostic tools for at-risk family members to mitigate disease progression in a timely manner.⁵⁰ However, it is important to note that recurrence risks are empiric risks derived from population studies and vary according to several factors. Detailed studies should be conducted to verify these opinions in future. Moreover, KC was identified as a multifactorial disease that might be influenced by the interaction of genetic and environmental factors. There are no effective methods to attain genetic control, hence controlling environmental factors, such as avoiding eye rubbing and antiallergy treatment, may be optimum measures to prevent and manage polygenic disease, especially with guidance from genetic information. Therefore, identifying patients who may benefit from the management of environmental factors, based on genetic risk stratification, has important significance in formulating precise prevention and control strategies for KC. Furthermore, the polygenic risk score is a method of assessing incidence risk of polygenic disease according to an individual's genotype.⁵¹ Understanding the multifactorial nature of KC and calculating the polygenic risk score may help to guide personalized prevention strategies for individuals. The present study indicated that an emphasis of future research could be polygenic risk score and gene–environment interactions, which may provide a new perspective for the prevention and management of KC.^52,53

Several limitations of the study needed to be addressed. First, this study only included KC probands and their first-degree relatives owing to the difficulty of recruitment. To make the findings of study more credible, all first-, second-, and third-degree relatives of KC probands should be recruited into future research. Expanding the cohort to include extended relatives could enhance our understanding of genetic contributions and better account for familial aggregation patterns. Second, owing to the lack of molecular data from participants, distinguishing the effects of environmental and genetic factors, as well as the effects of a single locus and two or more independently acting loci, is difficult. Moreover, determining the genetic model in a specific pedigree is similarly challenging. We are planning to collect molecular data for further studies in future. Third, this is a single-center study. A multicenter study should be performed to verify our findings.

Conclusions

The genetic model of KC was identified as a pattern of multifactorial inheritance in present study using genetics epidemiological analyses. These findings provide initial guidance for the prevention and management of KC, and point out the research direction for future study.