New Keratoconus Risk Factors: A Cross-Sectional Case—Control Study

Purpose To evaluate risk factors associated with keratoconus in a monocentric cross-sectional case-control study. Methods This observational study occurred from June 2019 to February 2021 in a university hospital (France). The case group consisted of 195 patients with keratoconus in at least one eye who were followed up by a corneal specialist. The control group consisted of 195 patients without any evidence of keratoconus on slit-lamp examination and corneal topography, who were matched 1 : 1 to controls by age and sex. Data were collected by a self-completed paper questionnaire before the consultation, and a multivariate logistic regression was performed. Results Multivariate analysis revealed significant associations of keratoconus with family history (odds ratio [OR] = 22.2, p < 0.001), rubbing eyes (OR = 10.9, p < 0.001), allergy (any kind) (OR = 3.80, p < 0.001), smoking exposure (OR = 2.08, p=0.017), and dry eyes (OR = 1.77, p=0.045f). The worst eye was associated with the more rubbed eye (p < 0.001) and the more pressed eye during the night according to sleeping position (p < 0.041). Conclusion This study confirmed the association between keratoconus and eye rubbing, family history, and allergy. It highlighted the role of pressure on the eyes during sleep. Other less known risk factors such as dry eyes and smoking exposure should be explored in future studies.


Introduction
Keratoconus is a bilateral and asymmetric corneal disease [1]. It is characterized by a progressive noninflammatory ectasia, leading to thinning and protrusion of the cornea, which assumes a conical shape. It usually results in vision loss because of refractive errors (irregular astigmatism and myopia) and corneal opacification. It is a public health issue because it usually affects young people and leads to social impairment [2][3][4]. e disease may progress acutely or be stable. People at a young age seem to be at a major risk factor for disease progression [5]. e disease is multifactorial with genetic [6,7] and environmental risk factors [8][9][10]. e genetic contribution is supported by familial aggregation [11,12], in particular, parental consanguinity [13], and monozygotic twins may have higher ectasia concordance than dizygotic twins [14]. Many genes have been implicated but none alone explains the genesis of the disease [15]. e most recognized environmental risk factor is eye rubbing [16][17][18]. e repeated mechanical pressure on the cornea causes structural modifications of the collagen, leading to ectasia [19]. Molecular modifications of tears that could explain these histologic changes have been reported in experimental eye rubbings for 60 seconds (increase in levels of metalloproteinases and inflammatory mediators) [20]. is hypothesis is supported by case reports of unilateral or very asymmetric keratoconus when the eye rubbing is unilateral [21][22][23].
Many ocular surface diseases, leading to itching, have been associated with corneal ectasia. Atopy [24,25] and allergic conjunctivitis [26,27] are classical diseases, but many other conditions such as dry eyes [28] and blepharitis [29] have been suspected. Computer vision syndrome [30] is a new entity caused by too much screen time resulting in ocular surface irritation. Studies have found that this syndrome could be associated with keratoconus [28].
Sleeping characteristics in keratoconus patients have been explored. Sleeping position (leading to pressure on the eyes) [23,28], snoring, and sleep apnea [31,32] could affect the occurrence of keratoconus, but more investigations are needed. Also, we lack knowledge about the role of the characteristics of eye rubbing in the genesis of keratoconus, such as frequency [11], duration [33], intensity [33], time of the day [23], and part of the hand used for rubbing [28].
is study aimed to more precisely characterize risk factors for keratoconus, especially eye rubbing, in a monocentric population in France.

Ethics Statement.
A written informed consent was obtained from each patient in accordance with the 1995 Declaration of Helsinki. e study was approved by the medical ethics committee of the University Hospital of Montpellier (Institutional review board no. 18032020).

Design.
is study was a cross-sectional monocentric case-control study performed in Gui de Chauliac University Hospital Center, Montpellier, France, from June 2019 to February 2021.

Study Population.
All patients underwent bilateral slit lamp examination and corneal topography using Orbscan II (Placido disc and slit-scanning technology, Bausch and Lomb, Rochester, NY) or Pentacam HR (Scheimpflug imagery, Oculus, Wetzlar, Germany). e primary outcome (keratoconus diagnosis) was determined by two corneal specialists unaware of the patient's risk factors. e inclusion criterion for the keratoconus group was clinical or topographic keratoconus in at least one eye. Patients who already had a corneal graft for keratoconus were also included.
Except for grafted eyes, each eye with keratoconus was classified from stage I to stage IV using the modified Amsler-Krumeich classification [34,35]. For each case, we defined which eye was the worst according to the classification. For cases with grafted eyes with no previous topography available in the medical file, we defined the first eye operated as the worst. e inclusion criterion for the control group was no clinical or topographic keratoconus in any eye. e control group was sex and age-matched, with a 2-year tolerance, to the case. Controls were patients consulting an ophthalmologist in our center for reasons other than keratoconus mostly for a routine examination.
Exclusion criteria were doubtful topography or slit lamp examination or refusing participation.

Data Source.
Each enrolled patient had to complete a paper questionnaire in the waiting room before an ophthalmologist consultation. is questionnaire was created by the National Center of Keratoconus and is currently subjected to a validation study. e data concerned demographics (age, sex, social status, and dominant hand), medical history (time since keratoconus diagnosis, family history of keratoconus, allergies, eye diseases, dry eye based on OSDI questionnaire), eye rubbing characteristics (frequency, part of the hand used, duration, intensity, time of the day, causes, noise, etc.), sleeping characteristics (regularity, position, snoring or apnea), and environmental exposure (screen time, weather, tobacco exposure, animals, irritant products, etc.). Potential factors associated with keratoconus were compared on univariate analysis (chi-square test or Fisher's exact test). ereafter, a multivariate logistic model was built based on selected parameters from the univariate analysis (level of significance set at p < 0.20 for selection). In addition, three procedures for selecting variables (forward, backward, and stepwise) by the minimization of Akaike's information criterion were used to obtain the most appropriate logistic regression model. e absence of multicollinearity has been verified by using the variance inflation factor. Odds ratios (ORs) and 95% confidence intervals (CIs) were estimated. p < 0.05 was considered statistically significant, and all statistical tests were two-sided. Statistical analyses were performed with R 3.6.1 (Vienna, Austria). No imputation was used for missing values because less than 1% were missing. (Figure 1 and Table 1). We included 195 patients with keratoconus who were followed up by a corneal specialist, and 195 healthy patients were matched on age and sex. Most participants were men (67%), which is similar to the literature [36]. e mean age was 32 ± 12 years. e mean time since diagnosis was 7 ± 9 years.

Participants
In the case population, the mean maximum keratometry (K max ) was 49.1 ± 5 diopters and the thinnest pachymetry was 441 ± 66 microns. In the control population, these parameters were 43.7 ± 1.5 diopters and 543 ± 37 microns. Eyes with corneal grafts were excluded from these analyses. ese parameters were all statistically different (p < 0.001) between the two groups.
Of the 390 eyes with keratoconus enrolled, 152 (39%) eyes were at stage I, 85 (22%) at stage II, 75 (19%) at stage III, 35 (9%) at stage IV, and 41 (11%) with corneal graft according to the modified Amsler-Krumeich classification. Figure 2). e first univariate analysis concerned the whole population of cases and controls. In total, 164 (84%) cases and 66 (34%) controls had a history of eye rubbing (p < 0.001) and 20% (n � 39) of cases had a family history of keratoconus (p < 0.001). Eye dryness was significantly associated with keratoconus (p < 0.001). e only sleeping position associated with keratoconus was the contact of the hand or the forearm with the eyes (p � 0.010). Allergies, in particular asthma and allergic conjunctivitis, were associated with the disease (p < 0.001). Also, cases were more frequently exposed to smoke (tobacco or cannabis) than controls (p < 0.015). e parameters selected from univariate analysis for a multivariate logistic model were family history, eye rubbing history, allergy (any kind), dry eyes, smoking exposure, screen time, hand or forearm contact with the eyes during sleeping and snoring, or obstructive sleep apnea (OSA  (Table 4 and Figure 3). We analyzed rubbing characteristics in the population of people who rubbed their eyes (164 cases and 66 controls). On univariate analysis of the eye rubbing population, On multivariate analysis, keratoconus remained associated with rubbing "often, 1-10 times a day" (OR � 9.48, 95% CI: 4.62-20.6; p < 0.001) and "very often ≥10 times a day" (OR � 57.3, 95% CI: 11.0-1,059; p < 0.001). e only way of rubbing significantly associated with multivariate analysis was protective: "use of fingertips" (OR � 0.27, 95% CI: 0.13-0.55; p < 0.001). (Table 5).

"More Advanced Side" Analysis
e side of the worst eye was not linked to the dominant hand used for rubbing (p � 0.43) but was associated with the preferential eye rubbed (p < 0.001) and the side with more mechanical pressure during the night according to the sleeping position (p � 0.041).

Discussion
e main point of our study is that it supports the mechanical etiology of keratoconus. According to the literature, we found a strong association between eye rubbing and multivariate analysis [11,28,36,37]. e association between the preferential eye rubbed and the worst diseased eye confirms a strong link. New risk factors have been described: tobacco, probably due to the dryness and the induced rubbing and the way of rubbing the eye. Using knuckles and the back of the hand for rubbing was associated with keratoconus, with less use of the index fingertips for rubbing as compared with controls; knuckles and the back of the hand expose the eye to harder mechanical power by direct bone contact. e main rubbing characteristic leading to keratoconus in the eye rubbing population was high frequency rather than intensity or duration of rubbing. e worst diseased eye was associated with more eye rubbing and more pressed eyes during the night, according to the sleeping position. Mazharian et al. [23] also found that patients with unilateral or highly asymmetric keratoconus exhibited homolateral eye rubbing and slept on the same side.
Concerning the sleeping position in the total population, the only statistically positive result was contact with the hand or forearm on the eyes during sleep on univariate analysis only. We did not confirm any correlation with the prone, supine, and side sleep positions or screen time as found in the Moran et al. study [28], nor any already suspected associations with sleep apnea [31,32], perhaps because of a lack of controls included. e sleeping position could be a risk factor for the pathogenesis of keratoconus itself or for the progression of the disease. e genesis of keratoconus is linked to an anatomical predisposition to thin corneas that are less resistant to mechanical stress. Keratoconus, therefore, could only occur in certain patients who have a predisposition and who apply a mechanical constraint on their eyes during sleep or eye rubbing. Also, on multivariable analysis, we confirmed classical known associations such as family history [11,12,36], dry eyes [28], and allergy [38] at rates similar to the literature.
A 2010 meta-analysis considered that the most important risk factors for keratoconus were eye rubbing, positive family history, allergy, asthma, and eczema [9]. e latter factor was the only one not found in our study, perhaps because we analyzed "skin allergies" in general and not eczema in particular. In addition, allergic conjunctivitis, which is a classical risk and disease severity factor [26], was also positive in our univariate analysis (p < 0.001).
We found an association between smoking and keratoconus but other studies usually did not [36,39]. An Australian study in 2021 found a correlation between severity and smoking cigarettes on univariate analysis, which was not confirmed in the multivariate model [40]. Different hypotheses may explain the association between smoking and keratoconus. e dysfunction of the meibomian glands and the dry syndrome induced by smoking could increase friction. e increase in rubbing could also be explained by the personality type of anxiety more frequent in this population. Eye rubbing would decrease the stress felt by the patient via the stimulation of the Vagus X nerve by the oculocardiac reflex.
We did not explain this association and other studies should be conducted to confirm this. Journal of Ophthalmology 5 Our study has some weaknesses and is prone to bias. First, the data were obtained from a questionnaire. Hence, there is a risk of reporting bias depending on the precision of the questions, the knowledge of patients, and their willingness to cooperate. For example, the fact that cases were aware of the association between keratoconus and eye rubbing may have influenced their response. Keratoconus is an acquired disorder, and we selected our control group with normal topographic and clinical criteria at the time of inclusion. Because of no follow-up,    Journal of Ophthalmology 7 some controls could have exhibited keratoconus during the study, which could imply selection bias. We controlled that bias by excluding all patients with doubtful corneal topography.

Conclusion
Our study confirmed the most important risk factors for keratoconus: eye rubbing, family history, and allergy. It allows for more understanding of the eyes rubbing habits of our patients by detailing many of the rubbing characteristics associated with the genesis of corneal ectasia. We also found that the sleeping position could play a key role in the pathophysiology of keratoconus. We recommend that all patients be systematically screened for inappropriate sleeping positions to identify the use of night-time eye protection if necessary. ese elements are crucial in care because we can use them to advise our patients and help them avoid their habits. Indeed, helping people to stop eye rubbing could be an effective treatment on its own to stop the progression of ectasia. In addition, it can be a target for primary prevention by educating the population to not adopt these behaviors.

Data Availability
e dataset analyzed during the current study is available from the corresponding author on reasonable request.

Ethical Approval
is study was approved by the ethics committee of Montpellier University Hospital (Institutional review board no. 18032020) in accordance with the Declaration of Helsinki and all participants have given their written informed consent.

Conflicts of Interest
e authors declare that there are no conflicts of interest.