Ophthalmological manifestations in inflammatory bowel disease under the watchful eye of a gastroenterologist from a tertiary centre

Introduction

The inflammatory bowel disease (IBD) group includes ulcerative colitis (UC) and Crohn’s disease (CD), which are incurable autoimmune diseases that may have extraintestinal manifestations (EIM) in up to 50% of patients [1–6]. Considering the type of IBD, the prevalence of EIM is variable, ranging from 12% to 35% in UC and 25% to 70% in CD [5–8]. Up to 6–40% of patients with IBD have one or more EIM, which can often be more problematic than the underlying IBD itself [5, 6]. Although EIM have a multifactorial pathogenesis, this is not well understood. It appears to be related to an immune response to toxins and antigens that enter the bloodstream from the gastrointestinal tract, leading to the deposition of an antigen-antibody complex in various extraintestinal tissues [5–8].
Among patients with IBD, ophthalmological manifestations (O-EIM) occur in 0.3–13.0% of cases, and in 1.6–5.4% of patients with UC and 3.5–6.8% of patients with CD [5, 8–12]. Due to the possibility of asymptomatic ophthalmological tissue inflammation, routine ophthalmological follow-up is recommended for all patients with IBD, mainly before changing treatment, because some drugs may cause ophthalmological side effects [5, 11]. It is also reasonable to inform patients who recurrently use systemic corticosteroids about the risk of cataracts and glaucoma [5, 6, 8, 11]. It is important for the patient to be aware of possible ophthalmological involvement as a manifestation of the underlying disease, to better understand the disease and contribute to improved treatment [5, 6, 9].
Visual impairment may be related to IBD, pharmacotherapy, or other factors such as age,genetics, and other comorbidities [13, 14]. Cataracts and open-angle glaucoma are complications of long-term ocular inflammation or long-term corticosteroid use [10, 14, 15]. Some O-EIM have been associated with drugs used to treat IBD, such as corneal immune infiltration and diffuse retinopathy associated with adalimumab; anterior optic nerve neuropathy and retinal vein thrombosis developing after infliximab; and cyclosporine, used in UC, causing a rare optic nerve neuropathy [16–20]. Methotrexate levels in tears are similar to serum levels after short-term use, which can lead to irritation of the conjunctiva, cornea, and eyelids [20].
In light of these data, the diagnosis of O-EIM appears to be multifactorial, and the fact that gastroenterologists do not have sufficient resources to carry out this diagnosis makes the process much more difficult. It therefore seems necessary to create a tool to isolate patients who require deeper diagnostics to assess O-EIM in the course of the disease. We conducted this study to assess the usefulness of a questionnaire that could potentially facilitate the identification of patients with O-EIM.

Aim

The primary outcome of the study was to assess the frequency of EIM, including those confirmed by an ophthalmologist (O-EIM), and ophthalmological symptoms not confirmed by an ophthalmologist. The secondary endpoints were to assess potential correlations between the occurrence of O-EIM and the activity of the underlying disease, medication, and family history.

Material and methods

This was a single-centre retrospective study conducted at the Departments of Gastroenterology and Ophthalmology at the National Medical Institute of the Ministry of the Interior and Administration in Warsaw, Poland between January 2020 and April 2022. We included a total of 538 patients in the study, who were at least 18 years of age, of whom 436 had a histologically confirmed diagnosis of IBD according to the criteria of the Polish Society of Gastroenterology and European Crohn’s and Colitis Organisation (ECCO), and 102 had no IBD [1–4]. The inclusion criterion for the IBD group was confirmed IBD, and for the non-IBD group it was the absence of IBD. Thus, the patients in the group without IBD came from a gastroenterology outpatient clinic, where IBD was clinically excluded. Because the peak incidence of IBD is in the second to fourth decade of life and ophthalmological symptoms are much more common in older patients, a group of patients of similar age without IBD was selected. The patients in the group without IBD came from a hospital-based gastroenterology outpatient clinic and from private facilities, where they presented with symptoms of irritable bowel syndrome (IBS) and small intestinal bacterial overgrowth syndrome (SIBO). During the diagnostic process, IBD was excluded in these patients. Due to the nature of the observational study, there were no exclusion criteria.
We used a questionnaire that had been prepared in collaboration with the Department of Ophthalmology at the National Medical Institute of the Ministry of the Interior and Administration. Two versions were used: one for the group with IBD and the other for the group without IBD. The questions explored the parameters of gender, age, weight, and height, and for the IBD patients the type, extent, and activity (clinical remission or flare) of the disease, the number of flare-ups, any surgical procedures performed in the last 12 months, the presence of EIM, and any medications used. The second part of the questionnaire was identical for both groups: current ophthalmological symptoms (red eye, ocular pain/discomfort, photophobia, visible haze or persistent visual disturbances, ocular bleeding, ophthalmia, or vision deterioration), confirmed diagnoses (conjunctivitis, dry eye syndrome, uveitis, episcleritis, scleritis, or others), ophthalmological consultation, number of hours spent in front of a computer, family history of ophthalmological involvement (cataracts, glaucoma, retinal detachment, or uveitis), and medications that may impair vision (-blockers, anti-coagulants, amiodarone, tuberculostatics, chemotherapeutics, antiepileptic drugs, anti-rheumatic drugs, metronidazole, -lactam antibiotics, tetracycline, or hormonal medicines). A positive questionnaire result suggesting ophthalmological involvement was defined as the occurrence of at least one ophthalmological symptom in the last 3 months or a confirmed diagnosis of ocular disease in the last 12 months. Complete questionnaires were obtained from 436 of the 480 IBD patients and from 102 of the 120 non-IBD patients, yielding response rates of 90.8% and 85.0%, respectively.
Statistical analysis was performed for the patients with IBD (n = 436) and the group without IBD (n = 102). Categorical variables are described with the number of observations and the frequency percentage. Numerical variables are described with mean and standard deviation or median and first and third quartiles, depending on normality. Normality was tested with the Shapiro-Wilk test and verified with skewness and kurtosis. Histograms were used for further normality verification in justified cases. Homogeneity was tested with Levene’s test. Differences between groups for continuous parameters were assessed with Student’s independent t-test or the Mann-Whitney U test, as appropriate. Differences between groups for categorical parameters were assessed with Pearson’s 2 test or Fisher’s exact test, as appropriate, and were described with Cramer’s V. A logistic 2-step regression approach was used to identify significant predictors of eye disease activity (univariable models for each variable in the first step and multivariable model as the second step). Variable selection for the multivariate model was based on the p-value produced in the first step, under the additional condition that the p-value was no higher than 0.250, and further on stepwise selection. Multivariable models were assessed with Nagelkerke’s R2, the Hosmer-Lemeshow test, and VIF indicators. The predictors’ effect was described with coefficient, standard error, and odds ratio, along with 95% confidence intervals. All statistical calculations assumed an  of 0.05. The analysis was run in the software programme R, version 4.1.2.

Results

Two groups were enrolled in the study: those with IBD and those without. The group with IBD consisted of 436 patients; among this group, UC was present in 55.3% (241) and CD in 44.7% (195) of cases. Clinical remission was declared by 38.3% (167) of the IBD patients, defined as a Total Mayo Score of ≤ 3 or a Crohn’s Disease Activity Index (CDAI) of ≤ 150 points. The average Total Mayo Score was 3.24, whereas for CDAI the average was 104.97 points. At least one EIM in the last 12 months was reported by 48.6% of patients (212). The most common EIM was axial/non-axial spondyloarthropathy, present in 16.3% of patients (71), followed by fatigue in 14.4% (63), dermal lesions in 10.8% (47), anaemia in 8.9% (39), and primary sclerosing cholangitis or autoimmune hepatis in 2.5% (11) of cases. Mesalazine was used by 77.8% of patients (339), steroids by 44.5%, immunosuppressive therapy – defined as thiopurine therapy – by 48.6% (212), methotrexate by 4.6% (20), and cyclosporine by 7.3% (32). Due to disease flare-ups, steroids were administered in 44.5% (194) and budesonide in 28.7% (125) of patients. Multiple steroid treatments in the last 12 months were declared by 79.1% of patients (345). Most of the IBD patients (87.4% [381]) were treated with biologics or Janus kinase inhibitors. The medication most often prescribed was vedolizumab, which was used in 48.8% of cases (186). Tumour necrosis factor (TNF) inhibitors (infliximab or adalimumab) were used sequentially in 47.5% of cases (181), followed by ustekinumab in 12.9% (49) and tofacitinib in 1.8% (7) (Table I).
The group without IBD consisted of 102 patients, and IBS was present in 56.9% (58) and SIBO in 43.1% (44) of cases. Among concomitant diseases the most common was hypertension (7.8% [8]), followed by venous thromboembolism (3.9% [4]), active neoplasm (2% [2]), and one case each of epilepsy and rheumatological disease (1% each). Antibiotic use was declared by 13.7% of patients (14) and contraception use by 11.8% (12) (Table II).
In Table II we present the percentage of declared ophthalmological diseases in relation to a history of ophthalmology consultation in the last 3 months. A recent need for ophthalmology consultation was reported by 30.5% of IBD patients (133), whilst those without IBD were half as likely to report one (V = 0.15, 95% CI [0.05, 0.26], p = 0.001). Among the ophthalmological diseases diagnosed during such consultations, the frequency of uveitis and episcleritis significantly differed between the groups. Uveitis was more common in the IBD group than in the non-IBD group (3.9% [17] vs. 2.9% [3]; V = 0.03, 95% CI [0.00, 0.11], p = 0.025). Similarly, episcleritis was more often diagnosed in the group with IBD (4.8% [21] vs. 2% [2]; V = 0.03, 95% CI [0.00, 0.14], p = 0.013).
The comparisons between IBD patients and patients without IBD regarding ophthalmological symptoms, time spent in front of a computer screen, familial history of ocular diseases, and treatment are presented in Table III. There was no difference in the prevalence of ocular symptoms in the last 3 months between the IBD and non-IBD groups (50.7% [221] vs. 48% [49]; V = 0.07, 95% CI [0.03, 0.21], p = 0.63). Other variables, such as the presentation of ocular symptoms in the last 3 months (including red eye, ocular pain, photophobia, visible haze or persistent visual disturbances for at least 24 h, ocular bleeding, ophthalmia, or vision deterioration), familial history of ophthalmological disease, or usage of certain drugs that can cause adverse effects on vision, did not differ significantly. However, the patients without IBD spent more time in front of a computer screen than those with IBD (6.45 vs. 5.59 h, MD = –0.86, 95% CI [–1.51, –0.21], p = 0.010).
In Table IV we present the association between ophthalmological symptoms and disease activity, which was confirmed for both subgroups of patients (UC and CD). Active UC, defined as a Total Mayo Score of > 3, was significantly more common among patients with ocular symptoms than among those without (40.0% [62] vs. 25.6% [22]; V = 0.14, 95% CI [0.03, 0.26], p = 0.035). Likewise, active CD, defined as a CDAI of > 150 points, among patients with ocular symptoms was significantly more common than among those without (24.1% [34] vs. 9.3% [5]; V = 0.17, 95% CI [0.05, 0.28], p = 0.034).
The proportion of patients with ophthalmological symptoms among all patients with IBD was 50.7% (221), 46.5% (112) among those with UC, and 55.9% (109) among those with CD (V = 0.09 95% CI [0.01, 0.18], p = 0.094; Table V).
In addition, Table VI shows the analysis of whether there was an association between ophthalmological symptoms and the usage of selected medicines. An association was confirmed for -blockers (13.2% vs. 5.0%; V = 0.12, 95% CI [0.04, 0.19], p = 0.015), anti-coagulants (12.2% vs. 4.3%; V = 0.12, 95% CI [0.04, 0.19], p = 0.015), tuberculostatics (4.7% vs. 0.0%; V = 0.13, 95% CI [0.09, 0.16], p = 0.007), -lactam antibiotics (6.4% vs. 0.7%; V = 0.13, 95% CI [0.07, 0.17], p = 0.016), steroids (54.7% vs. 22.9%; V = 0.30, 95% CI [0.21, 0.38], p < 0.001), thiopurines (43.6% vs. 59.3%; V = 0.15, 95% CI [0.05, 0.24], p = 0.003), methotrexate (6.4% vs. 0.7%; V = 0.13, 95% CI [0.07, 0.17]; p = 0.016), and ustekinumab (15.8% vs. 6.1%; V = 0.13, 95% CI [0.04, 0.20], p = 0.015). The usage of other medicines – amiodarone, chemotherapeutics, anti-epileptic, anti-rheumatic, metronidazole, tetracycline, hormonal medicines, mesalazine, budesonide, cyclosporine, TNF, vedolizumab, and tofacitinib – did not differ between IBD patients with and without ophthalmological symptoms.
Multivariable regression step identified some variables having a significant impact on the risk of ophthalmological symptoms: anaemia (OR = 13.00, 95% CI [2.38, 244.71], p = 0.017), axial/non-axial spondyloarthropathy (OR = 4.62, 95% CI [2.01, 11.68], p = 0.001), steroids (OR = 5.75, 95% CI [3.22, 10.65], p < 0.001), thiopurines (OR = 0.37, 95% CI [0.21, 0.66], p = 0.001), methotrexate (OR = 13.88, 95% CI [2.40, 265.88], p = 0.016), smoking (OR = 0.44, 95% CI [0.23, 0.85], p = 0.015), positive family history of glaucoma (OR = 4.56, 95% CI [1.82, 13.00], p = 0.002), and time spent in front of a computer (OR = 1.14, 95% CI [1.03, 1.25], p = 0.009). The other predictors did not have a significant impact on the risk of ocular symptoms when considered working at the same time (in the multivariable model step) (Table VII).
The Hosmer-Lemeshow test resulted in a p-value of 0.779 and indicated a good model fit. The fit of the model was additionally assessed with Nagelkerke’s R2, which was 45.6% and confirmed a good ability of the model to predict the odds of eye disease. The collinearity of the independent variables was assessed with VIF parameters, which ranged from 1.0 to 1.3 and indicated no collinearity.

Discussion

The main findings from our study include the presence of EIM in 48.6% of IBD patients, with O-EIM confirmed by an ophthalmologist being declared by 8.9% (uveitis 3.9% + episcleritis 4.8% + scleritis 0.2%). Furthermore, in the last 3 months 50.7% of IBD patients reported ophthalmological symptoms, whilst 30.5% of IBD patients reported having an ophthalmological consultation.
The first report of ophthalmological involvement in IBD was published by Crohn in 1925; he suggested that 2 patients he was treating suffered from keratomalacia and xerophthalmia [21]. The prevalence of O-EIM varies depending on the population, ranging from 0.3% to 13.0% among all patients with IBD, 1.6–5.4% of those with UC and 3.5–6.8% of those with CD [14, 21]. In our study, there was a similar prevalence of O-EIM, which had been identified in previous ophthalmological consultations. At the same time, most studies to date address the under-reporting of O-EIM, which is mainly linked to its sparse symptomatic course and the lack of good screening tools [8, 14, 21]. In our study the total number of O-EIM is underestimated because some of the patients have not yet had an ophthalmological consultation. Those patients who had not yet had a confirmed ophthalmological diagnosis but whose ophthalmological symptoms were confirmed in the questionnaire were referred for ophthalmological consultation.
In subsequent reports, ophthalmological changes in IBD have been associated with symptoms that appear in the early years after IBD diagnosis [8, 21]. This relationship was confirmed in our study: patients with active IBD (in UC defined as a Total Mayo Score of > 3 [p = 0.035] or in CD as a CDAI of > 150 points [p = 0.034]) were more likely to have eye involvement. This indicates a possible extension of the inflammatory process from the gut to the organ of vision. Some previous studies do not clearly explain whether disease activity or location can be responsible for ophthalmological symptoms [6, 22–24]. Other studies show a greater tendency for ocular inflammation in patients with CD – mainly those with colitis or ileitis – and in UC patients with pancolitis [6, 15, 21–26]. This was also confirmed in our study, as ophthalmological symptoms predominated among patients with CD in 46.5% than with UC in 55.9% cases. The pathogenesis of O-EIM in IBD remains unclear [23, 27–29]. It has been suggested that the local action of antigen-antibody complexes produced against the vessels of the intestinal wall and transported through the bloodstream may be responsible for ocular involvement [25–31]. However, Santeford et al. suggested that a disruption of autophagy mediated by macrophages and their impaired function may cause uveitis [32].
Risk factors for ophthalmological involvement among IBD patients included a coexisting EIM (anaemia or axial/non-axial spondyloarthropathy), the use of certain medications (steroids or methotrexate), a family history of ocular disease (glaucoma), and spending more time in front of a computer screen. As risk factors for the development of ophthalmological symptoms in IBD, female gender and the presence of arthritis or joint pain in patients with CD have been reported [28]. In our study we did not confirm gender as a risk factor but did confirm an association with the presence of an EIM, including anaemia and axial/non-axial spondyloarthropathy. A protective association between smoking and ophthalmological symptoms in UC patients has been shown in some studies, whilst our results confirmed this relationship for the entire IBD group [22, 29, 30]. Further confirmed protective factors in the context of ophthalmological involvement were a history of surgery within the past 12 months and the use of thiopurines, which could potentially be associated with improved disease control and reduced inflammation. Good control of IBD activity appears so far to be the best protective factor against the development or progression of O-EIM [29, 30]. Lin et al., in a large retrospective analysis, suggested that a family history of IBD alone may independently confer an increased susceptibility to the development of an O-EIM, which is supported by our results [33]. In our study, a positive family history of glaucoma was a risk factor for the development of ophthalmological involvement in patients with IBD. We found that risk factors for the development of ophthalmological involvement also included the presence of another EIM, the most common of which were anaemia and axial/non-axial spondyloarthropathy. Similar relationships were described by Bandyopadhyay et al. and Manser et al. [31, 34]. The significant incidence of ophthalmological involvement, present in 78.2% of IBD patients, is certainly related to the overlap with other diseases unrelated to IBD. A similar finding was observed by Felekis et al., who reported a high prevalence (43%) of ophthalmological involvement, which they linked primarily to a high proportion of dry eye syndrome and eyelid inflammation [35].
We also performed an analysis of the correlation between ophthalmological involvement and the use of selected medicines in the group of patients with IBD; it was confirmed for -blockers, anti-coagulants, tuberculostatics, -lactam antibiotics, steroids, thiopurines, methotrexate, and ustekinumab. Many of these have been shown to affect vision as a side effect, but at the same time this cannot be assumed to be the only explanation [19, 23]. Some of the drugs where this relationship has been confirmed are frequently used in therapy for IBD. Perhaps therapy with these drugs was related to IBD activity, and they were included because of IBD progression, which may indirectly explain the relationship with the drugs as well as with disease activity [26, 29, 33]. Cataracts and open-angle glaucoma are complications of long-term ocular inflammation or long-term corticosteroid use [36–38]. Some ophthalmological manifestations have been associated with drugs used to treat IBD, such as corneal immune infiltration and diffuse retinopathy associated with adalimumab; anterior optic nerve neuropathy and retinal vein thrombosis developing after infliximab; and cyclosporine, used in UC, causing a rare optic nerve neuropathy [9, 21, 39–41]. Methotrexate levels in tears are similar to serum levels after short-term use, which can lead to irritation of the conjunctiva, cornea, and eyelids, but in our study this relationship was not confirmed [42]. Uveitis has been associated with the use of TNF biologics; it has been described in combination with etanercept, infliximab, adalimumab, and rifabutin [8, 43]. Inflammation decreases when the drug is discontinued, which is recommended, and the use of topical corticosteroids may be necessary for complete resolution of the inflammation [8, 44]. In addition, the neurological side effects of pharmacotherapy may cause visual disturbances without directly affecting the eyes [45]. Following bowel resection in the context of IBD, short bowel and malabsorption syndromes can lead to vitamin A deficiency, which can result in dusk blindness (nyctalopia) and dry keratoconjunctivitis, but in our study the surgery performed was protective [22, 46].
Ophthalmological symptoms in IBD are mainly non-specific, and their significance may not be recognised by the clinician. In addition, there are data indicating that asymptomatic inflammation of ocular tissues may occur, so routine ophthalmological follow-up is recommended for all patients with IBD (with or without ophthalmological symptoms) – mainly before changing IBD treatment, because some drugs may cause ophthalmological side effects [8, 47]. It is important to remember that most ophthalmological symptoms can be treated without sequelae if recognised promptly [25, 48]. Ophthalmologists need to consider that ophthalmological symptoms of IBD may precede systemic disease, and a systematic history should be taken in chronic uveitis of unknown aetiology [49, 50]. Patients with chronic or recurrent use of systemic corticosteroids should be warned of the risk of cataracts and glaucoma [50]. Collaboration among the IBD clinical care team, which includes ophthalmologists, is crucial to improve the quality of care for these patients.
This study has several limitations, including its retrospective nature and the fact that not all patients underwent a simultaneous ophthalmological examination. Thus, there is no assessment of the sensitivity and specificity of this tool. This study represents a pilot experience, but the tool needs validation on larger samples. Given the much rarer occurrence of O-EIM in IBD without ophthalmological examination and confirmation of their presence, the severity is defined by a specific number of them among all ocular manifestations.

Conclusions

Ophthalmological manifestations are common in patients with IBD and require increased awareness of their occurrence among clinicians and patients with IBD. Patients at risk of developing them include not only those requiring treatment (with steroids or immunosuppressants, after surgery, or for another EIM), but also those in clinical remission and with a positive family history. Strict control of inflammation with medication or possibly surgery is probably the most effective strategy to manage both intestinal symptoms and extraintestinal complications. The screening tool used in this initial study can be recommended for use by IBD practitioners in their daily practice to select patients for referral to an ophthalmologist. The significant prevalence of ocular manifestations confirms the need to routinely send patients for ophthalmology consultation to truly assess the frequency of ophthalmological manifestations. In this regard, it is also important to highlight the positive aspect of creating an integrated multidisciplinary team to improve IBD therapy targeting disability and quality of life.

Funding

No external funding.

Ethical approval

The study was approved by the Ethics Committee of the National Medical Institute of the Ministry of the Inferior and Administration in Warsaw, Poland (approval code 11//2020).

Conflict of interest

The authors declare no conflict of interest.

References