Severe vernal keratoconjunctivitis requiring trabeculectomy with mitomycin C for corticosteroid-induced glaucoma


  • Conflict/competing interest: None declared.

  • Funding sources: None declared.

Dr Cordelia Chan, Singapore National Eye Centre, 11 Third Hospital Avenue, Singapore 168751. Email:


Background:  To describe clinical features of severe vernal keratoconjunctivitis with steroid response in Asian children and risk factors for glaucoma filtration surgery.

Design:  Retrospective non-controlled, comparative case series.

Participants:  Patients with severe vernal keratoconjunctivitis seen at a single centre over 6 years.

Methods:  Clinical features, symptoms and treatment modalities were recorded for patients (i) diagnosed with severe VKC (clinical grade ≥3); (ii) had >2 recordings of increased intraocular pressures of >21 mmHg; (iii) and a minimum follow-up period of 1 year post-presentation.

Main Outcome Measure:  Corticosteroid-induced glaucoma requiring trabeculectomy with mitomycin-C.

Results:  Six patients (eight eyes) of 36 patients required trabeculectomy/mitomycin-C. All were male. Mean age of disease onset was 9.3 ± 4.5 years for a mean duration of 6.08 ± 3.5 years. Mean intraocular pressures increase from baseline was 29.0 ± 8.2 mmHg and all required >2 anti-glaucoma medications. The main risk factor for trabeculectomy was a greater increase in intraocular pressures from baseline (odds ratio 1.3; 95% confidence interval, 1.0–1.5; P = 0.011), which was independent of potential confounders such as type and duration of corticosteroid use. Comparing eyes pre- and post-trabeculectomy, all improved in clinical severity of vernal keratoconjunctivitis (mean clinical grade improvement 2.1; 95% confidence interval, 1.3–3.0; P < 0.001) and reduced dependence on topical corticosteroids for mean duration of 22.5 ± 15.3 months.

Conclusion:  In our study, patients with a ‘greater steroid response’, that is, higher increase in intraocular pressures from baseline are associated with a 30% higher risk toftrabeculectomy.


Vernal keratoconjunctivitis (VKC) is a severe form of ocular allergy that affects mostly children and young adults living in areas with warm climates.1 VKC has a reported incidence of 0.1–0.5% of all ophthalmic patients and a prevalence of up to 5–15% among children.1–4 However, the pattern and severity of this disease vary widely, depending on racial and environmental differences – most patients with VKC in temperate countries suffer from seasonal exacerbations, but the disease is often chronic and persistent in warm, tropical climates around Asia.2 Although most cases of VKC are self-limiting, chronic forms in Asian eyes may be at higher risk of permanent visual impairment because of complications such as corneal scarring, cataract formation and glaucoma secondary to corticosteroid therapy.5

The reported incidence of glaucoma in patients with VKC receiving corticosteroid therapy is 2–7%.4–6‘Steroid response’ or corticosteroid-induced ocular hypertension is due to decreased trabecular outflow causing a rise of intraocular pressure (IOP).7 Resultant glaucomatous nerve damage may persist even after corticosteroid therapy is discontinued and the IOP normalizes.8 Proposed risk factors for steroid response from population-based studies in elderly include the type and potency of the corticosteroid, pre-existing open-angle glaucoma (POAG) and a family history of POAG.7–9 However, we did not find any studies that specifically looked into risk factors for steroid response in patients with VKC. Such patients are usually of a younger age group, who do not have the risk factors mentioned, such as previous POAG. Furthermore, no studies have specifically described risk factors for glaucoma filtration surgery in eyes of patients with VKC that suffered from steroid response glaucoma.

The purpose of our study is to conduct a comparative case series of patients with severe VKC and corticosteroid-induced glaucoma in our Asian population, as well as describe the clinical profile of cases requiring trabeculectomy with mitomycin C (MMC). The risk factors for trabeculectomy, clinical disease patterns, visual outcomes and IOP control following surgery are documented.


We retrospectively reviewed all patients diagnosed with VKC at the Singapore National Eye Center within a 6-year period (1 January 2003 to 31 December 2008) with approval from our institutional review board. We included patients who were (i) diagnosed with VKC clinical grade 3 and 4 (‘Severe’ and ‘Very severe’ VKC) as defined by Bonini et al.10 (ii) Had more than two consecutive recordings of increased IOP of more than 21 mmHg on Goldmann applanation tonometry; for the younger patients who were uncooperative with Goldmann applanation tonometry measurement, the Tono-pen XL (Reichert Ophthalmic Instruments, Depew, New York, USA) was used; (iii) and a minimum follow-up period of 1 year post-presentation. Corticosteroid-induced glaucoma was diagnosed in cases with glaucomatous optic nerve head changes or, in patients who could perform reliably, glaucomatous visual field defects on Humphrey 24-2 full-threshold test. In these patients, they were required to have at least two reliable baseline visual fields (false-positive, false-negative and fixation loss rates all below 15%). A glaucomatous field defect was considered if two consecutive reliable visual field tests revealed at least three contiguous points on the pattern deviation plot with sensitivity depressed below P < 5.0%. Patients with history of ocular hypertension, POAG, previous ocular surgery or trauma or other possible secondary causes of glaucoma were excluded from the study.

We reviewed all medical records and performed data entry in a standardized data collection form, which was designed before the study commenced. We collected information including patient demographics, presenting and worst logMAR best corrected visual acuity (BCVA), systemic associations such as history of atopy, asthma, eczema or significant family history of glaucoma. We also recorded clinical features such as chronicity, laterality, total duration of disease, recurrent episodes requiring a visit and surgeries required. Ocular symptoms such as itch, redness, mucoid discharge, glare and photophobia, but detailed slit-lamp examination findings of the eyelids, conjunctiva, cornea and optic nerve head were recorded. These clinical findings were collated and the severity of the VKC graded based on the classification as described by Bonini et al.10 The minimum, maximum and mean IOP readings on Goldmann applanation tonometry were recorded. Investigations such as skin tests, serum eosinophil levels and histology were included if they were performed. Treatment modalities used, such as lubricants, mast-cell stabilizers, topical or systemic corticosteroids and antihistamines, including the dosage, frequency and duration of use were recorded. All data forms were then collated and entered by two separate personnel into separate databases. Both databases were then compared with detect data entry errors, which were immediately rectified.

In patients who underwent trabeculectomy with MMC, clinical grading of VKC and BCVA were also recorded on three separate clinic visits within 8 weeks pre- and postoperatively. The mean clinical grading score of these visits and difference in clinical severity was calculated in both eyes, and compared between pre- and post-trabeculectomy. The type and dosage of steroid use was also noted within 8 weeks pre- and post-trabeculectomy. Patients were also analysed according to their treatment modalities used to control the steroid-response, that is, surgery or topical medications. The statistical software spss 17.0 (SPSS Inc., Chicago, IL, USA) was used for statistical analyses. Statistical analysis included descriptive statistics, where the mean and standard deviation with 95% confidence intervals (CI) were calculated for the continuous variables; whereas frequency distribution and percentages were used for categorical variables. We used the Student's t-test (unpaired) to compare means and chi-square test was used to find associations between the categorical variables. Trend analyses within variables were assessed using two-way anova for both the groups separately. Multivariate logistic regression was used to determine significant risk factors and the odds ratios (OR) with 95% CI were calculated. A P-value <0.05 has been considered statistically significant.


A total of 36 patients were found to meet the inclusion criteria, of which six patients (eight eyes) eventually underwent trabeculectomy with topical MMC. The overall mean age was 10.3 (standard deviation ± 5.0) years (range 4–24 years), and most patients were male (83.3%) and Chinese (77.8%). The mean logMAR baseline BCVA was 0.33 ± 0.29. All patients had severe and persistent symptoms throughout the year, with a mean duration of disease of 6.08 ± 3.5 years, during which they experienced a mean number of recurrences requiring a visit at 5.9 ± 3.0 per year. The demographics and clinical characteristics are described in Table 1.

Table 1.  Comparison baseline demographics and characteristics of patients on medications versus those requiring surgery
CharacteristicsTotal (n = 36)Medication (n = 30)Trabeculectomy (n = 6)P-value
  1. SD, standard deviation; VKC, vernal keratoconjunctivitis.

Age, years (SD)10.33 (5.02)10.53 (5.17)9.3 (4.50)0.856
Gender (%)    
 Male30 (83.30)24 (80.00)6 (100.00)0.561
 Female6 (16.70)6 (20.00)0 (0.00) 
Race (%)    
 Chinese28 (77.80)24 (80.00)4 (66.70) 
 Malay5 (13.90)4 (13.30)1 (16.70)0.686
 Indian3 (8.30)2 (6.70)1 (16.70) 
Visual acuity (logMAR)    
 Presenting (SD)0.33 (0.29)0.310 (0.29)0.400 (0.27)0.816
 Worst (SD)0.68 (0.38)0.667 (0.39)0.767 (0.33)0.561
Asthma (%)    
 Yes13 (36.10)10 (33.30)3 (50.00)0.645
 No23 (63.90)20 (66.70)3 (50.00) 
Eczema (%)    
 Yes21 (58.30)18 (60.00)3 (50.00)0.677
 No15 (41.70)12 (40.00)3 (50.00) 
Allergic rhinitis (%)    
 Yes19 (52.80)17 (56.70)2 (33.30)0.391
 No17 (47.20)13 (43.30)4 (66.70) 
Duration of topical corticosteroid use, months (SD)10.47 (8.51)8.9 (6.39)18.33 (13.47)0.003
Ratio of number of recurrences: duration of disease, per year (SD)5.92 (3.07)5.7 (2.96)7.0 (3.688)0.382
Intraocular pressure (mmHg)    
 Peak30.03 (7.10)28.27 (5.57)38.83 (7.89)0.268
 Increase from baseline18.50 (7.68)16.40 (5.69)29.00 (8.17)0.086
Type of VKC (%)    
 Mixed28 (77.80)23 (76.70)5 (83.30)0.879
 Palpebral7 (19.40)6 (20.00)1 (16.70) 
 Limbal1 (2.80)1 (3.30)0 (0.00) 
Use of topical cyclosporine (%)    
 No12 (33.30)9 (30.00)3 (50.00)0.378
 Yes24 (66.70)21 (70.00)3 (50.00) 
Mean number of anti-glaucoma medications used (SD)1.58 (0.81)1.4 (0.72)2.5 (0.55)0.0013
Type of topical corticosteroid used (%)    
 Dexamethasone 0.1%28 (77.80)23 (76.70)5 (83.30)0.243
 Fluoromethalone 0.1%6 (16.70)6 (20.00)0 (0.00) 
 Prednisolone acetate 1.0%2 (5.60)1 (3.30)1 (16.70) 
Oral Corticosteroids (%)    
 No34 (94.40)29 (96.70)5 (83.30)0.310
 Yes2 (5.60)1 (3.30)1 (16.70) 

None of the patients had a history of POAG or ocular hypertension, and none gave a positive family history of glaucoma. The mean baseline IOP was 11.3 ±2.7 mmHg. All patients had open angles on gonioscopy and normal baseline visual fields if reliably performed. The most common type of topical corticosteroids prescribed were dexamethasone 0.1% in 28 patients (77.8%), fluorometholone 0.1% in six (16.7%) and prednisolone acetate 1% in two patients (5.6%). Two patients (3.6%) received oral prednisolone to control severe eczema, but these were prescribed and monitored by an internal physician. The overall mean duration of corticosteroid usage was 10.5 ± 8.5 months. The mean duration of corticosteroid use was 8.9 ± 6.4 months in eyes that were controlled on anti-glaucoma medications, and 18.3 ± 13.5 months in those that needed surgery (P = 0.003). The mean baseline IOP in 30 patients who required only medical therapy was comparable to the remaining six patients who underwent surgery (11.3 ±2.6 mmHg vs. 11.6 ±2.4 mmHg [P = 0.37]).

Patients diagnosed with ‘steroid response’ were monitored closely and started on anti-glaucoma medication in a stepwise fashion, until they achieved their individualized target IOP. Thereafter, their drug therapy was tapered down until the IOP was controlled without medication. Patients with glaucomatous optic nerve head changes or glaucomatous visual field defects were referred to a glaucomatologist for further management (n = 8, mean increase in cup : disc diameter ratio was 0.3 ± 0.1). In patients whose IOP could not be controlled on maximal anti-glaucoma therapy and surgery was indicated, trabeculectomy with adjuvant topical MMC (0.02% applied for 3–5 min) was performed.

Six patients (eight eyes) had inadequate control of IOP leading to surgery, with demographics and clinical characteristics described in Table 2. The mean age was 9.3 ± 4.5 years and all were male. The mean peak IOP was 38.8 ± 7.9 mmHg and the mean increase in IOP from baseline was 29.0 ± 8.2 mmHg. The mean preoperative BCVA was 0.29 ± 0.21 whereas the mean final BCVA was 0.10 ± 0.09 (P = 0.02). The mean final vertical cup : disc diameter ratio was 0.6 (range 0.3–0.9). The estimated risk of requiring surgery after 18 months duration of steroid therapy was OR 6.3; 95% CI, 1.0–78.5; P = 0.04. All eyes were noted to have significant improvement in severity of VKC post-trabeculectomy (mean clinical grade improvement 2.1; 95% CI, 1.3–3.0; P < 0.001) and reduced dependence on topical steroids for a mean duration of 22.5 ± 5.3 months following trabeculectomy. In four of eight eyes, there was a reduction in frequency of topical corticosteroid use after trabeculectomy and in two eyes there was no change, with a mean reduction of 1.25 ± 2.87 drops per day. Seven of eight eyes had successful trabeculectomies, with a mean postoperative IOP of 9.74 ± 4.10 mmHg for the duration of follow up. One patient had a failed trabeculectomy after 18 months for which a second trabeculectomy wasdone and IOP control has been successful for 4 months thus far.

Table 2.  Patients who required trabeculectomy with topical mitomycin C
Patient (age/sex/race)AsthmaEczemaOther atopic diseases or family historyDuration of disease (years)Number of RecurrencesPeak IOP (mmHg)Duration of corticosteroid use (weeks)Main steroid usedNumber of anti-glaucoma medicationsSurgeryMean Clinical grade of VKC PreoperativeMean Clinical grade of VKC PostoperativeP-value
  1. Demographics:age, presenting age; M, male; c, Chinese; i, Indian. Medical history:Y, yes; N, no. Main steroid used:D, dexamethasone 0.1% topical; PF, predforte 1.0% topical; R, right; L, left. Clinical grade of VKC:based on modified clinical grade by Bonini et al. 8 weeks preoperative and duration of postoperative period. IOP, intraocular pressure; VKC, vernal keratoconjunctivitis.

5/M/iYYN3123118D3R & LL:3.666L:1.000 
13/M/cYNY9105330D2R & LL:3.668L:0.934 

We compared our patients with severe VKC with steroid response who were treated medically with those that required surgery (Table 3). Age- and gender-adjusted logistic regression revealed that the risk factors that lead to trabeculectomy include: longer duration of steroid use (OR 1.1; 95% CI, 1.0–1.3; P = 0.035), higher peak IOP (OR 1.3; 95% CI, 1.0–1.5; P = 0.017) and a greater increase in IOP from baseline (OR 1.3; 95% CI, 1.0–1.5; P = 0.011). Other risk factors such as type of VKC, type of corticosteroid used and corneal involvement or neovascularization were not found to be significant. Multivariate analysis revealed that a greater increase in IOP from baseline as the most significant risk factor (OR 1.3; 95% CI, 1.0–1.5; P = 0.011).

Table 3.  Risk factors for trabeculectomy in Asian patients with severe vernal keratoconjunctivitis
CharacteristicsnTrabeculectomy, OR (95% CI)
n (%)Age-gender adjustedPMultivariableP
  • Adjusted for age, sex and IOP diff.

  • Dexa, dexamethasone 0.1%, topical; FML, fluorometholone 0.1%, topical; PF, prednisolone acetate 1.0%, topical. CSA, topical cyclosporine A (0.5%); NA, not applicable; CI, confidence interval; IOP, intraocular pressure.

Age, years366 (16.7)0.840 (0.590, 1.210)0.3520.840 (0.590, 1.210)0.352
 Male246 (20.0)NA NA 
 Female60 (100.0)    
Ratio of recurrences: duration, per year366 (16.7)1.132 (0.863, 1.490)0.3700.810 (0.510, 1.270)0.357
Type of vernal keratoconjunctivitis      
 Mixed285 (17.9)0.440 (0.048, 4.070)0.4711.710 (0.094, 31.350)0.717
 Palpebral71 (14.3)1.000 1.000 
 Limbal10 (0.0)NA NA 
Corneal involvement      
 Yes266 (23.1)    
 No100 (0.0)NA NA 
Type of corticosteroid      
 Dexa285 (17.9)1.060 (0.223, 4.988)0.9460.660 (0.098, 4.460)0.672
 FML60 (0.0)NA NA 
 PF21 (50.0)1.000 1.000 
Duration of corticosteroid use (months)366 (16.7)1.140 (1.010, 1.290)0.0351.030 (0.880, 1.210)0.739
Peak IOP366 (16.7)1.280 (1.060, 1.580)0.0170.960 (0.540, 1.700)0.889
IOP increase3661.270 (1.060, 1.520)0.0111.270 (1.060, 1.520)0.011
Use of CSA      
 Yes243 (12.5)0.250 (0.033, 1.900)0.1800.462 (0.034, 6.260)0.561
 No123 (25.0)1.000 1.000 


Corticosteroid-induced glaucoma is an iatrogenic secondary open-angle glaucoma. The raised IOP in such eyes has been shown to reduce with time after stopping steroid use, but the time frame for this to happen and baseline factors that help identify eyes that would respond spontaneously versus those requiring more aggressive medical therapy or even surgery are not known.9 In Asian eyes suffering from VKC, the risk of corticosteroid-induced glaucoma may be higher due to the chronic use of topical corticosteroids.2 While most of these patients may be controlled with topical medications (83.3%), it is important to identify clinical characteristics, which increase the risk of progression and surgery. We found that risk factors for eyes that have severe VKC with steroid response to eventually require surgery to be related to a longer duration of topical corticosteroid use, higher peak IOP and greater increase in IOP from baseline IOP, with the latter being the most significant risk factor.

In patients who required surgery, the mean duration of topical corticosteroid use was 18.3 ± 13.5 months as compared with patients who were controlled with topical anti-glaucoma medications (8.9 ± 6.4 months) and this could be correlated with the necessity for surgery. Most studies have reported links between duration of topical corticosteroid therapy and risk of developing steroid-response glaucoma – however, this relationship has been inconsistent.11,12 One study reported that topical corticosteroids instilled for 8 weeks or less recovered normal IOP after discontinuation; but use for more than 4 years increased the risk of glaucoma.12 In most patients the average time to steroid response is 6 weeks – although the IOP may still rise after this time period.9 We did not find any other studies that reported whether the duration of therapy would lead to increased risk of surgery.

In patients with severe VKC and steroid response, the monitoring of IOP is important throughout the duration of therapy.13 Previous studies have suggested that a higher baseline IOP could increase the risk of steroid response, but there was no correlation with risk of surgery.14 We found that both the increase in IOP from baseline and peak IOP were important indicators for increased risk of surgery. Moreover, a greater rise in IOP from baseline is significant on multivariate analysis (P = 0.011). This may reflect the severity of steroid response in the individual, which explains poor control with topical anti-glaucoma medications alone. All patients (n = 6) required more than three anti-glaucoma medications before requiring surgery.

Previous studies have studied the relationship of IOP rise with the type and potency of topical corticosteroid used.15–17 The magnitude of IOP rise depends on the potency of topical corticosteroids, with dexamethasone 0.1% the most potent, prednisolone acetate 1.0% followed by fluorometholone 0.1%.15 New corticosteroids such as loteprednol 0.5% have reported reduced effects on IOP but have lower treatment efficacy.2,14 Other studies have suggested that mixed forms of VKC have a higher risk of steroid response.14 However, we did not find that any of these factors such as type and potency of topical corticosteroid used, type of VKC, recurrence of disease or presence of corneal involvement increased the risk for surgery. Cyclosporine is an immunosuppressive drug that has been shown to reduce corticosteroid dependence in patients with VKC, by blocking Th2 lymphocyte proliferation, histamine release from mast cells and eosinophil infiltration.18–20 However, its use also did not affect the risk of steroid response and need for surgery (P = 0.512).

Most of our patients were on one anti-glaucoma medication, with most receiving Timolol 0.5% (n = 21, 58.3%). Latanoprost is shown to be most efficacious at lowering IOP with a 28% decrease in IOP in eyes having corticosteroid-induced glaucoma.21 However, it is not used often as it has been found to be not well tolerated in patients with VKC.3 Surgery was required in six of 36 patients (16.7%) in our study, which is lower than previous reports of 26.5% (9/34 patients) and 36.4% (4/11 patients) among cases of corticosteroid-induced glaucoma among VKC patients.11,23 Reported surgical outcomes for corticosteroid-induced glaucoma are generally good (83.6% success at 5 years)23 and there was surgical success in for the mean seven out of eight eyes duration of follow up (17.6 ± 14.5 months).

Mitomycin C is an alkylator that selectively inhibits DNA synthesis and has a mechanism similar to radiation.22,23 It has been found to be effective for short-term use if used topically, at low concentrations of 0.01%.22 By inhibiting both inflammatory cells and fibroblasts, it is effective when steroids or mast-cell stabilizers cannot control symptoms of VKC. However, it can cause increase in punctate keratitis if used topically and can worsen epithelial defects due to its inhibitory effects on healing.22 Thus, it is has not been used extensively in clinical practice. However, we noticed that MMC 0.02% soaked in surgical sponges applied directly to the bare sclera in the superior fornix during trabeculectomy had a substantial effect on the clinical course of VKC in our patients. In most of the cases (six of eight eyes), there was either no change or reduction in frequency of topical steroid use post-trabeculectomy, with an overall mean reduction in steroid use comparing 8 weeks pre- and postoperatively. Moreover, we observed a general improvement of the ocular surface and reduced corneal epitheliopathy post-trabeculectomy with MMC, which led to improvement in our patients' BCVA postoperatively. Thus, we postulate that MMC at this concentration, when applied directly, inhibited the fibroblasts on the ocular surface sufficiently to reduce the signs and symptoms of VKC via the subconjunctival route. This may point to a possible future application of MMC subconjunctivally, instead of topically to eliminate the risk of toxicity to the cornea.

We specifically analysed risk factors for surgery in Asian patients with severe VKC, which may have different clinical courses and outcomes from studies in existing literature. However, the limitations of our paper include the small numbers and its retrospective nature.

In conclusion, we found prolonged duration of topical corticosteroid use and greater increase in IOP from baseline to be significant risk factors for severe steroid response requiring trabeculectomy in Asian patients with severe forms of VKC. Significant improvement in the signs and symptoms of VKC is seen in eyes following trabeculectomy with MMC. This could be related to the after-effects of MMC on the ocular surface. This supports the usefulness of MMC in the management of severe, refractory forms of VKC.