<h3>Introduction</h3>
Keratoconus is a progressive, bilateral, asymmetric, non-inflammatory corneal ectasia. Biomechanical instability of corneal collagen fibres results in progressive increased corneal curvature and thinning, causing irregular astigmatism and loss of visual acuity. It typically presents in puberty and stabilises in the fourth decade. Keratoconus in children (<18 years of age) is often more advanced at diagnosis and may progress more rapidly. Collagen cross-linking (CXL) treats keratoconus by strengthening corneal stromal collagen bonds with riboflavin activated by ultraviolet A (UVA).
Originally introduced by Wollensak and colleagues, CXL is now a first-line treatment for progressive keratoconus, and its ability to halt keratoconus progression in adults has been well established. CXL in paediatric keratoconus patients has been much less extensively studied. This Irish study aimed to assess the visual, refractive and topographic outcomes of CXL in patients ≤18 years of age treated at a single centre.
<h3>Patients and methods</h3>
The medical records of consecutive patients ≤18 years of age who underwent CXL treatment for keratoconus at the Mater Private Hospital between December 2009 and August 2013 were retrospectively reviewed. A diagnosis of keratoconus was based on clinical findings and/or corneal images generated by the Orbscan IIz Corneal Topographer (Orbtek; Bausch and Lomb, US). Exclusion criteria were as follows: history of a significant ophthalmic condition other than keratoconus or its associated comorbidities, and/or previous ophthalmic surgery, and preoperative corneal thickness <400microns (µm). Data analysis focused on sex, age at presentation, relevant comorbidities (eg, atopy), age at surgery, and length of follow-up. We analysed baseline and follow-up measurements of visual acuity, refractive error and corneal topographic values. Uncorrected distance visual acuity was assessed preoperatively and at all follow-up visits. Best-corrected distance visual acuity was recorded preoperatively and at yearly follow-up visits. Both were assessed with the Early Treatment Diabetic Retinopathy Study (ETDRS) logarithm of the minimal angle of resolution (logMAR) charts (Lighthouse International, New York, NY). We also recorded refractive indices preoperatively and at yearly follow-up.
Corneal topography was assessed and the following parameters were analysed: keratometry values — max (Kmax), min (Kmin), and mean values (Kmean); and thinnest corneal depth. Follow-up intervals were at three months, six months, and one year for all patients and at two and three years where available. Changes in Kmax, as described by Koller and colleagues, were defined as Kmax regression (>1 dioptre (D) decrease in Kmax), Kmax stabilisation (<1 D change in Kmax), and Kmax progression (>1 D increase in Kmax).
<h3>CXL procedure</h3>
CXL was performed under aseptic technique in the operating theatre under topical anaesthesia, if possible, with proxymetacaine 1 per cent eye drops.
After insertion of a lid speculum, alcohol 20 per cent solution was applied for 30 seconds in a 7mm diameter well (as per the Dresden protocol) placed on the corneal surface. The cornea was then rinsed with balanced salt solution and the central epithelium was removed.
Between December 2009 and February 2012 we used the UV-X illumination system, version 1,000 (IROC AG, Switzerland) cross-linking protocol. Starting five minutes prior to irradiation, we applied riboflavin 0.1 per cent solution (10mg riboflavin-5-phosphate in 10ml dextran solution) and continued for 30 minutes. A dose of 5.4J/cm2 UVA was applied at 1cm from the cornea for 30 minutes.
From March 2012, we used the VibeX Rapid (Avedro, US) cross-linking protocol, which entails a 10-minute soak time with riboflavin 0.1 per cent solution (100mL of solution contains: riboflavin 0.1g; HPMC; disodium hydrogen phosphate; sodium phosphate monobasic dihydrate; sodium chloride; water for injectable solution); and four minutes of exposure with 30mw/cm2 UVA.
Postoperatively, all patients had chloramphenicol 0.5 per cent eye drops instilled and a bandage contact lens inserted, which was removed after three days, and antibiotics were continued four times daily for one week. Dexamethasone 0.1 per cent eye drops were commenced on postoperative day seven and continued for six weeks.
<h3>Statistical analysis</h3>
Visual acuity was converted to a logarithm of the minimal angle of resolution units (logMAR) to facilitate analysis. Statistical analysis was performed using the STATA computer package (StataCorp, XXX, TX). Results are reported in mean values with standard deviations, unless otherwise stated. Comparisons were made using paired t tests. A P value of <0.05 was considered statistically significant.
<h3>Results</h3>
A total of 25 eyes of 14 patients (11 males) were included. Mean age at surgery was 16.2 ± 1.6 years (range, 13-to-18 years). Preoperative and follow-up data at three months, six months and one year was available for all patients; two-year follow-up was available for 11 patients; and three-year data for eight patients. All but one patient (13 years old) underwent CXL under topical anaesthesia. The UV-X Illumination System protocol was used for seven eyes; the Vibex Rapid for 18 eyes. Five patients (35.7 per cent) had a documented history of atopy (asthma, hay fever), four patients (28.6 per cent) had a history of isolated eye-rubbing behaviour, and the remaining five patients had no history of either.
<h3>Visual acuity and refractive data</h3>
Mean uncorrected visual acuity was 0.53 ± 0.32 at baseline; 0.55 ± 0.36 at three months; 0.51 ± 0.37 at six months; and 0.46 ± 0.36 at one year. There was no significant difference between the preoperative and one-year follow-up values (P = 0.07). Mean best-corrected visual acuity was 0.3 ± 0.26 at baseline and 0.15 ± 0.12 at one year. This improvement was significant (P = 0.01).
The mean preoperative spherical equivalent was −4.42 ± 3.9 D; the mean cylinder value, 3.65 ± 4.2 D. These values were not significantly changed at one year; mean spherical equivalent was −5.5 ± 3.4 D (P = 0.09); mean cylinder, −4.5 ± 2.2 D (P = 0.32).
<h3>Tomography</h3>
Tomographic data includes astigmatism, keratometry values and thinnest corneal area (thinnest).
Statistical analysis was performed to assess the significance of changes in data from preoperative to three-month, six-month and one-year values.
The mean astigmatic value was 4.76 D (± 2.73), preoperatively. This was statistically unchanged at 4.63 ± 2.7 D (P = 0.85) at one year. Mean baseline Kmax, Kmin, and Kmean values were 49.62 ± 4.5D, 44.68 ± 3.5D, and 46.3 ± 2.84D, respectively. There was no significant change in the mean values of Kmax, Kmin or Kmean after one year.
At one year, five eyes (20 per cent) showed, compared with baseline values, Kmax regression; 13 (52 per cent) showed stabilisation; seven (18 per cent) showed progression. Of the seven eyes whose Kmax progressed at one year, five had two-year follow-up showing stabilisation in three and regression in two eyes. In three of these patients, three-year follow-up showed Kmax stabilisation.
There was a significant reduction in corneal thickness at the thinnest area measured on the Orbscan at three and six months’ follow-up.
The mean preoperative value was 473.6 ± 37.68µm, which fell to 383.47 ±76.9µm at three months and recovered to 424.55 ± 70.2µm at six months and 452.82 ± 53.5µm at one year after CXL. There was no incidence of postoperative infective keratitis or scarring.
<h3>Challenges</h3>
Paediatric keratoconus poses unique management challenges. It is recognised that there is an inverse relationship between age and severity of keratoconus and that the disease presents at a more advanced stage (27.8 per cent being Stage 4 compared with 7.8 per cent of adults). It also progresses more frequently and rapidly in children with subsequent advanced vision loss.
There is also an inverse relationship between age and risk of development of acute hydrops, as demonstrated by Suhaibani and colleagues, and younger age is recognised as an independent risk factor for requiring penetrating keratoplasty in keratoconus.
Corneal biomechanical rigidity is weaker at a younger age; this may contribute to the severity of progression of the disease in children. Eye-rubbing in conditions such as vernal kerato-conjunctivitis and other allergic eye diseases may be a more significant challenge in children, both because of increased incidence and their more compulsive behaviour. Children are also less likely to tolerate correction of keratoconus with contact lenses, posing further challenges to the treatment of keratoconus in this group.
While uncommon, keratoconus may also be associated with multisystem conditions, such as Turner syndrome, trisomy 21, mitral valve prolapse and Ehlers-Danlos syndrome. Treatment of keratoconus in patients with trisomy 21 and other causes of learning disability poses significant challenges, with higher rates of graft failure in penetrating keratoplasty; therefore, early intervention with the less-invasive CXL may be of particular benefit in this population.
<h3>International studies</h3>
An early, small case series reported promising outcomes in treating paediatric keratoconus patients with CXL. The Siena Paediatrics CXL trial remains the largest study of CXL in children to date. This prospective study of 152 eyes of 77 patients (10-to-18 years) showed improvement in uncorrected and best-corrected visual acuity of +0.18 and 0.16 Snellen lines, respectively.
Improvement in K readings and asymmetry index values was also demonstrated. There were no adverse events with 36 months’ follow-up.
Vinciguerra and colleagues conducted a prospective study of 40 eyes of 40 patients with two years’ follow-up and also demonstrated improvement in uncorrected and best-corrected visual acuity, refractive indices and keratometry. Endothelial cell count (ECC) values were unchanged, and they also showed improvement in total and higher-order aberrations for a 3mm pupil.
In a retrospective study of 46 eyes treated with CXL in patients aged nine-to-19 years, Chatzis and Hafezi showed improvement in corrected visual acuity, maximal keratometry and keratoconus index, although significant Kmax improvement at 24 months lost significance at 36 months.
They also demonstrated a preoperative progression rate of 88 per cent in their patients and therefore recommend treatment with CXL as soon as the diagnosis has been made.
Soeters et al recently compared one-year outcomes of CXL in three age groups: paediatric (<18 years), adolescent (18-to-26 years), and adult (>26 years). The 31 eyes in the paediatric group showed significantly greater improvement in corrected visual acuity than did the adolescent eyes. The investigators also found that cones were more likely to be centrally located in paediatric patients than in adolescents. Kmax improved in all groups, and there was no change in ECC.
While epithelium-off CXL remains by far the most accepted CXL method, transepithelial CXL (TE-CXL) has also been recently studied in paediatric populations. Magli et al compared standard protocol with TE-CXL and concluded that both methods were effective treatments and that TE-CXL was associated with less pain and fewer complications.
In a prospective, comparative study of 22 eyes, Salman concluded that TE-CXL was XXX in a paediatric population. In contrast, Buzzonetti et al, in a small case series of 13 paediatric eyes treated with CXL, showed improvement in corrected visual acuity but deterioration in K readings and higher-order aberrations.
There are no reports of serious complications in any of the studies listed.
<h3>Conclusion</h3>
In our case series, CXL stabilised uncorrected visual acuity, refractive indices and K values and improved best-corrected visual acuity at one year in the treatment of keratoconus. Eighteen per cent of eyes showed progression of Kmax at one year, but this stabilised or improved at two and three years’ follow-up.
In our experience, younger patients tolerate the treatment very well under local anaesthetic, usually when accompanied by their parents in the operating theatre. (notably, only one patient required general aesthetic). The application of the faster VibeX Rapid protocol improves this tolerability. In addition, though it was not objectively recorded, we noted better postoperative pain tolerance among children vs adults. It is our policy to treat paediatric keratoconus following initial diagnosis rather than to wait for documented progression, as advocated by Chatzis and Hafezi .
CXL is a new and exciting non-invasive procedure. It halts the progression of keratoconus in paediatric patients and should be considered at initial diagnosis of the condition. Special consideration should be given to patients with Down syndrome to prevent progression to more advanced and often complicated stages and treatments of keratoconus.
<strong>References available on request</strong>
