Remember me
Photorefractive surgeries have evolved over the years from photorefractive keratectomy (PRK) to laser in situ keratomileusis (LASIK) and recently introduced small-incision lenticule extraction (SMILE). They are a common clinical practice for treating myopia, hyperopia, and astigmatism.1 Each of these procedures modifies the keratometry and central corneal thickness to improve the corneal refractive status and, eventually, visual outcomes.2 However, each poses a risk of keratectasia, a rare but serious complication.3
Corneal crosslinking (CXL) was introduced as a treatment option for patients with keratoconus (KCN), but it is also used to treat post-LASIK ectasia and pellucid marginal degeneration.4,5 Riboflavin (vitamin B) and UV-A are used in CXL to enhance the strength of the corneal collagen and make it resistant to deformation. This process significantly increases the rigidity of stromal collagen fibers by strengthening the covalent bonds among them.6,7 Many studies have come up with the prophylactic use of CXL with refractive surgery.8,9 The purpose of this procedure was to prevent high-risk corneas from developing ectasia.
We performed this research to compare refractive surgeries with vs without prophylactic CXL.
METHODSThis systematic review and meta-analysis was performed and reported in accordance with the Cochrane Collaboration Handbook for Systematic Review of Interventions and PRISMA statement guidelines.10,11
Eligibility CriteriaInclusion criteria for the studies in our meta-analysis were (1) randomized controlled trials or observational studies, (2) comparing refractive surgeries (SMILE, LASIK, and PRK) with and without CXL, (3) enrolling patients who underwent refractive surgery for myopic correction, and (4) studies reporting any of the clinical outcomes of interest. The study with a longer follow period or having the outcome of interest was selected among the studies that had an overlapping patient population.
The studies were excluded based on the following criteria: (1) no control group; (2) patients with a history of KCN, keratectasia, post-LASIK ectasia, and hyperopia; (3) patients with previous refractive eye surgery; (4) review articles; (5) letters to the editor; (6) conference abstracts; and (7) animal studies.
Search Strategy and Data ExtractionWe systematically searched 3 databases including PubMed, EMBASE, and the Cochrane Central Register of Controlled Trials from inception to April 2023 with the following search strategy: “SMILE,” “small incision lenticule extraction,” “LASIK,” “femtoLASIK,” “laser-assisted in situ keratomileusis,” “PRK,” “photorefractive keratectomy,” “XTRA,” “cross linking,” “crosslinking,” and “CXL.”
Three authors (S.H., K.K., G.D.V.) extracted the data independently according to the predefined inclusion and exclusion criteria. The protocol for our meta-analysis was registered on PROSPERO on April 26, 2023, under protocol ID CRD42023417551.
End Points and Subgroup AnalysesOutcomes included postoperative corrected distance visual acuity (CDVA) logMAR; uncorrected distance visual acuity (UDVA) of 20/20 or better; mean refractive spherical equivalent (MRSE) at ≥12 months; mean keratometry; central corneal thickness (CCT); efficacy index; safety index; safety, that is, CDVA same or better; predictability, that is, spherical equivalent within ±0.50 diopter (D); and cylindrical correction within ±0.50 D at mean follow-up.
Prespecified subgroup analyses included data restricted to (1) SMILE Xtra vs SMILE only, (2) LASIK Xtra vs LASIK only, and (3) PRK Xtra vs PRK only for each outcome.
Quality AssessmentWe assessed the quality of the randomized controlled trial using the RoB-2 tool by Cochrane.12 Nonrandomized studies were assessed with the ROBINS-I tool.13
Two authors (S.H. and K.S.) completed the risk of bias assessment independently. The disagreements were resolved through a consensus after discussing the reasons for the discrepancy.
Publication bias was investigated by funnel plot analysis of the primary outcome and evaluation of point estimates in relation to study weights.
Statistical AnalysisRisk ratios (RRs) with 95% CIs were used to compare treatment effects for dichotomous outcomes. Mean differences were used for continuous outcomes. Heterogeneity was assessed with I2 statistics and the Cochrane Q test; P values <0.10 and I2 > 25% were considered significant for heterogeneity. We used the DerSimonian and Laird random-effects model.
Review Manager v. 5.4.1 was used for statistical analysis.14 Sensitivity analysis was performed by removing individual studies from the outcomes assessment by using R v. 4.3.1.15
Safety and efficacy indices were calculated using CDVA and UDVA values. The efficacy index was determined using the formula: mean postoperative UDVA (decimal) divided by mean preoperative CDVA (decimal). Similarly, the safety index was calculated by dividing mean postoperative CDVA (decimal) by mean preoperative CDVA (decimal).
RESULTS Study Selection and Baseline CharacteristicsAs shown in Supplementary Figure 1 (available at https://links.lww.com/JRS/B78), our search strategy generated 1065 results. Fifty-seven studies remained after removing the duplicates and ineligible studies based on the title and abstracts. These 57 studies were reviewed in full, and 28 studies were selected to be included in the meta-analysis. There were 7 studies for SMILE Xtra vs SMILE only, 15 studies for LASIK Xtra vs LASIK only, 2 studies for PRK Xtra vs PRK only, and 4 studies for transepithelial PRK (tPRK) Xtra vs tPRK only. There were 4 randomized controlled trials (RCTs) and 24 observational studies. A total of 1386 eyes (49.14%) underwent refractive surgery with CXL, and 1434 eyes (50.85%) underwent refractive surgery without CXL.
Baseline study characteristics are summarized in Table 1. Follow-up ranged from 1 to 24 months. The mean age (years), mean refractive spherical equivalent MRSE (D), central corneal thickness CCT (μm), mean keratometry (D), and ablation depth (μm) values are mentioned for both groups.
Table 1. - Baseline characteristics of included studies Study Type Surgery Follow-up (months) I/C (n) Age (years)CCT = central corneal thickness; CXL = corneal crosslinking; HPMC = hydroxypropyl methylcellulose; I/C = intervention (CXL group)/control (no CXL group); MRSE = mean refractive spherical equivalent; NRCT = nonrandomized controlled trial; PRK = photorefractive keratectomy; RCT = randomized controlled trial; tPRK = transepithelial photorefractive keratectomy
aMitomycin 0.02% for 20 seconds
The riboflavin concentration ranged from 0.1% to 0.25% and soaking time ranged from 45 seconds to 15 minutes in various studies. All studies used UV-A irradiation of 30 mW/cm2, except for 5 studies, which used 9, 18, and 45 mW/cm2. The UV-A irradiation time ranged from 45 seconds to 5 minutes. The energy dose ranged from 0.8 to 5.4 J/cm2.
Visual Acuity and Refractive OutcomesThe postoperative CDVA (logMAR) at ≥12 months favored the no CXL group (mean difference [MD], 0.01 D; 95% CI, 0.00-0.02 D; P = .02; I2 = 13%), whereas there was no difference in performing subgroup analysis for SMILE (MD, 0.01 D; 95% CI, −0.01 to 0.02 D; P = .45; I2 = 0%) and LASIK (MD, 0.02 D; 95% CI, −0.00 to 0.03 D; P = .06; I2 = 38%) groups. The test for subgroup differences showed I2 = 0% (Figure 1).
Figure 1.: Postoperative CDVA (logMAR) at ≥ 12 months. The CDVA logMAR favored the no CXL group (MD, 0.01 D; 95% CI, 0.00-0.02 D; P = .02; I 2 = 13%). CXL = corneal crosslinking; MD = mean difference
There was no difference in postoperative UDVA of 20/20 or better at ≥12 months in the overall comparison of both groups (RR, 1.01; 95% CI, 0.99-1.04; P = .34; I2 = 0%) and subgroup analysis of SMILE (RR, 1.00; 95% CI, 0.96-1.03; P = .83; I2 = 0%), LASIK (RR, 1.06; 95% CI, 0.99-1.13; P = .11; I2 = 11%), and PRK (RR, 1.01; 95% CI, 0.97-1.06; P = .55; I2 = 0%). The test for subgroup differences showed I2 = 10.0% (Figure 2).
Figure 2.: Postoperative UDVA of 20/20 or better at ≥12 months. The UDVA was comparable in both groups (RR, 1.01; 95% CI, 0.99-1.04; P = .34; I 2 = 0%). RR = risk ratio
There was no difference in postoperative MRSE (D) in the overall comparison of both groups (MD, −0.01 D; 95% CI, −0.07 to 0.05 D; P = .74; I2 = 85%) and subgroup analysis for SMILE (MD, −0.07 D; 95% CI, −0.16 to 0.01 D; P = .07; I2 = 49%), LASIK (MD, 0.02 D; 95% CI, −0.09 to 0.13 D; P = .70; I2 = 84%), and PRK (MD, 0.01 D; 95% CI, −0.10 to 0.11 D; P = .89; I2 = 91%). The test for subgroup differences showed I2 = 20.0% (Supplementary Figure 2, available at https://links.lww.com/JRS/B79).
There was no difference in postoperative cylindrical correction within ±0.50 D in the overall comparison of both groups (RR, 0.99; 95% CI, 0.97-1.02; P = .57; I2 = 0%) and in subgroup analysis for SMILE (RR, 0.96; 95% CI, 0.90-1.02; P = .23; I2 = 0%) and LASIK (RR, 0.99; 95% CI, 0.97-1.02; P = .63; I2 = 0%). The test for subgroup differences showed I2 = 0% (Supplementary Figure 3, available at https://links.lww.com/JRS/B80).
Topographic OutcomesThere was a significant difference in postoperative CCT in the overall comparison of both groups (MD, −11.37 μm; 95% CI, −15.10 to −7.64 μm; P < .00001; I2 = 12%) and in subgroup analysis for LASIK (MD, −13.24 μm; 95% CI, −18.17 to −8.31 μm; P < .00001; I2 = 0%) and PRK (MD, −10.41 μm; 95% CI, −15.97 to −4.84 μm; P = .0002; I2 = 0%). However, there was no difference in SMILE (MD, −8.18 μm; 95% CI, −22.66 to 6.30 μm; P = .27; I2 = 58%). The test for subgroup differences showed I2 = 0% (Figure 3).
Figure 3.: Postoperative CCT. The postoperative central corneal thickness was considerably less in the crosslinking group (MD, −11.37 μm; 95% CI, −15.10 to −7.64 μm; P < .00001; I 2 = 12%). CCT = central corneal thickness; MD = mean difference
There was no difference in postoperative mean keratometry value in the overall comparison of both groups (MD, −0.19 D; 95% CI, −0.41 to 0.04 D; P = .11; I2 = 0%) and subgroup analysis for SMILE (MD, −0.30 D; 95% CI, −0.91 to 0.32 D; P = .34; I2 = 0%), LASIK (MD, −0.34 D; 95% CI, −0.69 to 0.01 D; P = .05; I2 = 0%), and PRK (MD, 0.00 D; 95% CI, −0.34 to 0.34 D; P = 1.00; I2 = 0%). The test for subgroup differences showed I2 = 2.7% (Supplementary Figure 4, available at https://links.lww.com/JRS/B81).
Safety and Efficacy IndicesThere was a significant difference in postoperative safety index in the overall comparison of both groups (MD, −0.02; 95% CI, −0.04 to −0.01; P = .001; I2 = 15%) and subgroup analysis for SMILE (MD, −0.03; 95% CI, −0.05 to −0.01; P = .0006; I2 = 0%) and LASIK (MD, −0.02; 95% CI, −0.04 to −0.01; P = .01; I2 = 0%). However, it was not significant in PRK (MD, 0.03; 95% CI, −0.04 to 0.09; P = .45; I2 = 57%). The test for subgroup differences showed I2 = 25.2% (Figure 4).
Figure 4.: Postoperative safety index. The postoperative safety index was significant between both groups (MD, −0.02; 95% CI, −0.04 to −0.01; P = .001; I 2 = 15%). MD = mean difference
There was a significant difference in the efficacy index of both groups in the overall comparison (MD, −0.02; 95% CI, −0.05 to 0.00; P = .02; I2 = 42%) and subgroup analysis for SMILE (MD, −0.04; 95% CI, −0.06 to −0.01; P = .003; I2 = 16%). However, it was nonsignificant in LASIK (MD, −0.02; 95% CI, −0.07 to 0.02; P = .27; I2 = 53%) and PRK (MD, 0.01; 95% CI, −0.05 to 0.07; P = .79; I2 = 65%). The test for subgroup differences showed I2 = 0% (Figure 5).
Figure 5.: Postoperative efficacy index. The postoperative efficacy index was significant between both groups (MD, −0.02; 95% CI, −0.05 to −0.00; P = .02; I 2 = 42%). MD = mean difference
Safety and PredictabilityThere was no difference in safety, that is, postoperative CDVA same or better, in the overall comparison of both groups (RR, 0.99; 95% CI, 0.97-1.01; P = .33; I2 = 47%) and subgroup analysis for SMILE (RR, 0.95; 95% CI, 0.85-1.06; P = .34; I2 = 80%), LASIK (RR, 0.98; 95% CI, 0.96-1.02; P = .32; I2 = 48%), and PRK (RR, 1.00; 95% CI, 0.98-1.02; P = .81; I2 = 0%). The test for subgroup differences showed I2 = 0% (Supplementary Figure 5, available at https://links.lww.com/JRS/B82).
There was no difference in postoperative spherical equivalent predictability within ±0.50 D of both groups in the overall comparison (RR, 1.02; 95% CI, 0.98-1.07;
Comments (0)