The classic Dresden CXL using ultraviolet light of 3mw/cm2 illumination is a time-consuming procedure, which can cause high percentage of haze  and the risk of progressive corneal flattening , especially in simultaneous combined treatments. The accelerated CXL uses ultraviolet light of high irradiation intensity. According to the Bunsen-Roscoe law of photochemical effect, the higher the illumination, the shorter the exposure time .The exposure time required is greatly shortened, which improves the treatment efficiency and increases patients’ compliance. Additionally, accelerated CXL was proved to be effective and have fewer complications [4, 5]. So, recently, accelerated CXL has been widely used by most surgeons for CXL combined treatment [13, 16,17,18, 20]. In the current study, 30mW/cm2 ultraviolet light intensity of illumination for 4 min was used for the first time.
Initially, epithelial removal prior to CXL was performed using manual debridement (with or without alcohol) [13,14,15]. Recently, transepithelial phototherapeutic keratectomy (PTK) has also been employed as a method for removing the epithelial [16,17,18]. Because of a doughnut-shaped model of the corneal epithelium in keratoconus , PTK removes some stromal tissue from the central cone, which flattens the cornea more than manual debridement , but consumes more corneal stromal tissue at the cone apex. The corneal surface following PTK is more consistent with the preoperative topography, which makes the subsequent TG-PRK become more accurate. PTK epithelial removal leaves some epithelium around the cone, which may reduce the ablation volume of the corneal stromal tissue during TG-PRK, but can cause under-correction. In the present study, manual debridement with alcohol was used, which could reduce the loss of corneal stroma at the cone apex to the greatest extent, and reduced the excessive friction stimulation of corneal stroma by mechanical scraping.
Topography-guided customized ablation attempts to maintain the aspheric shape of the cornea and neutralize corneal irregularities . It has been shown to be effective in treating irregular astigmatism caused by iatrogenic corneal irregularities. Since topography-guided ablation for normalizing the anterior cornea can bring in refractive change, especially astigmatism change , the clinical refraction should be adjusted to keep it neutral after refractive correction. Kanellopoulos found that topography-modified refraction (TMR) offered superior refractive and visual outcomes to standard clinical refraction in myopic topography-guided LASIK . In the current study, we used TMR for reference to compensate for partial cylinder measured by topographer without refractive correction while topography-guided ablation of irregular corneas on the basis of ensuring the depth of corneal ablation.
In the current study, UDVA improved slightly after surgery, but there was no statistical significance. BSCDVA improved significantly from 0.32 ± 0.20 logMAR to 0.15 ± 0.14 logMAR at postoperative 12 months (P < 0.05). Manifest refraction, flat K, steep K and corneal cylinder all decreased slightly after surgery, but no statistical significance was found. However, Kapexflattened significantly after surgery. Generally speaking, the improvement of UDVA, the decrease of refraction and corneal curvature were not obvious, which were different from most previous studies [10,11,12,13,14,15,16,17,18,19,20]. Because of partial correction of clinical refraction, there were improvements in UDVA [8,9,10,11,12,13, 15,16,17,18], curvature readings [13, 15, 17,18,19,20] and manifest refraction [13, 15, 16, 18,19,20] after surgery. The CXL plus TG-PRK in the present study aimed at halting the progression of keratoconus and reducing corneal HOA, so we did not carry out refractive correction, which was a main difference between the present study and previous studies [10,11,12,13,14,15,16,17,18,19,20]. Despite of the different TG-PRK protocol, the present study still showed improvements in BSCDVA and Kapexsimilar to the previous studies [10,11,12,13,14,15,16,17,18,19,20], which increased the correction effect of frame glasses or ICL implantation when contact lenses were not tolerated or helped to regain a good contact lenses fitting.
The current study also showed that SIf, KVf and BCVf, which reflected the irregularity of the anterior corneal surface, reduced significantly after surgery and remained in a reduced state up to 12 months after surgery. Those indicated that the cornea became regular after surgery and resulted in the improvement of BSCDVA. Some previous studies showed similar results that index of surface variance (ISV) and index of height decentration (IHD) decreased significantly after surgery. The current study also found that corneal aberrations including corneal HOA and coma decreased continuously within 12 months after surgery, which also indicated the improvement of corneal regularity due to TG ablation. Ahmet et al. and Rechichi et al.  found that corneal HOA, coma and spherical aberration (SA) all significantly decreased at 24 months after surgery. The difference of the change of corneal SA might related to the different excimer laser system (Schwind Amaris, Germany) and different ablation mode (Transepithelial TG-PRK).
Like some previous studies [10, 12, 16, 18], MMC was not used in the present study. Kymionis GD et al. found that CXL could destroy the regeneration of corneal anterior stromal cells by confocal microscopy, so it was not necessary to use MMC after PRK combined CXL . The synergistic effect of CXL and MMC may cause more cell death and more corneal haze .Although haze was observed at each case in postoperative one month in the present study, it gradually faded away 3 to 12 months after operation. This slightly obvious haze may be related to alcohol deepithelialization and 4-week treatment of glucocorticoid eyedrops after surgery.
The limitations of this study are as follows. First, this study had a small sample and was lack of a control group. Second, we didn’t estimate the impact of corneal posterior surface when planning TG ablation. In future, software based on calculation of mean pupillary power or raytracing technology compensating anterior and posterior corneal surfaces refractive and aberrometric contributions are mandatory to optimize visual and refractive outcomes. Third, like most previous studies, we used uniform protocol in all cases. Pachymetry-based or topography-guided customized CXL might be more effective and safer [18, 31, 32]. Rechichi et al.offset ablation center towards the location of cone apex, which might better reduce coma. These customized procedures provided a good reference for our surgical protocol in the future.
It has been reported that a small number of keratoconus (around 8 %) may still progress after CXL . Thus regular and long-term follow-up of refraction and corneal topography/tomography at different time points after surgery is necessary, especially after the combined surgery with reduced corneal thickness. The research of long-term safety is warrant.