This study proved that SMILE and FS-LASIK had comparable efficacy and safety and in correcting high myopic astigmatism. UDVA of 20/25 or better was 98% in SMILE and 100% in FS-LASIK, and CDVA was no loss in any eye for both groups. Chan T.C.Y. et al.  described no difference in outcomes of vision and refraction after FS-LASIK and SMILE for 3months, and Liu M. et al.’s  comparative study gave the similar result. Our current study showed the predictability within ±0.50 was 85% and 66% of spherical equivalent (SE) after SMILE and FS-LASIK (P=0.031), respectively, similar to Han T et al.  The predictability of SE after SMILE was better than FS-LASIK, which myopia was additional higher for FS-LASIK. The stability of postoperative spherical equivalent (SE) was no significant difference, and shift forward myopia was fewer at 12-months for FS-LASIK and SMILE, paralleled previous results. [13, 14]
We assessed astigmatic correction (≥2.0 diopters) by using the Alpins method, showed TIA, SIA, the magnitude of error (ME), difference vector (DV), correction index (CI), and angle of error (AE) had no statistically significant differences, other than the index of success (IOS), between FS-LASIK and SMILE group at 12-month follow-up. To the best of our knowledge, this study was the first study comparing a long-term result (12 months) including patients with high myopic astigmatism (≥2.0 diopters) by vector analysis between FS-LASIK (with compensation of cyclotorsion by iris-registration) and SMILE (with a stringent head positioning and manual limbal marking to correct).
In our results, mean preoperative cylinder of -2.65D, TIA of 2.65D in FS-LASIK, -2.51 D, and TIA of 2.51 D in the SMILE group were estimated. Postoperatively undercorrection of astigmatism was observed with a mean cylinder of -0.46 and -0.57 D, the magnitude of error (ME) of -0.22 and -0.33 D, correction index (CI) of 0.92 and 0.87 in FS-LASIK and SMILE at 12-month respectively. Zhang et al.  reported that their correction index was 0.94 in wavefront-guided FS-LASIK and 0.88 in SMILE by the analysis for a mean preoperative cylinder of TIA 2.48 to 2.65D at three months. Previous results showed a certain extent undercorrection after both surgical procedures. [6, 9, 15, 16]
The axis of the cylinder aligned was critical during the treatment. Ganesh et al.  accomplished manual cyclotorsion compensation directed by preoperative limbal marking, and also applied a 10% overcorrection nomogram. They had a mean preoperative cylinder of -2.48 D and TIA of 2.19 D, then the postoperative cylinder of -0.31 D, the magnitude of error for -0.149 D, correction index of 0.93, and index of success of 0.14 at three months. Pedersen IB et al.  suggested that a nomogram adjustment by 10% in the magnitude of astigmatism correction could be beneficial. Our study only applied about a 10% overcorrection nomogram for sphere, not for cylinder in SMILE, and long-term follow-up might be the extra favorite undercorrection. Myopic ablation was related to corneal epithelial remodeling and increased after FS-LASIK at 1-month and up to 1-year. Our results were consistent with previous that high astigmatic correction by FS-LASIK seemed better than SMILE, but no statistically significant difference in difference vector (DV), magnitude of error (ME) and Correction index (CI).
12-month post-operation in our result, an average difference of vector (DV) was 0.24@176° in FS-LASIK and 0.31@12° in SMILE. The average AE value was positive (2.14) after FS-LASIK and negative (-1.13) after SMILE. The absolute AE reported in the current study was 7.76 degrees in FS-LASIK and 8.11 degrees in SMILE, comparable to Pedersen et al.  reported AE 0.34 and absolute mean AE 8.94. Zhang JM et al.  showed corresponding results of AE -3.04 and absolute mean AE 6.08 after SMILE at 12-month follow-up. The mean absolute AE value was slightly higher after SMILE than after FS-LASIK, but our results showed no significant difference. When calculated by vector analysis, the proportion of loss of flattening effect (FE) is 1.5% when treatment is 5 degrees misaligned, 13.4% when 15 degrees, 50% when 30 degrees et al.  The residual cylinder of 0.25 D or less was 46.3% of eyes after FS-LASIK, and 24.5% of eyes after SMILE in the high astigmatic correction (≥2.0 D) (P=0.027) for 12-month in ours study. In Kanellopoulos’s study, 82% of eyes in LASIK and 50% of eyes in SMILE had a residual cylinder of 0.25 D or less for 3-month follow-up with correction of astigmatism (≥1.5 D). The difference might attribute to the disparity of astigmatic degree and shorter follow-up time. FS-LASIK by compensation of cyclotorsion expressed the better predictability of astigmatic correction.
In this study, three steps to control the compensation of cyclotorsion and others in the SMILE procedure were performed: preoperative marking on the limbal, a stringent head positioning, and limbal marking matching the horizontal line of the microscope cross, but did not apply gently rotating the cone on the eye when started suction, as the manual limbal marking method might introduce varying degrees from 3.8 to 6.0 degrees inherently. Prickett AL et al. observed that most of the rotations previously attributed to torsional components were probably due to non-cyclotorsion components, such as postural misalignments. Chan T.C.Y et al. also emphasized the position of the head to both eyes align along an imagined horizontal line without the manual limbal marking, further reason might be 86% of eyes for high astigmatism within 5 degrees or less cyclotorsion in the survey by Ganesh et al. Manual limbal marking and stringent head positioning both affirmed a safe, feasible, and effective strategy to perfect the results of high astigmatic correction with SMILE.
Shen EP et al.’s  compared manual limbal markings versus iris-registration systems in LASIK and concluded manual limbal marking was a safe alternative when automated systems are not available. Zhao XH et al.  also showed that wavefront-guided FS-LASIK and optimized SMILE achieved similar outcomes for astigmatism correction. Though various methods aligned cyclotorsion for SMILE procure [16, 17, 25] to improve astigmatism correction. In contrast, the system of iris-registration for the Visumax platform is warranted in the future to enhance its capability in astigmatic correction.
We got less efficacy in correcting astigmatism after SMILE by a stringent head positioning than FS-LASIK by an iris-registration system with compensation of cyclotorsion. An IOS of 0.00 indicates complete success in astigmatism treatment, a result of 1.00 shows no improvement on the preoperative status, and greater than 1.00 shows a deterioration in astigmatism. Index of success (IOS) was various from 0.07 to 0.17 of FS-LASIK and 0.09 to 0.15 of SMILE for 3 to 6 months follow-up in the previous result. [9, 15] Our study showed IOS was 0.18 of FS-LASIK and 0.24 of SMILE, with a significant difference between the two groups at 12 months. The IOS was higher than the previous might be long-term follow-up in our report. Ivarsen A et al.  also demonstrated a tendency for greater under-correction over time with higher degrees of astigmatic correction. Higher undercorrection with high astigmatic correction over time might be associate with corneal epithelial remodeling. 
The limitation of this study was its small sample size of high astigmatic correction. Varma R et al.  showed the prevalence of astigmatism was only 3.7% (3.1%-4.3%) for high astigmatism (defined as over 2.25 D) in adults of Chinese Americans. From the beginning of the study, we enrolled 840 right eyes of 840 patients for high astigmatism (≥2.0 diopters); only 94 eyes (11.2%) finished the process at 12-month follow-up.
Second, we got postoperative refraction using an automatic optometer which might be impacted by the eye accommodation for troublesome to acquire the manifested and cycloplegic refraction at each follow-up. Because of the age enrolled in our study (adult), the accommodation might less affect the results. Pesudovs K  proved excellent agreement between autorefraction and subjective refraction, and autorefraction is a valid outcome measure of refractive surgery.
Third, the current study was retrospective design; thus, the potential selection bias could not be excluded in a random clinical trial. Even more prospective, randomized clinical trials should therefore be advanced with more sample size.
In conclusion, SMILE with stringent head positioning provided fewer favorable high astigmatic corrections (≥2.0 diopters) at 12-months than FS-LASIK with compensation of cyclotorsion. Further study should improve SMILE with an eye-tracking system and cyclotorsion compensation to correct astigmatism accurately in the future.