Previous studies have demonstrated that SMILE has shown excellent efficacy, probable safety, and predictability for correcting myopia and myopic astigmatism [14-17]. Here, we demonstrate that SMILE surgery is effective, safe, and predictable for myopia astigmatism of >2.00 D. At 3 months post-surgery, 48 eyes (87.27%) had UCVA of 20/20 or better, and 48 eyes (87.27%) had SE within ±0.50 D. The postoperative SE and UCVA in our study are similar to recently published results [16-18].
Only a few studies have evaluated correction of high astigmatism after SMILE, especially in the vector method. Alpins vector analysis can comprehensively evaluate the outcomes of corneal refractive surgery for correcting myopic astigmatism by using the amount of astigmatism and the axial direction at the same time. Here, the vector analysis showed that the mean astigmatism in vector form was -2.12 D × 7.06° preoperatively, -0.11 D × 41.19° at 1 month post-surgery, and -0.09 D × 6.34° at 3 months post-surgery. These results indicate a reduction in the cylinder value and that the axis of astigmatism was rotated at 1 month after the operation, which is considered to be related to early postoperative wound healing and the inflammatory response . As the corneal healing response stabilized at 3 months, the axis of astigmatism mostly reversed, but the overall axis deviation of astigmatism was small.
Some studies have shown that there is a slight tendency toward undercorrection when treating astigmatism with SMILE [4,8,20-21]. Ivarsen and Hjortdal  reported an undercorrection of 13% per diopter of attempted cylinder correction in low astigmatic eyes (<2.5 D) and 16% per diopter in highly astigmatic eyes (≥2.5 D) after SMILE; they believed that the greater the preoperative astigmatism, the higher the degree of undercorrection. Pedersen et al.  reported that SMILE treatment of astigmatism seems to be predictable and effective, but with an astigmatic undercorrection of approximately 11%. Therefore, some researchers have suggested that the TIA should be increased by 10% based on the original cylindrical diopter before surgery when correcting astigmatism with SMILE . Chan et al.  reported that, in eyes with high myopic astigmatism, SMILE offered the same astigmatic correction efficacy as LASIK (laser in situ keratomileusis). The authors mentioned that the perfect astigmatism treatment is attributed to strict center positioning during the operation and higher measurement accuracy of the preoperative cylindrical diopter. Our results indicate the desirable astigmatic correction (≥2.00 D). Postoperative astigmatism vector analysis demonstrated only a slight undercorrection. The main factor affecting the logMAR UCVA at 3 months post-surgery was the preoperative astigmatism axis (P < 0.05). The influencing factors of the absolute error vector value at 3 months post-surgery were preoperative spherical diopter, preoperative cylindrical diopter, intraoperative lens thickness, lens diameter, and preoperative anterior corneal surface Km (P < 0.05), which further suggests the importance of preoperative diopter measurement accuracy and strict central positioning during the operation. Also, our study indicates that the residual astigmatism axis in the early postoperative period turned clockwise from the expected correction, which is different from the results of Pedersen et al. , who reported that the astigmatism axis rotated counterclockwise. Chan et al.  observed a slight rotation of the cylinder axis (−6.9°) in eyes with a temporal opening incision, although this was not statistically different from eyes with a superior incision (−0.39°). In the present study, the position of the side incision in both eyes was set at 120°, and that in the study of Pedersen et al.  was 30–60°, which suggests that the slight axial rotation in the early postoperative period of SMILE might be related to the surgical incision location and the healing response of the corneal incision. However, the exact reasons remains to be studied further via increased sample sizes and different incision positions.
Previous research has shown that the effect of astigmatic correction is mostly affected by the magnitude and direction of astigmatism corrected during the operation, and the type and source of astigmatism, wound healing response, laser energy, cutting center positioning, and cutting depth might also account for this [19,24-27]. In the present study, we show that the influencing factors of the absolute DV value at 3 months post-surgery are preoperative spherical diopter, preoperative cylindrical diopter, intraoperative lens thickness, lens diameter, and preoperative anterior corneal surface curvature Km (P < 0.05). Some studies have also reported that the position of the patient’s head, the rotation of their eyeballs during surgery, and the displacement of the pupil center might be the influencing factors that cause the axis rotation. As the body position changes, the eyeball rotates unconsciously, which would cause a deviation between the axis set before surgery and the axis corrected during surgery. If the eyeball were rotated >2° without correction, it would not only affect the correction of astigmatism, but would also induce significant aberrations [28-29]. When correcting astigmatism, inaccurate positioning of the astigmatism axis might cause undercorrection. An astigmatism axial deviation of 4°, 6°, 10°, 15°, and 30° would cause 14%, 20%, 35%, 52%, and complete astigmatism undercorrection, respectively [30-31]. In our study, the correlation analysis also showed that the absolute AofE value correlated positively with the absolute values of IOS and DV, and correlated negatively with the FI, indicating that accurate axial alignment and twist inspection are the key factors to achieving good visual quality after surgery. Ganesh et al.  observed that 86% of 81 highly astigmatic eyes demonstrated ≤5° cyclotorsion, and none of the eyes had ≥10° cyclotorsion; the mean magnitude was 5.5°. Based on this, they recommended manual compensation of cyclotorsion error during SMILE under the guidance of preoperative limbal markings, and observed improved results in the high-astigmatism subgroup (>1.5 D). Then, they concluded that manual compensation of the eyeball rotation angle during surgery was effective for solving the problem of cutting deviation caused by the astigmatism axial position change caused by the patient’s eyeball rotation. However, it has also been reported  that the manual marking method itself could introduce inconsistency with range of 3.8–6.0°. Here, we did not use the manual compensation method of corneal marking for intraoperative rotation error in SMILE surgery for astigmatic correction. However, the surgeon paid great attention during the operation to the correct positioning of the patient’s posture and head position and strict watermark center positioning for correcting astigmatism. The postoperative error angle of our study is similar to that of Ganesh et al. . How the accuracy of astigmatism correction in SMILE surgery can improved remains a worthwhile topic for further discussion.
For HOA, the t-HOA, spherical aberration, vertical coma aberration, and trefoil 30° all increased significantly 3 months postoperatively (P < 0.05), which is consistent with the research of Jin et al.  and Liu et al. . Jin et al.  observed the results of 196 eyes and found that after SMILE, the t-HOA of the anterior corneal surface increased, the magnitude of the horizontal coma and spherical aberration were more obvious, and the change of aberration was correlated to preoperative SE. Liu et al.  revealed that the difference in lenticule center positioning during SMILE surgery was likely to influence the postoperative HOA changes. The VN (vertex normal center) would be a better choice of reference for the optic zone center for SMILE compared with the PC (pupil center) in HOA production. In our study, the correlation analysis showed that the t-HOA was increased at 3 months post-surgery.
SMILE was mainly positively related to the preoperative spherical diopter and astigmatism, suggesting that as the expected correction degree before surgery increases, so does the thickness of the lens that has to be removed, and more corneal tissue needs to be cut, with the corresponding change in the corneal surface morphology, which leads to an increase in total postoperative HOA. The increase in spherical aberration and vertical coma after surgery was mainly related to preoperative astigmatism, suggesting that patients with high astigmatism might be more likely to have poor visual quality after surgery. Therefore, the aspheric design of the SMILE operation and strict alignment during the operation are very important, and can reduce the introduction of spherical aberration and vertical coma to a certain extent. Besides, the introduction of postoperative HOA might also be related to corneal cell apoptosis, hyperplasia and healing reactions, and poor tear film stability in the early postoperative period [19,37]. The different inclusion criteria and differences in pupil size and measuring instruments can lead to disparate results; furthermore, the proficiency of the surgeon and the setting of the surgical parameters would also affect the correlation between postoperative aberration and diopter. Therefore, many clinical studies are needed to further explore the correlation between the two.
There were a few limitations in this study. First, we included only 37 patients (55 eyes), and the follow-up duration was relatively short. A larger sample size and longer observation durations are needed in the future. Second, for bilaterally treated patients, although the two eyes of the same patient cannot be considered independent, the variance between eyes is usually less than that between subjects. Hence, the overall variance of a sample of measurements combined from both eyes was likely to be an underestimation of the true variance, resulting in increased risk of type 1 error.