Preoperative Corneal Astigmatic Direction is Associated with Stromal Mechanical Anisotropy in SMILE-Derived Lenticules Authors: Pengfei Han, Peng Chen, Xiaolong Fang, Hongwei Qin, Yuan Wu, Shuangshuang Chen, Yan Li, Jie Zhang, Qizhi Zhou Published: 2026 ABSTRACT PurposeDirection-dependent mechanical behavior exists in the human corneal stroma, but fixed-axis biomechanical testing may be influenced by preoperative astigmatic direction. This study evaluated whether corneal astigmatic direction, a routinely measurable clinical phenotype, corresponds to superior–inferior (SI) vs nasal–temporal (NT) mechanical anisotropy in human stromal lenticules obtained during small-incision lenticule extraction (SMILE).Patients and MethodsFifty-three spherical stromal lenticules with a programmed cylinder correction of 0 diopters were prospectively collected. Preoperative anterior corneal astigmatism was classified as with-the-rule (WTR) or against-the-rule (ATR), and the steep-axis information was decomposed into a vertical–horizontal component matched to the SI–NT mechanical testing coordinates. Equibiaxial testing was performed along the SI and NT axes, and tangent moduli were calculated at 0.03 MPa. The signed meridional difference index (MDI) was defined as ESI − ENT.ResultsWTR lenticules predominantly showed higher SI modulus, whereas ATR lenticules predominantly showed higher NT modulus. The vertical–horizontal astigmatic component was strongly correlated with signed MDI (Spearman rho = 0.875, P < 0.001), and this association remained significant after multivariable adjustment (β = 0.673 MPa/diopter, 95% CI 0.537–0.809, P < 0.001). Corneal astigmatism magnitude was also associated with direction-independent anisotropy magnitude.ConclusionPreoperative corneal astigmatic direction may be a clinically measurable stratification variable for interpreting fixed-axis stromal mechanical anisotropy measurements in SMILE-derived lenticules. These findings support incorporating directional clinical phenotypes into corneal biomechanical research and patient-specific modeling, while requiring validation in larger independent samples. Introduction The mechanical response of the human corneal stroma is direction dependent.1–3 Stromal lenticules removed during small-incision lenticule extraction (SMILE) provide surgically derived human tissue that can be tested directly to investigate the mechanical behavior of the central corneal stroma.4–9 Previous uniaxial and biaxial studies have compared mechanical responses along the superior–inferior (SI) and nasal–temporal (NT) directions in human corneas or SMILE lenticules, but their findings have not been entirely consistent. Some studies reported a higher tangent modulus or greater overall macroscopic stiffness in the vertical or SI direction,1,4,7–9 whereas others did not detect a clear SI–NT tensile difference.10,11 Brillouin microscopy studies have also demonstrated direction-dependent corneal mechanical responses using a different measurement approach.2,3 In addition to differences in testing methods, stress ranges, and mechanical metrics, directional clinical heterogeneity that has not been adequately considered may influence group-level findings. Corneal astigmatism contains both the magnitude of curvature difference and axial information. With-the-rule (WTR) astigmatism is characterized predominantly by a steep meridian close to the vertical direction, whereas against-the-rule (ATR) astigmatism is characterized predominantly by a steep meridian close to the horizontal direction.12 However, the actual steep axis is continuous and is not always exactly aligned with the fixed 90° or 180° directions. Representative previous studies of corneal stromal mechanics have generally compared tissue responses along fixed anatomic axes,1,4,7–11 without further stratifying specimens by preoperative astigmatic direction or expressing the astigmatic axis in the same coordinate system as SI–NT mechanical testing. If different astigmatic directions correspond to different SI–NT mechanical differences, direct pooling may strengthen, weaken, or partially cancel contributions of opposite sign. Classical astigmatic vector analysis uses a double-angle representation to decompose astigmatism into vertical–horizontal and oblique components.13 A signed vertical–horizontal component matched to the fixed SI–NT testing axes can retain the astigmatic magnitude, direction, and deviation of the actual axis from the cardinal meridians. The present study included only lenticules from spherical SMILE treatments with a programmed cylinder correction of 0 D. This design maintained approximate rotational symmetry of the removed lenticules and reduced confounding from cylinder-related lenticule geometry. We first compared the direction of SI–NT mechanical advantage between prespecified WTR and ATR groups and examined the overall directional response in the unstratified sample. We then performed an exploratory coordinate-matched analysis of the association between the vertical–horizontal component of preoperative corneal astigmatism and a signed meridional difference index (signed MDI = ESI − ENT). Secondary analyses evaluated the relationship between corneal astigmatism magnitude and the direction-independent anisotropy magnitude |MDI| and examined whether the main association remained stable after adjustment for age, sex, and ocular biometric parameters. The study was designed to assess whether preoperative corneal astigmatic direction should be considered as a stratification and interpretive variable in studies of corneal stromal mechanics. Materials and Methods Ethics This study adhered to the tenets of the Declaration of Helsinki and relevant medical ethical standards and was approved by the Ethics Committee of Shanxi Aier Eye Hospital (approval no. EYESX-20240415-04). All participants provided written informed consent before study procedures. Study data were de-identified before analysis. Study Participants Patients undergoing SMILE at Shanxi Aier Eye Hospital between April 2024 and April 2026 were prospectively enrolled. Inclusion criteria were age 18–40 years; stable refraction, defined as a change of no more than 0.50 D per year during the preceding 2 years; eligibility for routine SMILE surgery; no irregular astigmatism or subclinical keratoconus on corneal topography; and a surgical plan correcting only the spherical component, with a programmed cylinder correction of 0 D. Restricting enrollment to spherical treatments was intended to maintain approximate rotational symmetry of the removed stromal lenticules and reduce confounding from cylinder-related lenticule geometry. Exclusion criteria were incomplete lenticule dissection, tearing or fragmentation during surgery, pregnancy or lactation, and incomplete preoperative biometric data. When both eyes were eligible, only the right eye was included. Astigmatism Grouping Preoperative corneal astigmatism was classified according to the anterior corneal steep-meridian axis (K2 axis) measured with Pentacam. WTR astigmatism was defined as a steep axis of 60°–120°, ATR astigmatism as 0°–30° or 150°–180°, and oblique astigmatism as 31°–59° or 121°–149°.12 Eyes with oblique astigmatism were excluded according to the prespecified criteria. The final sample comprised 53 spherical SMILE lenticules, including 39 in the WTR group and 14 in the ATR group. Throughout this article, “astigmatism type” refers to the preoperative corneal astigmatism of the source eye rather than the geometric shape of the lenticule itself. Surgery and Specimen Handling All procedures were performed by the same surgeon using a 500-kHz VisuMax femtosecond laser system (Carl Zeiss Meditec, Germany). Laser pulse energy was 130 nJ, cap thickness was 120 μm, and optical-zone diameter was 6.5 mm. Immediately after removal, the 12-o’clock orientation of each lenticule was marked under the operating microscope. The marked 12-o’clock orientation was used to align the lenticule with the device Y-axis corresponding to the SI direction before mounting, and the orientation was visually rechecked before testing. The specimen was placed in Optisol-GS at 4°C and mechanically tested within 6 hours after surgery. During preparation and testing, all specimens were handled using the same storage and immersion protocol to reduce protocol-related variability. Equibiaxial Mechanical Testing and Tangent Modulus Equibiaxial quasi-static tensile testing was performed with a BioTester 5000 equipped with 5-N load cells (CellScale, Canada). Each specimen was immersed in lactated Ringer solution and mounted with four sets of BioRakes around a central 4 × 4 mm testing region. The device X-axis corresponded to the NT direction and the device Y-axis to the SI direction; therefore, E_X = ENT and E_Y = ESI. Full-field digital image correlation was not used. This displacement-based testing approach was selected to avoid additional surface preparation, such as applying a speckle pattern, to the thin and transparent stromal lenticules before mechanical testing. The biaxial testing protocol was based on previous studies of human SMILE stromal tissue and adapted for the present study.7,8 After seven preconditioning cycles at a displacement of 80 μm, corresponding to approximately 2% strain, specimens were loaded at 10 μm/s while force–displacement curves were recorded simultaneously in both directions. Engineering stress was calculated as σ = F/(t × w), where F was the recorded load, w was specimen width, and t was the central lenticule cutting depth specified in the SMILE surgical plan. Engineering strain was calculated from BioRake displacement relative to the initial gauge length rather than from full-field digital image correlation. This displacement-based estimate represents boundary displacement rather than local full-field strain, and this limitation is considered in the Discussion. The central lenticule cutting depth used for engineering-stress calculation and the thinnest corneal thickness included as a clinical covariate were distinct variables. The tangent modulus was defined as the local slope of the stress–strain curve at a specified point. Because corneal stroma exhibits nonlinear and time-dependent behavior and because the zero-strain reference length after preconditioning may be influenced by the initial relaxed state and changes in reference length,14–16 tangent modulus was evaluated at a fixed engineering stress rather than at a fixed strain to improve comparability of local slopes across specimens. Tangent modulus was determined at an engineering stress of 0.03 MPa during the loading phase after preconditioning. This stress was within a relatively stable region of the curves and was of the same order of magnitude as corneal wall stress under physiological intraocular pressure. This comparison was used only to indicate the physiological relevance of the selected point and does not equate engineering stress in an isolated lenticule with wall stress in the intact cornea. Curves were first smoothed using a Savitzky–Golay filter with an 11-point moving window and a second-order polynomial.17 Within each moving window, a second-order polynomial was fitted locally to estimate the smoothed value at the central data point. This procedure was used to reduce random high-frequency noise while preserving the nonlinear curve shape, local trends, and slope characteristics as much as possible. The same window length, polynomial order, and processing workflow were applied to all specimens and to both SI and NT directions to avoid introducing artificial differences between specimens or directions through inconsistent smoothing parameters. Local least-squares linear regression was then performed over the 0.025–0.035 MPa stress window. The fitted slope was used as the tangent modulus, and all local fits had R2 > 0.99. Mechanical and Astigmatic Directional Indices Tangent moduli along the two testing directions were denoted ESI and ENT. The direction of in-plane mechanical anisotropy was represented by the signed meridional difference index:\documentclass[12pt]{minimal} \usepackage{wasysym} \usepackage[substack]{amsmath} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage[mathscr]{eucal} \usepackage{mathrsfs} \DeclareFontFamily{T1}{linotext}{} \DeclareFontShape{T1}{linotext}{m}{n} {linotext }{} \DeclareSymbolFont{linotext}{T1}{linotext}{m}{n} \DeclareSymbolFontAlphabet{\mathLINOTEXT}{linotext} \begin{document}$$signed\,MDI = {{\mathrm{E}}_{SI}} - {{\mathrm{E}}_{NT}}$$\end{document} A signed MDI > 0 indicated SI mechanical advantage, whereas a signed MDI < 0 indicated NT mechanical advantage. The direction-independent magnitude of anisotropy was represented by |MDI|. Because the two directions of each specimen were tested simultaneously under identical mounting, immersion, and loading conditions, systematic errors shared by both directions may be partially reduced in the within-specimen difference. For the exploratory coordinate-matched analysis, preoperative corneal astigmatism was decomposed into vertical–horizontal and oblique coordinates using the double-angle representation of classical astigmatic vector analysis:13\documentclass[12pt]{minimal} \usepackage{wasysym} \usepackage[substack]{amsmath} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage[mathscr]{eucal} \usepackage{mathrsfs} \DeclareFontFamily{T1}{linotext}{} \DeclareFontShape{T1}{linotext}{m}{n} {linotext }{} \DeclareSymbolFont{linotext}{T1}{linotext}{m}{n} \DeclareSymbolFontAlphabet{\mathLINOTEXT}{linotext} \begin{document}$${{\mathrm{A}}_{VH}} = {\mathrm{C}} \times \cos \left[{2 \times \left({{{\mathrm{\theta\ }}_{steep}} - 90^\circ } \right)} \right]$$\end{document}\documentclass[12pt]{minimal} \usepackage{wasysym} \usepackage[substack]{amsmath} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage[mathscr]{eucal} \usepackage{mathrsfs} \DeclareFontFamily{T1}{linotext}{} \DeclareFontShape{T1}{linotext}{m}{n} {linotext }{} \DeclareSymbolFont{linotext}{T1}{linotext}{m}{n} \DeclareSymbolFontAlphabet{\mathLINOTEXT}{linotext} \begin{document}$${{\mathrm{A}}_{OBL}} = {\mathrm{C}} \times \sin \left[{2 \times \left({{{\mathrm{\theta\ }}_{steep}} - 90^\circ } \right)} \right]$$\end{document} Where \documentclass[12pt]{minimal} \usepackage{wasysym} \usepackage[substack]{amsmath} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage[mathscr]{eucal} \usepackage{mathrsfs} \DeclareFontFamily{T1}{linotext}{} \DeclareFontShape{T1}{linotext}{m}{n} {linotext }{} \DeclareSymbolFont{linotext}{T1}{linotext}{m}{n} \DeclareSymbolFontAlphabet{\mathLINOTEXT}{linotext} \begin{document}${\mathrm{C}} = {{\mathrm{K}}_2} - {{\mathrm{K}}_1}$\end{document} was the corneal astigmatism magnitude and θsteep was the K2 steep-meridian axis. AVH > 0 represented a vertical steep component and AVH < 0 a horizontal steep component; AOBL represented the 45°/135° oblique component. These expressions are mathematically equivalent to conventional J0/J45-type components, with corresponding changes in sign and scale because the K2 steep axis, positive corneal astigmatism magnitude, and full rather than half scale were used. AVH was matched to signed MDI for directional analysis. AOBL was included only to describe the completeness of the orthogonal decomposition and could not be directly compared with oblique mechanical directions that were not tested in this study. Preoperative Ocular Measurements Anterior corneal curvature, corneal astigmatism magnitude, K2 axis, thinnest corneal thickness, anterior chamber depth, and corneal diameter were measured under standardized dim-light conditions using Pentacam HR. Only examinations with an “OK” quality score were accepted. Axial length was measured with an OA-2000 optical biometer (NIDEK, Japan). Each measurement was repeated three times by the same technician and averaged. The lenticule thickness ratio (LTR) was defined as central lenticule cutting depth divided by thinnest corneal thickness and was therefore dimensionless. Statistical Analysis Statistical analyses were performed using GraphPad Prism 10.0 and Python 3.13.5 (SciPy 1.17.0; statsmodels 0.14.6). Normality of continuous variables was assessed using the Shapiro–Wilk test. Normally distributed data are presented as mean ± standard deviation and were compared between groups using Welch’s t-test. Non-normally distributed data are presented as median and interquartile range and were compared using the Mann–Whitney U-test. Categorical variables were analyzed using Fisher’s exact test. All tests were two-sided, with statistical significance set at 0.05. For the directional group analysis, ESI and ENT were compared within the WTR and ATR groups using the paired Wilcoxon signed-rank test. Signed MDI was compared between groups using the Mann–Whitney U-test, and the rank-biserial correlation coefficient was reported as an effect-size measure. ESI and ENT were also compared in the unstratified sample using the paired Wilcoxon signed-rank test. Agreement between the direction of higher modulus and the WTR/ATR classification, together with |MDI| among directionally discordant specimens, was reported descriptively without additional significance testing. After the prespecified WTR/ATR directional analysis, an exploratory coordinate-matched analysis was performed with AVH and signed MDI treated as continuous variables. Spearman rank correlation was used to assess their relationship, with Pearson correlation and simple linear regression as consistency analyses. The regression slope, 95% confidence interval (CI), and R2 were reported. To determine whether the overall correlation was driven primarily by separation between WTR and ATR groups, Spearman correlations were also calculated within each group, and a linear model containing AVH, astigmatism type, and their interaction was used to test for differences in slope. For the direction-independent magnitude analysis, Spearman correlation was used to evaluate the relationship between raw corneal astigmatism magnitude and |MDI|, with Pearson correlation and simple linear regression as consistency analyses. As a sensitivity analysis, |AVH| was substituted for raw corneal astigmatism magnitude, and 10,000 paired bootstrap resamples were used to compare the two Spearman correlation coefficients with |MDI|. To assess the adjusted association between AVH and signed MDI, a core model including AVH, age, and sex was first fitted. An extended sensitivity model additionally included axial length, thinnest corneal thickness, anterior chamber depth, and corneal diameter. Results are reported as unstandardized regression coefficients (β), 95% CIs, and P values. Variance inflation factors (VIFs) were used to evaluate multicollinearity, with values below 5 considered to indicate no important multicollinearity. No formal a priori sample-size calculation was performed. The sample size was determined by the availability of qualifying spherical SMILE lenticules meeting the prespecified WTR/ATR criteria during the enrollment period. Given the sample size and the number of predictors, the extended model was regarded as an exploratory sensitivity analysis. Subgroup, multivariable, and additional correlation analyses were interpreted as exploratory and precision-limited, without adjustment for multiple comparisons. This report was prepared with reference to the STROBE statement for observational studies, and the completed checklist is provided as online supplementary material. Results Participant Characteristics and Orthogonal Decomposition of Corneal Astigmatism The study included 53 spherical SMILE stromal lenticules: 39 in the WTR group (73.6%) and 14 in the ATR group (26.4%). Median age was 24.0 years (interquartile range, 20.0–30.0 years) in the WTR group and 20.0 years (18.25–20.75 years) in the ATR group, with a significant between-group difference (P = 0.007). Sex distribution did not differ significantly between groups (P = 0.330). Other preoperative ocular characteristics are presented in Table 1.Table 1Participant Characteristics and Orthogonal Decomposition of Corneal Astigmatism in the WTR and ATR GroupsParameterWTR (n = 39)ATR (n = 14)P valueAge, years24.0 (20.0–30.0)20.0 (18.25–20.75)0.007Sex, male/female23/1611/30.330K1-flat, D42.53 ± 0.9841.90 ± 1.090.070K2-steep, D43.29 ± 1.0742.44 ± 1.080.019Corneal astigmatism magnitude, D0.76 ± 0.290.54 ± 0.180.002AVH, D0.73 ± 0.29−0.50 ± 0.16—AOBL, D0.005 ± 0.225−0.007 ± 0.2280.868Thinnest corneal thickness, μm545.21 ± 23.93540.14 ± 24.740.514Central lenticule cutting depth, μm104.62 ± 19.4197.71 ± 21.130.296LTR0.192 ± 0.0350.181 ± 0.0380.350Axial length, mm25.52 ± 0.8825.73 ± 0.820.423Corneal diameter, mm11.62 ± 0.4211.83 ± 0.390.111Anterior chamber depth, mm3.20 ± 0.303.36 ± 0.180.027Notes: Data are presented as mean ± standard deviation, except age, which is presented as median (interquartile range). Age was compared using the Mann–Whitney U-test, sex using Fisher’s exact test, and other continuous variables using Welch’s t-test. AVH denotes the vertical–horizontal astigmatic component; AOBL, the oblique astigmatic component; and LTR, central lenticule cutting depth divided by thinnest corneal thickness. Because the sign of AVH and WTR/ATR classification were derived from the same axis information, no inferential between-group comparison was performed for AVH. Table 1Participant Characteristics and Orthogonal Decomposition of Corneal Astigmatism in the WTR and ATR GroupsParameterWTR (n = 39)ATR (n = 14)P valueAge, years24.0 (20.0–30.0)20.0 (18.25–20.75)0.007Sex, male/female23/1611/30.330K1-flat, D42.53 ± 0.9841.90 ± 1.090.070K2-steep, D43.29 ± 1.0742.44 ± 1.080.019Corneal astigmatism magnitude, D0.76 ± 0.290.54 ± 0.180.002AVH, D0.73 ± 0.29−0.50 ± 0.16—AOBL, D0.005 ± 0.225−0.007 ± 0.2280.868Thinnest corneal thickness, μm545.21 ± 23.93540.14 ± 24.740.514Central lenticule cutting depth, μm104.62 ± 19.4197.71 ± 21.130.296LTR0.192 ± 0.0350.181 ± 0.0380.350Axial length, mm25.52 ± 0.8825.73 ± 0.820.423Corneal diameter, mm11.62 ± 0.4211.83 ± 0.390.111Anterior chamber depth, mm3.20 ± 0.303.36 ± 0.180.027Notes: Data are presented as mean ± standard deviation, except age, which is presented as median (interquartile range). Age was compared using the Mann–Whitney U-test, sex using Fisher’s exact test, and other continuous variables using Welch’s t-test. AVH denotes the vertical–horizontal astigmatic component; AOBL, the oblique astigmatic component; and LTR, central lenticule cutting depth divided by thinnest corneal thickness. Because the sign of AVH and WTR/ATR classification were derived from the same axis information, no inferential between-group comparison was performed for AVH. Participant Characteristics and Orthogonal Decomposition of Corneal Astigmatism in the WTR and ATR Groups Notes: Data are presented as mean ± standard deviation, except age, which is presented as median (interquartile range). Age was compared using the Mann–Whitney U-test, sex using Fisher’s exact test, and other continuous variables using Welch’s t-test. AVH denotes the vertical–horizontal astigmatic component; AOBL, the oblique astigmatic component; and LTR, central lenticule cutting depth divided by thinnest corneal thickness. Because the sign of AVH and WTR/ATR classification were derived from the same axis information, no inferential between-group comparison was performed for AVH. Mean K1 and K2 values in the WTR group were 42.53 ± 0.98 D and 43.29 ± 1.07 D, respectively, and mean raw corneal astigmatism magnitude was 0.76 ± 0.29 D. Corresponding values in the ATR group were 41.90 ± 1.09 D, 42.44 ± 1.08 D, and 0.54 ± 0.18 D. After orthogonal decomposition, mean AVH was 0.73 ± 0.29 D in the WTR group and −0.50 ± 0.16 D in the ATR group, representing vertical and horizontal steep components, respectively. Mean AOBL was close to zero in both groups: 0.005 ± 0.225 D in the WTR group and −0.007 ± 0.228 D in the ATR group. The difference between raw corneal astigmatism magnitude and |AVH| was small, averaging 0.031 ± 0.039 D in the WTR group and 0.040 ± 0.070 D in the ATR group. This finding was consistent with the steep axes being predominantly close to the vertical or horizontal directions and indicated that most corneal astigmatism remained in the vertical–horizontal component matched to the SI–NT mechanical testing coordinates after axis decomposition. Overall and Within-Group Comparisons of SI and NT Tangent Moduli Across all 53 stromal lenticules, the mean SI tangent modulus was higher than the mean NT tangent modulus (2.69 ± 0.68 MPa vs 2.40 ± 0.59 MPa; P < 0.001). This overall result reflected the average directional response of the study sample; however, because WTR cases accounted for 73.6% of the sample, the overall SI advantage may have been influenced by the distribution of astigmatism types. After stratification by preoperative corneal astigmatism type, the two groups showed opposite directions of mechanical advantage. In the WTR group, mean SI tangent modulus was 2.82 ± 0.67 MPa and exceeded the NT value of 2.32 ± 0.55 MPa (P < 0.001). Median signed MDI was 0.463 MPa (interquartile range, 0.273–0.658 MPa), and 35 of 39 lenticules had a higher SI modulus. In contrast, mean NT tangent modulus in the ATR group was 2.62 ± 0.67 MPa and exceeded the SI value of 2.34 ± 0.61 MPa (P = 0.002). Median signed MDI was −0.366 MPa (interquartile range, −0.423 to −0.146 MPa), and 11 of 14 lenticules had a higher NT modulus (Figure 1). The directional mechanical results are summarized in Table 2. Figure 1Paired comparisons of NT and SI tangent moduli in the WTR group, ATR group, and unstratified sample. Each thin line represents one lenticule; thick lines and larger endpoints indicate group means. (A) WTR group, shown as blue circles, with a higher SI than NT modulus. (B) ATR group, shown as orange-red triangles, with a higher NT than SI modulus. (C) Unstratified sample, shown in dark gray, with an overall higher SI than NT modulus. All panels use the same vertical-axis scale. Within-group comparisons were performed using the paired Wilcoxon signed-rank test.Image A (WTR) shows a paired line graph with categories NT and SI on the horizontal axis and ′Tangent modulus at 0.03 MPa (MPa)′ on the vertical axis, ranging from 1.5 to 4.5. The text above reads ′P < 0.001′. Many thin lines connect NT to SI, with a thick mean line from (NT, 2.3) to (SI, 2.8). Image B (ATR) has similar axes and labels, with ′P = 0.002′ above. The thick mean line connects (NT, 2.6) to (SI, 2.3). Image C (Pooled) also features the same axes and labels, with ′P < 0.001′ above. The thick mean line runs from (NT, 2.4) to (SI, 2.7).A multi-plot line graph showing paired NT and SI tangent modulus for WTR, ATR and pooled groups.Table 2Directional Mechanical Results in the WTR Group, ATR Group, and Unstratified SampleGroupNT Modulus, MPaSI Modulus, MPaSI vs NT P valueSigned MDI,Median (IQR), MPaPredominantDirectionWTR (n = 39)2.32 ± 0.552.82 ± 0.67<0.001+0.463 (+0.273 to +0.658)SI: 35/39ATR (n = 14)2.62 ± 0.672.34 ± 0.610.002−0.366 (−0.423 to −0.146)NT: 11/14Unstratified (N = 53)2.40 ± 0.592.69 ± 0.68<0.001+0.386 (−0.135 to +0.635)SI: 38/53Notes: NT and SI tangent moduli are presented as mean ± standard deviation. Within-group comparisons were performed using the paired Wilcoxon signed-rank test.Abbreviation: IQR, interquartile range. Paired comparisons of NT and SI tangent moduli in the WTR group, ATR group, and unstratified sample. Each thin line represents one lenticule; thick lines and larger endpoints indicate group means. (A) WTR group, shown as blue circles, with a higher SI than NT modulus. (B) ATR group, shown as orange-red triangles, with a higher NT than SI modulus. (C) Unstratified sample, shown in dark gray, with an overall higher SI than NT modulus. All panels use the same vertical-axis scale. Within-group comparisons were performed using the paired Wilcoxon signed-rank test. Table 2Directional Mechanical Results in the WTR Group, ATR Group, and Unstratified SampleGroupNT Modulus, MPaSI Modulus, MPaSI vs NT P valueSigned MDI,Median (IQR), MPaPredominantDirectionWTR (n = 39)2.32 ± 0.552.82 ± 0.67<0.001+0.463 (+0.273 to +0.658)SI: 35/39ATR (n = 14)2.62 ± 0.672.34 ± 0.610.002−0.366 (−0.423 to −0.146)NT: 11/14Unstratified (N = 53)2.40 ± 0.592.69 ± 0.68<0.001+0.386 (−0.135 to +0.635)SI: 38/53Notes: NT and SI tangent moduli are presented as mean ± standard deviation. Within-group comparisons were performed using the paired Wilcoxon signed-rank test.Abbreviation: IQR, interquartile range. Directional Mechanical Results in the WTR Group, ATR Group, and Unstratified Sample Notes: NT and SI tangent moduli are presented as mean ± standard deviation. Within-group comparisons were performed using the paired Wilcoxon signed-rank test. Abbreviation: IQR, interquartile range. Signed MDI distributions differed significantly between the WTR and ATR groups (Mann–Whitney U = 527, P < 0.001; rank-biserial r = 0.930), indicating not only opposite directions of SI–NT mechanical difference but also substantial separation between the groups (Figure 2). Figure 2Distribution of signed MDI in the WTR and ATR groups. Blue circles represent WTR cases and orange-red triangles represent ATR cases. Signed MDI = ESI − ENT; positive values indicate SI advantage and negative values indicate NT advantage. Groups were compared using the Mann–Whitney U-test. The gray dashed line indicates MDI = 0.A box and scatter plot showing signed mechanical difference index distributions for WTR and ATR groups. Distribution of signed MDI in the WTR and ATR groups. Blue circles represent WTR cases and orange-red triangles represent ATR cases. Signed MDI = ESI − ENT; positive values indicate SI advantage and negative values indicate NT advantage. Groups were compared using the Mann–Whitney U-test. The gray dashed line indicates MDI = 0. Overall, the direction of higher modulus agreed with the WTR or ATR classification in 46 of 53 lenticules (86.8%), including 35 of 39 in the WTR group and 11 of 14 in the ATR group. Among the seven directionally discordant lenticules, |MDI| ranged from 0.061 to 0.189 MPa, with a median of 0.132 MPa. These discordant cases all showed relatively limited SI–NT modulus differences, although their causes could not be determined from the present data. Because a minimum meaningful directional difference threshold was not established from repeatability data, these findings are presented descriptively. Association Between the Vertical–Horizontal Astigmatic Component and the Signed Meridional Mechanical Difference Across all 53 stromal lenticules, AVH was strongly and positively correlated with signed MDI (Spearman rho = 0.875, P < 0.001; Pearson r = 0.842, P < 0.001). As the astigmatic direction shifted from a horizontal steep component toward a vertical steep component, the SI tangent modulus increased relative to the NT tangent modulus. Simple linear regression indicated that each 1-D increase in AVH was associated with an average 0.674-MPa increase in signed MDI (95% CI, 0.552–0.795 MPa; P < 0.001). The model explained approximately 70.9% of the variance in signed MDI (R2 = 0.709) (Figure 3). Figure 3Relationship between the vertical–horizontal astigmatic component AVH and signed MDI. Blue circles represent WTR cases and orange-red triangles represent ATR cases. The dark gray solid line indicates the overall linear regression, and the gray band indicates the 95% confidence interval. The horizontal and vertical light gray dashed lines indicate signed MDI = 0 and AVH = 0, respectively.A scatter plot showing signed MDI and vertical-horizontal astigmatic component, AVH. Relationship between the vertical–horizontal astigmatic component AVH and signed MDI. Blue circles represent WTR cases and orange-red triangles represent ATR cases. The dark gray solid line indicates the overall linear regression, and the gray band indicates the 95% confidence interval. The horizontal and vertical light gray dashed lines indicate signed MDI = 0 and AVH = 0, respectively. In group-specific sensitivity analyses, AVH and signed MDI remained positively correlated within both the WTR and ATR groups, with Spearman coefficients of 0.760 (P < 0.001) and 0.546 (P = 0.044), respectively. The interaction between astigmatism type and AVH was not significant (P = 0.902), providing no evidence of a marked difference in slope between the WTR and ATR groups. These findings indicate that the overall association was not generated solely by separation of group means; a correlation in the same direction was also observed within each group. Association Between Corneal Astigmatism Magnitude and Mechanical Anisotropy Magnitude Raw corneal astigmatism magnitude was positively correlated with |MDI| (Spearman rho = 0.772, P < 0.001; Pearson r = 0.721, P < 0.001). Simple linear regression explained approximately 52.0% of the variance in |MDI| (R2 = 0.520). Thus, greater corneal astigmatism magnitude was generally accompanied by a larger absolute difference between SI and NT tangent moduli, although this analysis did not address which direction had the higher modulus (Figure 4). Figure 4Relationship between corneal astigmatism magnitude and the direction-independent mechanical anisotropy magnitude |MDI|. Blue circles represent WTR cases and orange-red triangles represent ATR cases. The dark gray solid line indicates the overall linear regression, and the gray band indicates the 95% confidence interval.A scatter plot showing corneal astigmatism magnitude and vertical bar M D I vertical bar. Relationship between corneal astigmatism magnitude and the direction-independent mechanical anisotropy magnitude |MDI|. Blue circles represent WTR cases and orange-red triangles represent ATR cases. The dark gray solid line indicates the overall linear regression, and the gray band indicates the 95% confidence interval. As a sensitivity analysis, |AVH| was substituted for raw corneal astigmatism magnitude. After axis decomposition, |AVH| was lower than raw corneal astigmatism magnitude by an average of 0.033 ± 0.049 D. The mean reductions were 0.031 ± 0.039 D in the WTR group and 0.040 ± 0.070 D in the ATR group. Because the steep axes of the included cases were predominantly close to the vertical or horizontal directions, this numerical change was small. The correlation of |AVH| with |MDI| (Spearman rho = 0.776, P < 0.001) was essentially identical to that of raw corneal astigmatism magnitude with |MDI|. The difference between the two coefficients was not significant (Δrho = 0.004; 95% bootstrap CI, −0.059 to 0.073; P = 0.957), indicating that axis decomposition did not materially alter the association between astigmatism magnitude and mechanical anisotropy magnitude. Multivariable Adjustment and Other Ocular Parameters Because age distributions differed between the WTR and ATR groups, age was included in the multivariable models together with sex and ocular biometric parameters. In the core model containing AVH, age, and sex, each 1-D increase in AVH was associated with an average 0.664-MPa increase in signed MDI (unstandardized β = 0.664; 95% CI, 0.533–0.795 MPa; P < 0.001). In the extended multivariable linear regression model additionally containing axial length, thinnest corneal thickness, anterior chamber depth, and corneal diameter, each 1-D increase in AVH was associated with an average 0.673-MPa increase in signed MDI (unstandardized β = 0.673; 95% CI, 0.537–0.809 MPa; P < 0.001). The extended model had an R2 of 0.728 and an adjusted R2 of 0.686. The 95% CIs for age, sex, axial length, thinnest corneal thickness, anterior chamber depth, and corneal diameter all included zero, and none showed a statistically significant independent association with signed MDI. All VIFs were below 2, indicating no concerning multicollinearity. In unadjusted univariable analyses, age showed a weak positive correlation with signed MDI (Spearman rho = 0.320, P = 0.020), but the association was no longer significant after adjustment for AVH and other covariates (β = −0.003 MPa/year, P = 0.748). Univariable correlations of axial length, thinnest corneal thickness, anterior chamber depth, and corneal diameter with signed MDI were not statistically significant (Figure 5). AVH was the only variable showing a stable and consistent association, whereas the other demographic and ocular biometric parameters showed no consistent independent relationship. Figure 5Unadjusted relationships of signed MDI with age and ocular biometric parameters. (a) Age; this panel reflects only the difference in modulus between directions and does not represent absolute corneal stiffness. (b) Axial length. (c) Thinnest corneal thickness. (d) Anterior chamber depth. (e) Corneal diameter. Blue circles represent WTR cases and orange-red triangles represent ATR cases. The dark gray solid line and gray shading indicate the simple linear fit and 95% confidence interval, respectively. Spearman rho and P values are shown; independent associations are presented in Table 3.Image A: Scatter plot of Signed MDI vs Age (17.5-35 years). Linear fit rises from 0.2 to 0.6 MPa. Circles range 0.0-1.2 MPa, triangles -0.6-0.1 MPa. Rho=0.320, P=0.020. Image B: Signed MDI vs Axial Length (24-27 mm). Linear fit slightly declines from 0.35 to 0.25 MPa. Circles -0.2-1.6 MPa, triangles -0.6-0.1 MPa. Rho=-0.057, P=0.686. Image C: Signed MDI vs Thinnest Corneal Thickness (500-600 microm). Linear fit rises from 0.1 to 0.5 MPa. Circles -0.2-1.6 MPa, triangles -0.6-0.1 MPa. Rho=0.166, P=0.236. Image D: Signed MDI vs Anterior Chamber Depth (2.6-3.8 mm). Linear fit declines from 0.6 to 0.1 MPa. Circles -0.2-1.2 MPa, triangles -0.6-0.1 MPa. Rho=-0.255, P=0.066. Image E: Signed MDI vs Corneal Diameter (10.75-12.25 mm). Linear fit declines from 0.45 to 0.15 MPa. Circles -0.2-1.2 MPa, triangles -0.6-0.1 MPa. Rho=-0.180, P=0.197. Legend: Circles=WTR, Triangles=ATR, Line=Linear fit.Five scatter plots of signed MDI versus ocular biometric parameters, with linear fit trends.Table 3Extended Multivariable Linear Regression Model for Signed MDI (n = 53)PredictorUnstandardized β95% CIP valueVIFAVH, D0.6730.537 to 0.809<0.0011.18Age, years−0.003−0.022 to 0.0160.7481.76Male sex (reference: female)−0.066−0.251 to 0.1200.4791.40Axial length, mm0.083−0.020 to 0.1860.1111.37Thinnest corneal thickness, μm−0.0004−0.0038 to 0.00300.8321.15Anterior chamber depth, mm−0.121−0.487 to 0.2440.5081.84Corneal diameter, mm−0.029−0.261 to 0.2030.8031.61Notes: The dependent variable was signed MDI in MPa. Model R2 = 0.728; adjusted R2 = 0.686; overall F-test P < 0.001. Coefficients are unstandardized and represent the mean change in signed MDI for a one-unit increase in each continuous predictor. Female sex was the reference category.Abbreviations: CI, confidence interval; VIF, variance inflation factor. Unadjusted relationships of signed MDI with age and ocular biometric parameters. (a) Age; this panel reflects only the difference in modulus between directions and does not represent absolute corneal stiffness. (b) Axial length. (c) Thinnest corneal thickness. (d) Anterior chamber depth. (e) Corneal diameter. Blue circles represent WTR cases and orange-red triangles represent ATR cases. The dark gray solid line and gray shading indicate the simple linear fit and 95% confidence interval, respectively. Spearman rho and P values are shown; independent associations are presented in Table 3. Table 3Extended Multivariable Linear Regression Model for Signed MDI (n = 53)PredictorUnstandardized β95% CIP valueVIFAVH, D0.6730.537 to 0.809<0.0011.18Age, years−0.003−0.022 to 0.0160.7481.76Male sex (reference: female)−0.066−0.251 to 0.1200.4791.40Axial length, mm0.083−0.020 to 0.1860.1111.37Thinnest corneal thickness, μm−0.0004−0.0038 to 0.00300.8321.15Anterior chamber depth, mm−0.121−0.487 to 0.2440.5081.84Corneal diameter, mm−0.029−0.261 to 0.2030.8031.61Notes: The dependent variable was signed MDI in MPa. Model R2 = 0.728; adjusted R2 = 0.686; overall F-test P < 0.001. Coefficients are unstandardized and represent the mean change in signed MDI for a one-unit increase in each continuous predictor. Female sex was the reference category.Abbreviations: CI, confidence interval; VIF, variance inflation factor. Extended Multivariable Linear Regression Model for Signed MDI (n = 53) Notes: The dependent variable was signed MDI in MPa. Model R2 = 0.728; adjusted R2 = 0.686; overall F-test P < 0.001. Coefficients are unstandardized and represent the mean change in signed MDI for a one-unit increase in each continuous predictor. Female sex was the reference category. Abbreviations: CI, confidence interval; VIF, variance inflation factor. For descriptive visualization, the unadjusted scatterplots of signed MDI against age and the ocular biometric parameters are presented in Figure 5. This figure is intended only to display the univariable distributions; independent associations are evaluated by the multivariable regression results in Table 3. Discussion This study demonstrated a significant statistical correspondence between the direction of preoperative corneal astigmatism and the SI–NT mechanical difference in SMILE stromal lenticules. The WTR group predominantly showed a higher SI tangent modulus, whereas the ATR group predominantly showed a higher NT tangent modulus. In the coordinate-matched analysis, AVH was strongly and positively associated with signed MDI, and the relationship remained stable after adjustment for age, sex, and multiple ocular biometric parameters. Raw corneal astigmatism magnitude was also positively associated with |MDI|, indicating that corneas with greater astigmatism generally had more pronounced SI–NT mechanical differences. Together, these findings suggest that group-level results for corneal anisotropy may be influenced by the distribution of preoperative astigmatic directions when direction is not considered. The present study was a cross-sectional investigation of ex vivo tissue mechanics; therefore, the observed relationship represents statistical correspondence between corneal morphology and stromal mechanical directionality and does not establish temporal sequence or causality. A methodological feature of this study was the conversion of preoperative corneal astigmatic axis information into the coordinate system used for biaxial mechanical testing. Conventional WTR and ATR categories summarize the predominant astigmatic direction but discretize an intrinsically continuous axis. Corneas classified within the same category may still differ in the deviation of their actual steep axes from 90° or 180°. The vertical–horizontal component AVH, derived from classical astigmatic vector decomposition, simultaneously retained astigmatism magnitude, direction, and deviation from the cardinal meridians while matching its sign to the SI–NT definition of signed MDI.13 Previous representative studies primarily compared corneal or SMILE stromal mechanical responses along fixed anatomic directions.1,4,7–11 Song et al also identified the absence of marked anatomic orientation as a limitation and highlighted the need to investigate directional mechanics in astigmatic cases.6 Thus, compared with a simple comparison of WTR and ATR group means, treating both AVH and signed MDI as continuous variables provides a more direct assessment of statistical matching between corneal morphological direction and the corresponding mechanical direction. The strong positive association observed here remained directionally consistent within both groups, and the interaction between astigmatism type and AVH was not significant. These results suggest that the overall association was not solely attributable to differences in group means. Because mechanical testing was performed only along fixed SI and NT directions, the vertical–horizontal component was the directly corresponding astigmatic coordinate; the oblique component was used only to document the completeness of orthogonal decomposition and could not be compared with mechanical responses along untested 45° and 135° directions. The unstratified sample showed a higher SI than NT tangent modulus, but this overall finding should be interpreted in the context of the astigmatism-type distribution. WTR cases constituted 73.6% of the study sample and predominantly showed SI mechanical advantage, whereas ATR cases showed the opposite NT advantage. Consequently, the overall SI advantage more accurately represents a weighted average of directional phenotypes in this particular sample than a universal, fixed SI mechanical advantage in all corneas. Some previous studies of human corneas and SMILE stromal tissue reported a higher tangent modulus or greater overall macroscopic stiffness in the vertical or SI direction,1,4,7–9 whereas others did not detect a clear SI–NT tensile difference within their testing frameworks.10,11 The present results do not imply that preoperative astigmatic direction fully explains these differences across studies. Testing method, stress or strain range, loading ratio, tissue depth, hydration state, and strain-estimation method remain important sources of variation. Rather, astigmatic-direction composition may represent an additional clinical and geometric source of heterogeneity that has not always been recorded or incorporated into fixed-axis analyses. Depending on the proportions of WTR, ATR, and other axial phenotypes, mechanical differences of opposite sign may be strengthened, weakened, or partially canceled. Future studies of corneal anisotropy should therefore record and consider corneal astigmatic direction and steep-axis distribution when reporting group-average findings. Beyond directional correspondence, raw corneal astigmatism magnitude was positively associated with the mechanical anisotropy magnitude |MDI|, indicating that greater astigmatism was generally accompanied by a larger difference between SI and NT moduli. Because |MDI| removes the sign of mechanical advantage, this result concerns the size of the intermeridional difference rather than whether SI or NT was stronger. Replacing raw astigmatism magnitude with |AVH| produced an essentially unchanged correlation, probably because the steep axes of the included cases were predominantly close to the vertical or horizontal directions and axis decomposition caused only a small change in magnitude. This sensitivity analysis supports that the association between astigmatism magnitude and mechanical anisotropy magnitude was not dependent on the particular expression used for astigmatism. Nevertheless, such covariation does not demonstrate that greater astigmatism causes stronger mechanical anisotropy or vice versa. A more cautious interpretation is that corneal astigmatism magnitude and direction may serve as clinically measurable features of a directional mechanical phenotype and may be considered as candidate stratification variables or inputs in future corneal biomechanical research and patient-specific finite element modeling.18–21 Recent corneal modeling studies, including reduced-order plate formulations and surrogate or sensitivity-analysis approaches, further illustrate the importance of defining patient-specific geometry, material parameters, and modeling assumptions when translating tissue-level measurements into corneal-scale predictions.22,23 The immediate clinical significance of the present findings is therefore interpretive rather than prescriptive: preoperative astigmatic direction may help explain variability in fixed-axis stromal mechanical measurements, but it should not yet be used to guide routine clinical decisions or individualized surgical planning without external validation. Age was weakly correlated with signed MDI in unadjusted analysis, and age distributions differed between the WTR and ATR groups, but age no longer showed an independent association after adjustment for AVH, sex, and other ocular biometric parameters. Because participants were prospectively and consecutively enrolled according to the eligibility criteria and were not selected on the basis of age or sex, the between-group age difference was treated as a covariate imbalance rather than as a sampling criterion. This finding should not be interpreted as contradicting previously reported age-related corneal stiffening,24–26 because signed MDI measures the difference between SI and NT moduli rather than absolute corneal stiffness. If both directions stiffen to a similar extent with age, their difference may remain largely unchanged. In addition, the participants were young refractive-surgery candidates with a relatively narrow age range, limiting sensitivity to age-related mechanical changes. Axial length, thinnest corneal thickness, anterior chamber depth, and corneal diameter likewise showed no stable independent associations. These nonsignificant results do not establish that these factors are unrelated to stromal mechanical directionality; they indicate only that, within the present sample range, selected mechanical metric, and model, no clear independent relationship beyond the variance explained by AVH was detected. Several limitations should be considered. First, the ATR group contained only 14 lenticules and was substantially smaller than the WTR group. This reduced the precision of effect estimates within the ATR subgroup, limited the ability to characterize internal heterogeneity and smaller effects, and restricted generalizability to broader ATR populations. No formal a priori sample-size or power calculation was performed because the study was constrained by the availability of qualifying spherical SMILE lenticules with prespecified astigmatism directions. Although paired within-specimen comparisons were used, 11 of 14 ATR lenticules showed NT advantage, and the direction of this result was consistent with the between-group signed MDI difference and the overall positive relationship between AVH and signed MDI, such internal consistency cannot eliminate the limitation of the small subgroup. The ATR findings, subgroup analyses, and multivariable sensitivity analyses should therefore be interpreted as exploratory and require confirmation in larger independent samples with more balanced distributions of astigmatism types. Second, only lenticules from spherical SMILE treatments with a programmed cylinder correction of 0 D were included. This strict criterion reduced confounding from geometric asymmetry introduced by cylinder correction but also created a selected study population. The findings may not apply to lenticules from SMILE procedures that include cylinder correction, eyes with higher levels of astigmatism, or the general corneal population. Third, ex vivo SMILE lenticules primarily reflect the material response of the central corneal stroma under specific experimental conditions and cannot be equated directly with the behavior of the intact living cornea under intraocular pressure, boundary constraints, and interactions among tissue layers. All tested samples were central SMILE lenticules and the mechanical analysis was performed within a central 4 × 4 mm region. The relative stromal location of the lenticule was partly reflected by the lenticule thickness ratio (LTR), defined as central lenticule cutting depth divided by thinnest corneal thickness. In this cohort, the lenticules were obtained from the anterior-to-middle stroma, the variation in LTR was limited, and LTR did not differ significantly between the WTR and ATR groups (0.192 ± 0.035 vs 0.181 ± 0.038; P = 0.350; Table 1). Central lenticule cutting depth also did not differ significantly between groups (P = 0.296; Table 1). These findings reduce the likelihood that between-group imbalance in lenticule depth was a major explanation for the observed directional differences. Nevertheless, the exact stromal depth, lenticule thickness, and tissue layer represented by each specimen were determined by the individualized surgical plan and were not completely identical across eyes. The structure and mechanical properties of the central anterior stroma vary by direction and depth;27–36 therefore, findings from SMILE lenticules should not be extrapolated directly to the full-thickness cornea. Mechanical testing was also limited to fixed SI and NT directions and therefore could not directly assess responses along oblique 45° and 135° axes. The oblique astigmatic component described only the completeness of vector decomposition and did not correspond to unmeasured oblique mechanical properties. The present study used an equibiaxial loading protocol and a tangent modulus at 0.03 MPa to provide a standardized local comparison across specimens at a fixed stress level. A recent planar biaxial study of porcine corneas showed that corneal mechanical behavior depended on the applied stretch ratio, supporting that biaxial measurements are protocol dependent.37 Therefore, different loading ratios may change the quantitative magnitude of the SI–NT difference and the local tangent modulus. However, loading-ratio-dependent changes do not necessarily imply reversal of the predominant directional response, and the present data cannot determine whether the direction of mechanical advantage would remain unchanged under all possible loading protocols. Accordingly, the conclusions should be interpreted as applying to the standardized equibiaxial condition used here and should not be generalized to all biaxial loading ratios or higher stress/strain ranges. The 0.03-MPa sampling stress was chosen as a fixed local stress level within a relatively stable portion of the curves and at the same order of magnitude as corneal wall stress estimated under physiological intraocular pressure using a thin-shell approximation; it was not intended to reproduce the full in vivo stress state of the intact cornea. Hydration is another common challenge in ex vivo corneal and stromal-tissue mechanical studies. Corneal tensile behavior and stiffness estimates can depend on hydration state, including in thin corneal strip experiments and cross-linking studies.38,39 In the present study, all lenticules were stored in Optisol-GS and tested while immersed in lactated Ringer solution using the same protocol, which was chosen to maintain tissue hydration during testing and to improve comparability with previous ex vivo SMILE lenticule studies.7,8 However, hydration level, thickness change during handling, and potential direction-dependent hydration effects were not directly measured or actively controlled. Hydration may therefore have affected absolute modulus estimates. The within-specimen signed MDI may partially reduce systematic effects shared by both directions, but it cannot eliminate hydration-related uncertainty. Engineering strain was calculated from grip displacement rather than full-field digital image correlation to minimize additional manipulation of the thin transparent lenticules before testing. This choice avoided surface speckle preparation but limited strain estimation to boundary displacement. Finally, all specimens underwent the same preconditioning protocol to reduce loading-history effects; nevertheless, specimen-specific or group-specific responses to preconditioning cannot be excluded. Strip or isolated-tissue testing and intact-corneal inflation testing differ in boundary conditions and strain estimation,8,40–42 which may affect absolute modulus estimates. Moreover, no minimum detectable difference or minimum meaningful directional-difference threshold was established from repeatability experiments. For discordant cases with signed MDI close to zero, the present study could not distinguish a true weak directional difference from individual variability or measurement uncertainty. Finally, the coordinate-matched analysis of AVH and signed MDI emerged from further data reanalysis and was exploratory rather than a preregistered primary endpoint. Although the association showed good internal consistency across stratified analyses, multivariable adjustment, and sensitivity analyses, it should be regarded as a hypothesis-generating finding that requires external replication. Because the study used a cross-sectional ex vivo design, all observations are limited to statistical associations and cannot determine the temporal sequence, mechanism, or causal direction between the astigmatic phenotype and stromal mechanical directionality. Conclusion This study found a clear statistical correspondence between preoperative corneal astigmatic direction and the corresponding SI–NT stromal mechanical difference in spherical SMILE lenticules. WTR and ATR corneas showed opposite predominant directions of higher modulus, and the vertical–horizontal component derived from classical astigmatic vector decomposition was strongly associated with signed MDI when both were expressed in the same coordinate system. The findings also suggest that pooled analyses that do not consider astigmatic direction may be influenced by the distribution of astigmatism types in the sample. Preoperative steep axis and astigmatic direction should therefore be considered important recording, stratification, and interpretive variables in future studies of corneal anisotropy. These results do not establish causality between corneal morphology and stromal mechanical directionality and should be regarded as hypothesis-generating rather than directly practice-changing. Further studies should replicate these findings in larger independent samples with more balanced astigmatism-type distributions and incorporate additional testing directions, repeatability assessment, hydration control, and intact-corneal models to clarify their range of applicability.