Risk Factors and Prediction of Medial Canthal Angle Regression after Modified V-Shaped Concealed Flap Canthoplasty

Summary

This protocol aims to standardize intraoperative measurement of folding parameters of the anterior limb of the medial canthal tendon during modified V-shaped concealed flap canthoplasty and to assess their association with postoperative medial canthal angle regression.

Abstract

This study aimed to define a reproducible intraoperative measurement protocol for modified V-shaped concealed flap canthoplasty and to characterize the association between quantitative folding parameters of the anterior limb of the medial canthal tendon (MCT) and postoperative medial canthal angle regression. In this prospective cohort, 331 consecutive patients undergoing the procedure were enrolled. Intraoperative folding length (L), folding angle (θ), fold-point position (P), fixation plane, and suture tension grade were recorded using a standardized protocol. Medial canthal angle regression at 6 months was assessed from standardized photographs relative to the 1-week postoperative baseline. Multivariable logistic regression was used to explore associations between intraoperative measurements and postoperative regression, and the resulting model was internally validated with bootstrap resampling. The 6-month incidence of medial canthal angle regression was 24.6%. Greater L, θ, and P were protective, whereas tendon/soft-tissue fixation, low suture tension, greater epicanthal fold severity, thick soft tissue, larger preoperative medial canthal angle, and longer operative time were associated with increased risk. The apparent area under the receiver operating characteristic (ROC) curve (AUC) of the internally validated model was 0.85 (95% confidence interval [CI], 0.81–0.89), and the optimism-corrected AUC was 0.81 (95% CI, 0.77–0.85). These findings support the feasibility of intraoperative quantitative assessment but should be interpreted as internally validated single-center evidence requiring external validation before broader clinical application.

Introduction

Epicanthal folds are common medial canthal anatomical features in East Asian populations and represent an important consideration in Asian eyelid surgery¹. Previous studies have reported a high prevalence of epicanthal folds among East Asian populations and evaluated postoperative cosmetic outcomes following epicanthoplasty². The fold can visually shorten the horizontal palpebral fissure, widen the perceived intercanthal distance, and obscure the lacrimal caruncle or medial canthal contour. Modified V-shaped concealed flap canthoplasty is performed to improve medial canthal morphology, but postoperative medial canthal angle regression may reduce the durability of the initial correction³. Because even small changes in medial canthal position can affect aesthetic satisfaction, reproducible measurement of postoperative regression and intraoperative factors associated with that regression is clinically important.

Previous studies of epicanthoplasty have mainly addressed flap design, scar concealment, and postoperative scar optimization^4,5. These refinements are important, but they do not fully explain why some patients experience delayed medial canthal angle regression during scar remodeling. Recent ophthalmic outcome studies and evidence syntheses have further emphasized the importance of standardized imaging, quantitative endpoints, and reproducible risk stratification methods for evaluating surgical stability and predicting postoperative outcomes⁶. Recent studies in ophthalmic outcome assessment have highlighted the importance of standardized imaging, quantitative measurements, and reproducible evaluation frameworks for improving the objectivity and comparability of surgical outcome assessment^7,8,9. These approaches provide a methodological basis for developing more reproducible surgical evaluation protocols. In this context, the anterior limb of the MCT may be considered a deep structural support that can be measured and modified intraoperatively, rather than an implicit technical detail left solely to surgeon's experience^10,11.

The modified V-shaped concealed flap technique combines hidden incision placement with reconstruction of medial canthal curvature, consistent with recent anatomical approaches aiming to restore medial canthal configuration¹². The principal methodological contribution of the present protocol is the intraoperative measurement sequence used to quantify deep support, including identification of the anterior MCT limb, standardized traction, measurement of L, θ, and P, and documentation of fixation plane and suture tension (Figure 1). The predictive model represents a secondary application of these reproducible measurements rather than the primary focus of the protocol. Standardizing these steps may reduce inter-surgeon variability and allow postoperative regression to be analyzed against measurable intraoperative parameters¹³.

Therefore, this study was designed to standardize the intraoperative quantification of anterior MCT limb folding during modified V-shaped concealed flap canthoplasty and to examine how these measurements relate to 6-month medial canthal angle regression. The protocol links procedural steps, measurement definitions, postoperative image-based outcome assessment, and internal model validation, providing a structured workflow that can be reproduced and further evaluated in different surgical settings.

Protocol

Perform all human-participant procedures in accordance with the Declaration of Helsinki and institutional requirements. Written informed consent was obtained from all participants before enrollment. Separate written consent for the publication of clinical photographs was obtained from participants whose images are included in the manuscript. Images were cropped or limited to the periocular region whenever possible to minimize patient identifiability. The study was approved by the Ethics Committee of Tianjin Medical University Eye Hospital (approval No. 2025KY-35). 

1. Participant selection and preoperative evaluation

1. Screen consecutive adult patients (18 years or older) who seek modified V-shaped concealed flap canthoplasty for epicanthal fold correction or medial canthal contour improvement. Eligible cases may include unilateral or bilateral procedures and mild, moderate, or severe epicanthal fold severity when standardized photography and 6-month follow-up can be completed.
2. Exclude patients presenting a history of surgery affecting medial orbit anatomy.
3. Exclude cases with severe congenital or post-traumatic deformities requiring simultaneous complex medial orbital reconstruction.
4. Exclude individuals with active infection, significant inflammatory skin conditions, or a confirmed abnormal keloid tendency.
5. Capture standardized anterior and local close-up photographs using the same digital camera system equipped with a fixed lens, a fixed patient-to-camera distance of approximately 1.0 m, diffuse frontal lighting, a neutral background, and the camera lens aligned horizontally with the interpupillary line. Record the camera model, focal length or zoom setting, and distance so that the same setup can be repeated during follow-up.
6. Maintain uniform light intensity and exposure settings. Ask the patient to sit upright with the Frankfort horizontal plane level, the head centered, both eyes naturally open, and facial expression relaxed. Use the medial canthal point, pupil center, medial orbital rim reference, and palpebral margin as repeatable landmarks for image measurement.
7. Instruct the patient to maintain a centered head position and a level gaze.
8. Ask the patient to keep their eyes naturally open and facial expressions relaxed to minimize fluctuations in soft tissue tension.

2. Preoperative marking and surgical preparation 

NOTE: Figure 2 shows the step-by-step intraoperative photographs of the procedure, and Figure 3 shows the intraoperative quantification workflow for anterior MCT limb folding parameters and medial canthal angle measurement.

1. Mark the intended new medial canthal point, the existing medial canthal point, and the planned concealed V-shaped incision limbs with the patient seated upright. 
2. Design each limb of the concealed V-shaped incision with a length of approximately 5–8 mm and an included angle of approximately 60–90°. Adjust limb length and angle according to epicanthal fold severity, tissue redundancy, and the intended medial displacement of the new medial canthal point. 
3. Record the selected incision dimensions before incision and reproduce the same parameters on the contralateral side in bilateral procedures. 
4. Confirm left-right symmetry using the facial midline, interpupillary line, palpebral fissure length, and intercanthal distance before local anesthesia or incision.
5. Position the necessary reference points to establish the baseline for subsequent symmetry assessment and medial canthal angle regression measurement (Figure 2A).
6. Verify the anatomical landmarks and confirm the designated fold axis (Figure 2B).
7. Ensure the planned folding direction and fixation points align with predefined anatomical references.

3. Surgical incision and ligament exposure

1. Create the incision strictly along the preoperative design line.
2. Dissect through the subcutaneous and orbicularis oculi layers in the premarked plane, staying superficial to the lacrimal drainage structures and avoiding unnecessary medial orbital trauma. 
3. Identify the anterior MCT limb as a firm fibrous band extending from the medial canthal soft tissue toward the medial orbital rim/periosteal region, while keeping the lacrimal canaliculus, medial canthal vessels, and caruncular tissue outside the dissection field.
   CAUTION: Carefully identify and preserve the lacrimal canaliculus, medial canthal vessels, and adjacent soft tissues throughout the dissection. Excessively deep or misdirected dissection may result in injury to the lacrimal drainage system, increased bleeding, postoperative scarring, or distortion of the medial canthal contour.
4. Identify the anterior branch of the medial canthal tendon within the medial orbital tissues (Figure 1A).
5. Expose the anterior fibrous band clearly prior to initiating the folding procedure (Figure 2D).

4. Intraoperative folding and parameter quantification

1. Apply standardized medial-to-lateral traction with fine forceps or a skin hook until the medial fold is flattened without visible blanching, tearing, or eyelid-margin distortion. Because direct force measurement is not routinely available during surgery, define acceptable traction operationally as the minimum force required to achieve complete fold flattening while maintaining stable anatomical landmarks.
2. Execute the anterior branch folding maneuver along the confirmed axis (Figure 2E).
3. Define folding length (L, mm) as the linear distance from the folding origin on the exposed anterior MCT limb to the intended fixation point along the folded vector; measure it with a digital surgical caliper to the nearest 0.1 mm. 
4. Define fold-point position (P, mm) as the distance from the medial canthal point reference to the fold apex or fixation-related turning point, measured intraoperatively using the same digital surgical caliper. 
5. Measure folding angle (θ, degrees) using calibrated image analysis based on an intraoperative photograph obtained perpendicular to the surgical field, with the angle defined between the folded MCT limb vector and the medial orbital rim/fixation reference line. 
   NOTE: The measurement geometry and anatomical reference points are shown in Figure 1C and Figure 3C,D. Repeat each measurement three times and use the mean value for analysis.

5. Fixation and wound closure

1. Fix the folded anterior MCT limb using 6-0 nylon or 6-0 polypropylene on a fine ophthalmic needle. Place one or two interrupted fixation sutures according to tissue thickness and intraoperative stability, anchoring the fold to the periosteal/deep fascial plane when safely accessible or to the tendon/soft-tissue plane when deep anchoring is not feasible. 
2. Tie knots away from the skin edge and verify that the medial canthal contour is stable without excessive blanching, eyelid-margin distortion, or canalicular traction.
3. Record the specific fixation plane utilized during the procedure.
4. Grade suture tension before closure using predefined categories: low tension indicates visible recoil or unstable fold apposition after knot tying; moderate tension indicates stable apposition with preserved capillary color and no eyelid-margin distortion; high tension indicates firm fixation accompanied by tissue blanching, fold compression, or contour distortion. 
   NOTE: For modeling, moderate and high tension were grouped as moderate/high tension and compared with low tension.
5. Redrape the flap and perform wound closure while ensuring tension equilibrium (Figure 2G).
6. Capture immediate postoperative photographs to document the initial anatomical reduction (Figure 2H).
   NOTE: At this stage, the surgical procedure has been completed, and the patient may recover in accordance with institutional postoperative care protocols. Standardized follow-up photography and outcome measurements should be performed at scheduled postoperative visits, beginning approximately 1 week after surgery.

6. Postoperative follow-up and image measurement

1. Schedule postoperative follow-up visits at approximately 1 week, 1 month, 3 months, and 6 months.
2. Obtain standardized anterior and close-up images at each follow-up appointment utilizing the standardized imaging protocol described in section 1.
3. Define the measurements obtained at the 1-week follow-up point as the baseline reference state.
4. Define the medial canthal angle (MCA) as the angle formed at the medial canthal point by the superior and inferior palpebral-margin reference lines extending along the upper and lower eyelid margins. Identify the medial canthal point and the corresponding eyelid-margin reference landmarks according to the predefined anatomical landmark system (Figure 3A)¹⁴.
5. Calibrate each photograph using the standardized imaging conditions described in section 1, including the fixed patient-to-camera distance and camera alignment. Measure MCA using ImageJ software (version 1.54) based on the calibrated image. 
6. Have two independent masked observers perform all measurements. When the difference between observers is ≤2°, use the mean value for analysis; when the difference exceeds 2°, obtain a third independent measurement and use the median of the three values.
7. Define medial canthal angle regression as a postoperative increase in MCA of at least 3° and/or posterior displacement of the medial canthal point of at least 1.0 mm at 6 months compared with the 1-week postoperative baseline. 
8. Grade severity as mild (1.0 to <2.0 mm displacement or 3° to <5° MCA increase), moderate (2.0 to <3.0 mm displacement or 5° to <8° MCA increase), or severe (at least 3.0 mm displacement or at least 8° MCA increase). When angle and displacement categories differ, assign the higher severity grade.

7. Statistical analysis and predictive modeling

1. Express continuous variables as mean and standard deviation or median and interquartile range.
2. Represent categorical variables as frequencies and percentages.
3. Compile baseline demographic, anatomic, and surgical characteristics to compare the regression and non-regression cohorts (Table 1).
4. Summarize intraoperative quantitative folding parameters and calculate measurement reliability metrics (Table 2).
5. Document the incidence and severity grading of postoperative medial canthal angle regression across all evaluation timepoints (Table 3).
6. Conduct univariable analysis to identify preliminary associations between candidate variables and the primary endpoint (Table 4).
7. Perform multivariable logistic regression to identify independent predictors and estimate adjusted effects (Table 5).
8. Assess missing data before modeling. Exclude variables with more than 20% missing data from the primary model. 
9. For variables with ≤20% missingness and considered plausibly missing at random, perform multiple imputation using chained equations with 20 imputed datasets, including the outcome variable and all candidate predictors in the imputation model. Compare the pooled estimates from the imputed datasets with results from a complete-case sensitivity analysis.
10. Select candidate predictors using clinical pre-specification and univariable screening (P < 0.10), while avoiding highly collinear variables in the same model. Evaluate collinearity with variance inflation factors and preserve an acceptable event-per-variable ratio for the 74 regression events.
11. Perform internal validation with 1,000 bootstrap resamples to estimate optimism-corrected discrimination, calibration slope and intercept, Brier score, and decision-curve net benefit. Conduct all analyses in an appropriate data analysis software application (here, SPSS version 26.0) and verify graphics and additional model diagnostics using the corresponding statistical software workflow.
12. Evaluate the discriminative ability of the model using the area under the ROC curve (AUC).
13. Summarize predictive performance, calibration parameters, and net clinical benefit through decision curve analysis (Table 6).

Results

Baseline characteristics

According to the primary outcome definition, 331 patients underwent modified V-shaped concealed flap canthoplasty and were included in the analytic cohort. At the 6-month primary endpoint, 74 patients (24.6%) met the criteria for medial canthal angle regression and 257 did not. The cohort was predominantly female (286/331, 86.4%), with a mean age of 26.8 ± 5.9 years. Patients with regression were older (28.4 ± 6.2 vs 26.3 ± 5.7 years, P = 0.012). Age should therefore be interpreted as a candidate risk marker requiring further investigation.

Anatomical and morphological differences were more unfavorable in the regression group: severe epicanthal folds were more common (33.8% vs 19.8%), thick-soft tissue was more frequent (24.3% vs 17.5%), intercanthal distance was greater (35.3 ± 2.7 mm vs 34.4 ± 2.8 mm, P = 0.019), and preoperative medial canthal angle (MCA) was larger (43.2° ± 4.7° vs 41.4° ± 4.5°, p = 0.006). These findings suggest that baseline morphological load contributes to postoperative stability, although the strongest modifiable signals were intraoperative.

The regression group had longer operative time (45.9 ± 10.4 vs 41.8 ± 9.5 min, P = 0.003) and a higher proportion of tendon/soft-tissue fixation than the no-regression group. These findings support reporting the fixation plane as a reproducible procedural variable rather than describing it only as a surgeon-dependent technical impression.

Distribution and reliability of intraoperative folding parameters

As shown in Table 2, folding length, folding angle, and fold-point position differed clearly between groups. Compared with patients without regression, patients with regression had smaller L (5.8 +/- 1.2 vs 6.9 +/- 1.2 mm), smaller θ (32.8 +/- 7.4 vs 38.1 +/- 7.6 degrees), and smaller P (3.7 +/- 0.9 vs 4.5 +/- 0.8 mm; all p < 0.001). The intra-rater intraclass correlation coefficients (ICCs) ranged from 0.91 to 0.93, and the inter-rater ICCs ranged from 0.88 to 0.90 for the three primary measurements, supporting reproducibility when the traction and landmark protocol is followed.

Incidence, timing, and severity grading of regression

As shown in Table 3, medial canthal angle regression accumulated mainly during the first 3-6 months. Any regression was observed in 22/324 patients (6.8%) at 1 month, 49/312 (15.7%) at 3 months, and 74/301 (24.6%) at the 6-month primary endpoint. At 6 months, 41 cases were mild, 24 were moderate, and 9 were severe; the median displacement was 1.8 mm (interquartile range [IQR], 1.1-2.7mm), and the median MCA increase was 4.5 degrees (IQR, 3.0-6.4). The 12-month extended follow-up rate was similar (52/214, 24.3%), suggesting that most measurable regression occurred before or around the 6-month endpoint.

Candidate predictor screening

Univariable screening (Table 4) showed that regression risk was associated with both baseline morphology and intraoperative reconstruction. Age (odds ratio [OR], 1.05 per year, 95% confidence interval [CI], 1.01–1.10), body mass index (BMI) (OR 1.13 per kg/m2, 95% CI 1.02-1.24), epicanthal fold severity (OR 1.63 per grade, 95% CI 1.18–2.25), thick soft tissue (OR 1.71, 95% CI 1.01–2.88), intercanthal distance (OR 1.12 per mm, 95% CI 1.01–1.25), preoperative MCA (OR 1.07 per degree, 95% CI 1.02–1.13), operative time (OR 1.28 per 10 min, 95% CI 1.07–1.53), tendon/soft-tissue fixation (OR 1.91, 95% CI 1.12–3.25), low suture tension (OR 2.13, 95% CI 1.23–3.69), and left-right measurement differences were candidate risk markers. In contrast, larger L, θ, and P were protective in univariable analysis.

Multivariable risk factors and effect estimation

Building on univariable screening, the multivariable model incorporated preoperative morphology, perioperative factors, and intraoperative quantitative measurements of anterior MCT limb folding. The same endpoint and anatomical terminology were used throughout the analysis: medial canthal angle regression and the anterior limb of the MCT.

Figure 4A presents adjusted odds ratios and confidence intervals from the multivariable model. Larger L, θ, and P were associated with lower odds of medial canthal angle regression, whereas tendon/soft-tissue fixation and low suture tension were associated with higher odds. Baseline morphology variables retained independent contributions, indicating that preoperative tissue load and intraoperative structural reconstruction both influence postoperative stability.

Figure 4B illustrates the modeled relationship between folding parameters and predicted medial canthal angle regression probability. The curves suggest clinically useful target ranges for L, θ, and P, but they should be interpreted as model-based trends that require validation rather than as fixed universal cutoffs.

Figure 4C summarizes the relative weights of each predictive factor in the model by ranking them according to standardized contribution. The intraoperative folding parameters contributed substantially to the internally validated model and provided quantitative information for estimating regression risk within this cohort. In contrast, some preoperative demographic factors contributed less, suggesting that relying solely on preoperative characteristics is insufficient for effective stratification of regression risk. These findings indicate that intraoperative quantitative parameters were associated with postoperative medial canthal angle regression and provided additional information within the internally validated model.

Figure 4D demonstrates the actual incidence rate differences of regression stratified based on model-predicted probabilities. The regression rates exhibited progressive separation across low-risk, intermediate-risk, and high-risk tiers, suggesting that the internally validated model may provide preliminary risk differentiation within this cohort. However, these findings are based on internal validation only, and external validation is required before the model can be considered for routine intraoperative decision-making or individualized follow-up planning.

Prediction model derivation and interpretability

The final multivariable model (Table 5) retained intraoperative folding parameters as independent protective factors: L (adjusted OR 0.57 per 1 mm increase, 95% CI 0.45–0.73, P < 0.001), θ (adjusted OR 0.72 per 5° increase, 95% CI 0.60–0.86, P < 0.001), and P (adjusted OR 0.44 per 1 mm increase, 95% CI 0.31–0.63, P < 0.001). Higher risk was associated with tendon/soft-tissue fixation (adjusted OR 1.68, 95% CI 1.10–2.56), low suture tension (adjusted OR 1.84, 95% CI 1.15–2.93), greater epicanthal fold severity (adjusted OR 1.51 per grade, 95% CI 1.10-2.07), thick-soft tissue (adjusted OR 1.60, 95% CI 1.03–2.50), larger preoperative MCA (adjusted OR 1.06 per degree, 95% CI 1.00–1.12), and longer operative time (adjusted OR 1.23 per 10 min, 95% CI 1.01–1.49).

Model performance, calibration, and internal validation

Model performance was reported using discrimination, calibration, prediction error, internal validation, and clinical net benefit. The predictive model should be interpreted as a representative application of the measurement protocol rather than as a fully implemented clinical decision tool.

Figure 5A presents the ROC curve, with an apparent AUC of 0.85 (95% CI, 0.81–0.89) and an optimism-corrected AUC of 0.81 (95% CI, 0.77–0.85). Calibration was acceptable but not perfect: the apparent calibration slope was 0.94, and the intercept was 0.05, changing to 0.89 and 0.08 after optimism correction (Figure 5B). The Brier score increased from 0.12 to 0.15 after correction, indicating modest optimism in the apparent model.

Figure 5C summarizes apparent and optimism-corrected metrics. Figure 5D shows threshold-based classification performance at the prespecified 0.20 threshold, with corrected sensitivity of 0.78, specificity of 0.70, positive predictive value (PPV) of 0.65, and negative predictive value (NPV) of 0.82. These values suggest clinically useful discrimination but also show that misclassification remains possible.

Table 6 reports an overall net reclassification improvement of 0.22 (95% CI, 0.11–0.35) and a decision-curve net benefit of 0.23 at the 0.20 threshold, decreasing to 0.19 after optimism correction. These results indicate that the model may provide preliminary information for outcome assessment within this cohort; however, prospective evaluation and external validation are required before considering clinical implementation.

Clinical utility and representative outcome visualization

Figure 6A presents decision-curve analysis across threshold probabilities, with the main manuscript threshold set at 0.20. Figure 6B displays observed regression rates across low-, intermediate-, and high-risk strata; this stratification is useful for illustrating how the model separates risk groups, but it should not be interpreted as proof that intraoperative model-guided changes reduce regression. Figure 6C shows sensitivity and specificity at the 0.20 threshold, and Figure 6D summarizes net reclassification improvement (NRI). Because these are internally validated estimates from a single-center cohort, the values should be used to plan future validation rather than to mandate a universal operative threshold.

Figure 7 presents a representative case of postoperative medial canthal angle (MCA) regression after modified V-shaped concealed flap canthoplasty. Standardized preoperative and postoperative full-face and close-up photographs were used to assess changes in MCA and medial canthal point position. During follow-up, the case demonstrated MCA regression of 3.2° and medial canthal point displacement of 1.8 mm. This case illustrates the postoperative image-measurement workflow and should be interpreted as an illustrative outcome rather than independent validation of the prediction model. At the cohort level, relatively larger folding parameters and appropriate fixation characteristics were associated with postoperative stability, whereas smaller L, θ, or P values and unfavorable fixation characteristics were associated with increased regression risk. These patterns should be interpreted as reference patterns rather than universal thresholds because they require validation in independent populations.

Figure 1: Anatomical schematic of the medial canthal tendon (MCT) anterior limb and standardized definition of intraoperative folding parameters. (A) Medial canthus anatomy highlighting the MCT anterior limb and adjacent orbicularis layers; (B) intraoperative exposure showing folding trajectory and landmarks; (C) close-up defining folding length (L), angle (θ), and fold-point position (P). Please click here to view a larger version of this figure.

Figure 2: Step-by-step intraoperative photographs of modified V-shaped concealed flap canthoplasty with standardized V-design marking, anterior MCT limb folding, and fixation. (A) Preoperative concealed V-shaped incision design with reference points and planned medial canthal displacement; (B) confirmation of anatomical landmarks, symmetry, and fold axis before incision; (C) incision and exposure of the medial canthal region, identification of the anterior limb of the medial canthal tendon (MCT); (D) standardized folding maneuver of the anterior MCT limb, fixation of the folded structure; (E) flap redraping and wound closure; (F) immediate postoperative view. Please click here to view a larger version of this figure.

Figure 3: Intraoperative quantification workflow for anterior MCT limb folding parameters and medial canthal angle measurement: reference landmarks, traction standardization, and measurement geometry. (A) Reference landmarks defining the medial canthal angle (MCA), including the medial canthal point and superior/inferior palpebral-margin reference lines; (B) standardized traction direction to flatten the fold; (C) intraoperative measurement of folding length (L) and fold-point position (P); (D) close-up showing folding angle (θ) and fixation point geometry. Please click here to view a larger version of this figure.

Figure 4: Independent risk factors for postoperative medial canthal angle regression: multivariable effect estimates of intraoperative anterior MCT limb folding parameters and baseline covariates. (A) Multivariable forest plot of adjusted odds ratios (ORs) with 95% confidence intervals (CIs); (B) dose–response curves for L, θ, and P; (C) standardized predictor importance ranking; (D) observed regression rates across low/intermediate/high predicted-risk strata. Please click here to view a larger version of this figure.

Figure 5: Performance of the prediction model for 6-month medial canthal angle regression: discrimination, calibration, and internal validation. (A) Receiver operating characteristic (ROC) curve with apparent area under the curve (AUC) and 95% confidence interval (CI); (B) calibration plot with apparent calibration intercept, slope, and Brier score; (C) apparent and optimism-corrected metrics; (D) threshold-based classification performance at the 0.20 risk threshold. Please click here to view a larger version of this figure.

Figure 6: Clinical utility of the internally validated model: decision-curve analysis and risk stratification of predicted 6-month medial canthal angle regression probability. (A) Decision curve analysis; (B) observed regression rates by predicted-risk stratum; (C) sensitivity and specificity at the 0.20 threshold; (D) net reclassification improvement (NRI). Please click here to view a larger version of this figure.

Figure 7: Representative case demonstrating postoperative medial canthal angle regression after modified V-shaped concealed flap canthoplasty. (A–D) Standardized preoperative and postoperative full-face and close-up photographs used to evaluate the medial canthal angle (MCA) and medial canthal point position. During follow-up, the case showed MCA regression of 3.2° and medial canthal point displacement of 1.8 mm. Please click here to view a larger version of this figure.

Table 1: Baseline demographic, anatomic, and surgical characteristics of the study cohort. Abbreviations: BMI, Body Mass Index; ICD, Intercanthal Distance; MCA, Medial Canthal Angle. Please click here to download this Table.

Table 2: Intraoperative quantitative folding parameters of the anterior limb of the medial canthal tendon (MCT) and measurement reliability. Abbreviations: ICC, intraclass correlation coefficient. Please click here to download this Table.

Table 3: Definition and incidence of postoperative medial canthal angle regression across follow-up timepoints. Abbreviations: IQR, Interquartile Range; MCA, Medial Canthal Angle. Please click here to download this Table.

Table 4: Univariable analysis of candidate predictors for postoperative medial canthal angle regression. Abbreviations: ICD, Intercanthal Distance; MCA, Medial Canthal Angle; OR, odds ratio; CI, confidence interval. Please click here to download this Table.

Table 5: Final multivariable model coefficients and risk score derivation for predicting medial canthal angle regression. Abbreviations: CI, confidence interval; OR, odds ratio; MCA, Medial Canthal Angle. Please click here to download this Table.

Table 6: Predictive performance, calibration, and clinical utility of the model (internal validation). Abbreviations: NRI, Net reclassification improvement; CI, confidence interval; PPV, Positive Predictive Value; NPV, Negative Predictive Value. Please click here to download this Table.

Discussion

This study establishes a standardized intraoperative measurement workflow for quantifying anterior MCT limb folding during modified V-shaped concealed flap canthoplasty. The protocol integrates reproducible anatomical landmark identification, controlled traction, quantitative assessment of folding geometry, and structured documentation of fixation characteristics. The predictive model represents a secondary application of these standardized measurements rather than the primary methodological contribution of this work. The 6-month regression incidence was 24.6% in this cohort. Reported postoperative outcomes vary across different surgical techniques, including modified inverted L and other flap-based procedures15–18; therefore, direct comparisons with previous studies should be interpreted cautiously. Future comparative studies using standardized definitions and assessment protocols are needed to better contextualize the clinical performance of different epicanthoplasty approaches19. Smaller L, θ, and P, tendon/soft-tissue fixation, low suture tension, greater epicanthal fold severity, thick soft tissue, larger preoperative MCA, and longer operative time were associated with increased risk. The internally validated model demonstrated reasonable discrimination and calibration within this cohort; however, the principal methodological contribution of this work is the standardized and reproducible intraoperative measurement workflow. The most critical steps are standardized photography, accurate identification of the anterior MCT limb, reproducible traction during measurement, consistent measurement of L, θ, and P, and documentation of fixation plane and suture tension grade. Because these procedures require precise anatomical recognition and manual manipulation, standardized operator training and reference-based calibration are important for minimizing operator-dependent variability. Future multicenter studies should incorporate structured training procedures and standardized measurement criteria to improve reproducibility across surgeons and institutions. Errors in any of these steps may affect measurement reliability, postoperative outcome assessment, and the interpretation of associations between intraoperative parameters and surgical stability. If landmarks are poorly visible, repeat photography or delay measurement until edema, bleeding, or traction distortion is corrected. If left-right L differs by more than approximately 0.5 mm or θ differs by more than approximately 2° in bilateral cases, recheck the fold axis and traction direction before fixation. If tissue blanching, eyelid-margin distortion, or canalicular traction appear during knot tying, reduce suture tension or adjust the fixation plane. If follow-up photographs are inconsistent, repeat imaging under the standardized setup before classifying regression. This was a single-center cohort with internal validation only; therefore, calibration and threshold performance may differ in other centers, surgeons, patient populations, or imaging workflows. Suture tension grading remains partly subjective despite the operational definitions, and the protocol requires sufficient familiarity with medial canthal anatomy and measurement landmarks. Additional refinement, including objective force measurement and broader operator training, may further reduce variability before widespread adoption. The proposed thresholds for regression severity and risk stratification should be externally validated before being used as universal decision rules. Although increasing age was identified as an independent predictor of postoperative medial canthal angle regression, the underlying mechanism remains uncertain. Age-related changes in periocular skin elasticity, collagen remodeling, and medial canthal soft-tissue support may contribute to postoperative tissue remodeling after surgery. However, these potential explanations remain hypothetical because this study did not directly assess age-related anatomical or biomechanical alterations. Therefore, age should currently be interpreted as an observational risk marker rather than evidence of a causal mechanism, and further anatomical and biomechanical studies are required to clarify this association. In practice, the measurement protocol may provide a structured approach for documenting folding geometry before closure, maintaining intraoperative records, facilitating postoperative audit, and supporting standardized data collection for future multicenter studies. The representative case illustrates that the protocol is intended to facilitate standardized assessment of intraoperative geometry rather than define absolute surgical success criteria. Rather than promising precise real-time prevention of regression, the model should be used to support risk awareness, guide follow-up planning, and generate hypotheses for multicenter validation. Future application of this protocol should prioritize standardized operator training, reproducible identification of anatomical landmarks, and consistent intraoperative measurement procedures. Establishing shared reference materials, quality-control criteria, and multicenter validation frameworks will be essential for reducing operator-dependent variability and determining the generalizability of this workflow across different surgical settings. In conclusion, standardized intraoperative measurement of anterior MCT limb folding is feasible and reproducible when traction, landmarks, and fixation documentation are controlled. The principal contribution of this work is the standardized intraoperative measurement protocol for anterior MCT limb folding. The internally validated model serves as a representative application of these quantitative measurements and suggests associations among folding geometry, fixation strategy, and 6-month regression of the medial canthal angle. External validation is required before broader clinical application.

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
Digital Surgical Caliper	Mitutoyo Corporation	Series 500	Measurement of intraoperative folding length (L) and fold-point position (P)
ImageJ Software	National Institutes of Health (NIH)	Version 1.54	Measurement of medial canthal angle and landmark coordinates
SPSS Statistical Software	IBM Corporation	Version 26.0	Multivariable logistic regression and predictive model construction
Standardized Digital Camera System	Canon Inc.	EOS 90D	Acquisition of standardized perioperative photographs
Surgical Protractor	Shanghai Medical Instruments Co., Ltd.	Standard Surgical Grade	Measurement of intraoperative folding angle (θ)