Surgical Outcomes of Transposition Surgery for the Correction of Large-Angle Strabismus

Many potential surgical options exist to address large-angle deviations and head turns that result from various forms of paralytic strabismus. Muscle transposition surgeries serve as suitable alternatives to simple resection-recessions. Here, we report outcomes of augmented Hummelsheim and X-type transpositions for the correction of large-angle strabismus and provide insights for surgical planning.

Surgeons’ Corner Section Editors: Vivek R. Patel, MD Prem Subramanian, MD, PhD Surgical Outcomes of Transposition Surgery for the Correction of Large-Angle Strabismus Kimberly Kinga Gokoffski, MD, PhD, Jacob Lifton, MD, Benjamin Yixing Xu, MD, PhD, Vivek Ravindra Patel, MD Background: Many potential surgical options exist to address large-angle deviations and head turns that result from various forms of paralytic strabismus. Muscle transposition surgeries serve as suitable alternatives to simple resection–recessions. Here, we report outcomes of augmented Hummelsheim and X-type transpositions for the correction of large-angle strabismus and provide insights for surgical planning. Methods: We performed a retrospective chart review of 40 consecutive patients with strabismus who were treated with an augmented Hummelsheim or X-type transposition surgery at a single academic medical center. Etiologies included cranial nerve palsies (n = 26), monocular elevation palsy (n = 3), Duane syndrome (n = 1), traumatic extraocular muscle damage (n = 8), and chronic progressive external ophthalmoplegia (n = 2). All patients were followed for a minimum of 2 months postsurgery. Logistic regression analyses were performed to assess for predictors of surgical outcome. Results: Forty consecutive patients were enrolled in our series. The median preoperative deviation was 46.5D (interquartile range [IQR] 35–70). The median postoperative deviation 2 months after surgery was 0.5D (IQR 0–9.5), which represented a significant improvement (P , 0.001). Thirtythree patients (82.5%) experienced an improvement in range and/or centration of binocular single vision (BSV). More patients who underwent an augmented Hummelsheim procedure and had a small overcorrection at postoperative day 3 had a favorable result on postoperative month 2 (79%) comRoski Eye Institute (KKG, JL, BYX, VRP), University of Southern California, Los Angeles, California; Department of Ophthalmology (JL), University of California San Francisco, San Francisco, California; and Gavin Herbert Eye Institute (VRP), University of California, Irvine, Irvine, California. This work was supported by grants to K. K. Gokoffski (KL2 Career Development Award from the SC-CTSI (NCATS UL1TR001855)), B. Y. Xu (K23 EY029763), and an unrestricted grant to the USC Roski Eye Institute from Research to Prevent Blindness. The authors report no conflicts of interest. Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s Web site (www.jneuro-ophthalmology.com). Address correspondence to Kimberly Kinga Gokoffski, MD, PhD, Department of Ophthalmology, Roski Eye Institute University of Southern California, 1450 San Pablo Street 4th Floor, Los Angeles, CA 90033; E-mail: [e-mail] e806 pared with those that were initially under-corrected (38%). Multiple logistic regressions found larger preoperative deviation (P , 0.005) and esotropia (P , 0.021) to be predictors of a less favorable surgical outcome (C-statistic = 0.83). Subgroup analysis revealed that less, favorable outcome after X-type transposition occurred most frequently in patients undergoing correction of an esodeviation. Conclusion: Augmented Hummelsheim transposition techniques offer effective treatment options for paralytic strabismus with esotropic deviations, whereas X-type transpositions are effective for exotropic deviations and deviations from severe inferior rectus damage. In addition to potentially providing a wider field of BSV, improved centration is often achieved. Journal of Neuro-Ophthalmology 2021;41:e806–e814 doi: 10.1097/WNO.0000000000001372 © 2021 by North American Neuro-Ophthalmology Society M any potential surgical options exist to address large-angle deviations and head turns that result from various forms of paralytic strabismus. When the impaired muscle has substantial residual contractile force and near full ductions, combined resection–recession procedures are often sufficient to expand the field of binocular single vision (BSV). However, when the weak muscle has little to no contractile force, simple resection–recession procedures are unable to sufficiently improve the rotational forces of the eye and can actually decrease the range of BSV (1,2). Muscle transposition surgeries, such as the Hummelsheim procedure, can serve as suitable alternatives to simple resection– recessions. The traditional Hummelsheim surgery, originally described in 1908 as a surgical technique to correct severe abduction deficits (3), involves transposing half of the superior and inferior rectus (IR) muscles to the paralyzed lateral rectus. This procedure corrects esodeviations in paralytic strabismus by redirecting a portion of the tensile force of the remaining functioning muscles to that of the lateral rectus muscle. The advantage of the Hummelsheim procedure over traditional, full-tendon transposition surgeries such as the Knapp (4) is that it is relatively ciliary vessel sparing and likely carries a lower risk of anterior segment ischemic syndrome. Gokoffski et al: J Neuro-Ophthalmol 2021; 41: e806-e814 Copyright © North American Neuro-Ophthalmology Society. Unauthorized reproduction of this article is prohibited. Surgeons’ Corner Whether the traditional Hummelsheim transposition is as effective at correcting severe deficits in adduction, elevation, or depression is unknown. Given that convergence tone vastly outweighs divergence tone, the traditional Hummelsheim transposition may be insufficient to correct exodeviations. In consideration of this, several modifications to the original Hummelsheim procedure have been described to strengthen its effect on eye position. To maximize the transposition force while minimizing the risk of anterior segment ischemia, Rosenbaum proposed injecting botulinum toxin into the antagonist rectus muscle instead of performing a recession (5,6). Posterior muscle union combined with posterior scleral fixation sutures has been shown to augment the transposition effect by increasing the tonic contractile forces generated by the muscles and by redirecting the vector forces of the muscle so that they are more parallel to the action of the paralyzed muscle (7). To further increase the tonic forces of the transposed muscle segments, Brooks et al proposed concurrent resection of the transposed segment (8), whereas Cho et al found success with an X-type augmentation in which the transposed muscles are crossed under the paralytic muscle (9). Here, we present a case series describing patients with different etiologies for paralytic strabismus and their responses to the augmented Hummelsheim and X-type transposition surgeries. We also perform logistic regression analyses to identify risk factors associated with a less favorable surgical outcome. METHODS Data Collection This study was approved by the University of Southern California Institutional Review Board and conformed to the requirements of the Declaration of Helsinki and the US Health Insurance Portability and Accountability Act. We retrospectively reviewed the records of all consecutive patients older than 18 years who underwent an augmented Hummelsheim or X-type transposition procedure between 2014 and 2020 by 1 of 2 strabismus surgeons (K.K.G. or V.R.P.) at the Roski Eye Institute and the Los Angeles County Hospital-University of Southern California Hospitals (LAC). Patients were identified by reviewing the surgical log and operative reports of both physicians. Information was extracted from clinic notes and operative reports. We excluded patients for whom staged surgeries were planned to avoid underestimating the effect of each procedure. Patients who were found to have a slipped muscle on reoperation were also excluded from our analysis because this would also inaccurately reflect the effect of each procedure. Finally, any patient with less than a 2-month follow-up visit was excluded from the study. If more than 2-month followup data were available, this information was also collected. The following independent variables were collected as continuous data: patient age at the time of surgery (years), Gokoffski et al: J Neuro-Ophthalmol 2021; 41: e806-e814 duration of symptoms before surgical correction (years), preoperative manifest deviation on primary gaze (prism diopter [PD]), qualitative measurement of presurgical ductions, and postoperative day 3 manifest deviation (PD). Ocular alignment was measured with spectacle correction using alternate prism cover testing at distance (at 6 m) when possible. Krimsky light reflex was used when the patient could not fixate on the Snellen optotype with both eyes. Ductions were graded from 0 to 28 on a scale: 0 = full ductions, 21 = 75% ductions past midline, 22 = 50% ductions past midline, 23 = 25% ductions past midline, 24 = ability to move up to but not past midline, 25 = 25% deficit from midline, 26 = 50% deficit from midline, 27 = 75% deficit from midline, and 28 = complete absence of rotation toward the midline. The following independent variables were collected as categorical variables: biological gender (male or female), diagnosis (isolated sixth nerve palsy, isolated third nerve palsy, multiple cranial nerve [CN] palsies, monocular elevation [MOE] palsy, Duane syndrome, traumatic extraocular muscle damage, or chronic progressive external ophthalmoplegia [CPEO]), history of previous strabismus or orbital surgery (positive or negative), operative eye (right, left, or bilateral), presence of resistance during forced ductions (positive or negative), and various augmentations performed during the surgery (positive or negative for scleral pass, resection of transposed muscles, antagonist muscle recession, or botulinum injection). Our primary outcome was ocular alignment in primary gaze at 2 months postsurgery and was collected as a continuous variable (PD). For the purposes of analysis, the preoperative deviation was given a positive (+) value and overcorrections were deemed negative (2) deviations. Our primary outcome variable was then converted into a binary variable, favorable vs less favorable, for regression analysis. Patients were deemed to have a favorable surgical outcome if their residual horizontal deviation was less than 10D and their vertical deviation less than 2D on primary gaze. Conversely, patients were deemed to have an unfavorable outcome if either the horizontal was greater than 10D or the vertical deviation was greater than 2D. Secondary outcome variables included change in ductions (improvement, no change, or worsening) and change in centration of BSV (improvement, no change, or worsening). Both postoperative ductions and centration of BCV were recorded as discrete variables. The need for reoperation was recorded as a categorical variable. After recording presurgical and postsurgical ductions, we calculated the change in ductions and centration of binocular vision and converted these variables into categorical variables (no change, improved, or worsened). Surgical Procedure All patients underwent an augmented Hummelsheim procedure (3) or X-type transposition (9) with modifications as documented in Supplemental Digital Content 1 (see Supplemental Table 1, http://links.lww.com/WNO/A516). e807 Copyright © North American Neuro-Ophthalmology Society. Unauthorized reproduction of this article is prohibited. Surgeons’ Corner TABLE 1. Patient demographics Age (yrs) Duration of symptoms (yrs) Mean (SD) Median (IQR) Min. Max. 52.1 (16.3) 13.5 (17.9) 50.5 (40–66) 3 (1.1–22) 24 0.16 87 71 # (%) Gender Male Female N = 40 Operated eye OD OS OU N = 40 Diagnosis Cranial neuropathy Others N = 40 Previous eye surgery No Yes N = 39 # (%) 23 (57.5%) 17 (42.5%) 17 (42.5%) 18 (42.0%) 5 (12.5%) 26 (65.0%) 14 (35.0%) Deviation Esotropia Exotropia Hypertropia Hypotropia N = 40 Procedure type Traditional Hummelsheim X-type transposition N = 40 25 (62.5%) 6 (15%) 4 (10%) 5 (12.5%) 33 (82.5%) 7 (17.5%) Surgical augmentations* Scleral bite Botox injection Muscle resection 11/39 (28.2%) 11/40 (27.5%) 22/40 (55.0%) 31 (79.5%) 8 (20.5%) IQR, interquartile range. *Patients may have undergone multiple surgical augmentations. As such, percentages do not add up to 100%. Supplemental Digital Content 2 (see supplemental methods, http://links.lww.com/WNO/A515 for more details). Statistical Methods Statistical analysis was performed using version 14.2 of the Stata statistical software package (StataCorp LLC, College Station, TX). Analyses were conducted using a significance level of 0.05. Continuous data were described by calculating the mean, SD, median, and interquartile range (IQR). Categorical data were tabulated and percentages reported. The normality of data points was evaluated both graphically and using the Shapiro–Wilk test. Medians, ranges, and IQRs are reported herein because of nonnormal distributions of the data. TABLE 2. Patient demographics and clinical characteristics by procedure types Continuous Variables Age (yrs) Duration of symptoms (yrs) Preoperative deviation (PD) POD3 deviation (PD) POM2 deviation (PD) Categorical variables Final deviation final deviation . j10j PD horizontal or j2j PD vertical Female gender History of previous eye surgery Initial duction deficit worse than 24 Restricted forced ductions Worse postoperative ductions Worse postoperative centration Overall N = 40 Traditional Hummelsheim N = 33 X-type Transposition N=7 50.5 (40–66) 3 (1.1–22) 46.5 (35–70) 21 (-15–6) 0.5 (0–9.5) 51 (40–66) 3 (1.5–24) 45 (35–60) 0 (-6–7) 3 (0–14) 46 (29–60) 1.3 (1–3) 75 (35–90) 230 (240–212) 210 (-15–0) 0.49 0.16 0.13 0.003 0.002 15/40 (37.5%) 10/33 (30.3%) 5/7 (71.4%) 0.041 17/40 (42.5%) 8/39 (20.5%) 17/40 (42.5%) 29/35 (82.9%) 9/39 (23.1%) 2/39 (5.1%) 14/33 (42.4%) 5/32 (15.6%) 13/33 (39.4%) 25/28 (89.3%) 8/33 (24.2%) 2/33 (6.1%) 3/7 3/7 4/7 3/7 1/6 0/6 0.98 0.11 0.39 0.044 0.69 0.54 (42.9%) (42.9%) (57.1%) (42.9%) (16.7%) (0%) P* Values of continuous variables are listed as median (IQR), and categorical variables are tabulated with corresponding percentages. PD, prism diopters. *P value is derived using the Wilcoxon rank-sum test for continuous variables and the x2 test for categorical variables. e808 Gokoffski et al: J Neuro-Ophthalmol 2021; 41: e806-e814 Copyright © North American Neuro-Ophthalmology Society. Unauthorized reproduction of this article is prohibited. Surgeons’ Corner The Wilcoxon rank-sum, Kruskal–Wallis, and chi-squared tests were used to evaluate differences in continuous and categorical variables between diagnostic categories (see Supplemental Digital Content 3, Supplemental Table 2, http:// links.lww.com/WNO/A517), procedure types (Table 2), types of deviations (Table 3), and patients with favorable vs less favorable outcomes (see Supplemental Digital Content 4, Supplemental Table 3, http://links.lww.com/WNO/A518). Simple logistic regression analysis was used to evaluate age, biological gender, duration until surgery, diagnosis, preoperative deviation, type of deviation (hypertropia, hypotropia, esotropia, and exotropia), the presence of restricted forced ductions intraoperatively, and initial duction deficit worse than 24, along with procedure type and any surgical augmentations as potential predictors of less favorable surgical outcome as defined above (Table 4). This was repeated to evaluate for predictors of worsening ductions (see Supplemental Digital Content 3, Supplemental Table 2, http://links.lww.com/ WNO/A517). Our sample size was insufficient to perform the same analysis for improved centration. Multiple logistic regression analysis was then conducted to further evaluate predictors of unfavorable outcomes. Owing to our small sample size, we limited our multivariate logistic regression models to 2 independent variables each. Variables with simple regression P values ,0.2 were automatically considered for multiple regression analysis. Diagnostic category, type of deviation, initial duction deficit worse than 24, and positive forced ductions were forced into the models as well because of their presumed clinical relevance. Pearson correlation coefficients were calculated to evaluate collinearity between all independent variables (see Supplemental Digital Content 5, Supplemental Table 4, http://links.lww.com/ WNO/A519). If 2 variables were found to have a correlation coefficient greater in magnitude than 0.7, one of the 2 variables was excluded from the final multiple regression model at the discretion of the investigators. A series of multiple logistic regressions were then performed, and models were compared using the best subsets analysis (see Supplemental Digital Content 6, Supplemental Table 6, http://links.lww.com/ WNO/A521). Investigators examined each permutation’s C-statistic (the area under the receiver operating characteristic curve) to help select the most explanatory model (Table 5). Multiple regression analysis for worsening ductions and worsening centration (secondary outcomes) were omitted because of the small sample size. RESULTS Subject Characteristics Between 2014 and 2020, 43 patients were screened (29 from Roski Eye Institute and 14 from Los Angeles County Hospital), 5 of whom underwent bilateral surgery. One patient, who underwent a planned staged surgery, and 2 patients who developed slipped muscles postoperatively were excluded, yielding a total of 40 enrolled subjects. Demographic and clinical data are presented in Table 1. Patient age ranged from 24 to 87 years (median 50.5 years). Women comprised 42.5% of the patient population. Patients reported a duration of symptoms ranging from 2 TABLE 3. Patient demographics and clinical characteristics by quadrant of deviation Continuous Variables Age (yrs) Duration of symptoms (yrs) Preoperative deviation (PD) POD3 deviation (PD) POM2 deviation (PD) Categorical variables Final deviation final deviation . j10j PD horizontal or j2j PD vertical X-type transposition Female gender History of previous eye surgery Initial duction deficit worse than 24 Restricted forced ductions Worse postoperative ductions Worse postoperative centration Overall, N = 40 Hypertropia, N=4 Hypotropia, N=5 Exotropia, N=6 P* 37.5 (29–46) 2.1 (1.1–15.5) 34 (31.5–37.5) 25 (-16–2) 25 (212.5–0.5) 43 (40–71) 36 (1–43) 30 (25–40) 1 (0–10) 0 (0–17) 56 (40–66) 3 (1.5–15) 50 (40–70) 24 (-15–0) 0 (-2–8) 53.5 (41–69) 1.3 (0.5–2) 57.5 (50–60) 4 (-2–20) 22 (6–50) 0.33 0.34 0.057 0.28 0.013 15/40 (37.5%) 1/4 (25%) 2/5 (40%) 7/25 (28%) 5/6 (83.3%) 0.085 7/40 (17.5%) 3/4 (75%) 17/40 (42.5%) 1/4 (25%) 8/39 (20.5%) 3/4 (75%) 0/5 (0%) 2/5 (40%) 2/5 (40%) 3/25 (12%) 10/25 (40%) 2/24 (8.3%) 1/6 (16.7%) 0.013 4/6 (66.7%) 0.57 1/6 (16.7%) 0.014 17/40 (42.5%) 0/4 (0%) 2/5 (40%) 12/25 (48%) 3/6 (50%) 29/35 (82.9%) 2/4 (50%) 0/4 (0%) 20/22 (90.9%) 3/5 (60%) 9/39 (23.1%) 1/3 (33.3%) 2/39 (5.1%) 0/3 (0%) 2/5 (40%) 0/5 (0%) 4/25 (16%) 1/25 (4%) 50.5 (40–66) 3 (1.1–22) 46.5 (35–70) 21 (-15–6) 0.5 (0–9.5) Esotropia, N = 25 0.33 0.082 2/6 (33.3%) 0.56 1/5 (20%) 0.54 Values of continuous variables are listed as median (IQR), and categorical variables are tabulated with corresponding percentages. *P value is derived using the Kruskal–Wallis test for continuous variables and the x2 test for categorical variables. PD, prism diopters. Gokoffski et al: J Neuro-Ophthalmol 2021; 41: e806-e814 e809 Copyright © North American Neuro-Ophthalmology Society. Unauthorized reproduction of this article is prohibited. Surgeons’ Corner TABLE 4. Simple logistic regression analysis examining predictors of less favorable surgical outcome Independent Variables Age (yrs) Female gender Diagnosis (0 = cranial neuropathy, 1 = other) Previous eye surgery Duration of symptoms (yrs) Esotropia Exotropia Hypertropia Hypotropia Restricted forced ductions Preoperative deviation (PD) Initial duction deficit worse than 24 Intraoperative muscle resection Intraoperative Botox Intraoperative scleral bite Procedure type (0 = traditional, 1 = X-Type transposition) POD3 deviation (PD) Odds Ratio (95% Confidence Interval) P 0.95 0.29 0.61 0.95 0.88 0.11 0.032 0.59 0.90 0.16 0.014 0.088 0.62 0.18 0.48 0.057 0.74 0.99 2.0 1.4 0.95 1.0 0.34 11.9 0.52 1.13 0.26 1.0 3.2 1.4 2.7 0.58 5.8 1.0 (0.96–1.0) (0.55–7.5) (0.37–5.4) (0.19–4.7) (0.97–1.0) (0.09–1.3) (1.24–116) (0.049–5.5) (0.16–7.7) (0.041–1.7) (1.0–1.1) (0.84–12.1) (0.38–5.1) (0.64–11) (0.13–2.7) (0.95–34.8) (0.98–1.0) PD, prism diopters. months to 71 years (median 3 years). Diagnoses included sixth nerve palsy (n = 19), third nerve palsy (n = 2), multiple CN palsies (n= 5), MOE palsy (n = 3), Duane syndrome Type I (n = 1), traumatic extraocular muscle damage (n = 8), and CPEO (n = 2). Given the limited number of patients with non-neurogenic etiologies of strabismus, we grouped patients into 2 categories, neurogenic vs others, for analysis purposes. Demographics by diagnosis are presented in Supplemental Digital Content 3 (see Supplemental Table 2, http://links.lww.com/WNO/A517). Eight patients had orbital surgeries before their transposition procedure: 4 underwent previous orbital wall fracture repair, 2 had strabismus surgery as an infant (greater than 30 years before the transposition surgery), one underwent a previous muscle biopsy, and the final suffered pulled-in-two syndrome (PITS) 3 months before. The median preoperative deviation for the entire cohort was 46.5D (IQR 35 to 70). The median preoperative limitation in ductions was 24.0 (IQR 26 to 24). Procedure Details Of the 40 patients enrolled, 33 underwent an augmented Hummelsheim transposition. Of these, 4 patients underwent bilateral surgery simultaneously. Twenty-two patients had a concurrent 1–3 mm resection of the transposed muscle segment. All patients had a posterior muscle union suture placed 6 mm posterior to the muscle insertion, whereas 11 patients underwent a concurrent scleral fixation suture with their muscle union suture. All but 3 patients had a concurrent recession of the ipsilateral antagonist rectus muscle, 8 of which received concurrent botulinum toxin injection. Of the 40 patients enrolled, 7 underwent an X-type transposition. One patient underwent simultaneous bilat- e810 eral surgery. Four patients had a concurrent 1–3 mm resection of the transposed muscle segment. All but one underwent antagonist rectus recession, 3 of which received concurrent botulinum toxin. Of note, the median preoperative deviation in the X-type transposition group tended to be larger than that of the augmented Hummelsheim group, but this difference was not statistically significant (75D [IQR 35–90] compared with 45D [IQR 35–60]; P , 0.13; Table 2]). Surgical Outcomes At postoperative day 3, the median angle of deviation in primary gaze of all patients was 21.0D (IQR 215 to 6). The median deviation in primary gaze 2 months after surgery was 0.5D (IQR 0 to 9.5D), which represented a significant improvement when compared with the preoperative deviation (P , 0.001; Fig. 1). A total of 25 patients (62.5%) were corrected to within 10D of orthotropia horizontally and 2D vertically—what we defined to be a favorable outcome—whereas 5 patients (12.5%) were overcorrected and 10 patients (25%) were under-corrected. Three patients undergoing horizontal transpositions TABLE 5. Multiple logistic regression model predicting less favorable outcome Independent Variables (N = 35) Esotropia Preoperative deviation (PD) P Odds Ratio (95% Confidence Interval) 0.021 0.005 0.097 (0.013–0.70) 1.1 (1.0–1.1) PD, prism diopters. Gokoffski et al: J Neuro-Ophthalmol 2021; 41: e806-e814 Copyright © North American Neuro-Ophthalmology Society. Unauthorized reproduction of this article is prohibited. Surgeons’ Corner developed new vertical deviations, one in the augmented Hummelsheim group and 2 in the X-type transposition group. Nine patients (22.5%) required a subsequent surgical procedure. At postoperative month 2, 21 patients (52.5%) experienced a widening in their range of BSV, whereas 10 patients (25%) experienced no improvement and 9 patients (22.5%) experienced a worsening. Of the 19 patients who did not experience an improvement in BSV, 12 patients (63%) experienced an improvement in centration in their area of BSV (Fig. 2). A greater proportion of patients undergoing X-type transpositions experienced less favorable surgical outcomes than the augmented Hummelsheim group (Table 2). Moreover, the mean postoperative deviation at month 2 in the X-type transposition group was significantly more overcorrected than that in the augmented Hummelsheim group (Table 2). Subgroup analyses, however, suggested that this finding may be dependent on the direction (Table 3) and magnitude of the preoperative deviation. Although there are too few patients to perform formal statistical analyses, all patients with esodeviations from CN VI palsies who underwent an X-type transposition were markedly overcorrected, whereas those who had exodeviations or hyperdeviations fared well. Along the same lines, many patients with exodeviations who underwent an augmented Hummelsheim procedure were under-corrected. The percentage of patients with a duction deficit worse than 24 before surgery did not differ significantly between FIG. 1. Comparison of preoperative and postoperative deviations. *Paired Wilcoxon signed-rank test: P , 0.001. PD, prism diopters. Gokoffski et al: J Neuro-Ophthalmol 2021; 41: e806-e814 the 2 groups (4 of 7 in X-type transposition compared with 13 of 33 in augmented Hummelsheim group; P = 0.39) nor did the percentage of patients whose ductions worsened (P = 0.69) or who experienced worsening centration (P = 0.54) (Table 2). None of the variables that were analyzed were identified as risk factors for worsening of ductions (see Supplemental Digital Content 7, Supplemental Table 5, http://links.lww.com/WNO/A520). Twenty-eight patients had a follow-up visit beyond 2 months (range 4–46 months). The median deviation in primary gaze of this subgroup was 3.5D (IQR 0–10D). The median deviation of the 19 patients who did not undergo a subsequent surgery was 2D (IQR 0–10D), which did not represent a significant change from their deviation at postoperative month 2 (median 1D; IQR 0–9 D; P = 0.88, paired Wilcoxon signed-rank test). Five of these patients had botulinum injected into the antagonist rectus muscle during surgery, and only one was documented to have a change in their final deviation (14D 5 months after surgery). The median deviation of the 9 patients who ultimately underwent a subsequent strabismus procedure was 6D (IQR 6 to 14D), which did not represent a significant improvement from their deviation at postoperative month 2 (median 15D; IQR 6–30D; P = 0.34, paired Wilcoxon signed-rank test). Logistic Regression Analysis of Unfavorable Surgical Outcomes Logistic regression analysis was performed to identify variables associated with worse outcome. Of all of the independent variables analyzed, simple logistic regression analysis identified large preoperative deviation {odds ratio 1.04 (95% confidence interval [CI] 1.0–1.1), P = 0.014} and exotropia (odds ratio 11.9 [95% CI 1.24–116]) as the only factors significantly associated with an increased risk of a less favorable outcome (Table 4). Preoperative ductions being worse than 24 (odds ratio 3.2 [95% CI 0.38–5.1], P = 0.088), positive forced ductions (odds ratio 0.26 [0.041– 1.7], P = 0.16), esotropia (odds ratio = 0.34 [95% CI 0.09– 1.3], P = 0.11), and X-type procedure (odds ratio 5.8 [95% CI 0.95–34.8], P = 0.057) approached significance as predictors (Table 4). Age, gender, diagnostic category, and duration of symptoms were not shown to significantly predict the rate of less favorable outcomes. Multiple logistic regressions found preoperative deviation, X-type procedure, and exotropia to be the most consistently significant predictors of a less favorable outcome (see Supplemental Digital Content 6, Supplemental Table 6, http://links.lww.com/WNO/A521). The bestsubsets analysis (see Supplemental Digital Content 6, Supplemental Table 6, http://links.lww.com/WNO/ A521) found esotropia and preoperative deviation taken together to be the most explanatory model of patient outcomes (C-statistic = 0.83), with esotropia having a protective effect (odds ratio 0.097 (95% CI 0.013–0.7), P = 0.021) and greater preoperative deviations predicting less e811 Copyright © North American Neuro-Ophthalmology Society. Unauthorized reproduction of this article is prohibited. Surgeons’ Corner FIG. 2. Improved ductions and centration in patients undergoing augmented Hummelsheim and X-type transposition surgeries. Pictorial representation of patients’ preoperative and postoperative ductions. Most patients experienced an improvement in both range of ductions and centration of ductions. CN, cranial nerve; MOE, monocular elevation; LR, lateral rectus, MR, medial rectus; SR, superior rectus; IR, inferior rectus; CPEO, chronic progressive external ophthalmoplegia; MLF, medial longitudinal fasciculus. favorable outcomes (odds ratio 1.1 [95% CI 1.0–1.1], P = 0.005; Table 5). CONCLUSION The optimal surgical plan for paralytic strabismus varies case by case. For patients in whom the dysfunctional muscle has substantial residual contractile tone, demonstrated by nearly full ductions, simple resection–recession procedures can be generally effective. Deviations resulting from muscles with e812 weak residual contractile tone, however, are more likely to require muscle transposition surgery. Our case series supports previous findings that transposition surgeries are effective at correcting strabismus from paralytic etiologies: the median corrected deviation was 46.5D, similar to other published case series (10,11), and postoperative deviations were significantly improved when compared with preoperative deviations for all procedure types (P , 0.001). What is not clear from the literature, however, is what clinical features are predictive of worse surgical outcome. Gokoffski et al: J Neuro-Ophthalmol 2021; 41: e806-e814 Copyright © North American Neuro-Ophthalmology Society. Unauthorized reproduction of this article is prohibited. Surgeons’ Corner Moreover, there is little consensus regarding which surgical augmentation is suited for different types of strabismus. Our case series suggests that the direction and magnitude of the preoperative deviation is an important factor to consider when choosing between different transposition augmentations. Although our data showed that X-type transpositions were more likely to be associated with a less favorable outcome than the augmented Hummelsheim (P = 0.041) (Table 2), our analysis suggested that this association is dependent on the direction of the preoperative deviation (Table 3). X-type transpositions were found to be helpful and yielded favorable outcomes for patients undergoing inferior transpositions for complete IR palsy and medial transpositions for CN Ⅲ palsy. X-type transpositions, however, led to marked overcorrections when used for CN Ⅵ palsies. These patients also underwent concurrent botulinum injection in their antagonist muscle which further contributed to their overcorrection. We suspect that the reason why our outcomes with X-type transpositions for esodeviations differed from Phamonvaechavan et al is because the latter group allowed their crossed muscles to hang back, inducing less tension (11). Conversely, failure of the augmented Hummelsheim to correct exodeviations cannot solely be attributed to weakness of the transposed muscle segments because these under corrections were noted in patients with traumatic damage to the medial rectus muscle where the superior and IR muscles had intact function. Unfortunately, the limited number of cases in our cohort makes formal statistical analyses difficult to generalize. Future studies with more patients would be helpful to further validate or challenge our findings. Another interesting finding noted during subgroup analysis arose when the deviation from postoperative day 3 (POD3) was compared with that of postoperative month 2 (POM2). We found that more patients who underwent correction with the augmented Hummelsheim procedure and had a small overcorrection at POD3 had a successful outcome compared with those who were under-corrected. At POD3, 42% of eyes (14 of 33) were overcorrected an average of 12.1D after an augmented Hummelsheim procedure. However, by POM2, the correction regressed to an average deviation of 2.4D. This resulted in 79% (11 of 14) of initially overcorrected patients achieving a successful outcome, whereas the remaining patients regressed to an under-correction. This contrasts with the 11 eyes that were under-corrected at POD3, 38% of which (5 of 13) achieved a successful outcome at POM2, the rest remaining under-corrected. Although our sample size is too small to perform formal statistical analyses, these results suggest targeting a 10D overcorrection at POD3 will increase chances of a favorable outcome. Coupling the transposition surgery with an adjustable suture technique may help “titrate” the immediate postoperative deviation toward this degree of overcorrection, potentially improving longer term outcomes. Gokoffski et al: J Neuro-Ophthalmol 2021; 41: e806-e814 Decreased range of ductions occurs more commonly after simple resection–recession surgeries but has been reported after transposition surgeries as well: 100% with recession–resections compared with 30% with transpositions (2,12,13). Comparing preoperative and postoperative ductions in our cohort demonstrated improved range in 21 (52.5%) of our patients. Ten patients demonstrated no net change (25%) and 9 (22.5%) experienced a worsening of ductions. Interestingly, although X-type transpositions put the transposed muscle on increased tension compared with the augmented Hummelsheim, neither procedure was more likely to be associated with worsening of ductions than the other (Table 2). Although scleral fixation sutures strengthen the transposition and thereby could decrease ductions, this was not identified as a risk factor for worsening of ductions (see Supplemental Digital Content 7, Supplemental Table 5, http://links.lww.com/WNO/A520). This finding is supported by Foster and Miller who demonstrated that lateral fixation sutures increased the tonic abducting force of transposed muscles by 50% over transposition alone, without decreasing adduction (14,15). Increased abduction tone was attributed to 1) redirecting the vectors of the transposed muscles and associated pulleys toward the palsied lateral rectus and 2) by increasing the passive elastic force of the transposed muscle by lengthening its muscle path. In addition to overall improvement in range of ductions, 30 patients (75%) experienced improved centration of their range of BSV. Of the 19 patients who did not improve or experienced a worsening in their range of ductions, 12 still experienced an improvement in centration of their field of BSV. Thus, a total of 82.5% (33/40) of our patients experienced an improvement in range and/or centration of BSV. These findings suggest that although partial transposition techniques may cause a reduction in range of ductions, they often improve centration of BSV, conferring a meaningful functional benefit. Induced vertical deviation is a known complication of transposition procedures, ranging between 0% and 27% of patients undergoing partial or full tendon transpositions (6,11). In our series, 3 patients undergoing horizontal transpositions developed new vertical deviations, one in the augmented Hummelsheim group and 2 in the X-type transposition group. This complication most likely results from asymmetric muscle transposition and may be managed by placing the transposed vertical muscles on an adjustable suture, which has the added benefit of being able to offset horizontal overcorrections as well (16). Our study has a number of limitations. The retrospective nature of the study limits our ability to formalize subjective measurements (e.g., ductions). Moreover, we did not formally measure the preoperative and postoperative range of BSV with kinetic visual fields. Given that movements were full in the contralateral eye of 28 of our 40 patients, we assumed that the postoperative range of BSV is proportional e813 Copyright © North American Neuro-Ophthalmology Society. Unauthorized reproduction of this article is prohibited. Surgeons’ Corner to the range of ductions of the operated eye. This assumption, however, has a tendency to overestimate BSV. Rosembaum reported a diplopia-free range of binocular vision between 25 and 85° for patients who underwent a full tendon transposition surgery with botulinum injection into the antagonist rectus muscle (5). In addition, our follow-up period is shorter than many other publications. We chose 2 months as our end point because it was the most consistent follow-up recorded by all of our surgical patients and allowed for inclusion of all patients. Although the deviation in patients who received botulinum toxin injections could continue to change 2 months after surgery, our experience has shown this to not be the case. In fact, subset analysis demonstrated no significant change in deviation in patients with a longer follow-up interval. Our data set includes a number of outliers with large deviations at postoperative month 2. One of these patients had a CN Ⅵ palsy from a fourth ventricular tumor, leading to 90D esotropia and 28 abduction. Given the large deviation and poor ductions, we performed an X-type transposition with botulinum injection and antagonist muscle recession. This, however, led to marked overcorrection. Two patients who were markedly under-corrected included a patient with biopsy confirmed CPEO and another with bilateral nuclear CN Ⅲ palsies. We suspect that weak tone in the transposed rectus muscles contributed to their under-correction. Although our small study is not powered to predict which surgical modifications are suited for specific deviations, our experience lends to the following recommendations: 1) X-type transposition surgery is a suitable option for patients with $ 25 deficit in ductions, medial transpositions for CN III paresis or severe MR trauma, and inferior transpositions required for patients with complete IR paresis. 2) We advise to avoid botulinum toxin injection of antagonist muscle when performing X-type transposition surgery because this may be unnecessary and can lead to a large, persistent overcorrection. 3) Consider concurrent antagonist muscle recession if preoperative deviation .40D. 4) Consider intraoperative botulinum injection into the antagonist rectus muscle with augmented Hummelsheim procedures if positive forced ductions are encoun- e814 tered, signifying underlying antagonist muscle contracture. 5) If performing adjustable surgery, consider 10D overcorrection when performing an augmented Hummelsheim transposition. REFERENCES 1. Nelson LB. Severe adduction deficiency following a large medial rectus recession in Duane’s retraction syndrome. Arch Ophthalmol. 1986;104:859–862. 2. Pressman SH, Scott WE. Surgical treatment of Duane’s syndrome. Ophthalmology. 1986;93:29–38. 3. Hummelsheim E. Weitere Ertahunger mit partiellar Sehnenuberpflanzung an der Augenmuskln. Arch Augenheilkd. 1908;62:71. 4. Saunders RA, Sandall GS. Anterior segment ischemia syndrome following rectus muscle transposition. Am J Ophthalmol. 1982;93:34–38. 5. Rosenbaum AL, Kushner BJ, Kirschen D. Vertical rectus muscle transposition and botulinum toxin (Oculinum) to medial rectus for abducens palsy. Arch Ophthalmol. 1989;107:820– 823. 6. Rosenbaum AL. Adjustable vertical rectus muscle transposition surgery. Arch Ophthalmol. 1991;109:1346. 7. Britt MT. Partial rectus muscle-augmented transpositions in abduction deficiency. J AAPOS. 2003;7:325–332. 8. Brooks SE, Olitsky SE, de BRG. Augmented Hummelsheim procedure for paralytic strabismus. J Pediatr Ophthalmol Strabismus. 2000;37:189–195; quiz 226-187. 9. Cho YA, Rah SH, Kim MM, Lee JY. Vertical rectus muscles transposition in large exotropia with medial rectus muscle transection following endoscopic sinus surgery. Korean J Ophthalmol. 2008;22:104–110. 10. Couser NL, Lenhart PD, Hutchinson AK. Augmented Hummelsheim procedure to treat complete abducens nerve palsy. J AAPOS. 2012;16:331–335. 11. Phamonvaechavan P, Anwar D, Guyton DL. Adjustable suture technique for enhanced transposition surgery for extraocular muscles. J AAPOS. 2010;14:399–405. 12. Molarte AB, Rosenbaum AL. Vertical rectus muscle transposition surgery for Duane’s syndrome. J Pediatr Ophthalmol Strabismus. 1990;27:171–177. 13. Hussein MA, Coats DK. Limitation of opposing ductions following augmented horizontal rectus muscle transposition. Am J Ophthalmol. 2003;136:947–949. 14. Foster RS. Vertical muscle transposition augmented with lateral fixation. J AAPOS. 1997;1:20–30. 15. Miller JM, Bockisch CJ, Pavlovski DS. Missing lateral rectus force and absence of medial rectus co-contraction in ocular convergence. J Neurophysiol. 2002;87:2421–2433. doi:. 16. Laby DM, Rosenbaum AL. Adjustable vertical rectus muscle transposition surgery. J Pediatr Ophthalmol Strabismus. 1994;31:75–78. Gokoffski et al: J Neuro-Ophthalmol 2021; 41: e806-e814 Copyright © North American Neuro-Ophthalmology Society. Unauthorized reproduction of this article is prohibited.