It is known that the screws of the eightplate hemiepiphysiodesis construct diverge as growth occurs through the physis. Our objective was to investigate whether there is a correlation between the amount of change of the joint orientation angle (JOA) and that of the interscrew angle (ISA) of the eightplate hemiepiphysiodesis construct before and after correction.
After the institutional review board approval, medical charts and Xrays of all patients operated for either genu valgum or genu varum with eightplate hemiepiphysiodesis were analyzed retrospectively. All consecutive patients at various ages with miscellaneous diagnoses were included. JOA and ISA were measured before and after correction. After review of the Xrays, statistical analyses were performed which included Pearson correlation coefficient and regression analyses.
There were 53 segments of 30 patients included in the study. Eighteen were males, and 12 were females. Mean age at surgery was 9.1 (range 3–17). Mean followup time was 21.5 (range, 7–46) months. The diagnoses were diverse. A strong correlation was found between the delta JOA (dJOA) and delta ISA (dISA) of the eightplate hemiepiphysiodesis construct (r = 0.759 (0.615–0.854, 95%CI), p < 0.001). This correlation was independent of the age and gender of the patient.
There is a strong correlation between the dISA and the dJOA. The dISA follows the dJOA at a predictable amount through formulas which regression analysis yielded. This study confirms the clinical observation of the diverging angle between the screws is in correlation with the correction of the JOA.
Level IV, Therapeutic study.
Hemiepiphysiodesis ; Eightplate ; Correlation ; Screw divergence ; Joint orientation angle ; Interscrew angle
The eightplate is a tensionband plate and screws construct. It involves a twohole titanium plate with a figure of eight configuration used with two nonlocking 4.5 mm cannulated cortical screws. It shows its effect by limiting the growth of the hemiphysis over which it is applied, and as the contralateral side of the physis continues to grow and expand, the screws of the eightplate construct diverge.^{1} Since it was introduced to the literature by Stevens^{2} it has become popular in deformity correction in the growing skeleton. The surgeons using this implant observe that the screws diverge as the correction takes place.^{2} Divergence of the screws represent growth in the related physis.^{2 ; 3 ; 4} However it has not been studied before whether this divergence of the screws correlates with the amount of correction that takes place in the individual bone segment, femur or tibia. One of the problems that arises in the followup of patients who undergo eightplate hemiepiphysiodesis is that in order to assess the degree of correction the surgeon needs to see the Xrays of the entire individual bone segment; such as AP femur or AP tibia, or the entire lower extremity. At times, those Xrays might be missing the femoral head for the femur, or ankle for the tibia. Sometimes the patients live in areas with limited resources, and would be asked to travel back and forth in order to get proper Xrays. An alternative to that could be not taking Xrays and following patients with clinical judgment, however that could yield imprecise results. The rationale behind this study was that if there is a close enough correlation between the change in the joint orientation angle (JOA) and the change in the interscrew angle (ISA), the patients could be instructed to have only an AP knee Xray taken instead of a fulllength individual bone Xray until close to full correction. We investigated whether there is a correlation between the delta joint orientation angles (ΔJOA or dJOA), namely the mLDFA (mechanical lateral distal femoral angle) and MPTA (medial proximal tibial angle),^{5} and the delta interscrew angles (ΔISA or dISA). We aimed to derive formulas through regression analyses that would enable estimation of JOA when one knows the ISA. The hypothesis was that a close correlation exists between dISA and dJOA regardless of the remaining growth potential of the related physis.
This was a retrospective study to include consecutive patients treated with eightplate hemiepiphysiodesis at a single university hospital between 2009 and 2013. After the institutional review board approval, patients were identified retrospectively according to the surgery type. Inclusion criteria involved patients operated for genu valgum or genu varum, and treated by eightplate hemiepiphysiodesis in the distal femur and/or proximal tibia. Patients with all diagnoses at all ages were included. Diagnoses were miscellaneous from congenital, developmental, metabolic, posttraumatic and idiopathic causes. They included Blounts disease, DiamondBlackfan syndrome, cystinosis, diastematomyelia, fibular hemimelia, FanconiBickel syndrome, Stickler syndrome, rickets, multiple hereditary exostosis, congenital absence of patella, arthrogryposis, coxa vara, mucopolysaccharidosis, and posttraumatic (Cozen phenomenon). Exclusion criteria included double plate hemiepiphysiodesis of the same hemiphysis^{6} and simultaneous eightplate applications of both proximal tibia and distal tibia hemiphyses within the same tibia segment. Xrays were obtained before the surgery, at the immediate postoperative period and at the final followup. Accompanying surgeries at the same setting were noted. Ipsilateral proximal tibial osteotomies were performed in 5 cases together with distal femoral hemiepiphysiodesis. Middiaphyseal femoral osteotomies were done in two cases together with distal femoral hemiepiphysiodesis and a distal tibial osteotomy was performed in one case together with proximal tibial hemiepiphysiodesis (Table 1 ). The remaining 45 segments had eightplate hemiepiphysiodesis, only. In three patients where the osteotomy was performed within the same segment, correction through the osteotomy site was acute. The mLDFA or the MPTA in those patients were measured on the immediate postoperative Xray, instead of a preoperative Xray. Final followup Xray was determined as the last Xray with the eightplate on the bone. Mechanical axis of the lower extremity was not a parameter to be measured at any time during the study. All measurements were made within the same individual bone segment, either femur or tibia. Demographic characteristics of the patients along with the type of the frontal plane deformity present, the treatment modality, if any, accompanying surgeries, and duration of follow up are given in Table 1 . Patients received the same postoperative care. They were allowed to do range of motion exercises, and weightbear as tolerated immediately. All patients included in the study have completed the followup.
Number  Age  Sex  Bone segment  Frontal plane deformity  Diagnosis  Performed surgery  Side  Additional surgery  Followup (mos) 

1  11  M  Femur  Varus  compensatory, adolescent blounts disease  LDFH  L  PTO  17 
2  9  M  Tibia  Valgus  DiamondBlackfan syndrome  MPTH  R  –  46 
2  9  M  Tibia  Valgus  DiamondBlackfan syndrome  MPTH  L  –  46 
3  17  M  Femur  Valgus  Cystinosis  MDFH  R  –  20 
3  17  M  Femur  Valgus  Cystinosis  MDFH  L  MFO  20 
4  5  M  Tibia  Valgus  Cozen phenomenon  MPTH  L  –  18 
5  7  F  Tibia  Valgus  Diastematomyelia  MPTH  R  –  23 
6  10  F  Femur  Valgus  Fibular hemimelia  MDFH  R  –  27 
7  6  M  Femur  Valgus  Fibular hemimelia  MDFH  L  PTO  16 
8  5  F  Tibia  Valgus  Cozen phenomenon  MPTH  L  –  16 
9  8  M  Femur  Valgus  FanconiBickel syndrome  MDFH  R  –  25 
9  8  M  Femur  Valgus  FanconiBickel syndrome  MDFH  L  –  25 
9  8  M  Tibia  Valgus  FanconiBickel syndrome  MPTH  R  –  25 
9  8  M  Tibia  Valgus  FanconiBickel syndrome  MPTH  L  –  25 
10  15  M  Femur  Valgus  Status post liver transplantation  MDFH  R  MFO  17 
11  11  M  Tibia  Valgus  Stickler syndrome  MPTH  L  –  25 
12  7  F  Femur  Varus  Rickets  LDFH  L  –  16 
12  8  F  Tibia  Valgus  Rickets  MPTH  L  –  15 
13  12  M  Femur  Valgus  Cystinosis  MDFH  R  –  23 
13  12  M  Femur  Valgus  Cystinosis  MDFH  L  –  33 
13  12  M  Tibia  Valgus  Cystinosis  MPTH  R  –  23 
13  12  M  Tibia  Valgus  Cystinosis  MPTH  L  –  33 
14  14  M  Femur  Valgus  Idiopathic  MDFH  R  –  8 
14  14  M  Femur  Valgus  Idiopathic  MDFH  L  –  8 
15  11  M  Femur  Valgus  Compensatory, Proximal tibia growth arrest  MDFH  R  PTO  16 
16  10  F  Femur  Valgus  Cerebral palsy  MDFH  R  –  33 
16  10  F  Femur  Valgus  Cerebral palsy  MDFH  L  –  33 
17  7  F  Femur  Valgus  Multiple hereditary exostosis  MDFH  L  –  15 
17  7  F  Tibia  Valgus  Multiple hereditary exostosis  MPTH  L  –  23 
18  13  M  Femur  Valgus  Congenital absence of patella  MDFH  R  –  25 
19  8  F  Femur  Valgus  Fibular hemimelia  MDFH  R  PTO  16 
20  13  F  Tibia  Valgus  Arthrogryposis  MPTH  R  –  26 
21  8  F  Femur  Valgus  FanconiBickel syndrome  MDFH  R  –  24 
21  8  F  Femur  Valgus  FanconiBickel syndrome  MDFH  L  –  24 
21  9  F  Tibia  Valgus  FanconiBickel syndrome  MPTH  R  –  28 
21  8  F  Tibia  Valgus  FanconiBickel syndrome  MPTH  L  –  24 
22  5  M  Femur  Valgus  Fibular hemimelia  MDFH  R  –  12 
22  5  M  Tibia  Valgus  Fibular hemimelia  MPTH  R  –  12 
23  11  M  Femur  Valgus  Multiple Hereditary Exostosis  MDFH  R  –  10 
23  11  M  Tibia  Valgus  Multiple hereditary exostosis  MPTH  L  –  10 
24  4  M  Femur  Valgus  Fibular hemimelia  MDFH  R  PTO  10 
25  13  M  Femur  Valgus  Idiopathic  MDFH  R  –  12 
25  13  M  Tibia  Valgus  Idiopathic  MPTH  R  –  12 
26  6  F  Femur  Valgus  Rickets  MDFH  R  –  7 
27  4  F  Tibia  Valgus  Coxa vara  MPTH  R  –  15 
28  7  F  Tibia  Varus  Blounts disease  LPTH  R  –  16 
28  7  F  Tibia  Varus  Blounts disease  LPTH  L  –  16 
29  8  M  Femur  Valgus  Fibular hemimelia  MDFH  L  –  17 
29  8  M  Tibia  Valgus  Fibular hemimelia  MPTH  L  DTO  17 
30  6  M  Femur  Valgus  Mucopolysaccharidosis  MDFH  R  –  32 
30  6  M  Femur  Valgus  Mucopolysaccharidosis  MDFH  L  –  32 
30  6  M  Tibia  Valgus  Mucopolysaccharidosis  MPTH  R  –  32 
30  6  M  Tibia  Valgus  Mucopolysaccharidosis  MPTH  L  –  32 
LDFH: Lateral distal femoral hemiepiphysiodesis.
MDFH: Medial distal femoral hemiepiphysiodesis.
LPTH: Lateral proximal tibial hemiepiphysiodesis.
MPTH: Medial proximal tibial hemiepiphysiodesis.
PTO: Proximal tibial osteotomy, fixation with frame.
DTO: Distal tibial osteotomy, fixation with frame.
MFO: Middiaphyseal femoral osteotomy, fixation with plate.
Joint orientation angles, namely, the mLDFA and MPTA were measured on preoperative and postoperative AP Xrays of femur or tibia, respectively. Interobserver and intraobserver reliability of these angles were tested previously, and it was stated as good to very good.^{7} Two of the investigators who were trained in limb deformity analysis reviewed the Xrays on two separate occasions. Intraclass correlation coefficient was used in testing agreement. We described ‘Interscrew angle’ (ISA) as the angle between the long axes of the screws of the eightplate on each side of the physis (Fig. 1 ). Because most of the segments were valgus and few were varus, instead of the joint orientation angles themselves, the difference between the before and after measurements were obtained. So a delta value for each JOA was calculated. Immediate postoperative and final followup ISAs were measured and the difference was analyzed in comparison to the preoperative and final followup JOAs in order to find a correlation using Pearson correlation coefficient. Regression analyses were performed while investigating the effect of ISA on the JOA. Statistical analysis was performed with IBM SPSS Version 21.0 for Windows (SPSS Inc., Chicago, IL, USA). Continuous variables in this study were given with mean ± standard deviation or median [min − max] as appropriate. Categorical variables were summarized as frequencies and percentages. Correlation between the changes in angles was determined by Pearson correlation coefficient. Stepwise multiple linear regression analysis was used to verify the relation between JOA and ISA. Significance level was determined as p < 0.05. Confidence intervals at 95% were noted, as well. The resultant data were reviewed and analyzed by a biostatistician.

Fig. 1. Graphic illustration of interscrew angle (ISA).

There were 53 segments of 30 patients included in the study. Eighteen were males, and 12 were females. Mean age at surgery was 9.1 (range 3–17). There are 29 femurs, and 24 tibias included in the study. Four patients had all four segments operated. Fifteen patients had single segment operated, femur or tibia. The remainder had two segments operated horizontally or vertically. Thirtythree segments (18 femur and 15 tibia) were operated using eightplate (Orthofix, Verona, Italy), and 20 segments (11 femur and 9 tibia) were operated using eightplate (Ortopro, Istanbul, Turkey). Both products have same material properties and implant design. The two different brands did not show any significant difference in terms of correlation between the dJOA and dISA (p = 0.994). Mean followup time was 21.5 (range, 7–46) months. Intraclass correlation coefficient (ICC) yielded very good interobserver and intraobserver reliability (ICC was 0.950 for JOA and 0.993 for ISA for the interobserver reliability, and 0.980 for JOA and 0.997 for ISA for the intraobserver reliability). The ISA in the femur changed from 12.1 ± 10.1° at the immediate postoperative period to 32.3 ± 14.3° at the final followup and in the tibia it changed from 8.3 ± 8.0° at the immediate postoperative period to 25.4 ± 12.8° at the final followup. The dISA in the femur was 20.2 ± 9.8° and in the tibia it was 17.1 ± 11.4°. The dJOA in the femur was 14.9 ± 7.9° and in the tibia it was 9.2 ± 6.6°. Scatter Plots have been created for evaluating the correlation between the dJOA and the dISA, which revealed linear correlation. Using Pearson correlation coefficient, strong correlation was found between dJOA and dISA. (r = 0.759 (0.615–0.854, 95%CI), p < 0.001). Because strong correlation was found between the dJOA and dISA, stepwise multiple linear regression analyses were carried out in order to create formulas that will allow us to make some predictions. Age, gender and dISA were considered as independent factors to predict dJOA. As a result of the analysis, only dISA was found significant. Following are such formulas: For femurs (ΔJOA = 2.519 + 0.611 × ΔISA) and for tibias (ΔJOA = 1.385 + 0.457 × ΔISA). 95% Confidence Intervals were given along with these scatter plots that described strong correlation (Fig. 2 A and B).

Fig. 2. Scatter plots in femurs (A) and in tibias (B) for evaluating the change in interscrew angle (ISA) versus joint orientation angle (JOA) with 95% confidence interval marked.

Tension band hemiepiphysiodesis is a wellestablished technique that corrects deformities through the contralateral growing hemiphysis by tethering the ipsilateral hemiphysis after the application of the eightplate. The screws are inserted at an angle below and above the physis, which never stay at the same angle, as long as the physis continues to grow. For frontal plane deformities after the eightplate is implanted, the patient needs to be followedup with Xrays including individual bone in its fulllength, preferably every four months due to unpredictability associated with guided growth.^{8} At times, in order to avoid traveling, the patient has a knee Xray taken locally and sends it by mail. Before this study, it was not anticipatory to make a decision just by anteroposterior (AP) knee Xrays. We hypothesized that there would be a close correlation between the amount of divergence of the screws of eightplate and the amount of correction of the deformity, and the latter could be estimated by using certain formulae. Our results showed that the hypothesis was correct and there is correlation, between the delta joint orientation angle and the delta interscrew angle. This correlation is not related to the growth rate of the physis. It is a purely mathematical correlation, independent of the age and sex of the patient. However there is no one to one correlation. In other words 10 degrees of change in interscrew angle does not yield 10 degrees of change in the joint orientation angle. An explanation to this could be the slack of the screws within the plate construct when placed in situ.^{9 ; 10} The implant configuration described in this article that revealed a correlation between the dJOA and dISA applies to constructs, screws of which did not reach maximum divergent angle. In a situation where screws diverged at maximum, screws would theoretically impinge on the plate and the interscrew angle would not change anymore, while the change in the joint orientation angle could still occur. However this is rather unseen. In fact, Ballal et al speculated that due to the flexible nature of the implant, even after the screws reached at their maximal divergence there is facility within the flexible plate to bend.^{11} This can also be seen in Figure 5c of the paper by Boero et al.^{12} One case in our study who was treated for bilateral genu valgum didn't come to his followup visits. And when he showed up he developed varus deformity in the distal femur. Interestingly the ISA were found to be 43 and 78° on right and left femur, respectively. His immediate postoperative ISA values were found to be 8 and 38°, respectively. The differences between the interscrew angles were 35 and 40°, whereas the differences between the mLDFA angles were 34° in both sides. With careful analysis of the X–rays it was obvious that the divergence continued even after the screws reached their maximum at the expense of bending the plate. So the authors agree with Ballal et al^{11} that the titanium plate could bend being unable to resist the forces created by the physeal growth potential. This observation could falsify the assumption that the screws would stop diverging after they reach maximum divergence. We observed that the plate configuration facing the bone changed from being concave to convex after the bend due to its flexible nature.
An application of the formulas to estimate dJOA from dISA is depicted in Fig. 3 A and B. This is a case of Cozen phenomenon. The values are as follows: MPTA perioperative 103°, MPTA final followup 89°. ISA perioperative 6° and ISA final followup is 35°. By taking an AP knee Xray only, one would be able to estimate the difference in the joint orientation angle (MPTA). For that purpose the following formula for tibias should be chosen (ΔJOA = 1.385 + 0.457 × ΔISA).

Fig. 3. A 5yearold female patient who developed Cozen phenomenon after a proximal metaphyseal fracture. Her left medial proximal tibial angle (MPTA) was 103°, and the ISA was 6° when the eight plate was first applied (A  Perioperative). In 16 months the MPTA changed to 89° and the ISA changed to 35° (B – Final Followup).

dISA in this case is 35–6 = 29°.

14.6° is the amount that is estimated by this formula using the difference between the ISAs before and after correction at the time of the Xray.
(95% Confidence interval 0.615–0.854)
103–14.6 = 88.4° is the estimate for the current MPTA with the help of the formula and the difference of ISAs. The real MPTA was 89°.
Thus, this example shows that by knowing the change in the ISA, which was 29° in this case, it might be possible to predict the amount of change in joint orientation angle that is around 14 degrees of change in MPTA in this case.
The outliers have been shown in the scatterplot graphs. The outliers do not belong to a specific sex, age or disease. They are also not associated with an accompanying osteotomy. There are some underestimates and overestimates in using the formula. The formulas that the regression model yielded have been validated using bootstrap sampling in our dataset. The formulas need to be tested in other datasets for further validation. This study has some limitations. The number of participants included in the study were 30 involving their 53 segments. This does not jeopardize the results, since statistically this number is enough to investigate the correlation, run regression analyses and create formulas. However, further validating studies are needed to be performed in larger series of patient populations with eightplates. Twenty of the 53 segments were operated using a different brand of the eightplate implant. However both brands were composed of the same material. And there were no differences in the statistical analysis of the two groups (p = 0.994).
In our study, the angular divergence between the screws of the eightplate construct was found to be correlated with change in the joint orientation angle, mLDFA or MPTA. Regression analyses yielded formulas that enabled us make estimations for the change in interscrew angle if the change in joint orientation angle was known. The formulas depicted in this paper are applicable to constructs composed of a titanium twohole nonlocking plate with a figure of eight, and two cannulated 4.5 mm cortical screws. The formulas might or might not apply in other configurations of tensionband plates available in the market.
This study is important to confirm that there is close correlation between dISA and dJOA. This correlation exists in all patients with varying degrees of growth potential.
Published on 31/03/17
Licence: Other
Are you one of the authors of this document?