The most important finding of the present study was that PETS and TBP had different physeal behaviours according to the implant character and anatomical site after guided growth to correct idiopathic genu valgum. Physeal stapling was the first introduced technique for temporary epiphysiodesis [9]; however, the popularity of this technique is decreasing because of several problems, including physeal damage during the procedure, hardware breakage, migration, and extrusion [13, 21, 22]. Instead, transphyseal screw and tension-band plate have been introduced by Metaizeau et al. and Stevens, respectively [10, 11]. Although several studies have compared PETS versus stapling or TBP versus stapling [2, 3, 13, 15, 23], only two studies have compared PETS and TBP in patients with genu valgum [4, 5]. However, none of these described the physeal behaviour such as the stable correction, rebound phenomenon, or overcorrection after implant removal. In our study, 53.4% remained stable after removing the guided growth implant at an average age of 11.9 years; however, 30.1% showed rebound and 16.4% overcorrection.
We could achieve acceptable correction without any serious complications, such as permanent physeal damage or reoperation. PETS led to a faster deformity correction, in concurrence with a previous study [5], and this result was the same in a study for ankle valgus correction [24]. Stapling generally tended to be performed at an earlier age than PETS [2, 15], which is thought to be due to concerns about damage to the growth plate. For the same reason, TBP may be preferred to PETS at an earlier age with high growth remaining, and the younger age can lead to a slow rate of correction compared to the rapid growth rate at adolescents. However, there was no significant difference in age at the time of surgery between the two groups in this study and the number of cases in each group was similar (PETS 36 vs. TBP 37). Additionally, there was no correlation between the age at surgery and the correction rate. Therefore, the time for non-rigid TBP to act as a focal hinge at the perimeter of the physis is thought to be responsible for the difference in the pace of correction between the two implants [4, 11]. After implant removal, approximately half the cases demonstrated stable correction, but the rebound was more common than overcorrection after guided growth. In this predominant rebound group, TBP was more common, and the correction period was longer. The growing potential differs between the distal femur and proximal tibia, so this difference may relate to the correction period. The natural behaviours after implant removal should be separately evaluated according to the femur and tibia considering a different implant character.
Most studies about physeal behaviour after temporary hemiepiphysiodesis have been limited to the rebound phenomenon after TBP or stapling, and the incidence of the rebound has been widely reported between 0 and 60% [15, 17, 18, 25,26,27]. Such differences are probably due to the heterogeneity of the included patients, especially the etiology of deformity, the potency of growth after implant removal, and the definition of the rebound phenomenon. Our cohort was limited to patients with idiopathic genu valgum and > 1 year of remaining growth after implant removal. The incidences of rebound after PETS and TBP were 11.1 and 48.6% in our series, respectively. Our results showed that the risk factor of the rebound phenomenon was the type of implant. Although there was a difference in the correction period between the two groups, it seems to have originated from the difference in the proportion of implants in each group. Generally, TBP needs a longer correction period, and PETS accounted for 62.7% of the non-rebound group, whereas TBP accounted for 81.8% of the rebound group. TBP had a 4.4 times higher incidence of rebound phenomenon than PETS. Three recent studies focused on the rebound phenomenon reported 55, 42, and 56% of recurrent valgus after TBP, respectively [17, 26, 27], and these values are comparable with our results. A previous study reported a rebound rate of 29% after PETS, which was less than in stapling [2].
Transphyseal fixation of partially threaded screws suppresses the growth plate through direct mechanical compression, and the direct physeal involvement of PETS can make surgeons reluctant to use screws for patients with young age. Brauwer et al. reported a 31% rate of progression of correction after screw removal [2]. Although there were differences in the age at surgery and the definition of progression, we also found an overcorrection rate of 30.6% (11/36 limbs) after PETS. However, the overcorrection after implant removal does not mean permanent physeal arrest. The longest period of screw maintenance was 15 months in our series, but we did not observe any cases with abnormal radiographic findings indicating physeal arrest. Shin et al. also reported no case of physeal arrest after PETS [15]. The compression of the growth plate using PETS may cause a slower return of physeal function compared with TBP, even in the absence of permanent physeal arrest.
The physeal behaviour after implant removal was more pronounced in the distal femur than in the proximal tibia. In both PETS and TBP groups, the mLDFA changed significantly in different directions; overcorrection and rebound, respectively. The changes of mMPTA were in the same direction as the femur depending on the type of implant; however, there was no statistical significance. These findings may be due to different potentials of longitudinal growth or the effect of the fibula as resistant support. The remaining growth after implant removal will contribute significantly to the final coronal alignment, and this suggests that implants should be removed as close to the skeletal maturity as possible. Future studies with guided growth at the other location and younger patients are required.
There were several limitations to this study. First, we selected patients who had remaining growth for more than 12 months after implant removal, which might affect the incidence of rebound. Clinically, the proper guided growth is composed of an accurate prediction of correction and implant removal after skeletal maturation, in which physeal behaviour after implant removal is not expected. The incidence of rebound or overcorrection may be clinically different if we include patients who had undergone guided growth near skeletal maturation. Second, we included only idiopathic genu valgum as a single aetiology, and the small sample size may have weakened the statistical power to detect minor differences. However, single aetiology is more important because of the different effects of the fibula. Another limitation is the lack of randomization related to its retrospective design. The implants were not randomly selected, and the factors considered when choosing an implant could influence the outcomes of guided growth. Further studies with a prospective randomized trial and large cohort including genu varum or various causes of genu valgum are needed to better understand the natural behaviours after implant removal. Lastly, we could not analyse the effect of screw insertion angle in two techniques. The effect of hemiepiphysiodesis may be associated with the convergence angle of two screws in TBP and the angulation of the screw to the physis in PETS [2, 15]. Although we did not measure the inserted angle of each screw, at least three threads of all screws in PETS penetrated the medial quarter of the growth plate, and the two screws in TBP were inserted in parallel [28, 29]. Because all surgeries were performed in the same technique by a single surgeon, we believe that the positions of the screws were similar and that the implant type has a higher effect on the correction process and ensuing physeal behaviour than the minor differences in the position of screws.