Regarding the two hypotheses, relevant results showed that minimally invasive approaches would not yield better patellofemoral functions and no significant differences were found among the three groups in terms of clinical outcomes, radiographic evaluations, suggesting that no approach is superior to the other two regarding these results.
Compared with the MP approach, the MMP and QS approaches for TKA have two main differences in the present study. First, the arthrotomy extended into the quadriceps tendon less than 4 cm above the upper pole of the patella in the MMP and QS groups but extended 6-8 cm in the MP group; second, the patella was everted from the beginning of the femoral resection to the end when patella resurfacing was performed in the MP group, whereas in the MMP and QS groups, the patella was not everted during the surgery. Many comparative studies of the MMP, QS and MP approaches have reported good clinical results but inferior lower limb mechanical axis and malposition at the short-term follow-up for MMP [32, 33] and QS [8, 34-36] because of the limited working space compared with the MP approach. In addition, some studies indicated that the MP approach may affect patellar tracking, AKP and the patellar position [8, 34-36]. Therefore, we performed a comprehensive evaluation of patellofemoral function in addition to the overall knee score in the clinical investigations.
There are a number of studies that have compared the clinical and radiographic differences among the MMP, QS and MP approaches [34, 37, 41-44], but few of them had a follow-up of at least 10 years. In a retrospective study with a mean follow-up of 5 years, Huang et al. reported that the MMP and QS approaches led to more appropriate LPT and LPD than the conventional MP approach did. Chiang et al. conducted a prospective and randomized study with a mean follow-up of 2 years and reported that patients undergoing QS and MP TKA had comparable clinical outcomes, and that the QS TKA was more time consuming surgically and resulted in a less accurate prosthesis position. In a retrospective study with a mean follow-up of 2 years, Lin et al. showed that the QS approach provided inferior radiographic results with similar clinical outcomes when compared with the MP approach. In 2018, Kazarian et al.  conducted a meta-analysis of randomized controlled trials (RCTs) of the QS approach and MP approach, in which all the studies were no more than 2 years long ,and reported that the QS approach to TKA failed to demonstrate clinically significant advantages but showed increased lower limb mechanical axis and femoral and tibial prothesis frontal position. Our study is a more than 10-year study aimed at the long-term results, showing no statistically significant differences among the three approaches for TKA in terms of clinical outcomes, radiographic outcomes and prothesis survival rate.
Some studies have shown that the MP approach affects patellar tracking, AKP and the position of the patella [37-40] because the MMP and QS approaches preserve the majority of the extensor mechanism. Some studies have indicated that the position of the femoral and patellar protheses, especially when poor rotational alignment occurs, can lead to AKP . AKP is a common reason for reoperation or revision . Therefore, we conducted the current study and included the AKP questionnaire, Feller PFS, and patella physical examinations. Nevertheless, we did not find a statistically significant difference among them, which was consistent with the radiographic assessments. Furthermore, weakness of the quadriceps muscle is associated with patellofemoral function . When patients are discharged from the hospital and reexamined in outpatient clinics, we always advise them to exercise the quadriceps muscle, usually by squatting. There is a rough estimate that the amount of quadriceps atrophy is indicative of the amount of weakness present , so we measured the thigh circumference of every patient, and no statistically significant difference was noticed, which was consistent with the results of patellofemoral function.
Longevity is one of the concerns in TKA, and the survivorship of implants is dependent on the patient demographics, the surgical technique and implant-related factors [50, 51]. Incorrect positioning of the implant and improper mechanical axis of the lower limb can lead to accelerated implant wear and loosening, as well as suboptimal function. Aseptic loosening, instability, and malalignment were the most common reasons for late revision [52-55]. In this study, no differences were found in the value of or in the outliers of the radiographic assessment angles. We believe that the MMP and QS approaches not only do not increase the risk of malalignment and malposition of the components but also do not increase the reoperation rate of TKA. Our results are consistent with those from other studies [34, 56], indicating that radiographic outcomes were not compromised with the MMP and QS approaches. King et al.  demonstrated that a substantial learning curve (fifteen procedures) may be required for surgeons before reaching steady results using the QS approach, so we recommend that surgeons gradually decrease quadriceps exposure in order for patients to gain advantages of the TKA procedure.
In the present study, we conducted long-term follow-ups for a minimum of 10 years in which detailed evaluations of the postoperative clinical outcomes and radiographic assessments were measured. The new KSS system  has been recognized as a more efficient method to primarily differentiate the activities contributing to the function score compared with the old KSS system. In addition, some studies have demonstrated that the knee function score declines gradually  and that the alignment and prothesis position worsen during the long-term postoperative period , so we performed a multiple regression analysis, and the results were the same.
This study has several limitations. First, this study was limited by its retrospective design. Second, although we conducted a multiple linear regression, the significant difference in the follow-up time among groups could still introduce some degree of bias. Third, the small cohort sizes led to insufficient observations to determine the level of statistical significance. Fourth, only approximately 65% of the patients were able to directly be evaluated at the long-term follow-up. However, this follow-up rate is typical of the follow-up rate for TKA performed in elderly people [58, 59] due to a high number of patients faced with nonrelated TKA, including back pain, hip and ankle disease.
This study also has several strengths. This study is one of the few studies to report long-term postoperative clinical and radiographic outcomes for these three approaches to TKA. All surgeries were performed at a single center by the same surgical team using only one type of TKA, which allowed for better consistency among these three groups. Our study used the new KSS to assess the clinical outcomes, which provides sufficient detail to evaluate the functional capabilities of the knee .