Meniscal allograft transplantation: a meta-analysis

Purpose: This meta-analysis evaluates the mid- to long-term survival outcome of MAT (meniscal allograft transplantation). Potential prognosticators, with particular focus on chondral status and age of the patient at the time of transplantation, were also analysed. Study design: Meta-analysis. Methods: An online database search was performed using following search string: “meniscal allograft transplantation” and “outcome”. A total of 65 articles were analysed for a total of 3157 performed MAT with a mean follow-up of 5.4 years. Subjective and clinical data was analysed. Results: The subjective and objective results of 2977 patients (3157 allografts) were analysed; 70% were male, 30% were female. Thirty-eight percent received an isolated MAT. All other patients underwent at least one concomitant procedure. Lysholm, Knee injury and Osteoarthritis Outcome (KOOS), International Knee Documentation Committee (IKDC) and Visual Analogue Scale (VAS) scores were analysed. All scores showed a good patient satisfaction at long-term follow-up. The mean overall survival rate was 80.9%. Complication rates were comparable to standard meniscal repair surgery. There was a degenerative evolution in osteoarthritis with at least one grade in 1760 radiographically analysed patients. Concomitant procedures seem to have no effect on the outcome. Age at transplantation is a negative prognosticator. The body mass index (BMI) of the patient shows a slightly negative correlation with the outcome of MAT. Conclusions: MAT is a viable solution for the younger patient with chronic pain in the meniscectomised knee joint. The complications are not severe and comparable to meniscal repair. The overall failure rate at final follow-up is acceptable and the allograft heals well in most cases, but MAT cannot be seen as a definitive solution for post-meniscectomy pain. The correct approach to the chronic painful total meniscectomised knee joint thus requires consideration of all pathologies including alignment, stability, meniscal abnormality and cartilage degeneration. It requires possibly combined but appropriate action in that order.


Introduction
The meniscus is known to have an important role in distributing load, enhancing stability, facilitating congruency and contributing to lubrication and nutrition in the knee. Clinically and scientifically there is evidence that a meniscal deficiency can attribute to the development of premature osteoarthritis. Total meniscectomy increases the peak contact stresses in the affected compartment, accounting for an estimated 4% cartilage loss per annum (greater on the lateral side) [1]. The amount of removed meniscal tissue attributes to this increase in contact stress. Meniscus allograft transplan-tation is most commonly indicated in the symptomatic patient who is meniscus deficient. In particular, it is most suitable for patients under the age of 50 years with debilitating pain localized to the tibiofemoral articulation but who still have not developed advanced degenerative changes in the knee. It may also be indicated in patients undergoing reconstruction of the anterior cruciate ligament (where there is concomitant meniscal deficiency and a resultant risk of excessive forces acting through the reconstructed ligament). Meniscal allograft transplantation (MAT) may also be considered as a prophylactic chondroprotective procedure in the young but still asymptomatic patient with complete meniscal (mostly lateral after discoid meniscal resection) deficiency. While there is little evidence to directly support this third indication at present, long-term outcome data for the procedure may aid further discussion of its potential role here. Current intermediate term data demonstrates that meniscus allograft transplantation yields good to excellent results in terms of pain, function and activity levels in up to 85% of patients at up to eight years following transplantation [2,3]. The goal of this meta-analysis was to provide clinical and radiological outcome data, in addition to the survival analysis of meniscus allograft transplantation. A secondary goal was to identify prognosticators that may influence this process, with particular focus on chondral status and age of the patient at the time of transplantation. As it is stated that a meniscal allograft transplantation can be a solution for post-meniscectomy pain, our hypothesis is that there should be a difference in subjective results before index surgery and at final follow-up. As for the clinical data it was hypothesized that there should be little to no difference in outcome between short-term and long-term results for MAT to be a viable solution.

Materials and methods
From 2015 to 2016 the databases PubMed, Web of Science and Google Scholar were searched using the term ''Meniscal Allograft Transplantation''. An initial search resulted in 120 articles in PubMed, 57 in Web of Science and 504 in Google Scholar. The term ''outcome'' was added as it was the scope of this study to evaluate the outcome after MAT. This search was further refined by only searching for studies published in English in the last five years. This eventually gave a total of 87 articles, which were meticulously reviewed. It was then decided to exclude all articles that reported results with a follow-up of less than two years. Articles only containing data about meniscal extrusion were also excluded as this was not the main scope of the study. This delivered a total of 19 articles published from 2011 until 2016. To get an overview of the outcome after meniscal allograft transplantation in a time span of 32 years, all articles analysed by El Attar et al. were also included [2]. In July 2016, it was decided to perform a new search with the same variables to update our data pool with the most recently published data. This led to the inclusion of a further eight articles (Table 1). This means the total number of studies used in this analysis is 65, of which 36 were prospective cohort studies or case series and 29 were retrospective studies. Eight authors published studies from the same patient pool, but at different follow-up times [3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19] (Figure 1). Statistical analysis was performed using IBM Ò SPPS Ò Statistics 21. A confidence interval of 0.05 was used for all analyses. Initial data collection was done with Microsoft Ò Excel 2016. Tables were also computed with Excel. Cases with the same values were grouped in all graphs. Data reported in all these articles accounts for 3157 allografts in 2977 patients. Subjective and clinical outcomes were analysed. In case of the subjective findings only the results of the most used questionnaires were analysed, being Lysholm-, KOOS-, IKDC-and VAS-scores. Pre-operative scores were compared with scores at final follow-up for all of these questionnaires. For all questionnaires the hypothesis was the same: there should be a significant difference between the pre-operative scores and the scores at final follow-up if a MAT is a viable solution for post-meniscectomy pain. A Wilcoxon signed ranked test was used for these comparisons. Clinical survival rate was analysed using the Kruskal-Wallis test with mean time of follow-up as a grouping variable. As hypothesis, it was stated that there should be little to no difference between results of short-and long-term follow-up. A one-way analysis of variance (ANOVA) test was performed to define if there was a difference in outcome after MAT using different surgical techniques and differently preserved menisci. Complications, arthroscopic and radiologic findings were grouped and overall findings were stated, however they were not statistically analysed due to heterogeneity of these findings. Potential prognosticators (age, sex, BMI, number of concomitant procedures) were also investigated to search for a possible correlation between these prognosticators and the outcome of MAT. To evaluate these data, articles were grouped together.

Patient demographics
All articles together provided us with the results of 2977 patients of which 1982 were male, 898 were female and 97 were

Associated procedures
Information on associated procedures was available for 2742 patients (92.1%). Only 39.1% of this population received an isolated MAT. In all other cases, the patient was subject to associated procedures. Most frequently a concomitant anterior cruciate ligament reconstruction (ACLR) or corrective osteotomy was performed. Other procedures were related to optimizing the chondral surface ( Figure 2; see Appendix B) [3-19, 21-39, 41-68].

Outcome Subjective findings
Using the Lysholm-, VAS (visual analogue scale)-, IKDC (international knee documentation committee)-and KOOS (Knee injury and Osteoarthritis Outcome)-scores, the patient's subjective evaluation after MAT was analysed. A total of 840 patients (28.2% of the total population) filled in the Lysholm questionnaire before index surgery and at final follow-up. These questionnaires were uniform in all studies that reported these results. All patients gain a significant profit after MAT with a mean profit of 25% (7%-49%). Using a Wilcoxon signed rank test to compare pre-operative Lysholm scores and Lysholm scores at final follow-up showed that there is a significant difference between these scores (P < 0.005). However, it can be noted as shown in Figure 3 that patients who filled in the Lysholm questionnaire report less profit of MAT [5,9,11,14,17,21,23,29,40,45,46,54,57,61,62,64,66].
Furthermore, the VAS-and KOOS-scores were also analysed, respectively, represented by 463 and 406 patients out of 2977. For both scores again a Wilcoxon signed rank test was used to analyse the data pre-operative and at final followup. The VAS-score shows that patients feel an improvement after MAT (P = 0.001). Less pain and swelling of the knee joint allow for more of their daily activities. However, pain seems to return over time, as patients who filled in the VAS-chart tend to report more pain ( Figure 5) [4, 5, 15, 17, 21, 22, 27, 33-35, 37, 38, 47]. In the analysis of the KOOSscore, 12 papers were included. In all of the five categories questioned in the KOOS-questionnaire, there is a significant difference in mean scores between the pre-operative evaluation and evaluation at final follow-up (P < 0.001). As shown in Figure 6, there is a mean increase of 23.1 points at final follow-up. As the mean follow-up is six years for all these patients, the conclusion is that at a mid-term outcome patients experience a significant increase in their functionality after MAT [4,5,40,48,53,54,57].
In conclusion, the majority of patients experience a significant improvement in their daily living activities as in their sporting activities. At all times, being it a short-, mid-term or long-term outcome, the overall patient satisfaction is good. Although all of these subjective scoring questionnaires show a significant improvement, there is a declining trend in scores over time [3-11, 13-17, 19, 21-23, 25-54].

Clinical findings
The surgical technique was analysed first. In total, 53 of 65 studies investigated survival of MAT. This included 2677 patients representing 2835 transplants. The mean survival rate for these allografts was 80.9% (15.1%-97.9%). To further investigate the impact of time since surgery three mean follow-up categories were studied: <3 years, 3-6 years, >6 years. After analysis using a one-way ANOVA test, there seems to be a significant difference between these follow-up groups (P = 0.021). A further paired analysis of these groups using a Kruskal-Wallis non-parametric test showed no difference in outcome can be found between the mean short-term and mean mid-term follow-up group (P = 0.435). There is also no difference in outcome between mid-term and long-term follow-up groups (P = 0.074). However, a difference in outcome can be found between the mean short-term and mean long-term follow-up group (P = 0.04). There is a difference of 14.8% between these two groups in mean survival rate ( Figure 7). These results need to be interpreted with care. Most of the data are collected at a mean short-to mid-term follow-up ( Figure 8). It also shows outliers in mean mid-term to mean long-term follow-up survival data, which could have affected the results. These outliers were authors who used multiple endpoints as criteria for failure such as reoperation, extrusion, meniscal tears, pain and alternate findings after imaging. It is also important to point out that studies with a follow-up time of more than seven years report greater failure rates [6,8,13,37,40,51,65,66]. Most authors define failure as a clinical failure, being it a meniscectomy, a graft repair or an arthroplasty with or without a prosthesis. Some of them use more strict failure criteria such as structural damage, alternate imaging, pain or reoperation not related to the graft. As a result these authors report higher failure rates. These findings make it difficult to draw a definite conclusion, thus a more homogeneous approach of reporting survival data is needed. A total of 167 out of 1665 patients converted to a prosthesis at a mean time of 10.5 years [3-8, 12, 13, 16-19, 23-27, 30, 33, 39, 53, 62, 65]. The complication rate of MAT is comparable to

Arthroscopic findings
At index surgery for MAT a total of 999 knees were evaluated for chondral damage. Most of the authors report a definite change in the chondral surface, even though they used different selection criteria which excluded patients with high chondral wear. The pre-operative evaluation of the cartilage using MRI-imaging seems to be just an indication of possible osteoarthritis, but not a definite marker of the actual state of the cartilage. A total of 459 knees had moderate to severe osteoarthritis, so they did not meet the indication. Only a limited number of authors found no significant difference in survival rate, thus the osteoarthritis grade should still be considered pre-operatively [7, 9-19, 21-23, 27-30, 34-37, 39, 42-49, 53-59, 67, 69].
A post-operative arthroscopy was performed in 932 cases. Mostly because of a complication, only 479 patients had a post-operative arthroscopy for the sole purpose of evaluating the allograft and chondral surface. An important finding is that most of the patients had only a slight progression of osteoarthritis with a mean of one grade [20]. This restrains the chondroprotective effect suggested by many authors in long-term follow-up [7, 9, 11-19, 21-26, 29-32, 34-36, 38, 41-48, 51-58].
When evaluating the allograft most of the authors reported a good integration and healing of the allograft, which can also be seen histologically. Seven studies report the presence of shrinking of the meniscus allograft at evaluation. However, this is a difficult variable to analyse [3, 7, 11, 15, 26-31, 33, 35-37, 48, 54, 57].

Radiologic findings
A total of 1760 patients were available for analysis. Most of the patients were evaluated with a standing anteroposterior RX or with a Rosenberg view combined with MRI. To make a thorough analysis possible the best way to combine all the different classifications was used. Therefore it was chosen to merge all these classifications using the classification system introduced by Claes et al. [20]. This way pre-and post-operative osteoarthritis grades could be evaluated in a more coherent way. In Table 4, all available patients are classified into specific osteoarthritis groups. This confirms the findings found by the authors during arthroscopy. In all classifications, an overall decrease of one grade was found. These findings indicate that there is no long-term chondroprotective effect of MAT, but that the progression of osteoarthritis is delayed by a statistical mean of 10.5 years [6-9, 12, 17, 21, 22, 24, 35-38, 40, 43, 46, 51, 55-57, 60, 63, 67].
Some authors retain a narrowing of the joint in weightbearing evaluations (327 patients). Overall the authors report no significant joint space narrowing, but in most cases there is some narrowing. But when compared to the contralateral knee, the difference is not significant. These results indicate that MAT provides a good function and integration in the index knee [3,7,29,31,33,36,37,40,43,52,53].  It can be observed that data at long term followup has a bigger interval. Thus it seems there's a heterogeneity among these studies. Indeed it could be observed that in this group articles who used subjective problems such as pain tend to report a lower mean survival rate. Furthermore studies with a follow-up of more than 20 years are depicted in the two outliers. This indicates that more research has to be done to evaluate the lifetime of a meniscal allograft using more uniform failure criteria.
Furthermore, a total of 627 allografts were graded with MRI. Three grades were withheld: Grade 1: minimal hyper intensity in a specific meniscal area. Grade 2: a linear abnormal intensity in the complete allograft, divided into two subgroups (a) linear abnormal intensity; (b) linear abnormal intensity with a single image expansion and (c) globularwedge-shaped hyper intensity. Grade 3: completely abnormal hyper intensity expanded to at least one joint surface area. About 75% of these allografts had a grade one or two. However, this does not implicate that these allograft were failures. Most of them were abnormal on MRI, while they were clinically asymptomatic. These allografts also had some structural changes. Most of the time they healed well but with some minor effusion. In some cases, there is also some minor displacement of the allograft, depicted clinically by complaints of limited knee joint mobility. In most of these cases, the procedure has failed as the displacement is a result of loose stitches or bone plugs. Shrinking also occurs. Usually 25% or sometimes up to 50% although this is difficult to measure [3,7,29,31,33,36,37,40,43,52,53].
In total, five studies also evaluated the effect of the body mass index (BMI) on outcome after MAT. Only Saltzman et al. and McCormick et al. reported significant changes. They reported a minor negative effect on outcome after MAT when the patient's BMI was higher than 25. The other three authors did not find any significant difference [4,5,44,47,48,57].
A possible effect of concomitant procedures on the outcome after MAT was also considered. It appears there is no major difference between isolated MAT and MAT concomitant with other procedures. Only a small group of authors mentioned the effect on outcome. As all these data were already synthesized in all articles, a statistical analysis could not be done. However, most authors report they did not see a correlation between the number of procedures concomitantly performed with MAT and the outcome after MAT. Possibly because of the fact that these procedures are done because they stabilize the knee joint and optimize the chondral surface, which means the allograft can integrate properly. Furthermore, the time to transplant also does not have an influence on the outcome after MAT following a Mann-Whitney U test comparing the outcome between a short and long time to transplant (P = 0.445) [3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19]. It is important to note that all these prognosticators were analysed based on means, thus it is certainly not conclusive and needs more research.

Discussion
This meta-analysis included 2977 patients, 1982 males and 898 females and is thus the largest meta-analysis available in the current literature. The majority of patients had undergone lateral meniscal transplantation (1594 pts). The follow-up time ranges widely from two months to 25 years.
A number of variables have been used in defining proper patient selection criteria. The most commonly used indication was the younger age with pain after meniscectomy not responding to conservative treatment, with normal axial alignment, a stable knee joint and less than Grade 4 cartilage degeneration in the affected compartment. Routinely inflammatory joint diseases, infection and morbid obesity have been excluded. Interestingly, the majority of patients underwent a combined surgery including meniscus allograft transplantation and a wide range of associated procedures (including anterior cruciate ligament reconstruction, osteotomy and cartilage repair [5,7,22,24,28,30,63,66]). Only 38.8% underwent isolated meniscus allograft transplantation.
Most meniscal allograft transplantations have been performed using arthroscopic-assisted techniques. Very early on, open surgery was performed when dealing with open multiligament repair after trauma [66]. The preferred preservation technique was fresh frozen or cryopreserved in most cases. Lyophilization has been abandoned because of tissue deformation [66].
Regarding meniscal allograft fixation techniques, most studies (37.1%) have been using bone block fixation to obtain better hoop stress retaining biomechanics. In 9.1% the fixation technique was not explicitly described. No correlation could be found between surgical fixation technique and outcome.
A variety of subjective and objective clinical outcome scores have been used: Lysholm in 28.8%, IKDC in 14.2%, VAS in 15.5% and KOOS in 13.6%. All studies describe a significant clinical improvement after transplantation with a tendency to decline with follow-up time [3, 4, 8-19, 21, 22, 24, 26-28, 30-41, 43-55, 57-64]. Survival of MAT has been documented in 54 out of 65 studies. Most frequently, survival was defined as explantation of the graft or conversion to arthroplasty, although some authors used more strict criteria thus resulting in significantly lower survival rates. The mean survival rate in this meta-analysis was 80.9%. Ten per cent of the patients needed conversion to a prosthesis. Overall, the MAT procedure and the healing rate are comparable to standard arthroscopic meniscal repair surgery. When evaluating the allograft with arthroscopy most authors refer to good integration of the allograft in situ. Some studies refer to shrinking of the allograft [9,14,29,32,36,50,56].
No correlation could be identified between the mean patient age of the study and the outcome or survival rate. The patient's gender did not correlate with outcome, although the number of female patients was relatively smaller. There was a slightly minor negative effect on the outcome in BMI > 25 [35,46]. Concomitant surgery did not influence results negatively when compared with isolated meniscal allograft transplantation. This is possibly due to the fact that these additional surgeries are a clinical requirement and improve weight-bearing chondral properties and function in the index knee.
Time to index surgery from meniscectomy is only illustrated in one study. Jiang et al. [38] compared MAT after meniscectomy and meniscectomy alone and found better subjective results in the former patient group. All studies of this meta-analysis however did not show differences in results when comparing >5 years versus <5 years time delay between meniscectomy and index MAT surgery.
A weakness of the study is the fact that all conclusions were based on reported means. It is possible some of these results are over-or underestimated. Thus further research is definitely needed. Furthermore, studying cost-effectiveness of MAT could be useful since most patients are still part of the economically active population and MAT seems more of a temporary solution for post-meniscectomy pain. Furthermore, it has to be stated that due of no international guidelines some data were too heterogeneous to perform a thorough analysis of it such as the complications, arthroscopic and radiological findings. Therefore these data were compiled and an overall conclusion was stated. Study design did not differ significantly from each other among the analysed studies. However, failure criteria should be more uniform to get a better view of the significant difference in survival rate that this study found.
As it is clear that if a certain subjective threshold is used, the results are less favourable. It also needs to be noted that further long-term investigation is needed as there is quite a big incline in failure rate as the follow-up time is longer. There should also be an effort to improve the methodological quality of studying this procedure as most authors are in fact the performing surgeons. Furthermore, data was commonly selected out of a patient database which can also be seen as a flaw in this study, as it possibly consists of data with a possible selection bias.
In conclusion, it appears that MAT is a viable solution for the younger patient with chronic pain in the meniscectomised knee joint. The complications are not severe and comparable to meniscal repair. The overall failure rate at final follow-up is acceptable and the allograft heals well in most cases. However, it tends to have a greater failure rate at longer follow-up. Thus it should be seen as a more temporary solution of post-meniscectomy pain. Progression of osteoarthritis is acceptable at midterm evaluation. Increased signal on MRI is a common finding without clinical complaints. The correct approach to the chronic painful total meniscectomised knee joint thus requires consideration of all pathologies including alignment, stability, meniscal abnormality and cartilage degeneration. It requires possibly combined but appropriate action in that order. However, published data are of low methodologic quality, thus an effort should be made to improve it and give an opportunity to make more empowered conclusions about the subject.