Failures, Reoperations, and Improvement in Knee Symptoms Following Matrix-Assisted Autologous Chondrocyte Transplantation: A Meta-Analysis of Prospective Comparative Trials

Abstract

Objective

Though multiple high-level comparative studies have been performed for matrix-assisted autologous chondrocyte transplantation (MACT), quantitative reviews synthesizing best-available clinical evidence on the topic are lacking.

Design

A meta-analysis was performed of prospective randomized or nonrandomized comparative studies utilizing MACT. A total of 13 studies reporting 13 prospective trials (9 randomized, 5 nonrandomized) were included (658 total study participants at weighted mean 3.1 years follow-up, range 1-7.5 years).

Results

Reporting and methodological quality was moderate according to mean Coleman (59.4 SD 7.6), Delphi (3.0 SD 2.1), and MINORS (Methodological Index For Non-Randomized Studies) scores (20.2 SD 1.6). There was no evidence of small study or reporting bias. Effect sizes were not correlated with reporting quality, financial conflict of interest, sample size, year of publication, or length of follow-up (P > 0.05). Compared to microfracture, MACT had greater improvement in International Knee Documentation Committee (IKDC)-subjective and Knee Injury and Osteoarthritis Outcome Pain Subscale Score (KOOS)-pain scores in randomized studies (P < 0.05). Accelerated weight-bearing protocols (6 or 8 weeks) resulted in greater improvements in IKDC-subjective and KOOS-pain scores than standard protocols (8 or 11 weeks) for MACT in randomized studies (P < 0.05) with insufficient nonrandomized studies for pooled analysis.

Conclusions

Compared to microfracture, MACT has no increased risk of clinical failure and superior improvement in patient-reported outcome scores. Compared to MACT with standardized postoperative weight-bearing protocols, accelerated weight-bearing protocols have no increased risk of clinical failure and show superior improvement in patient-reported outcome scores. There is limited evidence regarding MACT compared to first-generation autologous chondrocyte implantation, mosaicplasty, and mesenchymal stem cell therapy without compelling differences in outcomes.

Introduction

High-grade chondral defects of the knee are frequently debilitating ¹ and, if left untreated, can lead to early osteoarthritis. ² Brittberg et al. first described using cultured autologous chondrocytes for the treatment of high-grade articular cartilage defects. ³ This technique eventually became known as first-generation autologous chondrocyte implantation (ACI) and has been shown to be an effective means of treating focal cartilage defects. ⁴ Follow-up studies of first-generation ACI showed significant clinical improvements in patients. ⁵ However, the original procedure was technically difficult because it required the harvested autologous chondrocytes be implanted in solution form under a periosteal patch secured with sutures upon the chondral defect. ³ Furthermore, rehabilitation times for these procedures were extensive, often more than a year before return to sports activity. ⁶

Matrix-assisted autologous chondrocyte transplantation (MACT) techniques were developed to address some of the shortcomings of first-generation ACI. MACT involves seeding the cultured chondrocytes onto a scaffold or membrane composed of a variety of materials. ⁶ In 2006, membrane-associated cartilage implantation (MACI) was the first MACT therapy to receive Food and Drug Administration approval for treatment of osteochondral and chondral lesions of the knee. ⁷ Though multiple high-level evidence studies have been performed for MACT,^3,8-10 currently there is no meta-analysis synthesizing best-available clinical evidence to assist providers with treatment decisions. Recent reviews on the topic have been limited to systematic reviews without application of meta-analytic techniques.^11-13

The purpose of this study was to systematically review and synthesize high-level clinical evidence on MACT trials. The study seeks to answer several questions: (1) Are clinical outcomes equivalent for standard versus accelerated weight bearing after MACT? (2) Is MACT superior to microfracture? (3) Is MACT superior to mesenchymal stem-cell therapy? (4) Is MACT superior to first-generation ACI? (5) Is MACT superior to mosaicplasty? (6) Does chondrocyte dose affect MACT outcomes?

Methods

Search Strategy and Study Selection

We conducted a systematic review based on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) ¹⁴ guidelines. An initial PubMed search using the terms (autologous chondrocyte) AND (transplantation OR implantation) in June 2018 yielded 1,231 results for original research studies reported in English. The Embase, SCOPUS, and Google Scholar databases were cross-referenced. After screening of study titles and abstracts, 1,108 articles were excluded due to failure to meet one or more study inclusion criteria ( Table 1 ). The reference lists of all articles were cross-referenced, and no additional studies were identified for review. The remaining 123 studies were reviewed, and 106 were excluded for failing to meet inclusion criteria. One study ¹⁵ was excluded due to use of an improper control group: MACT in adults was compared to third-generation ACI in adolescents. The remaining 16 studies reported outcomes from 13 clinical trials (9 randomized and 4 nonrandomized). Only one publication per trial was utilized; Wondrasch et al. ¹⁶ (mean 2-year follow-up) was excluded, and Wondrasch et al. ¹⁷ (mean 5-year follow-up for the same trial) was included in the final analysis. Becher et al. ¹⁸ and Niemeyer et al. ¹⁹ reported outcomes from the same trial; only Niemeyer et al. ¹⁹ was included in the analysis as Becher et al. ¹⁸ failed to report patient-reported outcomes. Two studies reported outcomes of the SUMMIT (Superiority of MACI Implant Versus Microfracture Treatment) trial,^20,21 comparing MACI versus microfracture. To avoid inclusion of redundant data, we included the report of 5-year outcomes (Brittberg et al. ²¹ ) and excluded the report of 2-year outcomes (Saris et al. ²⁰ ).

Table 1.

Systematic Review Inclusion Criteria.

Prospective cohort study or randomized trial consisting of an experiment that received matrix-assisted autologous chondrocyte implantation and a control group

Outcome measures include at least one standardized measure for patient-reported subjective knee symptoms

Less than 20% loss to follow-up

Minimum 1-year follow-up

Results are reported in a peer-reviewed academic journal

Results are reported in English

Nonrandomized studies with controls with clinically important differences in demographics or cartilage lesion characteristics compared to experimental group are to be excluded

Therefore, a total of 13 studies reporting 13 trials were included in the final analysis ( Fig. 1 ). This comprised data on 658 study participants at a weighted mean follow-up of 3.1 years (range = 1-7.5 years). Of note, 2 separate randomized trials are reported by the same first author; Ebert et al. ²² report the results of a trial of 11-week versus 8-week delay until full weight bearing following MACI, and Ebert et al. ¹⁰ report the results of a trial of 8-week versus 6-week delay.

Figure 1.

PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines. Fourteen studies were identified for inclusion.

Outcome scores of interest utilized in the analysis were the Knee Injury and Osteoarthritis Outcome Pain Subscale Score (KOOS-pain), ²³ International Knee Documentation Committee (IKDC) scores, and rates of clinical treatment failure or reoperation. Both IKDC and KOOS-pain scores are acceptable measures of knee disability in cartilage repair patients. ²⁴ Data were directly abstracted from the articles with the exception of KOOS-pain results reported by Akgun et al. ²⁵ and Ebert et al. ¹⁰ ; in these studies, reported KOOS-pain values had an abnormally small level of dispersion for all groups at all time points; this anomaly likely represented a common reporting error in which standard error of the mean was inadvertently reported rather than standard deviation. To address this issue, the mean reported standard deviation (SD) for KOOS-pain from the remaining trials (SD = 13) was assigned to avoid excessive weighting in the meta-analysis.

Risk of Bias and Quality Assessment

Several quality assessments were used to determine risk of bias and study reporting quality. The modified Coleman Methodology score ²⁶ (Appendix B), the Methodological Index for Non-Randomized Studies (MINORS) score ²⁷ (Appendix C), and the Delphi criteria ²⁸ (Appendix D) were all assessed by 2 independent reviewers (AD and SP). The modified Coleman Methodology score assessed the likelihood that biases and confounding factors were absent in the study (85-100 excellent; 70-84 good; 55-69 fair; and ≤54 poor). The MINORS scoring system was used to assess the overall quality of nonrandomized trials (scored 0 to 24). The Delphi criteria assessed the overall quality of randomized clinical trials using a 9-item checklist with scores of 1, −1, and 0 corresponding to “yes,” “no,” and “don’t know,” respectively. Agreement statistics were determined for all quality assessment tools. Discrepant scores were resolved by a third author. Reporting of a financial conflict of interest (COI), specifically funding or royalties received by one or more study authors, a manufacturer of MACT, or the control therapy, was also assessed.

Statistical Analysis

All statistical tests were performed with STATA 13.1 (StataCorp, College Station, TX). Interrater agreement for methodological quality scores was reported as the interclass correlation coefficient (ICC). Random effects meta-analyses using the Dersimian and Laird method ²⁹ were performed for each outcome (risk of clinical failure, risk of repeat operation, improvement in IKDC scores, and improvement in KOOS-pain scores) and each treatment-control combination. Due to fundamental differences in study design between randomized and nonrandomized studies, separate meta-analyses were performed for each study design. Effect heterogeneity was assessed using the I² measure as described by Higgins et al. ³⁰ For studies reporting either IKDC or KOOS-pain but not both outcome scores, a formula was derived (Appendix A) from the remaining studies reporting both IKDC and KOOS-pain to allow estimation of one score from the other. A positive transformation was applied (Appendix A) to allow inclusion of studies with zero failures or reoperations in both treatment and control groups in the statistical analysis. Treatment-control combinations with fewer than 2 dedicated randomized trials or 2 dedicated nonrandomized trials were not eligible for meta-analysis and best-available evidence for these studies are discussed qualitatively. Multiple assessments of bias were performed (Appendix A) including publication bias (via Egger’s test and visual inspection of funnel plots), bias due to reporting or methodological quality, and bias due to financial conflict of interest, sample size, year of publication, or length of follow-up

Funding Source

The authors received no funding for this study.

Results

Reporting Quality and Risk of Bias

Methodological quality was moderate with a mean Coleman score of 59.4 (SD 7.6, range 50-70), Delphi score of 3.0 (SD 2.1, range −2 to 6), and MINORS of 20.2 (SD 1.6, range 18-23). Interrater agreement for the Coleman score was good (ICC 0.79), and the interrater agreement was fair for the MINORS (ICC = 0.55) and Delphi scores (ICC = 0.48). The most common methodological shortcomings as assessed by these scoring systems were lack of blinding of either the patient (11/13 studies) or care provider (13/13 studies) and lack of prospective sample size estimation (7/13 studies).

There were no significant correlations between study effect size (as defined by difference between treatment and control groups in KOOS, IKDC, reoperation rates, or failures rates) and methodological quality scores, financial conflict of interest, sample size, year of publication, or length of follow-up (P > 0.05). There was no evidence of small study bias in any of the meta-analyses in the current study per Egger’s test of bias (P > 0.05, all analyses) or visual inspection of funnel plots with IKDC or KOOS-pain as outcome measures.

MACT versus Microfracture

Four trials compared clinical outcomes of MACT versus microfracture ( Table 2 ). Among randomized trials, there was no difference in risk of clinical failure for MACT versus microfracture (pooled relative risk [RR] 1.58, 95% CI 0.69, 3.61; I² = 0%; P = 0.28) ( Fig. 2 ). Meta-analyses of failure risk among nonrandomized studies or reoperation rates among randomized or nonrandomized studies could not be performed due to insufficient reporting. Among randomized trials, a greater improvement for MACT compared to microfracture was seen for KOOS-pain (pooled standardized mean distance [SMD] 1.83, 95% CI 1.46, 2.21; I² = 0%; P < 0.001) ( Fig. 3 ) and IKDC (pooled SMD 0.23, 95% CI 0.01, 0.45; P = 0.04; I² = 77%) ( Fig. 4 ). Meta-analyses among nonrandomized trials for improvement in KOOS-pain or IKDC could not be performed due to insufficient reporting.

Table 2.

Studies of MACT versus Microfracture and MACT with Differing Weight-Bearing Protocols.

Level of Evidence	Author	Implant, Mean Defect Size, and Sample Size	Mean Follow-up	Outcome	Financial Conflict of Interest	Mean Coleman Score	Mean MINORS Score	Mean Delphi Score
MACT versus microfracture
1	Crawford (2012) 44	21 NeoCart (mean 2.75 cm2); 9 microfracture (mean 2.52 cm2)	2 years	Randomized trial of distal femoral lesions. Greater IKDC and KOOS improvement at 2 years with NeoCart.	Yes	58	19.5	3.5
2	Kon (2011) 45	21 Hyalograft C (mean 2.1 cm2); 20 microfracture (mean 1.9 cm2)	7.5 years	Return to sport was a median 8 months for microfracture, 12.5 months for Hyalograft C. Symptom improvement with microfracture deteriorated with time whereas Hyalograft C was durable.	No	51.5	18.5	−2
1	Basad (2010) 35	40 MACI; 20 microfracture (range 4-10 cm2 for both groups, no mean size reported)	2 years	At 24 months, greater improvements seen with MACI in Tegner activity score, subjective symptoms scores and ICRS scores on second-look arthroscopy.	No	50	18	2
1	Brittberg (2018) 21	65 MACI (mean 5.1 cm2), 53 microfracture (mean 4.9 cm2)	5 years	MACI patients demonstrated statistically greater improvement in KOOS pain and function scores, and activities of daily living compared to microfracture. Quality of life and other symptoms were also improved in MACI patients relative to microfracture, but not statistically significant.	Yes	70	22	5
MACT with differing weight-bearing protocols
1	Ebert (2017) 10	19 MACI-standard WB (mean 2.89 cm2); 18 MACI-accelerated WB (mean 3.15 cm2)	2 years	Randomized trial of standard 8-week return to weight bearing versus accelerated 6-week return to weight bearing postoperatively. Two graft failures in standard WB group, none in accelerated group. No difference in symptom improvement or patients satisfaction scores.	Yes	65.5	22	5
1	Ebert (2012) 22	32 MACI-standard WB (mean 3.32 cm2), 31 MACI-accelerated WB (mean 3.21 cm2)	5 years	Trial of mean 11-week versus 8-week return to weight bearing. Less pain frequency at 5 years in the accelerated WB group based on (visual analog scale–frequency). No other significant differences between groups in subjective or functional measures.	Yes	70	21	1
1	Wondrasch (2015) 17	16 CaRes/Hyalograft C accelerated WB (mean 5.1 cm2); 15 CaReS/Hyalograft C standard WB (mean 4.6 cm2)	5 years	Randomized trial of 6-week versus 10-week return to weight bearing. No difference in symptom improvement between groups. MOCART score decreased from years 2 to 5 which did not correlate with symptom scores.	No	63.5	22.5	2.5

CAIS = Cartilage Autograft Implantation System; IKDC = International Knee Documentation Committee; KOOS = Knee Injury and Osteoarthritis Outcome Pain Subscale Score; MACI = matrix-induced autologous chondrocyte implantation; MRI = magnetic resonance imaging; MOCART = Magnetic Resonance Observation of Cartilage Repair Tissue; WB = weight bearing.

Figure 2.

Meta-analysis of risk of treatment failure among randomized studies. Too few nonrandomized studies reported treatment failure to perform a meta-analysis of nonrandomized studies (not shown). (A) There is no difference in risk of treatment failure for MACT versus microfracture (pooled RR 1.58, 95% CI 0.69, 3.61) (P = 0.28). (B) There is no difference in risk of treatment failure for MACT with early versus delayed weight bearing (pooled RR 0.78, 95% CI 0.48, 1.26) (P = 0.31).

Figure 3.

(A) Forest plot of KOOS-pain scores among randomized studies. A greater improvement in KOOS-pain was seen for MACT compared to microfracture (SMD 1.83, 95% CI 1.46, 2.21) (P < 0.001) as well as with MACT with early versus delayed weight bearing (SMD 1.07, 95% CI 0.67, 1.46) (P < 0.001). (B) Funnel plot of effect size (change in KOOS among treatments vs. controls) versus standard error among randomized trials; there is no visual evidence of reporting bias which would be represented by a lack of points in the lower left quadrant of the graph.

Figure 4.

(A) Forest plot of IKDC scores among randomized studies. A greater improvement in IKDC was seen for MACT compared to microfracture (SMD 0.72, 95% CI 0.39, 1.05) (P < 0.001) as well as with MACT with early versus delayed weight bearing (SMD 0.39, 95% CI 0.05, 0.74) (P = 0.03). (B) Funnel plot of effect size (change in IKDC among treatments vs. controls) versus standard error among randomized trials; there is no visual evidence of reporting bias which would be represented by a lack of points in the lower left quadrant of the graph.

MACT with Early versus Delayed Weight Bearing

Three trials compared early versus delayed weight bearing (WB) protocols following MACT ( Table 2 ). There was no difference between early versus delayed WB for risk of clinical failure among randomized studies (pooled RR 0.78, 95% CI 0.48, 1.26; I² = 0%; P = 0.31) ( Fig. 2 ) or reoperation among randomized studies (pooled RR 1.20, 95% CI 0.46, 3.16; I² = 0%; P = 0.71). Meta-analyses of failure risk among nonrandomized studies or reoperation rates among randomized or nonrandomized studies could not be performed due to insufficient reporting. Among randomized studies, MACT with early versus delayed WB had superior KOOS-pain (pooled SMD 1.07, 95% CI 0.67, 1.46; I² = 94%; P < 0.001) ( Fig. 4 ) and IKDC scores (pooled SMD 0.39, 95% CI 0.05, 0.74; I² = 0%; P = 0.03).

MACT versus Mesenchymal Stem Cell Therapy

Two trials compared outcomes of MACT against mesenchymal stem cell therapy (MSC) ( Table 3 ) with insufficient reporting for meta-analysis. Akgun et al. ²⁵ showed that MSC resulted in better pain, symptoms, and activities of daily living domain KOOS scores at 2-year follow-up with no other significant differences between the 2 therapies. Gobbi et al. ³¹ conducted a nonrandomized study that showed similar improvements in KOOS, visual analog scale (VAS), and Tegner scores between both groups at 2 years, but a greater improvement in IKDC scores in the MSC group.

Table 3.

Studies of MACT versus Non-Microfracture Cartilage Therapies.

Level of Evidence	Author	Implant, Mean Defect Size, and Sample Size	Mean Follow-up	Outcome	Financial Conflict of Interest	Mean Coleman Score	Mean MINORS Score	Mean Delphi Score
MACT versus mesenchymal stem cell therapy
1	Akgun (2015) 25	7 MACI (mean 3.0 cm2); 7 MSC (mean 2.9 cm2)	2 years	Small randomized trial of MACI versus stem cells (also seeded onto a collagen scaffold. Stem cell group showed significantly better subjective scores (pain, symptoms, activities of daily living KOOS) compared to MACI at final follow-up. Tegner score and VAS frequency scores were similar between the 2 groups at final follow-up.	No	59	20.5	5
2	Gobbi (2015) 31	19 Hyalograft C chondrocytes (mean 7.12 cm2); 18 Hyalograft C-MSCs (mean 5.54 cm2)	3 years	Nonrandomized prospective trial of patellofemoral lesions treated with autologous chondrocytes versus mesenchymal stem cells (bone marrow aspirate concentrate); Hyalograft C scaffold used in both groups. Similar improvement in KOOS, VAS, and Tegner scores with both groups; Statistically greater improvement in IKDC scores in MSC group.	Yes	51.5	19	1.5
MACT versus first-generation ACI
1	Bartlett (2005) 32	47 MACI (mean 6.1 cm2); 44 ACI-C (mean 6.0 cm2)	1 year	Similar increase in Cincinnati knee scores in both groups. Similar ICRS scores on second-look biopsy in 79.2% ACI-C, 66.6% MACI. Similar rates of hyaline cartilage found on second-look biopsy. Reoperation rate 9% in both groups.	Yes	52.5	18.5	2.5
2	Zeifang (2010) 33	11 BioSeed-C (mean 4.3 cm2); 9 ACI-P (mean 4.1 cm2)	2 years	No difference in IKDC, Tegner scores, and Short Form-36 scores after 24 months between groups; Better Lysholm and Gillquist scores observed in the first-generation ACI.	Yes	53	19.5	1.5
MACT versus mosaicplasty
1	Clavé (2016) 34	30 Cartipatch (mean 3.2 cm2); 25 mosaicplasty (mean 3.6 cm2)	2 years	Both groups showed improvement in IKDC scores over baseline though mosaicplasty had significantly greater symptom improvement than Cartipatch at 2 years for femoral lesions 2.5-7.5 cm2.	No	55.5	18	2.5
MACT with varying chondrocyte dose
1	Niemeyer (2016) 19	25 low dose (mean 4.8 cm2), 25 medium dose (mean 4.9 cm2), 25 high dose (mean 5.2 cm2)	1 year	Chondrosphere randomized to 3 groups. No difference in MOCART or KOOS scores was detected between the different doses.	Yes	70	20.5	5.5

ACI-C = autologous chondrocyte implantation with type I/III collagen as cover; ACI-P = autologous chondrocyte implantation with periosteal flap as cover; ICRS = International Cartilage Repair Society; IKDC = International Knee Documentation Committee; KOOS = Knee Injury and Osteoarthritis Outcome Pain Subscale Score; MACI = matrix-induced autologous chondrocyte implantation; MOCART = Magnetic Resonance Observation of Cartilage Repair Tissue; MSC = mesenchymal stem cell; VAS = visual analog scale.

MACT versus First-Generation ACI

Two trials ( Table 3 ) compared second-generation MACT compared to first-generation ACI techniques including ACI with type I/III collagen as a cover (ACI-C) and ACI with the original periosteal flap as cover (ACI-P), as described by Brittberg et al. ³ Reporting was insufficient for meta-analysis. Bartlett et al. ³² demonstrated similar outcomes and reoperation rates between MACT and ACI-C at 1-year follow-up. Zeifang et al. ³³ noted better Lysholm and Gillquist scores in ACI-P compared to MACT at 24 months, but other outcomes were similar between the 2 groups.

MACT versus Mosaicplasty

One study by Clavé et al. ³⁴ evaluated MACT against mosaicplasty ( Table 3 ). In this trial, mosaicplasty demonstrated significantly greater symptom improvement per IKDC scores compared to MACT at 2-year follow-up.

MACT at Varying Chondrocyte Doses

One trial by Niemeyer et al. ¹⁹ evaluated different chondrocyte dosing when performing MACT ( Table 3 ). Patients were randomized to low (3-7 spheroids/cm²), medium (10-30 spheroids/cm²), or high dose (40-70 spheroids/cm²) groups. No difference in magnetic resonance observation of cartilage repair tissue (MOCART) or KOOS-pain scores were observed between the different doses at 1-year follow-up.

Discussion

The purpose of this meta-analysis was to assess the available literature and determine (1) whether MACT is superior to other procedures, (2) whether clinical outcomes are altered with accelerated WB postoperatively after MACT, and (3) whether outcomes differ with different MACT chondrocyte doses. Our analyses reveal that accelerated WB (as early as 6 weeks) is safe for MACT, and that MACT is superior to microfracture for defects up to 10 cm² at up to 7.5 years follow-up.

Three trials evaluating accelerated WB after MACT were included in our review.^10,17,22 All 3 studies concluded that accelerated return to WB is noninferior to standard return to WB with return to weight bearing as soon as 6 weeks postoperatively and follow-up times ranging from 2 to 5 years. Of note, in all 3 studies, there were no major differences in magnetic resonance imaging (MRI) findings postoperatively with the exception of graft border integration and surface contour scoring better in the accelerated (6-week) weight-bearing group in Ebert et al. ¹⁰

We found that MACT is superior to microfracture for equivalent sized defects (up to 10 cm²). ³⁵ The evidence clearly favors MACT with regard to patient pain and functional scores. The limitation regarding these trials is that long-term outcome of MACT versus microfracture is still relatively unknown; the longest follow-up time included in our analysis was 7.5 years, with a weighted mean of 3.1 years. We believe that if MACT has similar durability to ACI-P or ACI-C (which has been shown to last well past 10 years^36-39), then it should be durable enough that the outcomes of MACT versus microfracture at longer follow-up times will be similar to what has been reported. There is also evidence against microfracture as a first-line treatment for isolated cartilage defects of the knee; several studies^37,40,41 have shown that second-line ACI after failed microfracture has inferior outcomes (higher failure rates, lower KOOS scores, higher rate of re-intervention) compared to when ACI is performed first.

Current high-level evidence is limited comparing MACT to first-generation ACI, MSC, and mosaicplasty. All of the trials that fit our inclusion criteria for these questions were small single-institution studies that may not be powered adequately to draw accurate conclusions. In all 3 comparison categories, there is insufficient evidence and power to conclude whether one is superior to the other. However, there is evidence in the literature to suggest that mosaicplasty is inferior to first-generation ACI in the treatment of larger cartilage defects ⁴² (average of 440.9 mm²). Ultimately, longer-term and adequately powered studies are needed.

Niemeyer et al. ¹⁹ concluded that there are no differences in outcomes between low, medium, and high chondrocyte dosing for MACT. Again, this single study is not enough to draw conclusions regarding dosing, but we believe that the answer to this question is most likely to be dependent on the specific commercial product. The chondrocyte dosing depends on the nature of the scaffold and delivery system, so different products will have different dosing requirements. Still, there is likely a threshold value of chondrocyte concentration below which graft maturity or fill will not occur, but currently this value is unknown.

The quality of this review is limited by the quality of studies analyzed. The mean Coleman score for our studies was 59.4, which falls into the “fair” category. However, 5 of the 13 trials had scores below 54, placing them in the “poor” category in terms of quality as determined by Cowan et al. ²⁶ In addition, several of our questions could not be adequately answered due to the small number of trials and small number of patients in those trials. Unfortunately, this is likely due to the fact that orthopedic clinical trials tend to be small, and often underpowered. In sports medicine trials, Khan et al. ⁴³ found a mean Fragility Index of 2, indicating reported significant results would become nonsignificant if the outcomes of 2 or more patients in the trial were to be changed. Additional adequately powered studies are needed to answer some questions proposed by our review.

Compared to microfracture, MACT has no increased risk of clinical failure or reoperation and has superior improvement in patient-reported outcome scores. Compared to MACT with standardized postoperative weight-bearing protocols, accelerated weight-bearing protocols have no increased risk of clinical failure but show superior improvement in patient-reported outcome scores. There is limited evidence regarding MACT compared to first-generation ACI, mosaicplasty, and MSC without compelling differences in outcomes. Evidence regarding the effect of chondrocyte dose at implantation is limited.

Appendix A

Paired Linear Regression Formulas for IKDC and KOOS-Pain Scores

The following formulas were derived from studies reporting both IKDC and KOOS-pain. The formulas were used to impute IKDC and KOOS-pain scores among studies that reported one but not both outcomes score (IKDC or KOOS-pain but not IKDC and KOOS-pain):

$$\begin{array}{l}\mathrm{IKDC}\mathrm{preop}\mathrm{average}=16.382511+0.4294749\ast \\ \mathrm{KOOS}-\mathrm{pain}\mathrm{preop}\mathrm{average}\\ \mathrm{KOOS}-\mathrm{pain}\mathrm{preop}\mathrm{average}=-6.349042+1.4942579\ast \\ \mathrm{IKDC}\mathrm{preop}\mathrm{average}\\ R=0.64;P=0.002\end{array}$$

$$\begin{array}{l}\mathrm{IKDC}\mathrm{postop}\mathrm{average}=-22.52936+1.1391425\ast \\ \mathrm{KOOS}-\mathrm{pain}\mathrm{postop}\mathrm{average}\\ \mathrm{KOOS}-\mathrm{pain}\mathrm{postop}\mathrm{average}=38.515244+0.6142721\ast \\ \mathrm{IKDC}\mathrm{postop}\mathrm{average}\\ R=0.70;P=0.001\end{array}$$

Among studies in this review, preoperative IKDC scores were reliably correlated with paired KOOS-pain scores form the same study (R = 0.64, P = 0.002), and postoperative IKDC scores were reliably correlated with paired postoperative KOOS-pain scores (R = 0.70; P = 0.001).

Positive Transformation to Include Studies with Zero Failures or Reoperations in Meta-Analysis

For studies reporting zero failures or reoperations in both treatment groups, rather than excluding the entire study from the meta-analysis due to inability to calculate confidence intervals, the nearest nonzero positive integer was assigned for failures to each treatment and control group to maintain relative risk point estimate between 0.95 and 1.05. The smallest positive integers needed to meet this criterion were utilized in order to avoid artificially shrinking the estimated confidence intervals for the outcome of interest.

Assessment of Bias

Publication bias was assessed with Egger’s test of bias ⁴⁶ as well as visual inspection of a funnel plot of effect size (improvement in IKDC or KOOS-pain) versus standard error. Finally, Spearman correlations between study outcomes (as defined by difference between treatment and control groups in KOOS-pain, IKDC, reoperation rates, or failures rates) and methodological quality scores, financial conflict of interest, sample size, year of publication, or length of follow-up were also performed to assess for potential bias due to these additional factors. A P value <0.05 was considered to be statistically significant, unless otherwise noted.

Appendix B

Modified Coleman Methodology Score ²³ .

Item	Points
Inclusion Criteria
Not described	0
Described without percentages given	3
Enrollment rate <80%	6
Enrollment rate >80%	9
Power
Not reported	0
>80%, methods not described	3
>80%, methods described	6
Alpha error
Not reported	0
<0.05	3
<0.01	6
Sample size
Not stated or <20	0
20-40	3
41-60	6
>60	9
Randomization
Not randomized	0
Modified/partial
Not blinded	2
Blinded	4
Complete
Not blinded	6
Blinded	8
Follow-up
Short-term (<6 months)
Patient retention <80%	0
Patient retention 80% to 90%	2
Patient retention >90%	4
Medium-term (6-24 months)
Patient retention <80%	2
Patient retention 80% to 90%	4
Patient retention >90%	6
Long term (>24 months)
Patient retention <80%	4
Patient retention 80% to 90%	6
Patient retention >90%	8
Patient analysis
Incomplete	0
Complete	3
Complete and intention-to-treat-based	6
Blinding
None	0
Single	2
Double	4
Triple	6
Similarity in treatment
No	0
Similar co-interventions	3
No co-interventions	6
Treatment description
None	0
Fair	3
Adequate	6
Group comparability
Not comparable	0
Partially comparable	3
Comparable	6
Outcome assessment
Written assessment by patient with assistance	0
Written assessment by patient without assistance	2
Independent investigator	4
Recruited patients	6
Description of rehabilitation protocol
Not reported	0
Not adequately described	2
Well described	4
Clinical effect measurement
Effect size
Not reported	0
<50%	2
50% to 75%	4
>75%	6
or Relative risk reduction
Not reported	0
<25%	3
>25%	6
or Absolute risk reduction
Not reported	0
<10%	3
>10%	6
Number of patients to treat
Not reported	0
Reported	4

Appendix C

Methodological Index for Non-Randomized Studies (MINORS) ²⁴ .

Item	Description
1	A clearly stated aim: The question addressed should be precise and relevant in the light of available literature.
2	Inclusion of consecutive patients: All patients potentially fit for inclusion (satisfying the criteria for inclusion) have been included in the study during the study period (no exclusion or details about the reasons for exclusion).
3	Prospective collection of data: Data were collected according to a protocol established before the beginning of the study.
4	Endpoints appropriate to the aim of the study: Unambiguous explanation of the criteria used to evaluate the main outcome which should be in accordance with the question addressed by the study. Also, the endpoints should be assessed on an intention-to-treat basis.
5	Unbiased assessment of the study endpoint: Blind evaluation of objective endpoints and double-blind evaluation of subjective endpoints. Otherwise the reasons for not blinding should be stated.
6	Follow-up period appropriate to the aim of the study: The follow-up should be sufficiently long to allow the assessment of the main endpoint and possible adverse events.
7	Loss to follow-up less than 5%: All patients should be included in the follow up. Otherwise, the proportion lost to follow up should not exceed the proportion experiencing the major endpoint.
8	Prospective calculation of the study size: Information of the size of detectable difference of interest with a calculation of 95% confidence interval, according to the expected incidence of the outcome event, and information about the level for statistical significance and estimates of power when comparing the outcomes.
9	An adequate control group: Having a gold standard diagnostic test or therapeutic intervention recognized as the optimal intervention according to the available published data.
10	Contemporary groups: Control and studied groups should be managed during the same time period (no historical comparison).
11	Baseline equivalence of groups: The groups should be similar regarding the criteria other than the studied endpoints. Absence of confounding factors that could bias the interpretation of the results.
12	Adequate statistical analyses: Whether the statistics were in accordance with the type of study with calculation of confidence intervals or relative risk.

Each item was assigned a score of 0, 1, or 2—0 (not reported), 1 (reported but inadequate) or 2 (reported and adequate).

Appendix D

Delphi Criteria ²⁵

Each item was given a response of “Yes” (1 point), “No” (−1 point), “don’t know” (0 point)

1. Treatment allocation

   1. Was a method of randomization performed?
   2. Was the treatment allocation concealed?

2. Were the groups similar at baseline regarding the most important prognostic indicators?

3. Were the eligibility criteria specified?

4. Was the outcome assessor blinded?

5. Was the care provider blinded?

6. Was the patient blinded?

7. Were point estimates and measures of variability presented for the primary outcome measures?

8. Did the analysis include an intention-to treat analysis?